# Auto Classes 多くの場合、`from_pretrained()`メソッドに与えられた事前学習済みモデルの名前やパスから、使用したいアーキテクチャを推測することができます。自動クラスはこの仕事をあなたに代わって行うためにここにありますので、事前学習済みの重み/設定/語彙への名前/パスを与えると自動的に関連するモデルを取得できます。 [AutoConfig](/docs/transformers/v5.6.0/ja/model_doc/auto#transformers.AutoConfig)、[AutoModel](/docs/transformers/v5.6.0/ja/model_doc/auto#transformers.AutoModel)、[AutoTokenizer](/docs/transformers/v5.6.0/ja/model_doc/auto#transformers.AutoTokenizer)のいずれかをインスタンス化すると、関連するアーキテクチャのクラスが直接作成されます。例えば、 ```python model = AutoModel.from_pretrained("google-bert/bert-base-cased") ``` これは[BertModel](/docs/transformers/v5.6.0/ja/model_doc/bert#transformers.BertModel)のインスタンスであるモデルを作成します。 各タスクごと、そして各バックエンド（PyTorch、TensorFlow、またはFlax）ごとに`AutoModel`のクラスが存在します。 ## 自動クラスの拡張 それぞれの自動クラスには、カスタムクラスで拡張するためのメソッドがあります。例えば、`NewModel`というモデルのカスタムクラスを定義した場合、`NewModelConfig`を確保しておけばこのようにして自動クラスに追加することができます： ```python from transformers import AutoConfig, AutoModel AutoConfig.register("new-model", NewModelConfig) AutoModel.register(NewModelConfig, NewModel) ``` その後、通常どおりauto classesを使用することができるようになります！ あなたの`NewModelConfig`が[PreTrainedConfig](/docs/transformers/v5.6.0/ja/main_classes/configuration#transformers.PreTrainedConfig)のサブクラスである場合、その`model_type`属性がコンフィグを登録するときに使用するキー（ここでは`"new-model"`）と同じに設定されていることを確認してください。 同様に、あなたの`NewModel`が[PreTrainedModel](/docs/transformers/v5.6.0/ja/main_classes/model#transformers.PreTrainedModel)のサブクラスである場合、その`config_class`属性がモデルを登録する際に使用するクラス（ここでは`NewModelConfig`）と同じに設定されていることを確認してください。 ## AutoConfig[[transformers.AutoConfig]] #### transformers.AutoConfig[[transformers.AutoConfig]] [Source](https://github.com/huggingface/transformers/blob/v5.6.0/src/transformers/models/auto/configuration_auto.py#L264) This is a generic configuration class that will be instantiated as one of the configuration classes of the library when created with the [from_pretrained()](/docs/transformers/v5.6.0/ja/model_doc/auto#transformers.AutoConfig.from_pretrained) class method. This class cannot be instantiated directly using `__init__()` (throws an error). from_pretrainedtransformers.AutoConfig.from_pretrainedhttps://github.com/huggingface/transformers/blob/v5.6.0/src/transformers/models/auto/configuration_auto.py#L287[{"name": "pretrained_model_name_or_path", "val": ": str | os.PathLike[str]"}, {"name": "**kwargs", "val": ""}]- **pretrained_model_name_or_path** (`str` or `os.PathLike`) -- Can be either: - A string, the *model id* of a pretrained model configuration hosted inside a model repo on huggingface.co. - A path to a *directory* containing a configuration file saved using the [save_pretrained()](/docs/transformers/v5.6.0/ja/main_classes/configuration#transformers.PreTrainedConfig.save_pretrained) method, or the [save_pretrained()](/docs/transformers/v5.6.0/ja/main_classes/model#transformers.PreTrainedModel.save_pretrained) method, e.g., `./my_model_directory/`. - a path to a saved configuration JSON *file*, e.g., `./my_model_directory/configuration.json`. - **cache_dir** (`str` or `os.PathLike`, *optional*) -- Path to a directory in which a downloaded pretrained model configuration should be cached if the standard cache should not be used. - **force_download** (`bool`, *optional*, defaults to `False`) -- Whether or not to force the (re-)download the model weights and configuration files and override the cached versions if they exist. - **proxies** (`dict[str, str]`, *optional*) -- A dictionary of proxy servers to use by protocol or endpoint, e.g., `{'http': 'foo.bar:3128', 'http://hostname': 'foo.bar:4012'}`. The proxies are used on each request. - **revision** (`str`, *optional*, defaults to `"main"`) -- The specific model version to use. It can be a branch name, a tag name, or a commit id, since we use a git-based system for storing models and other artifacts on huggingface.co, so `revision` can be any identifier allowed by git. - **return_unused_kwargs** (`bool`, *optional*, defaults to `False`) -- If `False`, then this function returns just the final configuration object. If `True`, then this functions returns a `Tuple(config, unused_kwargs)` where *unused_kwargs* is a dictionary consisting of the key/value pairs whose keys are not configuration attributes: i.e., the part of `kwargs` which has not been used to update `config` and is otherwise ignored. - **trust_remote_code** (`bool`, *optional*, defaults to `False`) -- Whether or not to allow for custom models defined on the Hub in their own modeling files. This option should only be set to `True` for repositories you trust and in which you have read the code, as it will execute code present on the Hub on your local machine. - **kwargs(additional** keyword arguments, *optional*) -- The values in kwargs of any keys which are configuration attributes will be used to override the loaded values. Behavior concerning key/value pairs whose keys are *not* configuration attributes is controlled by the `return_unused_kwargs` keyword parameter.0 Instantiate one of the configuration classes of the library from a pretrained model configuration. The configuration class to instantiate is selected based on the `model_type` property of the config object that is loaded, or when it's missing, by falling back to using pattern matching on `pretrained_model_name_or_path`: - **EvollaModel** -- `EvollaConfig` (EvollaConfig model) - **afmoe** -- `AfmoeConfig` (AfmoeConfig model) - **aimv2** -- `Aimv2Config` (Aimv2Config model) - **aimv2_text_model** -- `Aimv2TextConfig` (Aimv2TextConfig model) - **aimv2_vision_model** -- `Aimv2VisionConfig` (Aimv2VisionConfig model) - **albert** -- [AlbertConfig](/docs/transformers/v5.6.0/ja/model_doc/albert#transformers.AlbertConfig) (AlbertConfig model) - **align** -- [AlignConfig](/docs/transformers/v5.6.0/ja/model_doc/align#transformers.AlignConfig) (AlignConfig model) - **align_text_model** -- [AlignTextConfig](/docs/transformers/v5.6.0/ja/model_doc/align#transformers.AlignTextConfig) (AlignTextConfig model) - **align_vision_model** -- [AlignVisionConfig](/docs/transformers/v5.6.0/ja/model_doc/align#transformers.AlignVisionConfig) (AlignVisionConfig model) - **altclip** -- [AltCLIPConfig](/docs/transformers/v5.6.0/ja/model_doc/altclip#transformers.AltCLIPConfig) (AltCLIPConfig model) - **altclip_text_model** -- [AltCLIPTextConfig](/docs/transformers/v5.6.0/ja/model_doc/altclip#transformers.AltCLIPTextConfig) (AltCLIPTextConfig model) - **altclip_vision_model** -- [AltCLIPVisionConfig](/docs/transformers/v5.6.0/ja/model_doc/altclip#transformers.AltCLIPVisionConfig) (AltCLIPVisionConfig model) - **apertus** -- `ApertusConfig` (ApertusConfig model) - **arcee** -- `ArceeConfig` (ArceeConfig model) - **aria** -- `AriaConfig` (AriaConfig model) - **aria_text** -- `AriaTextConfig` (AriaTextConfig model) - **audio-spectrogram-transformer** -- [ASTConfig](/docs/transformers/v5.6.0/ja/model_doc/audio-spectrogram-transformer#transformers.ASTConfig) (ASTConfig model) - **audioflamingo3** -- `AudioFlamingo3Config` (AudioFlamingo3Config model) - **audioflamingo3_encoder** -- `AudioFlamingo3EncoderConfig` (AudioFlamingo3EncoderConfig model) - **autoformer** -- [AutoformerConfig](/docs/transformers/v5.6.0/ja/model_doc/autoformer#transformers.AutoformerConfig) (AutoformerConfig model) - **aya_vision** -- `AyaVisionConfig` (AyaVisionConfig model) - **bamba** -- `BambaConfig` (BambaConfig model) - **bark** -- [BarkConfig](/docs/transformers/v5.6.0/ja/model_doc/bark#transformers.BarkConfig) (BarkConfig model) - **bart** -- [BartConfig](/docs/transformers/v5.6.0/ja/model_doc/bart#transformers.BartConfig) (BartConfig model) - **beit** -- [BeitConfig](/docs/transformers/v5.6.0/ja/model_doc/beit#transformers.BeitConfig) (BeitConfig model) - **bert** -- [BertConfig](/docs/transformers/v5.6.0/ja/model_doc/bert#transformers.BertConfig) (BertConfig model) - **bert-generation** -- [BertGenerationConfig](/docs/transformers/v5.6.0/ja/model_doc/bert-generation#transformers.BertGenerationConfig) (BertGenerationConfig model) - **big_bird** -- [BigBirdConfig](/docs/transformers/v5.6.0/ja/model_doc/big_bird#transformers.BigBirdConfig) (BigBirdConfig model) - **bigbird_pegasus** -- [BigBirdPegasusConfig](/docs/transformers/v5.6.0/ja/model_doc/bigbird_pegasus#transformers.BigBirdPegasusConfig) (BigBirdPegasusConfig model) - **biogpt** -- [BioGptConfig](/docs/transformers/v5.6.0/ja/model_doc/biogpt#transformers.BioGptConfig) (BioGptConfig model) - **bit** -- [BitConfig](/docs/transformers/v5.6.0/ja/model_doc/bit#transformers.BitConfig) (BitConfig model) - **bitnet** -- `BitNetConfig` (BitNetConfig model) - **blenderbot** -- [BlenderbotConfig](/docs/transformers/v5.6.0/ja/model_doc/blenderbot#transformers.BlenderbotConfig) (BlenderbotConfig model) - **blenderbot-small** -- [BlenderbotSmallConfig](/docs/transformers/v5.6.0/ja/model_doc/blenderbot-small#transformers.BlenderbotSmallConfig) (BlenderbotSmallConfig model) - **blip** -- [BlipConfig](/docs/transformers/v5.6.0/ja/model_doc/blip#transformers.BlipConfig) (BlipConfig model) - **blip-2** -- [Blip2Config](/docs/transformers/v5.6.0/ja/model_doc/blip-2#transformers.Blip2Config) (Blip2Config model) - **blip_2_qformer** -- [Blip2QFormerConfig](/docs/transformers/v5.6.0/ja/model_doc/blip-2#transformers.Blip2QFormerConfig) (Blip2QFormerConfig model) - **blip_2_vision_model** -- [Blip2VisionConfig](/docs/transformers/v5.6.0/ja/model_doc/blip-2#transformers.Blip2VisionConfig) (Blip2VisionConfig model) - **blip_text_model** -- [BlipTextConfig](/docs/transformers/v5.6.0/ja/model_doc/blip#transformers.BlipTextConfig) (BlipTextConfig model) - **blip_vision_model** -- [BlipVisionConfig](/docs/transformers/v5.6.0/ja/model_doc/blip#transformers.BlipVisionConfig) (BlipVisionConfig model) - **bloom** -- [BloomConfig](/docs/transformers/v5.6.0/ja/model_doc/bloom#transformers.BloomConfig) (BloomConfig model) - **blt** -- `BltConfig` (BltConfig model) - **blt_global_transformer** -- `BltGlobalTransformerConfig` (BltGlobalTransformerConfig model) - **blt_local_decoder** -- `BltLocalDecoderConfig` (BltLocalDecoderConfig model) - **blt_local_encoder** -- `BltLocalEncoderConfig` (BltLocalEncoderConfig model) - **blt_patcher** -- `BltPatcherConfig` (BltPatcherConfig model) - **bridgetower** -- [BridgeTowerConfig](/docs/transformers/v5.6.0/ja/model_doc/bridgetower#transformers.BridgeTowerConfig) (BridgeTowerConfig model) - **bridgetower_text_model** -- [BridgeTowerTextConfig](/docs/transformers/v5.6.0/ja/model_doc/bridgetower#transformers.BridgeTowerTextConfig) (BridgeTowerTextConfig model) - **bridgetower_vision_model** -- [BridgeTowerVisionConfig](/docs/transformers/v5.6.0/ja/model_doc/bridgetower#transformers.BridgeTowerVisionConfig) (BridgeTowerVisionConfig model) - **bros** -- [BrosConfig](/docs/transformers/v5.6.0/ja/model_doc/bros#transformers.BrosConfig) (BrosConfig model) - **camembert** -- [CamembertConfig](/docs/transformers/v5.6.0/ja/model_doc/camembert#transformers.CamembertConfig) (CamembertConfig model) - **canine** -- [CanineConfig](/docs/transformers/v5.6.0/ja/model_doc/canine#transformers.CanineConfig) (CanineConfig model) - **chameleon** -- `ChameleonConfig` (ChameleonConfig model) - **chameleon_vqgan** -- `ChameleonVQVAEConfig` (ChameleonVQVAEConfig model) - **chinese_clip** -- [ChineseCLIPConfig](/docs/transformers/v5.6.0/ja/model_doc/chinese_clip#transformers.ChineseCLIPConfig) (ChineseCLIPConfig model) - **chinese_clip_text_model** -- [ChineseCLIPTextConfig](/docs/transformers/v5.6.0/ja/model_doc/chinese_clip#transformers.ChineseCLIPTextConfig) (ChineseCLIPTextConfig model) - **chinese_clip_vision_model** -- [ChineseCLIPVisionConfig](/docs/transformers/v5.6.0/ja/model_doc/chinese_clip#transformers.ChineseCLIPVisionConfig) (ChineseCLIPVisionConfig model) - **chmv2** -- `CHMv2Config` (CHMv2Config model) - **clap** -- [ClapConfig](/docs/transformers/v5.6.0/ja/model_doc/clap#transformers.ClapConfig) (ClapConfig model) - **clap_audio_model** -- [ClapAudioConfig](/docs/transformers/v5.6.0/ja/model_doc/clap#transformers.ClapAudioConfig) (ClapAudioConfig model) - **clap_text_model** -- [ClapTextConfig](/docs/transformers/v5.6.0/ja/model_doc/clap#transformers.ClapTextConfig) (ClapTextConfig model) - **clip** -- [CLIPConfig](/docs/transformers/v5.6.0/ja/model_doc/clip#transformers.CLIPConfig) (CLIPConfig model) - **clip_text_model** -- [CLIPTextConfig](/docs/transformers/v5.6.0/ja/model_doc/clip#transformers.CLIPTextConfig) (CLIPTextConfig model) - **clip_vision_model** -- [CLIPVisionConfig](/docs/transformers/v5.6.0/ja/model_doc/clip#transformers.CLIPVisionConfig) (CLIPVisionConfig model) - **clipseg** -- [CLIPSegConfig](/docs/transformers/v5.6.0/ja/model_doc/clipseg#transformers.CLIPSegConfig) (CLIPSegConfig model) - **clipseg_text_model** -- [CLIPSegTextConfig](/docs/transformers/v5.6.0/ja/model_doc/clipseg#transformers.CLIPSegTextConfig) (CLIPSegTextConfig model) - **clipseg_vision_model** -- [CLIPSegVisionConfig](/docs/transformers/v5.6.0/ja/model_doc/clipseg#transformers.CLIPSegVisionConfig) (CLIPSegVisionConfig model) - **clvp** -- [ClvpConfig](/docs/transformers/v5.6.0/ja/model_doc/clvp#transformers.ClvpConfig) (ClvpConfig model) - **clvp_decoder** -- [ClvpDecoderConfig](/docs/transformers/v5.6.0/ja/model_doc/clvp#transformers.ClvpDecoderConfig) (ClvpDecoderConfig model) - **clvp_encoder** -- [ClvpEncoderConfig](/docs/transformers/v5.6.0/ja/model_doc/clvp#transformers.ClvpEncoderConfig) (ClvpEncoderConfig model) - **codegen** -- [CodeGenConfig](/docs/transformers/v5.6.0/ja/model_doc/codegen#transformers.CodeGenConfig) (CodeGenConfig model) - **cohere** -- `CohereConfig` (CohereConfig model) - **cohere2** -- `Cohere2Config` (Cohere2Config model) - **cohere2_vision** -- `Cohere2VisionConfig` (Cohere2VisionConfig model) - **cohere_asr** -- `CohereAsrConfig` (CohereAsrConfig model) - **colmodernvbert** -- `ColModernVBertConfig` (ColModernVBertConfig model) - **colpali** -- `ColPaliConfig` (ColPaliConfig model) - **colqwen2** -- `ColQwen2Config` (ColQwen2Config model) - **conditional_detr** -- [ConditionalDetrConfig](/docs/transformers/v5.6.0/ja/model_doc/conditional_detr#transformers.ConditionalDetrConfig) (ConditionalDetrConfig model) - **convbert** -- [ConvBertConfig](/docs/transformers/v5.6.0/ja/model_doc/convbert#transformers.ConvBertConfig) (ConvBertConfig model) - **convnext** -- [ConvNextConfig](/docs/transformers/v5.6.0/ja/model_doc/convnext#transformers.ConvNextConfig) (ConvNextConfig model) - **convnextv2** -- [ConvNextV2Config](/docs/transformers/v5.6.0/ja/model_doc/convnextv2#transformers.ConvNextV2Config) (ConvNextV2Config model) - **cpmant** -- [CpmAntConfig](/docs/transformers/v5.6.0/ja/model_doc/cpmant#transformers.CpmAntConfig) (CpmAntConfig model) - **csm** -- `CsmConfig` (CsmConfig model) - **csm_depth_decoder_model** -- `CsmDepthDecoderConfig` (CsmDepthDecoderConfig model) - **ctrl** -- [CTRLConfig](/docs/transformers/v5.6.0/ja/model_doc/ctrl#transformers.CTRLConfig) (CTRLConfig model) - **cvt** -- [CvtConfig](/docs/transformers/v5.6.0/ja/model_doc/cvt#transformers.CvtConfig) (CvtConfig model) - **cwm** -- `CwmConfig` (CwmConfig model) - **d_fine** -- `DFineConfig` (DFineConfig model) - **dab-detr** -- `DabDetrConfig` (DabDetrConfig model) - **dac** -- `DacConfig` (DacConfig model) - **data2vec-audio** -- [Data2VecAudioConfig](/docs/transformers/v5.6.0/ja/model_doc/data2vec#transformers.Data2VecAudioConfig) (Data2VecAudioConfig model) - **data2vec-text** -- [Data2VecTextConfig](/docs/transformers/v5.6.0/ja/model_doc/data2vec#transformers.Data2VecTextConfig) (Data2VecTextConfig model) - **data2vec-vision** -- [Data2VecVisionConfig](/docs/transformers/v5.6.0/ja/model_doc/data2vec#transformers.Data2VecVisionConfig) (Data2VecVisionConfig model) - **dbrx** -- `DbrxConfig` (DbrxConfig model) - **deberta** -- [DebertaConfig](/docs/transformers/v5.6.0/ja/model_doc/deberta#transformers.DebertaConfig) (DebertaConfig model) - **deberta-v2** -- [DebertaV2Config](/docs/transformers/v5.6.0/ja/model_doc/deberta-v2#transformers.DebertaV2Config) (DebertaV2Config model) - **decision_transformer** -- [DecisionTransformerConfig](/docs/transformers/v5.6.0/ja/model_doc/decision_transformer#transformers.DecisionTransformerConfig) (DecisionTransformerConfig model) - **deepseek_v2** -- `DeepseekV2Config` (DeepseekV2Config model) - **deepseek_v3** -- `DeepseekV3Config` (DeepseekV3Config model) - **deepseek_vl** -- `DeepseekVLConfig` (DeepseekVLConfig model) - **deepseek_vl_hybrid** -- `DeepseekVLHybridConfig` (DeepseekVLHybridConfig model) - **deformable_detr** -- [DeformableDetrConfig](/docs/transformers/v5.6.0/ja/model_doc/deformable_detr#transformers.DeformableDetrConfig) (DeformableDetrConfig model) - **deit** -- [DeiTConfig](/docs/transformers/v5.6.0/ja/model_doc/deit#transformers.DeiTConfig) (DeiTConfig model) - **depth_anything** -- `DepthAnythingConfig` (DepthAnythingConfig model) - **depth_pro** -- `DepthProConfig` (DepthProConfig model) - **detr** -- [DetrConfig](/docs/transformers/v5.6.0/ja/model_doc/detr#transformers.DetrConfig) (DetrConfig model) - **dia** -- `DiaConfig` (DiaConfig model) - **dia_decoder** -- `DiaDecoderConfig` (DiaDecoderConfig model) - **dia_encoder** -- `DiaEncoderConfig` (DiaEncoderConfig model) - **diffllama** -- `DiffLlamaConfig` (DiffLlamaConfig model) - **dinat** -- [DinatConfig](/docs/transformers/v5.6.0/ja/model_doc/dinat#transformers.DinatConfig) (DinatConfig model) - **dinov2** -- `Dinov2Config` (Dinov2Config model) - **dinov2_with_registers** -- `Dinov2WithRegistersConfig` (Dinov2WithRegistersConfig model) - **dinov3_convnext** -- `DINOv3ConvNextConfig` (DINOv3ConvNextConfig model) - **dinov3_vit** -- `DINOv3ViTConfig` (DINOv3ViTConfig model) - **distilbert** -- `DistilBertConfig` (DistilBertConfig model) - **doge** -- `DogeConfig` (DogeConfig model) - **donut-swin** -- `DonutSwinConfig` (DonutSwinConfig model) - **dots1** -- `Dots1Config` (Dots1Config model) - **dpr** -- `DPRConfig` (DPRConfig model) - **dpt** -- `DPTConfig` (DPTConfig model) - **edgetam** -- `EdgeTamConfig` (EdgeTamConfig model) - **edgetam_video** -- `EdgeTamVideoConfig` (EdgeTamVideoConfig model) - **edgetam_vision_model** -- `EdgeTamVisionConfig` (EdgeTamVisionConfig model) - **efficientloftr** -- `EfficientLoFTRConfig` (EfficientLoFTRConfig model) - **efficientnet** -- `EfficientNetConfig` (EfficientNetConfig model) - **electra** -- `ElectraConfig` (ElectraConfig model) - **emu3** -- `Emu3Config` (Emu3Config model) - **emu3_text_model** -- `Emu3TextConfig` (Emu3TextConfig model) - **emu3_vqgan** -- `Emu3VQVAEConfig` (Emu3VQVAEConfig model) - **encodec** -- `EncodecConfig` (EncodecConfig model) - **encoder-decoder** -- `EncoderDecoderConfig` (EncoderDecoderConfig model) - **eomt** -- `EomtConfig` (EomtConfig model) - **eomt_dinov3** -- `EomtDinov3Config` (EomtDinov3Config model) - **ernie** -- `ErnieConfig` (ErnieConfig model) - **ernie4_5** -- `Ernie4_5Config` (Ernie4_5Config model) - **ernie4_5_moe** -- `Ernie4_5_MoeConfig` (Ernie4_5_MoeConfig model) - **ernie4_5_vl_moe** -- `Ernie4_5_VLMoeConfig` (Ernie4_5_VLMoeConfig model) - **ernie4_5_vl_moe_text** -- `Ernie4_5_VLMoeTextConfig` (Ernie4_5_VLMoeTextConfig model) - **ernie4_5_vl_moe_vision** -- `Ernie4_5_VLMoeVisionConfig` (Ernie4_5_VLMoeVisionConfig model) - **esm** -- `EsmConfig` (EsmConfig model) - **eurobert** -- `EuroBertConfig` (EuroBertConfig model) - **evolla** -- `EvollaConfig` (EvollaConfig model) - **exaone4** -- `Exaone4Config` (Exaone4Config model) - **exaone_moe** -- `ExaoneMoeConfig` (ExaoneMoeConfig model) - **falcon** -- `FalconConfig` (FalconConfig model) - **falcon_h1** -- `FalconH1Config` (FalconH1Config model) - **falcon_mamba** -- `FalconMambaConfig` (FalconMambaConfig model) - **fast_vlm** -- `FastVlmConfig` (FastVlmConfig model) - **fastspeech2_conformer** -- `FastSpeech2ConformerConfig` (FastSpeech2ConformerConfig model) - **fastspeech2_conformer_hifigan** -- `FastSpeech2ConformerHifiGanConfig` (FastSpeech2ConformerHifiGanConfig model) - **fastspeech2_conformer_with_hifigan** -- `FastSpeech2ConformerWithHifiGanConfig` (FastSpeech2ConformerWithHifiGanConfig model) - **flaubert** -- `FlaubertConfig` (FlaubertConfig model) - **flava** -- `FlavaConfig` (FlavaConfig model) - **flava_image_model** -- `FlavaImageConfig` (FlavaImageConfig model) - **flava_multimodal_model** -- `FlavaMultimodalConfig` (FlavaMultimodalConfig model) - **flava_text_model** -- `FlavaTextConfig` (FlavaTextConfig model) - **flex_olmo** -- `FlexOlmoConfig` (FlexOlmoConfig model) - **florence2** -- `Florence2Config` (Florence2Config model) - **florence_vision** -- `Florence2VisionConfig` (Florence2VisionConfig model) - **fnet** -- `FNetConfig` (FNetConfig model) - **focalnet** -- `FocalNetConfig` (FocalNetConfig model) - **fsmt** -- `FSMTConfig` (FSMTConfig model) - **funnel** -- `FunnelConfig` (FunnelConfig model) - **fuyu** -- `FuyuConfig` (FuyuConfig model) - **gemma** -- `GemmaConfig` (GemmaConfig model) - **gemma2** -- `Gemma2Config` (Gemma2Config model) - **gemma3** -- `Gemma3Config` (Gemma3Config model) - **gemma3_text** -- `Gemma3TextConfig` (Gemma3TextConfig model) - **gemma3n** -- `Gemma3nConfig` (Gemma3nConfig model) - **gemma3n_audio** -- `Gemma3nAudioConfig` (Gemma3nAudioConfig model) - **gemma3n_text** -- `Gemma3nTextConfig` (Gemma3nTextConfig model) - **gemma3n_vision** -- `Gemma3nVisionConfig` (Gemma3nVisionConfig model) - **gemma4** -- `Gemma4Config` (Gemma4Config model) - **gemma4_audio** -- `Gemma4AudioConfig` (Gemma4AudioConfig model) - **gemma4_text** -- `Gemma4TextConfig` (Gemma4TextConfig model) - **gemma4_vision** -- `Gemma4VisionConfig` (Gemma4VisionConfig model) - **git** -- `GitConfig` (GitConfig model) - **git_vision_model** -- `GitVisionConfig` (GitVisionConfig model) - **glm** -- `GlmConfig` (GlmConfig model) - **glm4** -- `Glm4Config` (Glm4Config model) - **glm46v** -- `Glm46VConfig` (Glm46VConfig model) - **glm4_moe** -- `Glm4MoeConfig` (Glm4MoeConfig model) - **glm4_moe_lite** -- `Glm4MoeLiteConfig` (Glm4MoeLiteConfig model) - **glm4v** -- `Glm4vConfig` (Glm4vConfig model) - **glm4v_moe** -- `Glm4vMoeConfig` (Glm4vMoeConfig model) - **glm4v_moe_text** -- `Glm4vMoeTextConfig` (Glm4vMoeTextConfig model) - **glm4v_moe_vision** -- `Glm4vMoeVisionConfig` (Glm4vMoeVisionConfig model) - **glm4v_text** -- `Glm4vTextConfig` (Glm4vTextConfig model) - **glm4v_vision** -- `Glm4vVisionConfig` (Glm4vVisionConfig model) - **glm_image** -- `GlmImageConfig` (GlmImageConfig model) - **glm_image_text** -- `GlmImageTextConfig` (GlmImageTextConfig model) - **glm_image_vision** -- `GlmImageVisionConfig` (GlmImageVisionConfig model) - **glm_image_vqmodel** -- `GlmImageVQVAEConfig` (GlmImageVQVAEConfig model) - **glm_moe_dsa** -- `GlmMoeDsaConfig` (GlmMoeDsaConfig model) - **glm_ocr** -- `GlmOcrConfig` (GlmOcrConfig model) - **glm_ocr_text** -- `GlmOcrTextConfig` (GlmOcrTextConfig model) - **glm_ocr_vision** -- `GlmOcrVisionConfig` (GlmOcrVisionConfig model) - **glmasr** -- `GlmAsrConfig` (GlmAsrConfig model) - **glmasr_encoder** -- `GlmAsrEncoderConfig` (GlmAsrEncoderConfig model) - **glpn** -- `GLPNConfig` (GLPNConfig model) - **got_ocr2** -- `GotOcr2Config` (GotOcr2Config model) - **gpt-sw3** -- `GPT2Config` (GPT2Config model) - **gpt2** -- `GPT2Config` (GPT2Config model) - **gpt_bigcode** -- `GPTBigCodeConfig` (GPTBigCodeConfig model) - **gpt_neo** -- `GPTNeoConfig` (GPTNeoConfig model) - **gpt_neox** -- `GPTNeoXConfig` (GPTNeoXConfig model) - **gpt_neox_japanese** -- `GPTNeoXJapaneseConfig` (GPTNeoXJapaneseConfig model) - **gpt_oss** -- `GptOssConfig` (GptOssConfig model) - **gptj** -- `GPTJConfig` (GPTJConfig model) - **granite** -- `GraniteConfig` (GraniteConfig model) - **granite_speech** -- `GraniteSpeechConfig` (GraniteSpeechConfig model) - **granite_speech_encoder** -- `GraniteSpeechEncoderConfig` (GraniteSpeechEncoderConfig model) - **granitemoe** -- `GraniteMoeConfig` (GraniteMoeConfig model) - **granitemoehybrid** -- `GraniteMoeHybridConfig` (GraniteMoeHybridConfig model) - **granitemoeshared** -- `GraniteMoeSharedConfig` (GraniteMoeSharedConfig model) - **grounding-dino** -- `GroundingDinoConfig` (GroundingDinoConfig model) - **groupvit** -- `GroupViTConfig` (GroupViTConfig model) - **groupvit_text_model** -- `GroupViTTextConfig` (GroupViTTextConfig model) - **groupvit_vision_model** -- `GroupViTVisionConfig` (GroupViTVisionConfig model) - **helium** -- `HeliumConfig` (HeliumConfig model) - **hgnet_v2** -- `HGNetV2Config` (HGNetV2Config model) - **hiera** -- `HieraConfig` (HieraConfig model) - **higgs_audio_v2** -- `HiggsAudioV2Config` (HiggsAudioV2Config model) - **higgs_audio_v2_tokenizer** -- `HiggsAudioV2TokenizerConfig` (HiggsAudioV2TokenizerConfig model) - **hubert** -- `HubertConfig` (HubertConfig model) - **hunyuan_v1_dense** -- `HunYuanDenseV1Config` (HunYuanDenseV1Config model) - **hunyuan_v1_moe** -- `HunYuanMoEV1Config` (HunYuanMoEV1Config model) - **hy_v3** -- `HYV3Config` (HYV3Config model) - **ibert** -- `IBertConfig` (IBertConfig model) - **idefics** -- `IdeficsConfig` (IdeficsConfig model) - **idefics2** -- `Idefics2Config` (Idefics2Config model) - **idefics2_perceiver** -- `Idefics2PerceiverConfig` (Idefics2PerceiverConfig model) - **idefics2_vision** -- `Idefics2VisionConfig` (Idefics2VisionConfig model) - **idefics3** -- `Idefics3Config` (Idefics3Config model) - **idefics3_vision** -- `Idefics3VisionConfig` (Idefics3VisionConfig model) - **idefics_perciever** -- `IdeficsPerceiverConfig` (IdeficsPerceiverConfig model) - **idefics_vision** -- `IdeficsVisionConfig` (IdeficsVisionConfig model) - **ijepa** -- `IJepaConfig` (IJepaConfig model) - **imagegpt** -- `ImageGPTConfig` (ImageGPTConfig model) - **informer** -- `InformerConfig` (InformerConfig model) - **instructblip** -- `InstructBlipConfig` (InstructBlipConfig model) - **instructblip_qformer** -- `InstructBlipQFormerConfig` (InstructBlipQFormerConfig model) - **instructblip_vision_model** -- `InstructBlipVisionConfig` (InstructBlipVisionConfig model) - **instructblipvideo** -- `InstructBlipVideoConfig` (InstructBlipVideoConfig model) - **instructblipvideo_qformer** -- `InstructBlipVideoQFormerConfig` (InstructBlipVideoQFormerConfig model) - **instructblipvideo_vision_model** -- `InstructBlipVideoVisionConfig` (InstructBlipVideoVisionConfig model) - **internvl** -- `InternVLConfig` (InternVLConfig model) - **internvl_vision** -- `InternVLVisionConfig` (InternVLVisionConfig model) - **jais2** -- `Jais2Config` (Jais2Config model) - **jamba** -- `JambaConfig` (JambaConfig model) - **janus** -- `JanusConfig` (JanusConfig model) - **janus_vision_model** -- `JanusVisionConfig` (JanusVisionConfig model) - **janus_vqgan** -- `JanusVQVAEConfig` (JanusVQVAEConfig model) - **jetmoe** -- `JetMoeConfig` (JetMoeConfig model) - **jina_embeddings_v3** -- `JinaEmbeddingsV3Config` (JinaEmbeddingsV3Config model) - **kosmos-2** -- `Kosmos2Config` (Kosmos2Config model) - **kosmos-2.5** -- `Kosmos2_5Config` (Kosmos2_5Config model) - **kosmos_2_5_text_model** -- `Kosmos2_5TextConfig` (Kosmos2_5TextConfig model) - **kosmos_2_5_vision_model** -- `Kosmos2_5VisionConfig` (Kosmos2_5VisionConfig model) - **kosmos_2_text_model** -- `Kosmos2TextConfig` (Kosmos2TextConfig model) - **kosmos_2_vision_model** -- `Kosmos2VisionConfig` (Kosmos2VisionConfig model) - **kyutai_speech_to_text** -- `KyutaiSpeechToTextConfig` (KyutaiSpeechToTextConfig model) - **lasr_ctc** -- `LasrCTCConfig` (LasrCTCConfig model) - **lasr_encoder** -- `LasrEncoderConfig` (LasrEncoderConfig model) - **layoutlm** -- `LayoutLMConfig` (LayoutLMConfig model) - **layoutlmv2** -- `LayoutLMv2Config` (LayoutLMv2Config model) - **layoutlmv3** -- `LayoutLMv3Config` (LayoutLMv3Config model) - **layoutxlm** -- `LayoutXLMConfig` (LayoutXLMConfig model) - **led** -- `LEDConfig` (LEDConfig model) - **levit** -- `LevitConfig` (LevitConfig model) - **lfm2** -- `Lfm2Config` (Lfm2Config model) - **lfm2_moe** -- `Lfm2MoeConfig` (Lfm2MoeConfig model) - **lfm2_vl** -- `Lfm2VlConfig` (Lfm2VlConfig model) - **lightglue** -- `LightGlueConfig` (LightGlueConfig model) - **lighton_ocr** -- `LightOnOcrConfig` (LightOnOcrConfig model) - **lilt** -- `LiltConfig` (LiltConfig model) - **llama** -- `LlamaConfig` (LlamaConfig model) - **llama4** -- `Llama4Config` (Llama4Config model) - **llama4_text** -- `Llama4TextConfig` (Llama4TextConfig model) - **llama4_vision_model** -- `Llama4VisionConfig` (Llama4VisionConfig model) - **llava** -- `LlavaConfig` (LlavaConfig model) - **llava_next** -- `LlavaNextConfig` (LlavaNextConfig model) - **llava_next_video** -- `LlavaNextVideoConfig` (LlavaNextVideoConfig model) - **llava_onevision** -- `LlavaOnevisionConfig` (LlavaOnevisionConfig model) - **longcat_flash** -- `LongcatFlashConfig` (LongcatFlashConfig model) - **longformer** -- `LongformerConfig` (LongformerConfig model) - **longt5** -- `LongT5Config` (LongT5Config model) - **luke** -- `LukeConfig` (LukeConfig model) - **lw_detr** -- `LwDetrConfig` (LwDetrConfig model) - **lw_detr_vit** -- `LwDetrViTConfig` (LwDetrViTConfig model) - **lxmert** -- `LxmertConfig` (LxmertConfig model) - **m2m_100** -- `M2M100Config` (M2M100Config model) - **mamba** -- `MambaConfig` (MambaConfig model) - **mamba2** -- `Mamba2Config` (Mamba2Config model) - **marian** -- `MarianConfig` (MarianConfig model) - **markuplm** -- `MarkupLMConfig` (MarkupLMConfig model) - **mask2former** -- `Mask2FormerConfig` (Mask2FormerConfig model) - **maskformer** -- `MaskFormerConfig` (MaskFormerConfig model) - **maskformer-swin** -- `MaskFormerSwinConfig` (MaskFormerSwinConfig model) - **mbart** -- `MBartConfig` (MBartConfig model) - **megatron-bert** -- `MegatronBertConfig` (MegatronBertConfig model) - **metaclip_2** -- `MetaClip2Config` (MetaClip2Config model) - **metaclip_2_text_model** -- `MetaClip2TextConfig` (MetaClip2TextConfig model) - **metaclip_2_vision_model** -- `MetaClip2VisionConfig` (MetaClip2VisionConfig model) - **mgp-str** -- `MgpstrConfig` (MgpstrConfig model) - **mimi** -- `MimiConfig` (MimiConfig model) - **minimax** -- `MiniMaxConfig` (MiniMaxConfig model) - **minimax_m2** -- `MiniMaxM2Config` (MiniMaxM2Config model) - **ministral** -- `MinistralConfig` (MinistralConfig model) - **ministral3** -- `Ministral3Config` (Ministral3Config model) - **mistral** -- `MistralConfig` (MistralConfig model) - **mistral3** -- `Mistral3Config` (Mistral3Config model) - **mistral4** -- `Mistral4Config` (Mistral4Config model) - **mixtral** -- `MixtralConfig` (MixtralConfig model) - **mlcd** -- `MLCDVisionConfig` (MLCDVisionConfig model) - **mlcd_vision_model** -- `MLCDVisionConfig` (MLCDVisionConfig model) - **mllama** -- `MllamaConfig` (MllamaConfig model) - **mllama_text_model** -- `MllamaTextConfig` (MllamaTextConfig model) - **mllama_vision_model** -- `MllamaVisionConfig` (MllamaVisionConfig model) - **mm-grounding-dino** -- `MMGroundingDinoConfig` (MMGroundingDinoConfig model) - **mobilebert** -- `MobileBertConfig` (MobileBertConfig model) - **mobilenet_v1** -- `MobileNetV1Config` (MobileNetV1Config model) - **mobilenet_v2** -- `MobileNetV2Config` (MobileNetV2Config model) - **mobilevit** -- `MobileViTConfig` (MobileViTConfig model) - **mobilevitv2** -- `MobileViTV2Config` (MobileViTV2Config model) - **modernbert** -- `ModernBertConfig` (ModernBertConfig model) - **modernbert-decoder** -- `ModernBertDecoderConfig` (ModernBertDecoderConfig model) - **modernvbert** -- `ModernVBertConfig` (ModernVBertConfig model) - **moonshine** -- `MoonshineConfig` (MoonshineConfig model) - **moonshine_streaming** -- `MoonshineStreamingConfig` (MoonshineStreamingConfig model) - **moonshine_streaming_encoder** -- `MoonshineStreamingEncoderConfig` (MoonshineStreamingEncoderConfig model) - **moshi** -- `MoshiConfig` (MoshiConfig model) - **moshi_depth** -- `MoshiDepthConfig` (MoshiDepthConfig model) - **mpnet** -- `MPNetConfig` (MPNetConfig model) - **mpt** -- `MptConfig` (MptConfig model) - **mra** -- `MraConfig` (MraConfig model) - **mt5** -- `MT5Config` (MT5Config model) - **musicflamingo** -- `MusicFlamingoConfig` (MusicFlamingoConfig model) - **musicgen** -- `MusicgenConfig` (MusicgenConfig model) - **musicgen_decoder** -- `MusicgenDecoderConfig` (MusicgenDecoderConfig model) - **musicgen_melody** -- `MusicgenMelodyConfig` (MusicgenMelodyConfig model) - **musicgen_melody_decoder** -- `MusicgenMelodyDecoderConfig` (MusicgenMelodyDecoderConfig model) - **mvp** -- `MvpConfig` (MvpConfig model) - **nanochat** -- `NanoChatConfig` (NanoChatConfig model) - **nemotron** -- `NemotronConfig` (NemotronConfig model) - **nemotron_h** -- `NemotronHConfig` (NemotronHConfig model) - **nllb-moe** -- `NllbMoeConfig` (NllbMoeConfig model) - **nomic_bert** -- `NomicBertConfig` (NomicBertConfig model) - **nougat** -- `NougatConfig` (NougatConfig model) - **nystromformer** -- `NystromformerConfig` (NystromformerConfig model) - **olmo** -- `OlmoConfig` (OlmoConfig model) - **olmo2** -- `Olmo2Config` (Olmo2Config model) - **olmo3** -- `Olmo3Config` (Olmo3Config model) - **olmo_hybrid** -- `OlmoHybridConfig` (OlmoHybridConfig model) - **olmoe** -- `OlmoeConfig` (OlmoeConfig model) - **omdet-turbo** -- `OmDetTurboConfig` (OmDetTurboConfig model) - **oneformer** -- `OneFormerConfig` (OneFormerConfig model) - **openai-gpt** -- `OpenAIGPTConfig` (OpenAIGPTConfig model) - **openai_privacy_filter** -- `OpenAIPrivacyFilterConfig` (OpenAIPrivacyFilterConfig model) - **opt** -- `OPTConfig` (OPTConfig model) - **ovis2** -- `Ovis2Config` (Ovis2Config model) - **owlv2** -- `Owlv2Config` (Owlv2Config model) - **owlv2_text_model** -- `Owlv2TextConfig` (Owlv2TextConfig model) - **owlv2_vision_model** -- `Owlv2VisionConfig` (Owlv2VisionConfig model) - **owlvit** -- `OwlViTConfig` (OwlViTConfig model) - **owlvit_text_model** -- `OwlViTTextConfig` (OwlViTTextConfig model) - **owlvit_vision_model** -- `OwlViTVisionConfig` (OwlViTVisionConfig model) - **paddleocr_vl** -- `PaddleOCRVLConfig` (PaddleOCRVLConfig model) - **paddleocr_vl_text** -- `PaddleOCRTextConfig` (PaddleOCRTextConfig model) - **paddleocr_vl_vision** -- `PaddleOCRVisionConfig` (PaddleOCRVisionConfig model) - **paligemma** -- `PaliGemmaConfig` (PaliGemmaConfig model) - **parakeet_ctc** -- `ParakeetCTCConfig` (ParakeetCTCConfig model) - **parakeet_encoder** -- `ParakeetEncoderConfig` (ParakeetEncoderConfig model) - **patchtsmixer** -- `PatchTSMixerConfig` (PatchTSMixerConfig model) - **patchtst** -- `PatchTSTConfig` (PatchTSTConfig model) - **pe_audio** -- `PeAudioConfig` (PeAudioConfig model) - **pe_audio_encoder** -- `PeAudioEncoderConfig` (PeAudioEncoderConfig model) - **pe_audio_video** -- `PeAudioVideoConfig` (PeAudioVideoConfig model) - **pe_audio_video_encoder** -- `PeAudioVideoEncoderConfig` (PeAudioVideoEncoderConfig model) - **pe_video** -- `PeVideoConfig` (PeVideoConfig model) - **pe_video_encoder** -- `PeVideoEncoderConfig` (PeVideoEncoderConfig model) - **pegasus** -- `PegasusConfig` (PegasusConfig model) - **pegasus_x** -- `PegasusXConfig` (PegasusXConfig model) - **perceiver** -- `PerceiverConfig` (PerceiverConfig model) - **perception_lm** -- `PerceptionLMConfig` (PerceptionLMConfig model) - **persimmon** -- `PersimmonConfig` (PersimmonConfig model) - **phi** -- `PhiConfig` (PhiConfig model) - **phi3** -- `Phi3Config` (Phi3Config model) - **phi4_multimodal** -- `Phi4MultimodalConfig` (Phi4MultimodalConfig model) - **phi4_multimodal_audio** -- `Phi4MultimodalAudioConfig` (Phi4MultimodalAudioConfig model) - **phi4_multimodal_vision** -- `Phi4MultimodalVisionConfig` (Phi4MultimodalVisionConfig model) - **phimoe** -- `PhimoeConfig` (PhimoeConfig model) - **pi0** -- `PI0Config` (PI0Config model) - **pix2struct** -- `Pix2StructConfig` (Pix2StructConfig model) - **pix2struct_text_model** -- `Pix2StructTextConfig` (Pix2StructTextConfig model) - **pix2struct_vision_model** -- `Pix2StructVisionConfig` (Pix2StructVisionConfig model) - **pixio** -- `PixioConfig` (PixioConfig model) - **pixtral** -- `PixtralVisionConfig` (PixtralVisionConfig model) - **plbart** -- `PLBartConfig` (PLBartConfig model) - **poolformer** -- `PoolFormerConfig` (PoolFormerConfig model) - **pop2piano** -- `Pop2PianoConfig` (Pop2PianoConfig model) - **pp_chart2table** -- `PPChart2TableConfig` (PPChart2TableConfig model) - **pp_doclayout_v2** -- `PPDocLayoutV2Config` (PPDocLayoutV2Config model) - **pp_doclayout_v3** -- `PPDocLayoutV3Config` (PPDocLayoutV3Config model) - **pp_lcnet** -- `PPLCNetConfig` (PPLCNetConfig model) - **pp_lcnet_v3** -- `PPLCNetV3Config` (PPLCNetV3Config model) - **pp_ocrv5_mobile_det** -- `PPOCRV5MobileDetConfig` (PPOCRV5MobileDetConfig model) - **pp_ocrv5_mobile_rec** -- `PPOCRV5MobileRecConfig` (PPOCRV5MobileRecConfig model) - **pp_ocrv5_server_det** -- `PPOCRV5ServerDetConfig` (PPOCRV5ServerDetConfig model) - **pp_ocrv5_server_rec** -- `PPOCRV5ServerRecConfig` (PPOCRV5ServerRecConfig model) - **prompt_depth_anything** -- `PromptDepthAnythingConfig` (PromptDepthAnythingConfig model) - **prophetnet** -- `ProphetNetConfig` (ProphetNetConfig model) - **pvt** -- `PvtConfig` (PvtConfig model) - **pvt_v2** -- `PvtV2Config` (PvtV2Config model) - **qianfan_ocr** -- `QianfanOCRConfig` (QianfanOCRConfig model) - **qianfan_ocr_vision** -- `QianfanOCRVisionConfig` (QianfanOCRVisionConfig model) - **qwen2** -- `Qwen2Config` (Qwen2Config model) - **qwen2_5_omni** -- `Qwen2_5OmniConfig` (Qwen2_5OmniConfig model) - **qwen2_5_omni_audio_encoder** -- `Qwen2_5OmniAudioEncoderConfig` (Qwen2_5OmniAudioEncoderConfig model) - **qwen2_5_omni_bigvgan** -- `Qwen2_5OmniBigVGANConfig` (Qwen2_5OmniBigVGANConfig model) - **qwen2_5_omni_dit** -- `Qwen2_5OmniDiTConfig` (Qwen2_5OmniDiTConfig model) - **qwen2_5_omni_talker** -- `Qwen2_5OmniTalkerConfig` (Qwen2_5OmniTalkerConfig model) - **qwen2_5_omni_text** -- `Qwen2_5OmniTextConfig` (Qwen2_5OmniTextConfig model) - **qwen2_5_omni_thinker** -- `Qwen2_5OmniThinkerConfig` (Qwen2_5OmniThinkerConfig model) - **qwen2_5_omni_token2wav** -- `Qwen2_5OmniToken2WavConfig` (Qwen2_5OmniToken2WavConfig model) - **qwen2_5_omni_vision_encoder** -- `Qwen2_5OmniVisionEncoderConfig` (Qwen2_5OmniVisionEncoderConfig model) - **qwen2_5_vl** -- `Qwen2_5_VLConfig` (Qwen2_5_VLConfig model) - **qwen2_5_vl_text** -- `Qwen2_5_VLTextConfig` (Qwen2_5_VLTextConfig model) - **qwen2_5_vl_vision** -- `Qwen2_5_VLVisionConfig` (Qwen2_5_VLVisionConfig model) - **qwen2_audio** -- `Qwen2AudioConfig` (Qwen2AudioConfig model) - **qwen2_audio_encoder** -- `Qwen2AudioEncoderConfig` (Qwen2AudioEncoderConfig model) - **qwen2_moe** -- `Qwen2MoeConfig` (Qwen2MoeConfig model) - **qwen2_vl** -- `Qwen2VLConfig` (Qwen2VLConfig model) - **qwen2_vl_text** -- `Qwen2VLTextConfig` (Qwen2VLTextConfig model) - **qwen2_vl_vision** -- `Qwen2VLVisionConfig` (Qwen2VLVisionConfig model) - **qwen3** -- `Qwen3Config` (Qwen3Config model) - **qwen3_5** -- `Qwen3_5Config` (Qwen3_5Config model) - **qwen3_5_moe** -- `Qwen3_5MoeConfig` (Qwen3_5MoeConfig model) - **qwen3_5_moe_text** -- `Qwen3_5MoeTextConfig` (Qwen3_5MoeTextConfig model) - **qwen3_5_moe_vision** -- `Qwen3_5MoeVisionConfig` (Qwen3_5MoeVisionConfig model) - **qwen3_5_text** -- `Qwen3_5TextConfig` (Qwen3_5TextConfig model) - **qwen3_5_vision** -- `Qwen3_5VisionConfig` (Qwen3_5VisionConfig model) - **qwen3_moe** -- `Qwen3MoeConfig` (Qwen3MoeConfig model) - **qwen3_next** -- `Qwen3NextConfig` (Qwen3NextConfig model) - **qwen3_omni_moe** -- `Qwen3OmniMoeConfig` (Qwen3OmniMoeConfig model) - **qwen3_omni_moe_audio_encoder** -- `Qwen3OmniMoeAudioEncoderConfig` (Qwen3OmniMoeAudioEncoderConfig model) - **qwen3_omni_moe_talker_code_predictor** -- `Qwen3OmniMoeTalkerCodePredictorConfig` (Qwen3OmniMoeTalkerCodePredictorConfig model) - **qwen3_omni_moe_talker_text** -- `Qwen3OmniMoeTalkerTextConfig` (Qwen3OmniMoeTalkerTextConfig model) - **qwen3_omni_moe_text** -- `Qwen3OmniMoeTextConfig` (Qwen3OmniMoeTextConfig model) - **qwen3_omni_moe_thinker** -- `Qwen3OmniMoeThinkerConfig` (Qwen3OmniMoeThinkerConfig model) - **qwen3_omni_moe_vision_encoder** -- `Qwen3OmniMoeVisionEncoderConfig` (Qwen3OmniMoeVisionEncoderConfig model) - **qwen3_vl** -- `Qwen3VLConfig` (Qwen3VLConfig model) - **qwen3_vl_moe** -- `Qwen3VLMoeConfig` (Qwen3VLMoeConfig model) - **qwen3_vl_moe_text** -- `Qwen3VLMoeTextConfig` (Qwen3VLMoeTextConfig model) - **qwen3_vl_moe_vision** -- `Qwen3VLMoeVisionConfig` (Qwen3VLMoeVisionConfig model) - **qwen3_vl_text** -- `Qwen3VLTextConfig` (Qwen3VLTextConfig model) - **qwen3_vl_vision** -- `Qwen3VLVisionConfig` (Qwen3VLVisionConfig model) - **rag** -- `RagConfig` (RagConfig model) - **recurrent_gemma** -- `RecurrentGemmaConfig` (RecurrentGemmaConfig model) - **reformer** -- `ReformerConfig` (ReformerConfig model) - **regnet** -- `RegNetConfig` (RegNetConfig model) - **rembert** -- `RemBertConfig` (RemBertConfig model) - **resnet** -- `ResNetConfig` (ResNetConfig model) - **roberta** -- `RobertaConfig` (RobertaConfig model) - **roberta-prelayernorm** -- `RobertaPreLayerNormConfig` (RobertaPreLayerNormConfig model) - **roc_bert** -- `RoCBertConfig` (RoCBertConfig model) - **roformer** -- `RoFormerConfig` (RoFormerConfig model) - **rt_detr** -- `RTDetrConfig` (RTDetrConfig model) - **rt_detr_resnet** -- `RTDetrResNetConfig` (RTDetrResNetConfig model) - **rt_detr_v2** -- `RTDetrV2Config` (RTDetrV2Config model) - **rwkv** -- `RwkvConfig` (RwkvConfig model) - **sam** -- `SamConfig` (SamConfig model) - **sam2** -- `Sam2Config` (Sam2Config model) - **sam2_hiera_det_model** -- `Sam2HieraDetConfig` (Sam2HieraDetConfig model) - **sam2_video** -- `Sam2VideoConfig` (Sam2VideoConfig model) - **sam2_vision_model** -- `Sam2VisionConfig` (Sam2VisionConfig model) - **sam3** -- `Sam3Config` (Sam3Config model) - **sam3_detr_decoder** -- `Sam3DETRDecoderConfig` (Sam3DETRDecoderConfig model) - **sam3_detr_encoder** -- `Sam3DETREncoderConfig` (Sam3DETREncoderConfig model) - **sam3_geometry_encoder** -- `Sam3GeometryEncoderConfig` (Sam3GeometryEncoderConfig model) - **sam3_lite_text** -- `Sam3LiteTextConfig` (Sam3LiteTextConfig model) - **sam3_lite_text_detr_decoder** -- `Sam3LiteTextDETRDecoderConfig` (Sam3LiteTextDETRDecoderConfig model) - **sam3_lite_text_detr_encoder** -- `Sam3LiteTextDETREncoderConfig` (Sam3LiteTextDETREncoderConfig model) - **sam3_lite_text_geometry_encoder** -- `Sam3LiteTextGeometryEncoderConfig` (Sam3LiteTextGeometryEncoderConfig model) - **sam3_lite_text_mask_decoder** -- `Sam3LiteTextMaskDecoderConfig` (Sam3LiteTextMaskDecoderConfig model) - **sam3_lite_text_text_model** -- `Sam3LiteTextTextConfig` (Sam3LiteTextTextConfig model) - **sam3_mask_decoder** -- `Sam3MaskDecoderConfig` (Sam3MaskDecoderConfig model) - **sam3_tracker** -- `Sam3TrackerConfig` (Sam3TrackerConfig model) - **sam3_tracker_video** -- `Sam3TrackerVideoConfig` (Sam3TrackerVideoConfig model) - **sam3_video** -- `Sam3VideoConfig` (Sam3VideoConfig model) - **sam3_vision_model** -- `Sam3VisionConfig` (Sam3VisionConfig model) - **sam3_vit_model** -- `Sam3ViTConfig` (Sam3ViTConfig model) - **sam_hq** -- `SamHQConfig` (SamHQConfig model) - **sam_hq_vision_model** -- `SamHQVisionConfig` (SamHQVisionConfig model) - **sam_vision_model** -- `SamVisionConfig` (SamVisionConfig model) - **seamless_m4t** -- `SeamlessM4TConfig` (SeamlessM4TConfig model) - **seamless_m4t_v2** -- `SeamlessM4Tv2Config` (SeamlessM4Tv2Config model) - **seed_oss** -- `SeedOssConfig` (SeedOssConfig model) - **segformer** -- `SegformerConfig` (SegformerConfig model) - **seggpt** -- `SegGptConfig` (SegGptConfig model) - **sew** -- `SEWConfig` (SEWConfig model) - **sew-d** -- `SEWDConfig` (SEWDConfig model) - **shieldgemma2** -- `ShieldGemma2Config` (ShieldGemma2Config model) - **siglip** -- `SiglipConfig` (SiglipConfig model) - **siglip2** -- `Siglip2Config` (Siglip2Config model) - **siglip2_text_model** -- `Siglip2TextConfig` (Siglip2TextConfig model) - **siglip2_vision_model** -- `Siglip2VisionConfig` (Siglip2VisionConfig model) - **siglip_text_model** -- `SiglipTextConfig` (SiglipTextConfig model) - **siglip_vision_model** -- `SiglipVisionConfig` (SiglipVisionConfig model) - **slanet** -- `SLANetConfig` (SLANetConfig model) - **slanext** -- `SLANeXtConfig` (SLANeXtConfig model) - **smollm3** -- `SmolLM3Config` (SmolLM3Config model) - **smolvlm** -- `SmolVLMConfig` (SmolVLMConfig model) - **smolvlm_vision** -- `SmolVLMVisionConfig` (SmolVLMVisionConfig model) - **solar_open** -- `SolarOpenConfig` (SolarOpenConfig model) - **speech-encoder-decoder** -- `SpeechEncoderDecoderConfig` (SpeechEncoderDecoderConfig model) - **speech_to_text** -- `Speech2TextConfig` (Speech2TextConfig model) - **speecht5** -- `SpeechT5Config` (SpeechT5Config model) - **speecht5_hifigan** -- `SpeechT5HifiGanConfig` (SpeechT5HifiGanConfig model) - **splinter** -- `SplinterConfig` (SplinterConfig model) - **squeezebert** -- `SqueezeBertConfig` (SqueezeBertConfig model) - **stablelm** -- `StableLmConfig` (StableLmConfig model) - **starcoder2** -- `Starcoder2Config` (Starcoder2Config model) - **superglue** -- `SuperGlueConfig` (SuperGlueConfig model) - **superpoint** -- `SuperPointConfig` (SuperPointConfig model) - **swiftformer** -- `SwiftFormerConfig` (SwiftFormerConfig model) - **swin** -- `SwinConfig` (SwinConfig model) - **swin2sr** -- `Swin2SRConfig` (Swin2SRConfig model) - **swinv2** -- `Swinv2Config` (Swinv2Config model) - **switch_transformers** -- `SwitchTransformersConfig` (SwitchTransformersConfig model) - **t5** -- `T5Config` (T5Config model) - **t5_gemma_module** -- `T5GemmaModuleConfig` (T5GemmaModuleConfig model) - **t5gemma** -- `T5GemmaConfig` (T5GemmaConfig model) - **t5gemma2** -- `T5Gemma2Config` (T5Gemma2Config model) - **t5gemma2_decoder** -- `T5Gemma2DecoderConfig` (T5Gemma2DecoderConfig model) - **t5gemma2_encoder** -- `T5Gemma2EncoderConfig` (T5Gemma2EncoderConfig model) - **t5gemma2_text** -- `T5Gemma2TextConfig` (T5Gemma2TextConfig model) - **table-transformer** -- `TableTransformerConfig` (TableTransformerConfig model) - **tapas** -- `TapasConfig` (TapasConfig model) - **textnet** -- `TextNetConfig` (TextNetConfig model) - **time_series_transformer** -- `TimeSeriesTransformerConfig` (TimeSeriesTransformerConfig model) - **timesfm** -- `TimesFmConfig` (TimesFmConfig model) - **timesfm2_5** -- `TimesFm2_5Config` (TimesFm2_5Config model) - **timesformer** -- `TimesformerConfig` (TimesformerConfig model) - **timm_backbone** -- `TimmBackboneConfig` (TimmBackboneConfig model) - **timm_wrapper** -- `TimmWrapperConfig` (TimmWrapperConfig model) - **trocr** -- `TrOCRConfig` (TrOCRConfig model) - **tvp** -- `TvpConfig` (TvpConfig model) - **udop** -- `UdopConfig` (UdopConfig model) - **umt5** -- `UMT5Config` (UMT5Config model) - **unispeech** -- `UniSpeechConfig` (UniSpeechConfig model) - **unispeech-sat** -- `UniSpeechSatConfig` (UniSpeechSatConfig model) - **univnet** -- `UnivNetConfig` (UnivNetConfig model) - **upernet** -- `UperNetConfig` (UperNetConfig model) - **uvdoc** -- `UVDocConfig` (UVDocConfig model) - **uvdoc_backbone** -- `UVDocBackboneConfig` (UVDocBackboneConfig model) - **vaultgemma** -- `VaultGemmaConfig` (VaultGemmaConfig model) - **vibevoice_acoustic_tokenizer** -- `VibeVoiceAcousticTokenizerConfig` (VibeVoiceAcousticTokenizerConfig model) - **vibevoice_acoustic_tokenizer_decoder** -- `VibeVoiceAcousticTokenizerDecoderConfig` (VibeVoiceAcousticTokenizerDecoderConfig model) - **vibevoice_acoustic_tokenizer_encoder** -- `VibeVoiceAcousticTokenizerEncoderConfig` (VibeVoiceAcousticTokenizerEncoderConfig model) - **vibevoice_asr** -- `VibeVoiceAsrConfig` (VibeVoiceAsrConfig model) - **video_llama_3** -- `VideoLlama3Config` (VideoLlama3Config model) - **video_llama_3_vision** -- `VideoLlama3VisionConfig` (VideoLlama3VisionConfig model) - **video_llava** -- `VideoLlavaConfig` (VideoLlavaConfig model) - **videomae** -- `VideoMAEConfig` (VideoMAEConfig model) - **videomt** -- `VideomtConfig` (VideomtConfig model) - **vilt** -- `ViltConfig` (ViltConfig model) - **vipllava** -- `VipLlavaConfig` (VipLlavaConfig model) - **vision-encoder-decoder** -- `VisionEncoderDecoderConfig` (VisionEncoderDecoderConfig model) - **vision-text-dual-encoder** -- `VisionTextDualEncoderConfig` (VisionTextDualEncoderConfig model) - **visual_bert** -- `VisualBertConfig` (VisualBertConfig model) - **vit** -- `ViTConfig` (ViTConfig model) - **vit_mae** -- `ViTMAEConfig` (ViTMAEConfig model) - **vit_msn** -- `ViTMSNConfig` (ViTMSNConfig model) - **vitdet** -- `VitDetConfig` (VitDetConfig model) - **vitmatte** -- `VitMatteConfig` (VitMatteConfig model) - **vitpose** -- `VitPoseConfig` (VitPoseConfig model) - **vitpose_backbone** -- `VitPoseBackboneConfig` (VitPoseBackboneConfig model) - **vits** -- `VitsConfig` (VitsConfig model) - **vivit** -- `VivitConfig` (VivitConfig model) - **vjepa2** -- `VJEPA2Config` (VJEPA2Config model) - **voxtral** -- `VoxtralConfig` (VoxtralConfig model) - **voxtral_encoder** -- `VoxtralEncoderConfig` (VoxtralEncoderConfig model) - **voxtral_realtime** -- `VoxtralRealtimeConfig` (VoxtralRealtimeConfig model) - **voxtral_realtime_encoder** -- `VoxtralRealtimeEncoderConfig` (VoxtralRealtimeEncoderConfig model) - **voxtral_realtime_text** -- `VoxtralRealtimeTextConfig` (VoxtralRealtimeTextConfig model) - **wav2vec2** -- `Wav2Vec2Config` (Wav2Vec2Config model) - **wav2vec2-bert** -- `Wav2Vec2BertConfig` (Wav2Vec2BertConfig model) - **wav2vec2-conformer** -- `Wav2Vec2ConformerConfig` (Wav2Vec2ConformerConfig model) - **wavlm** -- `WavLMConfig` (WavLMConfig model) - **whisper** -- `WhisperConfig` (WhisperConfig model) - **xclip** -- `XCLIPConfig` (XCLIPConfig model) - **xclip_text_model** -- `XCLIPTextConfig` (XCLIPTextConfig model) - **xclip_vision_model** -- `XCLIPVisionConfig` (XCLIPVisionConfig model) - **xcodec** -- `XcodecConfig` (XcodecConfig model) - **xglm** -- `XGLMConfig` (XGLMConfig model) - **xlm** -- `XLMConfig` (XLMConfig model) - **xlm-roberta** -- `XLMRobertaConfig` (XLMRobertaConfig model) - **xlm-roberta-xl** -- `XLMRobertaXLConfig` (XLMRobertaXLConfig model) - **xlnet** -- `XLNetConfig` (XLNetConfig model) - **xlstm** -- `xLSTMConfig` (xLSTMConfig model) - **xmod** -- `XmodConfig` (XmodConfig model) - **yolos** -- `YolosConfig` (YolosConfig model) - **yoso** -- `YosoConfig` (YosoConfig model) - **youtu** -- `YoutuConfig` (YoutuConfig model) - **zamba** -- `ZambaConfig` (ZambaConfig model) - **zamba2** -- `Zamba2Config` (Zamba2Config model) - **zoedepth** -- `ZoeDepthConfig` (ZoeDepthConfig model) Examples: ```python >>> from transformers import AutoConfig >>> # Download configuration from huggingface.co and cache. >>> config = AutoConfig.from_pretrained("google-bert/bert-base-uncased") >>> # Download configuration from huggingface.co (user-uploaded) and cache. >>> config = AutoConfig.from_pretrained("dbmdz/bert-base-german-cased") >>> # If configuration file is in a directory (e.g., was saved using *save_pretrained('./test/saved_model/')*). >>> config = AutoConfig.from_pretrained("./test/bert_saved_model/") >>> # Load a specific configuration file. >>> config = AutoConfig.from_pretrained("./test/bert_saved_model/my_configuration.json") >>> # Change some config attributes when loading a pretrained config. >>> config = AutoConfig.from_pretrained("google-bert/bert-base-uncased", output_attentions=True, foo=False) >>> config.output_attentions True >>> config, unused_kwargs = AutoConfig.from_pretrained( ... "google-bert/bert-base-uncased", output_attentions=True, foo=False, return_unused_kwargs=True ... ) >>> config.output_attentions True >>> unused_kwargs {'foo': False} ``` **Parameters:** pretrained_model_name_or_path (`str` or `os.PathLike`) : Can be either: - A string, the *model id* of a pretrained model configuration hosted inside a model repo on huggingface.co. - A path to a *directory* containing a configuration file saved using the [save_pretrained()](/docs/transformers/v5.6.0/ja/main_classes/configuration#transformers.PreTrainedConfig.save_pretrained) method, or the [save_pretrained()](/docs/transformers/v5.6.0/ja/main_classes/model#transformers.PreTrainedModel.save_pretrained) method, e.g., `./my_model_directory/`. - a path to a saved configuration JSON *file*, e.g., `./my_model_directory/configuration.json`. cache_dir (`str` or `os.PathLike`, *optional*) : Path to a directory in which a downloaded pretrained model configuration should be cached if the standard cache should not be used. force_download (`bool`, *optional*, defaults to `False`) : Whether or not to force the (re-)download the model weights and configuration files and override the cached versions if they exist. proxies (`dict[str, str]`, *optional*) : A dictionary of proxy servers to use by protocol or endpoint, e.g., `{'http': 'foo.bar:3128', 'http://hostname': 'foo.bar:4012'}`. The proxies are used on each request. revision (`str`, *optional*, defaults to `"main"`) : The specific model version to use. It can be a branch name, a tag name, or a commit id, since we use a git-based system for storing models and other artifacts on huggingface.co, so `revision` can be any identifier allowed by git. return_unused_kwargs (`bool`, *optional*, defaults to `False`) : If `False`, then this function returns just the final configuration object. If `True`, then this functions returns a `Tuple(config, unused_kwargs)` where *unused_kwargs* is a dictionary consisting of the key/value pairs whose keys are not configuration attributes: i.e., the part of `kwargs` which has not been used to update `config` and is otherwise ignored. trust_remote_code (`bool`, *optional*, defaults to `False`) : Whether or not to allow for custom models defined on the Hub in their own modeling files. This option should only be set to `True` for repositories you trust and in which you have read the code, as it will execute code present on the Hub on your local machine. kwargs(additional keyword arguments, *optional*) : The values in kwargs of any keys which are configuration attributes will be used to override the loaded values. Behavior concerning key/value pairs whose keys are *not* configuration attributes is controlled by the `return_unused_kwargs` keyword parameter. #### register[[transformers.AutoConfig.register]] [Source](https://github.com/huggingface/transformers/blob/v5.6.0/src/transformers/models/auto/configuration_auto.py#L424) Register a new configuration for this class. **Parameters:** model_type (`str`) : The model type like "bert" or "gpt". config ([PreTrainedConfig](/docs/transformers/v5.6.0/ja/main_classes/configuration#transformers.PreTrainedConfig)) : The config to register. ## AutoTokenizer[[transformers.AutoTokenizer]] #### transformers.AutoTokenizer[[transformers.AutoTokenizer]] [Source](https://github.com/huggingface/transformers/blob/v5.6.0/src/transformers/models/auto/tokenization_auto.py#L558) This is a generic tokenizer class that will be instantiated as one of the tokenizer classes of the library when created with the [AutoTokenizer.from_pretrained()](/docs/transformers/v5.6.0/ja/model_doc/auto#transformers.AutoTokenizer.from_pretrained) class method. This class cannot be instantiated directly using `__init__()` (throws an error). from_pretrainedtransformers.AutoTokenizer.from_pretrainedhttps://github.com/huggingface/transformers/blob/v5.6.0/src/transformers/models/auto/tokenization_auto.py#L572[{"name": "pretrained_model_name_or_path", "val": ""}, {"name": "*inputs", "val": ""}, {"name": "**kwargs", "val": ""}]- **pretrained_model_name_or_path** (`str` or `os.PathLike`) -- Can be either: - A string, the *model id* of a predefined tokenizer hosted inside a model repo on huggingface.co. - A path to a *directory* containing vocabulary files required by the tokenizer, for instance saved using the [save_pretrained()](/docs/transformers/v5.6.0/ja/internal/tokenization_utils#transformers.PreTrainedTokenizerBase.save_pretrained) method, e.g., `./my_model_directory/`. - a path to a single saved vocabulary file if and only if the tokenizer only requires a single vocabulary file (like Bert or XLNet), e.g.: `./my_model_directory/vocab.txt`. (Not applicable to all derived classes) - **inputs** (additional positional arguments, *optional*) -- Will be passed along to the Tokenizer `__init__()` method. - **config** ([PreTrainedConfig](/docs/transformers/v5.6.0/ja/main_classes/configuration#transformers.PreTrainedConfig), *optional*) -- The configuration object used to determine the tokenizer class to instantiate. - **cache_dir** (`str` or `os.PathLike`, *optional*) -- Path to a directory in which a downloaded pretrained model configuration should be cached if the standard cache should not be used. - **force_download** (`bool`, *optional*, defaults to `False`) -- Whether or not to force the (re-)download the model weights and configuration files and override the cached versions if they exist. - **proxies** (`dict[str, str]`, *optional*) -- A dictionary of proxy servers to use by protocol or endpoint, e.g., `{'http': 'foo.bar:3128', 'http://hostname': 'foo.bar:4012'}`. The proxies are used on each request. - **revision** (`str`, *optional*, defaults to `"main"`) -- The specific model version to use. It can be a branch name, a tag name, or a commit id, since we use a git-based system for storing models and other artifacts on huggingface.co, so `revision` can be any identifier allowed by git. - **subfolder** (`str`, *optional*) -- In case the relevant files are located inside a subfolder of the model repo on huggingface.co (e.g. for facebook/rag-token-base), specify it here. - **tokenizer_type** (`str`, *optional*) -- Tokenizer type to be loaded. - **backend** (`str`, *optional*, defaults to `"tokenizers"`) -- Backend to use for tokenization. Valid options are: - `"tokenizers"`: Use the HuggingFace tokenizers library backend (default) - `"sentencepiece"`: Use the SentencePiece backend - **trust_remote_code** (`bool`, *optional*, defaults to `False`) -- Whether or not to allow for custom models defined on the Hub in their own modeling files. This option should only be set to `True` for repositories you trust and in which you have read the code, as it will execute code present on the Hub on your local machine. - **kwargs** (additional keyword arguments, *optional*) -- Will be passed to the Tokenizer `__init__()` method. Can be used to set special tokens like `bos_token`, `eos_token`, `unk_token`, `sep_token`, `pad_token`, `cls_token`, `mask_token`, `additional_special_tokens`. See parameters in the `__init__()` for more details.0 Instantiate one of the tokenizer classes of the library from a pretrained model vocabulary. The tokenizer class to instantiate is selected based on the `model_type` property of the config object (either passed as an argument or loaded from `pretrained_model_name_or_path` if possible), or when it's missing, by falling back to using pattern matching on `pretrained_model_name_or_path`: - **aimv2** -- [CLIPTokenizer](/docs/transformers/v5.6.0/ja/model_doc/clip#transformers.CLIPTokenizer) (Aimv2Config model) - **albert** -- [AlbertTokenizer](/docs/transformers/v5.6.0/ja/model_doc/albert#transformers.AlbertTokenizer) (AlbertConfig model) - **align** -- [BertTokenizer](/docs/transformers/v5.6.0/ja/model_doc/bert#transformers.BertTokenizer) (AlignConfig model) - **audioflamingo3** -- `Qwen2Tokenizer` (AudioFlamingo3Config model) - **aya_vision** -- `CohereTokenizer` (AyaVisionConfig model) - **bark** -- [BertTokenizer](/docs/transformers/v5.6.0/ja/model_doc/bert#transformers.BertTokenizer) (BarkConfig model) - **bart** -- [RobertaTokenizer](/docs/transformers/v5.6.0/ja/model_doc/bart#transformers.RobertaTokenizer) (BartConfig model) - **bert** -- [BertTokenizer](/docs/transformers/v5.6.0/ja/model_doc/bert#transformers.BertTokenizer) (BertConfig model) - **bert-generation** -- [BertGenerationTokenizer](/docs/transformers/v5.6.0/ja/model_doc/bert-generation#transformers.BertGenerationTokenizer) (BertGenerationConfig model) - **big_bird** -- [BigBirdTokenizer](/docs/transformers/v5.6.0/ja/model_doc/big_bird#transformers.BigBirdTokenizer) (BigBirdConfig model) - **bigbird_pegasus** -- `PegasusTokenizer` (BigBirdPegasusConfig model) - **biogpt** -- [BioGptTokenizer](/docs/transformers/v5.6.0/ja/model_doc/biogpt#transformers.BioGptTokenizer) (BioGptConfig model) - **blenderbot** -- [BlenderbotTokenizer](/docs/transformers/v5.6.0/ja/model_doc/blenderbot#transformers.BlenderbotTokenizer) (BlenderbotConfig model) - **blenderbot-small** -- [BlenderbotSmallTokenizer](/docs/transformers/v5.6.0/ja/model_doc/blenderbot-small#transformers.BlenderbotSmallTokenizer) (BlenderbotSmallConfig model) - **blip** -- [BertTokenizer](/docs/transformers/v5.6.0/ja/model_doc/bert#transformers.BertTokenizer) (BlipConfig model) - **blip-2** -- `GPT2Tokenizer` (Blip2Config model) - **bridgetower** -- [RobertaTokenizer](/docs/transformers/v5.6.0/ja/model_doc/bart#transformers.RobertaTokenizer) (BridgeTowerConfig model) - **bros** -- [BertTokenizer](/docs/transformers/v5.6.0/ja/model_doc/bert#transformers.BertTokenizer) (BrosConfig model) - **camembert** -- [CamembertTokenizer](/docs/transformers/v5.6.0/ja/model_doc/camembert#transformers.CamembertTokenizer) (CamembertConfig model) - **canine** -- [CanineTokenizer](/docs/transformers/v5.6.0/ja/model_doc/canine#transformers.CanineTokenizer) (CanineConfig model) - **chameleon** -- [TokenizersBackend](/docs/transformers/v5.6.0/ja/main_classes/tokenizer#transformers.TokenizersBackend) (ChameleonConfig model) - **chinese_clip** -- [BertTokenizer](/docs/transformers/v5.6.0/ja/model_doc/bert#transformers.BertTokenizer) (ChineseCLIPConfig model) - **clap** -- [RobertaTokenizer](/docs/transformers/v5.6.0/ja/model_doc/bart#transformers.RobertaTokenizer) (ClapConfig model) - **clip** -- [CLIPTokenizer](/docs/transformers/v5.6.0/ja/model_doc/clip#transformers.CLIPTokenizer) (CLIPConfig model) - **clipseg** -- [CLIPTokenizer](/docs/transformers/v5.6.0/ja/model_doc/clip#transformers.CLIPTokenizer) (CLIPSegConfig model) - **clvp** -- [ClvpTokenizer](/docs/transformers/v5.6.0/ja/model_doc/clvp#transformers.ClvpTokenizer) (ClvpConfig model) - **codegen** -- `GPT2Tokenizer` (CodeGenConfig model) - **cohere** -- `CohereTokenizer` (CohereConfig model) - **cohere2** -- `CohereTokenizer` (Cohere2Config model) - **cohere_asr** -- [TokenizersBackend](/docs/transformers/v5.6.0/ja/main_classes/tokenizer#transformers.TokenizersBackend) (CohereAsrConfig model) - **colqwen2** -- `Qwen2Tokenizer` (ColQwen2Config model) - **convbert** -- [BertTokenizer](/docs/transformers/v5.6.0/ja/model_doc/bert#transformers.BertTokenizer) (ConvBertConfig model) - **cpmant** -- [CpmAntTokenizer](/docs/transformers/v5.6.0/ja/model_doc/cpmant#transformers.CpmAntTokenizer) (CpmAntConfig model) - **ctrl** -- [CTRLTokenizer](/docs/transformers/v5.6.0/ja/model_doc/ctrl#transformers.CTRLTokenizer) (CTRLConfig model) - **data2vec-audio** -- `Wav2Vec2CTCTokenizer` (Data2VecAudioConfig model) - **data2vec-text** -- [RobertaTokenizer](/docs/transformers/v5.6.0/ja/model_doc/bart#transformers.RobertaTokenizer) (Data2VecTextConfig model) - **dbrx** -- `GPT2Tokenizer` (DbrxConfig model) - **deberta** -- [DebertaTokenizer](/docs/transformers/v5.6.0/ja/model_doc/deberta#transformers.DebertaTokenizer) (DebertaConfig model) - **deberta-v2** -- [DebertaV2Tokenizer](/docs/transformers/v5.6.0/ja/model_doc/deberta-v2#transformers.DebertaV2Tokenizer) (DebertaV2Config model) - **deepseek_v2** -- [TokenizersBackend](/docs/transformers/v5.6.0/ja/main_classes/tokenizer#transformers.TokenizersBackend) (DeepseekV2Config model) - **deepseek_v3** -- [TokenizersBackend](/docs/transformers/v5.6.0/ja/main_classes/tokenizer#transformers.TokenizersBackend) (DeepseekV3Config model) - **deepseek_vl** -- [TokenizersBackend](/docs/transformers/v5.6.0/ja/main_classes/tokenizer#transformers.TokenizersBackend) (DeepseekVLConfig model) - **deepseek_vl_hybrid** -- [TokenizersBackend](/docs/transformers/v5.6.0/ja/main_classes/tokenizer#transformers.TokenizersBackend) (DeepseekVLHybridConfig model) - **dia** -- `DiaTokenizer` (DiaConfig model) - **distilbert** -- [BertTokenizer](/docs/transformers/v5.6.0/ja/model_doc/bert#transformers.BertTokenizer) (DistilBertConfig model) - **dpr** -- `DPRQuestionEncoderTokenizer` (DPRConfig model) - **electra** -- [BertTokenizer](/docs/transformers/v5.6.0/ja/model_doc/bert#transformers.BertTokenizer) (ElectraConfig model) - **emu3** -- `GPT2Tokenizer` (Emu3Config model) - **ernie** -- [BertTokenizer](/docs/transformers/v5.6.0/ja/model_doc/bert#transformers.BertTokenizer) (ErnieConfig model) - **esm** -- `EsmTokenizer` (EsmConfig model) - **falcon_mamba** -- `GPTNeoXTokenizer` (FalconMambaConfig model) - **fastspeech2_conformer** -- `None` (FastSpeech2ConformerConfig model) - **flaubert** -- `FlaubertTokenizer` (FlaubertConfig model) - **flava** -- [BertTokenizer](/docs/transformers/v5.6.0/ja/model_doc/bert#transformers.BertTokenizer) (FlavaConfig model) - **flex_olmo** -- `GPT2Tokenizer` (FlexOlmoConfig model) - **florence2** -- [BartTokenizer](/docs/transformers/v5.6.0/ja/model_doc/bart#transformers.RobertaTokenizer) (Florence2Config model) - **fnet** -- `FNetTokenizer` (FNetConfig model) - **fsmt** -- `FSMTTokenizer` (FSMTConfig model) - **funnel** -- `FunnelTokenizer` (FunnelConfig model) - **fuyu** -- [TokenizersBackend](/docs/transformers/v5.6.0/ja/main_classes/tokenizer#transformers.TokenizersBackend) (FuyuConfig model) - **gemma** -- `GemmaTokenizer` (GemmaConfig model) - **gemma2** -- `GemmaTokenizer` (Gemma2Config model) - **gemma3** -- `GemmaTokenizer` (Gemma3Config model) - **gemma3_text** -- `GemmaTokenizer` (Gemma3TextConfig model) - **gemma3n** -- `GemmaTokenizer` (Gemma3nConfig model) - **gemma3n_text** -- `GemmaTokenizer` (Gemma3nTextConfig model) - **git** -- [BertTokenizer](/docs/transformers/v5.6.0/ja/model_doc/bert#transformers.BertTokenizer) (GitConfig model) - **glm** -- [TokenizersBackend](/docs/transformers/v5.6.0/ja/main_classes/tokenizer#transformers.TokenizersBackend) (GlmConfig model) - **glm4** -- [TokenizersBackend](/docs/transformers/v5.6.0/ja/main_classes/tokenizer#transformers.TokenizersBackend) (Glm4Config model) - **glm4_moe** -- [TokenizersBackend](/docs/transformers/v5.6.0/ja/main_classes/tokenizer#transformers.TokenizersBackend) (Glm4MoeConfig model) - **glm4_moe_lite** -- [TokenizersBackend](/docs/transformers/v5.6.0/ja/main_classes/tokenizer#transformers.TokenizersBackend) (Glm4MoeLiteConfig model) - **glm4v** -- [TokenizersBackend](/docs/transformers/v5.6.0/ja/main_classes/tokenizer#transformers.TokenizersBackend) (Glm4vConfig model) - **glm4v_moe** -- [TokenizersBackend](/docs/transformers/v5.6.0/ja/main_classes/tokenizer#transformers.TokenizersBackend) (Glm4vMoeConfig model) - **glm_image** -- [TokenizersBackend](/docs/transformers/v5.6.0/ja/main_classes/tokenizer#transformers.TokenizersBackend) (GlmImageConfig model) - **glmasr** -- [TokenizersBackend](/docs/transformers/v5.6.0/ja/main_classes/tokenizer#transformers.TokenizersBackend) (GlmAsrConfig model) - **got_ocr2** -- [TokenizersBackend](/docs/transformers/v5.6.0/ja/main_classes/tokenizer#transformers.TokenizersBackend) (GotOcr2Config model) - **gpt-sw3** -- `GPTSw3Tokenizer` (GPT2Config model) - **gpt2** -- `GPT2Tokenizer` (GPT2Config model) - **gpt_bigcode** -- `GPT2Tokenizer` (GPTBigCodeConfig model) - **gpt_neo** -- `GPT2Tokenizer` (GPTNeoConfig model) - **gpt_neox** -- `GPTNeoXTokenizer` (GPTNeoXConfig model) - **gpt_neox_japanese** -- `GPTNeoXJapaneseTokenizer` (GPTNeoXJapaneseConfig model) - **gptj** -- `GPT2Tokenizer` (GPTJConfig model) - **granite** -- `GPT2Tokenizer` (GraniteConfig model) - **granitemoe** -- `GPT2Tokenizer` (GraniteMoeConfig model) - **granitemoehybrid** -- `GPT2Tokenizer` (GraniteMoeHybridConfig model) - **granitemoeshared** -- `GPT2Tokenizer` (GraniteMoeSharedConfig model) - **grounding-dino** -- [BertTokenizer](/docs/transformers/v5.6.0/ja/model_doc/bert#transformers.BertTokenizer) (GroundingDinoConfig model) - **groupvit** -- [CLIPTokenizer](/docs/transformers/v5.6.0/ja/model_doc/clip#transformers.CLIPTokenizer) (GroupViTConfig model) - **hubert** -- `Wav2Vec2CTCTokenizer` (HubertConfig model) - **ibert** -- [RobertaTokenizer](/docs/transformers/v5.6.0/ja/model_doc/bart#transformers.RobertaTokenizer) (IBertConfig model) - **idefics** -- `LlamaTokenizer` (IdeficsConfig model) - **idefics2** -- `LlamaTokenizer` (Idefics2Config model) - **instructblip** -- `GPT2Tokenizer` (InstructBlipConfig model) - **instructblipvideo** -- `GPT2Tokenizer` (InstructBlipVideoConfig model) - **internvl** -- `Qwen2Tokenizer` (InternVLConfig model) - **jais2** -- `GPT2Tokenizer` (Jais2Config model) - **jamba** -- [TokenizersBackend](/docs/transformers/v5.6.0/ja/main_classes/tokenizer#transformers.TokenizersBackend) (JambaConfig model) - **janus** -- [TokenizersBackend](/docs/transformers/v5.6.0/ja/main_classes/tokenizer#transformers.TokenizersBackend) (JanusConfig model) - **jina_embeddings_v3** -- `XLMRobertaTokenizer` (JinaEmbeddingsV3Config model) - **kosmos-2** -- `XLMRobertaTokenizer` (Kosmos2Config model) - **lasr_ctc** -- `LasrTokenizer` (LasrCTCConfig model) - **lasr_encoder** -- `LasrTokenizer` (LasrEncoderConfig model) - **layoutlm** -- [BertTokenizer](/docs/transformers/v5.6.0/ja/model_doc/bert#transformers.BertTokenizer) (LayoutLMConfig model) - **layoutlmv2** -- `LayoutLMv2Tokenizer` (LayoutLMv2Config model) - **layoutlmv3** -- `LayoutLMv3Tokenizer` (LayoutLMv3Config model) - **layoutxlm** -- `LayoutXLMTokenizer` (LayoutXLMConfig model) - **led** -- [LEDTokenizer](/docs/transformers/v5.6.0/ja/model_doc/bart#transformers.RobertaTokenizer) (LEDConfig model) - **lighton_ocr** -- `Qwen2TokenizerFast` (LightOnOcrConfig model) - **lilt** -- [RobertaTokenizer](/docs/transformers/v5.6.0/ja/model_doc/bart#transformers.RobertaTokenizer) (LiltConfig model) - **llava** -- [TokenizersBackend](/docs/transformers/v5.6.0/ja/main_classes/tokenizer#transformers.TokenizersBackend) (LlavaConfig model) - **llava_next** -- [TokenizersBackend](/docs/transformers/v5.6.0/ja/main_classes/tokenizer#transformers.TokenizersBackend) (LlavaNextConfig model) - **longformer** -- [RobertaTokenizer](/docs/transformers/v5.6.0/ja/model_doc/bart#transformers.RobertaTokenizer) (LongformerConfig model) - **luke** -- `LukeTokenizer` (LukeConfig model) - **lxmert** -- [LxmertTokenizer](/docs/transformers/v5.6.0/ja/model_doc/bert#transformers.BertTokenizer) (LxmertConfig model) - **m2m_100** -- `M2M100Tokenizer` (M2M100Config model) - **mamba** -- `GPTNeoXTokenizer` (MambaConfig model) - **mamba2** -- `GPTNeoXTokenizer` (Mamba2Config model) - **marian** -- `MarianTokenizer` (MarianConfig model) - **markuplm** -- `MarkupLMTokenizer` (MarkupLMConfig model) - **mbart** -- `MBartTokenizer` (MBartConfig model) - **megatron-bert** -- [BertTokenizer](/docs/transformers/v5.6.0/ja/model_doc/bert#transformers.BertTokenizer) (MegatronBertConfig model) - **metaclip_2** -- `XLMRobertaTokenizer` (MetaClip2Config model) - **mgp-str** -- `MgpstrTokenizer` (MgpstrConfig model) - **minimax_m2** -- [TokenizersBackend](/docs/transformers/v5.6.0/ja/main_classes/tokenizer#transformers.TokenizersBackend) (MiniMaxM2Config model) - **ministral** -- `MistralCommonBackend` (MinistralConfig model) - **ministral3** -- `MistralCommonBackend` (Ministral3Config model) - **mistral** -- `MistralCommonBackend` (MistralConfig model) - **mistral3** -- `MistralCommonBackend` (Mistral3Config model) - **mixtral** -- `MistralCommonBackend` (MixtralConfig model) - **mm-grounding-dino** -- [BertTokenizer](/docs/transformers/v5.6.0/ja/model_doc/bert#transformers.BertTokenizer) (MMGroundingDinoConfig model) - **mobilebert** -- [MobileBertTokenizer](/docs/transformers/v5.6.0/ja/model_doc/bert#transformers.BertTokenizer) (MobileBertConfig model) - **modernbert** -- [TokenizersBackend](/docs/transformers/v5.6.0/ja/main_classes/tokenizer#transformers.TokenizersBackend) (ModernBertConfig model) - **mpnet** -- `MPNetTokenizer` (MPNetConfig model) - **mpt** -- `GPTNeoXTokenizer` (MptConfig model) - **mra** -- [RobertaTokenizer](/docs/transformers/v5.6.0/ja/model_doc/bart#transformers.RobertaTokenizer) (MraConfig model) - **mt5** -- `T5Tokenizer` (MT5Config model) - **musicgen** -- `T5Tokenizer` (MusicgenConfig model) - **musicgen_melody** -- `T5Tokenizer` (MusicgenMelodyConfig model) - **mvp** -- [MvpTokenizer](/docs/transformers/v5.6.0/ja/model_doc/bart#transformers.RobertaTokenizer) (MvpConfig model) - **nemotron** -- [TokenizersBackend](/docs/transformers/v5.6.0/ja/main_classes/tokenizer#transformers.TokenizersBackend) (NemotronConfig model) - **nllb-moe** -- `NllbTokenizer` (NllbMoeConfig model) - **nomic_bert** -- [BertTokenizer](/docs/transformers/v5.6.0/ja/model_doc/bert#transformers.BertTokenizer) (NomicBertConfig model) - **nougat** -- `NougatTokenizer` (NougatConfig model) - **nystromformer** -- [AlbertTokenizer](/docs/transformers/v5.6.0/ja/model_doc/albert#transformers.AlbertTokenizer) (NystromformerConfig model) - **olmo** -- `GPTNeoXTokenizer` (OlmoConfig model) - **olmo2** -- `GPTNeoXTokenizer` (Olmo2Config model) - **olmo3** -- [TokenizersBackend](/docs/transformers/v5.6.0/ja/main_classes/tokenizer#transformers.TokenizersBackend) (Olmo3Config model) - **olmo_hybrid** -- [TokenizersBackend](/docs/transformers/v5.6.0/ja/main_classes/tokenizer#transformers.TokenizersBackend) (OlmoHybridConfig model) - **olmoe** -- `GPTNeoXTokenizer` (OlmoeConfig model) - **omdet-turbo** -- [CLIPTokenizer](/docs/transformers/v5.6.0/ja/model_doc/clip#transformers.CLIPTokenizer) (OmDetTurboConfig model) - **oneformer** -- [CLIPTokenizer](/docs/transformers/v5.6.0/ja/model_doc/clip#transformers.CLIPTokenizer) (OneFormerConfig model) - **openai-gpt** -- `OpenAIGPTTokenizer` (OpenAIGPTConfig model) - **opt** -- `GPT2Tokenizer` (OPTConfig model) - **ovis2** -- `Qwen2Tokenizer` (Ovis2Config model) - **owlv2** -- [CLIPTokenizer](/docs/transformers/v5.6.0/ja/model_doc/clip#transformers.CLIPTokenizer) (Owlv2Config model) - **owlvit** -- [CLIPTokenizer](/docs/transformers/v5.6.0/ja/model_doc/clip#transformers.CLIPTokenizer) (OwlViTConfig model) - **pegasus** -- `PegasusTokenizer` (PegasusConfig model) - **pegasus_x** -- `PegasusTokenizer` (PegasusXConfig model) - **perceiver** -- `PerceiverTokenizer` (PerceiverConfig model) - **phi** -- `GPT2Tokenizer` (PhiConfig model) - **phi3** -- [TokenizersBackend](/docs/transformers/v5.6.0/ja/main_classes/tokenizer#transformers.TokenizersBackend) (Phi3Config model) - **phimoe** -- [TokenizersBackend](/docs/transformers/v5.6.0/ja/main_classes/tokenizer#transformers.TokenizersBackend) (PhimoeConfig model) - **pix2struct** -- `T5Tokenizer` (Pix2StructConfig model) - **pixtral** -- `MistralCommonBackend` (PixtralVisionConfig model) - **plbart** -- `PLBartTokenizer` (PLBartConfig model) - **prophetnet** -- `ProphetNetTokenizer` (ProphetNetConfig model) - **qianfan_ocr** -- `Qwen2Tokenizer` (QianfanOCRConfig model) - **qwen2** -- `Qwen2Tokenizer` (Qwen2Config model) - **qwen2_5_omni** -- `Qwen2Tokenizer` (Qwen2_5OmniConfig model) - **qwen2_5_vl** -- `Qwen2Tokenizer` (Qwen2_5_VLConfig model) - **qwen2_audio** -- `Qwen2Tokenizer` (Qwen2AudioConfig model) - **qwen2_moe** -- `Qwen2Tokenizer` (Qwen2MoeConfig model) - **qwen2_vl** -- `Qwen2Tokenizer` (Qwen2VLConfig model) - **qwen3** -- `Qwen2Tokenizer` (Qwen3Config model) - **qwen3_5** -- `Qwen3_5Tokenizer` (Qwen3_5Config model) - **qwen3_5_moe** -- `Qwen3_5Tokenizer` (Qwen3_5MoeConfig model) - **qwen3_moe** -- `Qwen2Tokenizer` (Qwen3MoeConfig model) - **qwen3_next** -- `Qwen2Tokenizer` (Qwen3NextConfig model) - **qwen3_omni_moe** -- `Qwen2Tokenizer` (Qwen3OmniMoeConfig model) - **qwen3_vl** -- `Qwen2Tokenizer` (Qwen3VLConfig model) - **qwen3_vl_moe** -- `Qwen2Tokenizer` (Qwen3VLMoeConfig model) - **rag** -- `RagTokenizer` (RagConfig model) - **recurrent_gemma** -- `GemmaTokenizer` (RecurrentGemmaConfig model) - **reformer** -- `ReformerTokenizer` (ReformerConfig model) - **rembert** -- `RemBertTokenizer` (RemBertConfig model) - **roberta** -- [RobertaTokenizer](/docs/transformers/v5.6.0/ja/model_doc/bart#transformers.RobertaTokenizer) (RobertaConfig model) - **roberta-prelayernorm** -- [RobertaTokenizer](/docs/transformers/v5.6.0/ja/model_doc/bart#transformers.RobertaTokenizer) (RobertaPreLayerNormConfig model) - **roc_bert** -- `RoCBertTokenizer` (RoCBertConfig model) - **roformer** -- `RoFormerTokenizer` (RoFormerConfig model) - **rwkv** -- `GPTNeoXTokenizer` (RwkvConfig model) - **sam3** -- [CLIPTokenizer](/docs/transformers/v5.6.0/ja/model_doc/clip#transformers.CLIPTokenizer) (Sam3Config model) - **sam3_video** -- [CLIPTokenizer](/docs/transformers/v5.6.0/ja/model_doc/clip#transformers.CLIPTokenizer) (Sam3VideoConfig model) - **seamless_m4t** -- `SeamlessM4TTokenizer` (SeamlessM4TConfig model) - **seamless_m4t_v2** -- `SeamlessM4TTokenizer` (SeamlessM4Tv2Config model) - **shieldgemma2** -- `GemmaTokenizer` (ShieldGemma2Config model) - **siglip** -- `SiglipTokenizer` (SiglipConfig model) - **siglip2** -- `Siglip2Tokenizer` (Siglip2Config model) - **speech_to_text** -- `Speech2TextTokenizer` (Speech2TextConfig model) - **speecht5** -- `SpeechT5Tokenizer` (SpeechT5Config model) - **splinter** -- `SplinterTokenizer` (SplinterConfig model) - **squeezebert** -- [BertTokenizer](/docs/transformers/v5.6.0/ja/model_doc/bert#transformers.BertTokenizer) (SqueezeBertConfig model) - **stablelm** -- `GPTNeoXTokenizer` (StableLmConfig model) - **starcoder2** -- `GPT2Tokenizer` (Starcoder2Config model) - **switch_transformers** -- `T5Tokenizer` (SwitchTransformersConfig model) - **t5** -- `T5Tokenizer` (T5Config model) - **t5gemma** -- `GemmaTokenizer` (T5GemmaConfig model) - **tapas** -- `TapasTokenizer` (TapasConfig model) - **trocr** -- `XLMRobertaTokenizer` (TrOCRConfig model) - **tvp** -- [BertTokenizer](/docs/transformers/v5.6.0/ja/model_doc/bert#transformers.BertTokenizer) (TvpConfig model) - **udop** -- `UdopTokenizer` (UdopConfig model) - **umt5** -- `T5Tokenizer` (UMT5Config model) - **unispeech** -- `Wav2Vec2CTCTokenizer` (UniSpeechConfig model) - **unispeech-sat** -- `Wav2Vec2CTCTokenizer` (UniSpeechSatConfig model) - **vilt** -- [BertTokenizer](/docs/transformers/v5.6.0/ja/model_doc/bert#transformers.BertTokenizer) (ViltConfig model) - **vipllava** -- [TokenizersBackend](/docs/transformers/v5.6.0/ja/main_classes/tokenizer#transformers.TokenizersBackend) (VipLlavaConfig model) - **visual_bert** -- [BertTokenizer](/docs/transformers/v5.6.0/ja/model_doc/bert#transformers.BertTokenizer) (VisualBertConfig model) - **vits** -- `VitsTokenizer` (VitsConfig model) - **voxtral** -- `MistralCommonBackend` (VoxtralConfig model) - **voxtral_realtime** -- `MistralCommonBackend` (VoxtralRealtimeConfig model) - **wav2vec2** -- `Wav2Vec2CTCTokenizer` (Wav2Vec2Config model) - **wav2vec2-bert** -- `Wav2Vec2CTCTokenizer` (Wav2Vec2BertConfig model) - **wav2vec2-conformer** -- `Wav2Vec2CTCTokenizer` (Wav2Vec2ConformerConfig model) - **whisper** -- `WhisperTokenizer` (WhisperConfig model) - **xclip** -- [CLIPTokenizer](/docs/transformers/v5.6.0/ja/model_doc/clip#transformers.CLIPTokenizer) (XCLIPConfig model) - **xglm** -- `XGLMTokenizer` (XGLMConfig model) - **xlm** -- `XLMTokenizer` (XLMConfig model) - **xlm-roberta** -- `XLMRobertaTokenizer` (XLMRobertaConfig model) - **xlm-roberta-xl** -- `XLMRobertaTokenizer` (XLMRobertaXLConfig model) - **xlnet** -- `XLNetTokenizer` (XLNetConfig model) - **xlstm** -- `GPTNeoXTokenizer` (xLSTMConfig model) - **xmod** -- `XLMRobertaTokenizer` (XmodConfig model) - **yoso** -- [AlbertTokenizer](/docs/transformers/v5.6.0/ja/model_doc/albert#transformers.AlbertTokenizer) (YosoConfig model) Examples: ```python >>> from transformers import AutoTokenizer >>> # Download vocabulary from huggingface.co and cache. >>> tokenizer = AutoTokenizer.from_pretrained("google-bert/bert-base-uncased") >>> # Download vocabulary from huggingface.co (user-uploaded) and cache. >>> tokenizer = AutoTokenizer.from_pretrained("dbmdz/bert-base-german-cased") >>> # If vocabulary files are in a directory (e.g. tokenizer was saved using *save_pretrained('./test/saved_model/')*) >>> # tokenizer = AutoTokenizer.from_pretrained("./test/bert_saved_model/") >>> # Download vocabulary from huggingface.co and define model-specific arguments >>> tokenizer = AutoTokenizer.from_pretrained("FacebookAI/roberta-base", add_prefix_space=True) >>> # Explicitly use the tokenizers backend >>> tokenizer = AutoTokenizer.from_pretrained("hf-internal-testing/llama-tokenizer", backend="tokenizers") >>> # Explicitly use the sentencepiece backend >>> tokenizer = AutoTokenizer.from_pretrained("hf-internal-testing/llama-tokenizer", backend="sentencepiece") ``` **Parameters:** pretrained_model_name_or_path (`str` or `os.PathLike`) : Can be either: - A string, the *model id* of a predefined tokenizer hosted inside a model repo on huggingface.co. - A path to a *directory* containing vocabulary files required by the tokenizer, for instance saved using the [save_pretrained()](/docs/transformers/v5.6.0/ja/internal/tokenization_utils#transformers.PreTrainedTokenizerBase.save_pretrained) method, e.g., `./my_model_directory/`. - a path to a single saved vocabulary file if and only if the tokenizer only requires a single vocabulary file (like Bert or XLNet), e.g.: `./my_model_directory/vocab.txt`. (Not applicable to all derived classes) inputs (additional positional arguments, *optional*) : Will be passed along to the Tokenizer `__init__()` method. config ([PreTrainedConfig](/docs/transformers/v5.6.0/ja/main_classes/configuration#transformers.PreTrainedConfig), *optional*) : The configuration object used to determine the tokenizer class to instantiate. cache_dir (`str` or `os.PathLike`, *optional*) : Path to a directory in which a downloaded pretrained model configuration should be cached if the standard cache should not be used. force_download (`bool`, *optional*, defaults to `False`) : Whether or not to force the (re-)download the model weights and configuration files and override the cached versions if they exist. proxies (`dict[str, str]`, *optional*) : A dictionary of proxy servers to use by protocol or endpoint, e.g., `{'http': 'foo.bar:3128', 'http://hostname': 'foo.bar:4012'}`. The proxies are used on each request. revision (`str`, *optional*, defaults to `"main"`) : The specific model version to use. It can be a branch name, a tag name, or a commit id, since we use a git-based system for storing models and other artifacts on huggingface.co, so `revision` can be any identifier allowed by git. subfolder (`str`, *optional*) : In case the relevant files are located inside a subfolder of the model repo on huggingface.co (e.g. for facebook/rag-token-base), specify it here. tokenizer_type (`str`, *optional*) : Tokenizer type to be loaded. backend (`str`, *optional*, defaults to `"tokenizers"`) : Backend to use for tokenization. Valid options are: - `"tokenizers"`: Use the HuggingFace tokenizers library backend (default) - `"sentencepiece"`: Use the SentencePiece backend trust_remote_code (`bool`, *optional*, defaults to `False`) : Whether or not to allow for custom models defined on the Hub in their own modeling files. This option should only be set to `True` for repositories you trust and in which you have read the code, as it will execute code present on the Hub on your local machine. kwargs (additional keyword arguments, *optional*) : Will be passed to the Tokenizer `__init__()` method. Can be used to set special tokens like `bos_token`, `eos_token`, `unk_token`, `sep_token`, `pad_token`, `cls_token`, `mask_token`, `additional_special_tokens`. See parameters in the `__init__()` for more details. #### register[[transformers.AutoTokenizer.register]] [Source](https://github.com/huggingface/transformers/blob/v5.6.0/src/transformers/models/auto/tokenization_auto.py#L844) Register a new tokenizer in this mapping. **Parameters:** config_class ([PreTrainedConfig](/docs/transformers/v5.6.0/ja/main_classes/configuration#transformers.PreTrainedConfig)) : The configuration corresponding to the model to register. tokenizer_class : The tokenizer class to register (V5 - preferred parameter). slow_tokenizer_class : (Deprecated) The slow tokenizer to register. fast_tokenizer_class : (Deprecated) The fast tokenizer to register. ## AutoFeatureExtractor[[transformers.AutoFeatureExtractor]] #### transformers.AutoFeatureExtractor[[transformers.AutoFeatureExtractor]] [Source](https://github.com/huggingface/transformers/blob/v5.6.0/src/transformers/models/auto/feature_extraction_auto.py#L231) This is a generic feature extractor class that will be instantiated as one of the feature extractor classes of the library when created with the [AutoFeatureExtractor.from_pretrained()](/docs/transformers/v5.6.0/ja/model_doc/auto#transformers.AutoFeatureExtractor.from_pretrained) class method. This class cannot be instantiated directly using `__init__()` (throws an error). from_pretrainedtransformers.AutoFeatureExtractor.from_pretrainedhttps://github.com/huggingface/transformers/blob/v5.6.0/src/transformers/models/auto/feature_extraction_auto.py#L245[{"name": "pretrained_model_name_or_path", "val": ""}, {"name": "**kwargs", "val": ""}]- **pretrained_model_name_or_path** (`str` or `os.PathLike`) -- This can be either: - a string, the *model id* of a pretrained feature_extractor hosted inside a model repo on huggingface.co. - a path to a *directory* containing a feature extractor file saved using the [save_pretrained()](/docs/transformers/v5.6.0/ja/main_classes/feature_extractor#transformers.FeatureExtractionMixin.save_pretrained) method, e.g., `./my_model_directory/`. - a path to a saved feature extractor JSON *file*, e.g., `./my_model_directory/preprocessor_config.json`. - **cache_dir** (`str` or `os.PathLike`, *optional*) -- Path to a directory in which a downloaded pretrained model feature extractor should be cached if the standard cache should not be used. - **force_download** (`bool`, *optional*, defaults to `False`) -- Whether or not to force to (re-)download the feature extractor files and override the cached versions if they exist. - **proxies** (`dict[str, str]`, *optional*) -- A dictionary of proxy servers to use by protocol or endpoint, e.g., `{'http': 'foo.bar:3128', 'http://hostname': 'foo.bar:4012'}.` The proxies are used on each request. - **token** (`str` or *bool*, *optional*) -- The token to use as HTTP bearer authorization for remote files. If `True`, will use the token generated when running `hf auth login` (stored in `~/.huggingface`). - **revision** (`str`, *optional*, defaults to `"main"`) -- The specific model version to use. It can be a branch name, a tag name, or a commit id, since we use a git-based system for storing models and other artifacts on huggingface.co, so `revision` can be any identifier allowed by git. - **return_unused_kwargs** (`bool`, *optional*, defaults to `False`) -- If `False`, then this function returns just the final feature extractor object. If `True`, then this functions returns a `Tuple(feature_extractor, unused_kwargs)` where *unused_kwargs* is a dictionary consisting of the key/value pairs whose keys are not feature extractor attributes: i.e., the part of `kwargs` which has not been used to update `feature_extractor` and is otherwise ignored. - **trust_remote_code** (`bool`, *optional*, defaults to `False`) -- Whether or not to allow for custom models defined on the Hub in their own modeling files. This option should only be set to `True` for repositories you trust and in which you have read the code, as it will execute code present on the Hub on your local machine. - **kwargs** (`dict[str, Any]`, *optional*) -- The values in kwargs of any keys which are feature extractor attributes will be used to override the loaded values. Behavior concerning key/value pairs whose keys are *not* feature extractor attributes is controlled by the `return_unused_kwargs` keyword parameter.0 Instantiate one of the feature extractor classes of the library from a pretrained model vocabulary. The feature extractor class to instantiate is selected based on the `model_type` property of the config object (either passed as an argument or loaded from `pretrained_model_name_or_path` if possible), or when it's missing, by falling back to using pattern matching on `pretrained_model_name_or_path`: - **audio-spectrogram-transformer** -- [ASTFeatureExtractor](/docs/transformers/v5.6.0/ja/model_doc/audio-spectrogram-transformer#transformers.ASTFeatureExtractor) (ASTConfig model) - **audioflamingo3** -- `WhisperFeatureExtractor` (AudioFlamingo3Config model) - **clap** -- [ClapFeatureExtractor](/docs/transformers/v5.6.0/ja/model_doc/clap#transformers.ClapFeatureExtractor) (ClapConfig model) - **clvp** -- [ClvpFeatureExtractor](/docs/transformers/v5.6.0/ja/model_doc/clvp#transformers.ClvpFeatureExtractor) (ClvpConfig model) - **cohere_asr** -- `CohereAsrFeatureExtractor` (CohereAsrConfig model) - **csm** -- `EncodecFeatureExtractor` (CsmConfig model) - **dac** -- `DacFeatureExtractor` (DacConfig model) - **data2vec-audio** -- `Wav2Vec2FeatureExtractor` (Data2VecAudioConfig model) - **dia** -- `DiaFeatureExtractor` (DiaConfig model) - **encodec** -- `EncodecFeatureExtractor` (EncodecConfig model) - **gemma3n** -- `Gemma3nAudioFeatureExtractor` (Gemma3nConfig model) - **gemma4** -- `Gemma4AudioFeatureExtractor` (Gemma4Config model) - **glmasr** -- `WhisperFeatureExtractor` (GlmAsrConfig model) - **granite_speech** -- `GraniteSpeechFeatureExtractor` (GraniteSpeechConfig model) - **higgs_audio_v2_tokenizer** -- `DacFeatureExtractor` (HiggsAudioV2TokenizerConfig model) - **hubert** -- `Wav2Vec2FeatureExtractor` (HubertConfig model) - **kyutai_speech_to_text** -- `KyutaiSpeechToTextFeatureExtractor` (KyutaiSpeechToTextConfig model) - **lasr_ctc** -- `LasrFeatureExtractor` (LasrCTCConfig model) - **lasr_encoder** -- `LasrFeatureExtractor` (LasrEncoderConfig model) - **markuplm** -- `MarkupLMFeatureExtractor` (MarkupLMConfig model) - **mimi** -- `EncodecFeatureExtractor` (MimiConfig model) - **moonshine** -- `Wav2Vec2FeatureExtractor` (MoonshineConfig model) - **moshi** -- `EncodecFeatureExtractor` (MoshiConfig model) - **musicgen** -- `EncodecFeatureExtractor` (MusicgenConfig model) - **musicgen_melody** -- `MusicgenMelodyFeatureExtractor` (MusicgenMelodyConfig model) - **parakeet_ctc** -- `ParakeetFeatureExtractor` (ParakeetCTCConfig model) - **parakeet_encoder** -- `ParakeetFeatureExtractor` (ParakeetEncoderConfig model) - **pe_audio** -- `PeAudioFeatureExtractor` (PeAudioConfig model) - **pe_audio_video** -- `PeAudioFeatureExtractor` (PeAudioVideoConfig model) - **phi4_multimodal** -- `Phi4MultimodalFeatureExtractor` (Phi4MultimodalConfig model) - **pop2piano** -- `Pop2PianoFeatureExtractor` (Pop2PianoConfig model) - **qwen2_5_omni** -- `WhisperFeatureExtractor` (Qwen2_5OmniConfig model) - **qwen2_audio** -- `WhisperFeatureExtractor` (Qwen2AudioConfig model) - **qwen3_omni_moe** -- `WhisperFeatureExtractor` (Qwen3OmniMoeConfig model) - **seamless_m4t** -- `SeamlessM4TFeatureExtractor` (SeamlessM4TConfig model) - **seamless_m4t_v2** -- `SeamlessM4TFeatureExtractor` (SeamlessM4Tv2Config model) - **sew** -- `Wav2Vec2FeatureExtractor` (SEWConfig model) - **sew-d** -- `Wav2Vec2FeatureExtractor` (SEWDConfig model) - **speech_to_text** -- `Speech2TextFeatureExtractor` (Speech2TextConfig model) - **speecht5** -- `SpeechT5FeatureExtractor` (SpeechT5Config model) - **unispeech** -- `Wav2Vec2FeatureExtractor` (UniSpeechConfig model) - **unispeech-sat** -- `Wav2Vec2FeatureExtractor` (UniSpeechSatConfig model) - **univnet** -- `UnivNetFeatureExtractor` (UnivNetConfig model) - **vibevoice_acoustic_tokenizer** -- `VibeVoiceAcousticTokenizerFeatureExtractor` (VibeVoiceAcousticTokenizerConfig model) - **vibevoice_asr** -- `VibeVoiceAcousticTokenizerFeatureExtractor` (VibeVoiceAsrConfig model) - **voxtral** -- `WhisperFeatureExtractor` (VoxtralConfig model) - **voxtral_realtime** -- `VoxtralRealtimeFeatureExtractor` (VoxtralRealtimeConfig model) - **wav2vec2** -- `Wav2Vec2FeatureExtractor` (Wav2Vec2Config model) - **wav2vec2-bert** -- `Wav2Vec2FeatureExtractor` (Wav2Vec2BertConfig model) - **wav2vec2-conformer** -- `Wav2Vec2FeatureExtractor` (Wav2Vec2ConformerConfig model) - **wavlm** -- `Wav2Vec2FeatureExtractor` (WavLMConfig model) - **whisper** -- `WhisperFeatureExtractor` (WhisperConfig model) - **xcodec** -- `DacFeatureExtractor` (XcodecConfig model) Passing `token=True` is required when you want to use a private model. Examples: ```python >>> from transformers import AutoFeatureExtractor >>> # Download feature extractor from huggingface.co and cache. >>> feature_extractor = AutoFeatureExtractor.from_pretrained("facebook/wav2vec2-base-960h") >>> # If feature extractor files are in a directory (e.g. feature extractor was saved using *save_pretrained('./test/saved_model/')*) >>> # feature_extractor = AutoFeatureExtractor.from_pretrained("./test/saved_model/") ``` **Parameters:** pretrained_model_name_or_path (`str` or `os.PathLike`) : This can be either: - a string, the *model id* of a pretrained feature_extractor hosted inside a model repo on huggingface.co. - a path to a *directory* containing a feature extractor file saved using the [save_pretrained()](/docs/transformers/v5.6.0/ja/main_classes/feature_extractor#transformers.FeatureExtractionMixin.save_pretrained) method, e.g., `./my_model_directory/`. - a path to a saved feature extractor JSON *file*, e.g., `./my_model_directory/preprocessor_config.json`. cache_dir (`str` or `os.PathLike`, *optional*) : Path to a directory in which a downloaded pretrained model feature extractor should be cached if the standard cache should not be used. force_download (`bool`, *optional*, defaults to `False`) : Whether or not to force to (re-)download the feature extractor files and override the cached versions if they exist. proxies (`dict[str, str]`, *optional*) : A dictionary of proxy servers to use by protocol or endpoint, e.g., `{'http': 'foo.bar:3128', 'http://hostname': 'foo.bar:4012'}.` The proxies are used on each request. token (`str` or *bool*, *optional*) : The token to use as HTTP bearer authorization for remote files. If `True`, will use the token generated when running `hf auth login` (stored in `~/.huggingface`). revision (`str`, *optional*, defaults to `"main"`) : The specific model version to use. It can be a branch name, a tag name, or a commit id, since we use a git-based system for storing models and other artifacts on huggingface.co, so `revision` can be any identifier allowed by git. return_unused_kwargs (`bool`, *optional*, defaults to `False`) : If `False`, then this function returns just the final feature extractor object. If `True`, then this functions returns a `Tuple(feature_extractor, unused_kwargs)` where *unused_kwargs* is a dictionary consisting of the key/value pairs whose keys are not feature extractor attributes: i.e., the part of `kwargs` which has not been used to update `feature_extractor` and is otherwise ignored. trust_remote_code (`bool`, *optional*, defaults to `False`) : Whether or not to allow for custom models defined on the Hub in their own modeling files. This option should only be set to `True` for repositories you trust and in which you have read the code, as it will execute code present on the Hub on your local machine. kwargs (`dict[str, Any]`, *optional*) : The values in kwargs of any keys which are feature extractor attributes will be used to override the loaded values. Behavior concerning key/value pairs whose keys are *not* feature extractor attributes is controlled by the `return_unused_kwargs` keyword parameter. #### register[[transformers.AutoFeatureExtractor.register]] [Source](https://github.com/huggingface/transformers/blob/v5.6.0/src/transformers/models/auto/feature_extraction_auto.py#L373) Register a new feature extractor for this class. **Parameters:** config_class ([PreTrainedConfig](/docs/transformers/v5.6.0/ja/main_classes/configuration#transformers.PreTrainedConfig)) : The configuration corresponding to the model to register. feature_extractor_class (`FeatureExtractorMixin`) : The feature extractor to register. ## AutoImageProcessor[[transformers.AutoImageProcessor]] #### transformers.AutoImageProcessor[[transformers.AutoImageProcessor]] [Source](https://github.com/huggingface/transformers/blob/v5.6.0/src/transformers/models/auto/image_processing_auto.py#L442) This is a generic image processor class that will be instantiated as one of the image processor classes of the library when created with the [AutoImageProcessor.from_pretrained()](/docs/transformers/v5.6.0/ja/model_doc/auto#transformers.AutoImageProcessor.from_pretrained) class method. This class cannot be instantiated directly using `__init__()` (throws an error). from_pretrainedtransformers.AutoImageProcessor.from_pretrainedhttps://github.com/huggingface/transformers/blob/v5.6.0/src/transformers/models/auto/image_processing_auto.py#L456[{"name": "pretrained_model_name_or_path", "val": ""}, {"name": "*inputs", "val": ""}, {"name": "**kwargs", "val": ""}]- **pretrained_model_name_or_path** (`str` or `os.PathLike`) -- This can be either: - a string, the *model id* of a pretrained image_processor hosted inside a model repo on huggingface.co. - a path to a *directory* containing a image processor file saved using the [save_pretrained()](/docs/transformers/v5.6.0/ja/main_classes/image_processor#transformers.ImageProcessingMixin.save_pretrained) method, e.g., `./my_model_directory/`. - a path to a saved image processor JSON *file*, e.g., `./my_model_directory/preprocessor_config.json`. - **cache_dir** (`str` or `os.PathLike`, *optional*) -- Path to a directory in which a downloaded pretrained model image processor should be cached if the standard cache should not be used. - **force_download** (`bool`, *optional*, defaults to `False`) -- Whether or not to force to (re-)download the image processor files and override the cached versions if they exist. - **proxies** (`dict[str, str]`, *optional*) -- A dictionary of proxy servers to use by protocol or endpoint, e.g., `{'http': 'foo.bar:3128', 'http://hostname': 'foo.bar:4012'}.` The proxies are used on each request. - **token** (`str` or *bool*, *optional*) -- The token to use as HTTP bearer authorization for remote files. If `True`, will use the token generated when running `hf auth login` (stored in `~/.huggingface`). - **revision** (`str`, *optional*, defaults to `"main"`) -- The specific model version to use. It can be a branch name, a tag name, or a commit id, since we use a git-based system for storing models and other artifacts on huggingface.co, so `revision` can be any identifier allowed by git. - **use_fast** (`bool`, *optional*, defaults to `False`) -- **Deprecated**: Use `backend="torchvision"` instead. This parameter is kept for backward compatibility. Use a fast torchvision-based image processor if it is supported for a given model. If a fast image processor is not available for a given model, a normal numpy-based image processor is returned instead. - **backend** (`str`, *optional*, defaults to `None`) -- The backend to use for image processing. Can be: - `None`: Automatically select the best available backend (torchvision if available, otherwise pil) - `"torchvision"`: Use Torchvision backend (GPU-accelerated, faster) - `"pil"`: Use PIL backend (portable, CPU-only) - Any custom backend name registered via `register()` method - **return_unused_kwargs** (`bool`, *optional*, defaults to `False`) -- If `False`, then this function returns just the final image processor object. If `True`, then this functions returns a `Tuple(image_processor, unused_kwargs)` where *unused_kwargs* is a dictionary consisting of the key/value pairs whose keys are not image processor attributes: i.e., the part of `kwargs` which has not been used to update `image_processor` and is otherwise ignored. - **trust_remote_code** (`bool`, *optional*, defaults to `False`) -- Whether or not to allow for custom models defined on the Hub in their own modeling files. This option should only be set to `True` for repositories you trust and in which you have read the code, as it will execute code present on the Hub on your local machine. - **image_processor_filename** (`str`, *optional*, defaults to `"config.json"`) -- The name of the file in the model directory to use for the image processor config. - **kwargs** (`dict[str, Any]`, *optional*) -- The values in kwargs of any keys which are image processor attributes will be used to override the loaded values. Behavior concerning key/value pairs whose keys are *not* image processor attributes is controlled by the `return_unused_kwargs` keyword parameter.0 Instantiate one of the image processor classes of the library from a pretrained model vocabulary. The image processor class to instantiate is selected based on the `model_type` property of the config object (either passed as an argument or loaded from `pretrained_model_name_or_path` if possible), or when it's missing, by falling back to using pattern matching on `pretrained_model_name_or_path`: - **aimv2** -- `{'torchvision': 'CLIPImageProcessor', 'pil': 'CLIPImageProcessorPil'}` (Aimv2Config model) - **aimv2_vision_model** -- `{'torchvision': 'CLIPImageProcessor', 'pil': 'CLIPImageProcessorPil'}` (Aimv2VisionConfig model) - **align** -- `{'torchvision': 'EfficientNetImageProcessor', 'pil': 'EfficientNetImageProcessorPil'}` (AlignConfig model) - **altclip** -- `{'torchvision': 'CLIPImageProcessor', 'pil': 'CLIPImageProcessorPil'}` (AltCLIPConfig model) - **aria** -- `{'pil': 'AriaImageProcessorPil', 'torchvision': 'AriaImageProcessor'}` (AriaConfig model) - **aya_vision** -- `{'torchvision': 'GotOcr2ImageProcessor', 'pil': 'GotOcr2ImageProcessorPil'}` (AyaVisionConfig model) - **beit** -- `{'pil': 'BeitImageProcessorPil', 'torchvision': 'BeitImageProcessor'}` (BeitConfig model) - **bit** -- `{'pil': 'BitImageProcessorPil', 'torchvision': 'BitImageProcessor'}` (BitConfig model) - **blip** -- `{'pil': 'BlipImageProcessorPil', 'torchvision': 'BlipImageProcessor'}` (BlipConfig model) - **blip-2** -- `{'torchvision': 'BlipImageProcessor', 'pil': 'BlipImageProcessorPil'}` (Blip2Config model) - **bridgetower** -- `{'pil': 'BridgeTowerImageProcessorPil', 'torchvision': 'BridgeTowerImageProcessor'}` (BridgeTowerConfig model) - **chameleon** -- `{'pil': 'ChameleonImageProcessorPil', 'torchvision': 'ChameleonImageProcessor'}` (ChameleonConfig model) - **chinese_clip** -- `{'pil': 'ChineseCLIPImageProcessorPil', 'torchvision': 'ChineseCLIPImageProcessor'}` (ChineseCLIPConfig model) - **chmv2** -- `{'torchvision': 'CHMv2ImageProcessor'}` (CHMv2Config model) - **clip** -- `{'pil': 'CLIPImageProcessorPil', 'torchvision': 'CLIPImageProcessor'}` (CLIPConfig model) - **clipseg** -- `{'torchvision': 'ViTImageProcessor', 'pil': 'ViTImageProcessorPil'}` (CLIPSegConfig model) - **cohere2_vision** -- `{'torchvision': 'Cohere2VisionImageProcessor'}` (Cohere2VisionConfig model) - **colpali** -- `{'torchvision': 'SiglipImageProcessor', 'pil': 'SiglipImageProcessorPil'}` (ColPaliConfig model) - **colqwen2** -- `{'torchvision': 'Qwen2VLImageProcessor', 'pil': 'Qwen2VLImageProcessorPil'}` (ColQwen2Config model) - **conditional_detr** -- `{'pil': 'ConditionalDetrImageProcessorPil', 'torchvision': 'ConditionalDetrImageProcessor'}` (ConditionalDetrConfig model) - **convnext** -- `{'pil': 'ConvNextImageProcessorPil', 'torchvision': 'ConvNextImageProcessor'}` (ConvNextConfig model) - **convnextv2** -- `{'torchvision': 'ConvNextImageProcessor', 'pil': 'ConvNextImageProcessorPil'}` (ConvNextV2Config model) - **cvt** -- `{'torchvision': 'ConvNextImageProcessor', 'pil': 'ConvNextImageProcessorPil'}` (CvtConfig model) - **data2vec-vision** -- `{'torchvision': 'BeitImageProcessor', 'pil': 'BeitImageProcessorPil'}` (Data2VecVisionConfig model) - **deepseek_vl** -- `{'pil': 'DeepseekVLImageProcessorPil', 'torchvision': 'DeepseekVLImageProcessor'}` (DeepseekVLConfig model) - **deepseek_vl_hybrid** -- `{'pil': 'DeepseekVLHybridImageProcessorPil', 'torchvision': 'DeepseekVLHybridImageProcessor'}` (DeepseekVLHybridConfig model) - **deformable_detr** -- `{'pil': 'DeformableDetrImageProcessorPil', 'torchvision': 'DeformableDetrImageProcessor'}` (DeformableDetrConfig model) - **deit** -- `{'pil': 'DeiTImageProcessorPil', 'torchvision': 'DeiTImageProcessor'}` (DeiTConfig model) - **depth_anything** -- `{'torchvision': 'DPTImageProcessor', 'pil': 'DPTImageProcessorPil'}` (DepthAnythingConfig model) - **depth_pro** -- `{'torchvision': 'DepthProImageProcessor'}` (DepthProConfig model) - **detr** -- `{'pil': 'DetrImageProcessorPil', 'torchvision': 'DetrImageProcessor'}` (DetrConfig model) - **dinat** -- `{'torchvision': 'ViTImageProcessor', 'pil': 'ViTImageProcessorPil'}` (DinatConfig model) - **dinov2** -- `{'torchvision': 'BitImageProcessor', 'pil': 'BitImageProcessorPil'}` (Dinov2Config model) - **dinov3_vit** -- `{'torchvision': 'DINOv3ViTImageProcessor'}` (DINOv3ViTConfig model) - **donut-swin** -- `{'torchvision': 'DonutImageProcessor', 'pil': 'DonutImageProcessorPil'}` (DonutSwinConfig model) - **dpt** -- `{'pil': 'DPTImageProcessorPil', 'torchvision': 'DPTImageProcessor'}` (DPTConfig model) - **edgetam** -- `{'torchvision': 'Sam2ImageProcessor'}` (EdgeTamConfig model) - **efficientloftr** -- `{'pil': 'EfficientLoFTRImageProcessorPil', 'torchvision': 'EfficientLoFTRImageProcessor'}` (EfficientLoFTRConfig model) - **efficientnet** -- `{'pil': 'EfficientNetImageProcessorPil', 'torchvision': 'EfficientNetImageProcessor'}` (EfficientNetConfig model) - **emu3** -- `{'pil': 'Emu3ImageProcessor'}` (Emu3Config model) - **eomt** -- `{'pil': 'EomtImageProcessorPil', 'torchvision': 'EomtImageProcessor'}` (EomtConfig model) - **eomt_dinov3** -- `{'torchvision': 'EomtImageProcessor', 'pil': 'EomtImageProcessorPil'}` (EomtDinov3Config model) - **ernie4_5_vl_moe** -- `{'pil': 'Ernie4_5_VLMoeImageProcessorPil', 'torchvision': 'Ernie4_5_VLMoeImageProcessor'}` (Ernie4_5_VLMoeConfig model) - **flava** -- `{'pil': 'FlavaImageProcessorPil', 'torchvision': 'FlavaImageProcessor'}` (FlavaConfig model) - **florence2** -- `{'torchvision': 'CLIPImageProcessor', 'pil': 'CLIPImageProcessorPil'}` (Florence2Config model) - **focalnet** -- `{'torchvision': 'BitImageProcessor', 'pil': 'BitImageProcessorPil'}` (FocalNetConfig model) - **fuyu** -- `{'pil': 'FuyuImageProcessorPil', 'torchvision': 'FuyuImageProcessor'}` (FuyuConfig model) - **gemma3** -- `{'pil': 'Gemma3ImageProcessorPil', 'torchvision': 'Gemma3ImageProcessor'}` (Gemma3Config model) - **gemma3n** -- `{'torchvision': 'SiglipImageProcessor', 'pil': 'SiglipImageProcessorPil'}` (Gemma3nConfig model) - **gemma4** -- `{'pil': 'Gemma4ImageProcessorPil', 'torchvision': 'Gemma4ImageProcessor'}` (Gemma4Config model) - **git** -- `{'torchvision': 'CLIPImageProcessor', 'pil': 'CLIPImageProcessorPil'}` (GitConfig model) - **glm46v** -- `{'pil': 'Glm46VImageProcessorPil', 'torchvision': 'Glm46VImageProcessor'}` (Glm46VConfig model) - **glm4v** -- `{'pil': 'Glm4vImageProcessorPil', 'torchvision': 'Glm4vImageProcessor'}` (Glm4vConfig model) - **glm_image** -- `{'pil': 'GlmImageImageProcessorPil', 'torchvision': 'GlmImageImageProcessor'}` (GlmImageConfig model) - **glpn** -- `{'pil': 'GLPNImageProcessorPil', 'torchvision': 'GLPNImageProcessor'}` (GLPNConfig model) - **got_ocr2** -- `{'pil': 'GotOcr2ImageProcessorPil', 'torchvision': 'GotOcr2ImageProcessor'}` (GotOcr2Config model) - **grounding-dino** -- `{'pil': 'GroundingDinoImageProcessorPil', 'torchvision': 'GroundingDinoImageProcessor'}` (GroundingDinoConfig model) - **groupvit** -- `{'torchvision': 'CLIPImageProcessor', 'pil': 'CLIPImageProcessorPil'}` (GroupViTConfig model) - **hiera** -- `{'torchvision': 'BitImageProcessor', 'pil': 'BitImageProcessorPil'}` (HieraConfig model) - **idefics** -- `{'pil': 'IdeficsImageProcessorPil', 'torchvision': 'IdeficsImageProcessor'}` (IdeficsConfig model) - **idefics2** -- `{'pil': 'Idefics2ImageProcessorPil', 'torchvision': 'Idefics2ImageProcessor'}` (Idefics2Config model) - **idefics3** -- `{'pil': 'Idefics3ImageProcessorPil', 'torchvision': 'Idefics3ImageProcessor'}` (Idefics3Config model) - **ijepa** -- `{'torchvision': 'ViTImageProcessor', 'pil': 'ViTImageProcessorPil'}` (IJepaConfig model) - **imagegpt** -- `{'pil': 'ImageGPTImageProcessorPil', 'torchvision': 'ImageGPTImageProcessor'}` (ImageGPTConfig model) - **instructblip** -- `{'torchvision': 'BlipImageProcessor', 'pil': 'BlipImageProcessorPil'}` (InstructBlipConfig model) - **internvl** -- `{'torchvision': 'GotOcr2ImageProcessor', 'pil': 'GotOcr2ImageProcessorPil'}` (InternVLConfig model) - **janus** -- `{'pil': 'JanusImageProcessorPil', 'torchvision': 'JanusImageProcessor'}` (JanusConfig model) - **kosmos-2** -- `{'torchvision': 'CLIPImageProcessor', 'pil': 'CLIPImageProcessorPil'}` (Kosmos2Config model) - **kosmos-2.5** -- `{'torchvision': 'Kosmos2_5ImageProcessor', 'pil': 'Kosmos2_5ImageProcessorPil'}` (Kosmos2_5Config model) - **layoutlmv2** -- `{'pil': 'LayoutLMv2ImageProcessorPil', 'torchvision': 'LayoutLMv2ImageProcessor'}` (LayoutLMv2Config model) - **layoutlmv3** -- `{'pil': 'LayoutLMv3ImageProcessorPil', 'torchvision': 'LayoutLMv3ImageProcessor'}` (LayoutLMv3Config model) - **layoutxlm** -- `{'torchvision': 'LayoutLMv2ImageProcessor', 'pil': 'LayoutLMv2ImageProcessorPil'}` (LayoutXLMConfig model) - **levit** -- `{'pil': 'LevitImageProcessorPil', 'torchvision': 'LevitImageProcessor'}` (LevitConfig model) - **lfm2_vl** -- `{'torchvision': 'Lfm2VlImageProcessor'}` (Lfm2VlConfig model) - **lightglue** -- `{'pil': 'LightGlueImageProcessorPil', 'torchvision': 'LightGlueImageProcessor'}` (LightGlueConfig model) - **lighton_ocr** -- `{'torchvision': 'PixtralImageProcessor', 'pil': 'PixtralImageProcessorPil'}` (LightOnOcrConfig model) - **llama4** -- `{'torchvision': 'Llama4ImageProcessor'}` (Llama4Config model) - **llava** -- `{'pil': 'LlavaImageProcessorPil', 'torchvision': 'LlavaImageProcessor'}` (LlavaConfig model) - **llava_next** -- `{'pil': 'LlavaNextImageProcessorPil', 'torchvision': 'LlavaNextImageProcessor'}` (LlavaNextConfig model) - **llava_next_video** -- `{'torchvision': 'LlavaNextImageProcessor', 'pil': 'LlavaNextImageProcessorPil'}` (LlavaNextVideoConfig model) - **llava_onevision** -- `{'pil': 'LlavaOnevisionImageProcessorPil', 'torchvision': 'LlavaOnevisionImageProcessor'}` (LlavaOnevisionConfig model) - **lw_detr** -- `{'torchvision': 'DeformableDetrImageProcessor', 'pil': 'DeformableDetrImageProcessorPil'}` (LwDetrConfig model) - **mask2former** -- `{'pil': 'Mask2FormerImageProcessorPil', 'torchvision': 'Mask2FormerImageProcessor'}` (Mask2FormerConfig model) - **maskformer** -- `{'pil': 'MaskFormerImageProcessorPil', 'torchvision': 'MaskFormerImageProcessor'}` (MaskFormerConfig model) - **metaclip_2** -- `{'torchvision': 'CLIPImageProcessor', 'pil': 'CLIPImageProcessorPil'}` (MetaClip2Config model) - **mgp-str** -- `{'torchvision': 'ViTImageProcessor', 'pil': 'ViTImageProcessorPil'}` (MgpstrConfig model) - **mistral3** -- `{'torchvision': 'PixtralImageProcessor', 'pil': 'PixtralImageProcessorPil'}` (Mistral3Config model) - **mlcd** -- `{'torchvision': 'CLIPImageProcessor', 'pil': 'CLIPImageProcessorPil'}` (MLCDVisionConfig model) - **mllama** -- `{'pil': 'MllamaImageProcessorPil', 'torchvision': 'MllamaImageProcessor'}` (MllamaConfig model) - **mm-grounding-dino** -- `{'torchvision': 'GroundingDinoImageProcessor', 'pil': 'GroundingDinoImageProcessorPil'}` (MMGroundingDinoConfig model) - **mobilenet_v1** -- `{'pil': 'MobileNetV1ImageProcessorPil', 'torchvision': 'MobileNetV1ImageProcessor'}` (MobileNetV1Config model) - **mobilenet_v2** -- `{'pil': 'MobileNetV2ImageProcessorPil', 'torchvision': 'MobileNetV2ImageProcessor'}` (MobileNetV2Config model) - **mobilevit** -- `{'pil': 'MobileViTImageProcessorPil', 'torchvision': 'MobileViTImageProcessor'}` (MobileViTConfig model) - **mobilevitv2** -- `{'torchvision': 'MobileViTImageProcessor', 'pil': 'MobileViTImageProcessorPil'}` (MobileViTV2Config model) - **nougat** -- `{'pil': 'NougatImageProcessorPil', 'torchvision': 'NougatImageProcessor'}` (NougatConfig model) - **omdet-turbo** -- `{'torchvision': 'DetrImageProcessor', 'pil': 'DetrImageProcessorPil'}` (OmDetTurboConfig model) - **oneformer** -- `{'pil': 'OneFormerImageProcessorPil', 'torchvision': 'OneFormerImageProcessor'}` (OneFormerConfig model) - **ovis2** -- `{'pil': 'Ovis2ImageProcessorPil', 'torchvision': 'Ovis2ImageProcessor'}` (Ovis2Config model) - **owlv2** -- `{'pil': 'Owlv2ImageProcessorPil', 'torchvision': 'Owlv2ImageProcessor'}` (Owlv2Config model) - **owlvit** -- `{'pil': 'OwlViTImageProcessorPil', 'torchvision': 'OwlViTImageProcessor'}` (OwlViTConfig model) - **paddleocr_vl** -- `{'pil': 'PaddleOCRVLImageProcessorPil', 'torchvision': 'PaddleOCRVLImageProcessor'}` (PaddleOCRVLConfig model) - **paligemma** -- `{'torchvision': 'SiglipImageProcessor', 'pil': 'SiglipImageProcessorPil'}` (PaliGemmaConfig model) - **perceiver** -- `{'pil': 'PerceiverImageProcessorPil', 'torchvision': 'PerceiverImageProcessor'}` (PerceiverConfig model) - **perception_lm** -- `{'torchvision': 'PerceptionLMImageProcessor'}` (PerceptionLMConfig model) - **phi4_multimodal** -- `{'torchvision': 'Phi4MultimodalImageProcessor'}` (Phi4MultimodalConfig model) - **pi0** -- `{'torchvision': 'PI0ImageProcessor'}` (PI0Config model) - **pix2struct** -- `{'pil': 'Pix2StructImageProcessorPil', 'torchvision': 'Pix2StructImageProcessor'}` (Pix2StructConfig model) - **pixio** -- `{'torchvision': 'BitImageProcessor', 'pil': 'BitImageProcessorPil'}` (PixioConfig model) - **pixtral** -- `{'pil': 'PixtralImageProcessorPil', 'torchvision': 'PixtralImageProcessor'}` (PixtralVisionConfig model) - **poolformer** -- `{'pil': 'PoolFormerImageProcessorPil', 'torchvision': 'PoolFormerImageProcessor'}` (PoolFormerConfig model) - **pp_chart2table** -- `{'pil': 'PPChart2TableImageProcessorPil', 'torchvision': 'PPChart2TableImageProcessor'}` (PPChart2TableConfig model) - **pp_doclayout_v2** -- `{'torchvision': 'PPDocLayoutV2ImageProcessor'}` (PPDocLayoutV2Config model) - **pp_doclayout_v3** -- `{'torchvision': 'PPDocLayoutV3ImageProcessor'}` (PPDocLayoutV3Config model) - **pp_lcnet** -- `{'torchvision': 'PPLCNetImageProcessor'}` (PPLCNetConfig model) - **pp_ocrv5_mobile_det** -- `{'torchvision': 'PPOCRV5ServerDetImageProcessor'}` (PPOCRV5MobileDetConfig model) - **pp_ocrv5_mobile_rec** -- `{'torchvision': 'PPOCRV5ServerRecImageProcessor'}` (PPOCRV5MobileRecConfig model) - **pp_ocrv5_server_det** -- `{'torchvision': 'PPOCRV5ServerDetImageProcessor'}` (PPOCRV5ServerDetConfig model) - **pp_ocrv5_server_rec** -- `{'torchvision': 'PPOCRV5ServerRecImageProcessor'}` (PPOCRV5ServerRecConfig model) - **prompt_depth_anything** -- `{'pil': 'PromptDepthAnythingImageProcessorPil', 'torchvision': 'PromptDepthAnythingImageProcessor'}` (PromptDepthAnythingConfig model) - **pvt** -- `{'pil': 'PvtImageProcessorPil', 'torchvision': 'PvtImageProcessor'}` (PvtConfig model) - **pvt_v2** -- `{'torchvision': 'PvtImageProcessor', 'pil': 'PvtImageProcessorPil'}` (PvtV2Config model) - **qianfan_ocr** -- `{'torchvision': 'GotOcr2ImageProcessor', 'pil': 'GotOcr2ImageProcessorPil'}` (QianfanOCRConfig model) - **qwen2_5_omni** -- `{'torchvision': 'Qwen2VLImageProcessor', 'pil': 'Qwen2VLImageProcessorPil'}` (Qwen2_5OmniConfig model) - **qwen2_5_vl** -- `{'torchvision': 'Qwen2VLImageProcessor', 'pil': 'Qwen2VLImageProcessorPil'}` (Qwen2_5_VLConfig model) - **qwen2_vl** -- `{'pil': 'Qwen2VLImageProcessorPil', 'torchvision': 'Qwen2VLImageProcessor'}` (Qwen2VLConfig model) - **qwen3_5** -- `{'torchvision': 'Qwen2VLImageProcessor', 'pil': 'Qwen2VLImageProcessorPil'}` (Qwen3_5Config model) - **qwen3_5_moe** -- `{'torchvision': 'Qwen2VLImageProcessor', 'pil': 'Qwen2VLImageProcessorPil'}` (Qwen3_5MoeConfig model) - **qwen3_omni_moe** -- `{'torchvision': 'Qwen2VLImageProcessor', 'pil': 'Qwen2VLImageProcessorPil'}` (Qwen3OmniMoeConfig model) - **qwen3_vl** -- `{'torchvision': 'Qwen2VLImageProcessor', 'pil': 'Qwen2VLImageProcessorPil'}` (Qwen3VLConfig model) - **regnet** -- `{'torchvision': 'ConvNextImageProcessor', 'pil': 'ConvNextImageProcessorPil'}` (RegNetConfig model) - **resnet** -- `{'torchvision': 'ConvNextImageProcessor', 'pil': 'ConvNextImageProcessorPil'}` (ResNetConfig model) - **rt_detr** -- `{'pil': 'RTDetrImageProcessorPil', 'torchvision': 'RTDetrImageProcessor'}` (RTDetrConfig model) - **sam** -- `{'pil': 'SamImageProcessorPil', 'torchvision': 'SamImageProcessor'}` (SamConfig model) - **sam2** -- `{'torchvision': 'Sam2ImageProcessor'}` (Sam2Config model) - **sam2_video** -- `{'torchvision': 'Sam2ImageProcessor'}` (Sam2VideoConfig model) - **sam3** -- `{'torchvision': 'Sam3ImageProcessor'}` (Sam3Config model) - **sam3_lite_text** -- `{'torchvision': 'Sam3ImageProcessor'}` (Sam3LiteTextConfig model) - **sam3_tracker** -- `{'torchvision': 'Sam3ImageProcessor'}` (Sam3TrackerConfig model) - **sam3_tracker_video** -- `{'torchvision': 'Sam3ImageProcessor'}` (Sam3TrackerVideoConfig model) - **sam3_video** -- `{'torchvision': 'Sam3ImageProcessor'}` (Sam3VideoConfig model) - **sam_hq** -- `{'torchvision': 'SamImageProcessor', 'pil': 'SamImageProcessorPil'}` (SamHQConfig model) - **segformer** -- `{'pil': 'SegformerImageProcessorPil', 'torchvision': 'SegformerImageProcessor'}` (SegformerConfig model) - **seggpt** -- `{'pil': 'SegGptImageProcessorPil', 'torchvision': 'SegGptImageProcessor'}` (SegGptConfig model) - **shieldgemma2** -- `{'torchvision': 'Gemma3ImageProcessor', 'pil': 'Gemma3ImageProcessorPil'}` (ShieldGemma2Config model) - **siglip** -- `{'pil': 'SiglipImageProcessorPil', 'torchvision': 'SiglipImageProcessor'}` (SiglipConfig model) - **siglip2** -- `{'pil': 'Siglip2ImageProcessorPil', 'torchvision': 'Siglip2ImageProcessor'}` (Siglip2Config model) - **slanet** -- `{'torchvision': 'SLANeXtImageProcessor'}` (SLANetConfig model) - **slanext** -- `{'torchvision': 'SLANeXtImageProcessor'}` (SLANeXtConfig model) - **smolvlm** -- `{'pil': 'SmolVLMImageProcessorPil', 'torchvision': 'SmolVLMImageProcessor'}` (SmolVLMConfig model) - **superglue** -- `{'pil': 'SuperGlueImageProcessorPil', 'torchvision': 'SuperGlueImageProcessor'}` (SuperGlueConfig model) - **superpoint** -- `{'pil': 'SuperPointImageProcessorPil', 'torchvision': 'SuperPointImageProcessor'}` (SuperPointConfig model) - **swiftformer** -- `{'torchvision': 'ViTImageProcessor', 'pil': 'ViTImageProcessorPil'}` (SwiftFormerConfig model) - **swin** -- `{'torchvision': 'ViTImageProcessor', 'pil': 'ViTImageProcessorPil'}` (SwinConfig model) - **swin2sr** -- `{'pil': 'Swin2SRImageProcessorPil', 'torchvision': 'Swin2SRImageProcessor'}` (Swin2SRConfig model) - **swinv2** -- `{'torchvision': 'ViTImageProcessor', 'pil': 'ViTImageProcessorPil'}` (Swinv2Config model) - **t5gemma2** -- `{'torchvision': 'Gemma3ImageProcessor', 'pil': 'Gemma3ImageProcessorPil'}` (T5Gemma2Config model) - **t5gemma2_encoder** -- `{'torchvision': 'Gemma3ImageProcessor', 'pil': 'Gemma3ImageProcessorPil'}` (T5Gemma2EncoderConfig model) - **table-transformer** -- `{'torchvision': 'DetrImageProcessor', 'pil': 'DetrImageProcessorPil'}` (TableTransformerConfig model) - **textnet** -- `{'pil': 'TextNetImageProcessorPil', 'torchvision': 'TextNetImageProcessor'}` (TextNetConfig model) - **timesformer** -- `{'pil': 'VideoMAEImageProcessorPil', 'torchvision': 'VideoMAEImageProcessor'}` (TimesformerConfig model) - **timm_wrapper** -- `{'pil': 'TimmWrapperImageProcessor'}` (TimmWrapperConfig model) - **trocr** -- `{'torchvision': 'ViTImageProcessor', 'pil': 'ViTImageProcessorPil'}` (TrOCRConfig model) - **tvp** -- `{'pil': 'TvpImageProcessorPil', 'torchvision': 'TvpImageProcessor'}` (TvpConfig model) - **udop** -- `{'torchvision': 'LayoutLMv3ImageProcessor', 'pil': 'LayoutLMv3ImageProcessorPil'}` (UdopConfig model) - **upernet** -- `{'torchvision': 'SegformerImageProcessor', 'pil': 'SegformerImageProcessorPil'}` (UperNetConfig model) - **uvdoc** -- `{'torchvision': 'UVDocImageProcessor'}` (UVDocConfig model) - **video_llama_3** -- `{'pil': 'VideoLlama3ImageProcessorPil', 'torchvision': 'VideoLlama3ImageProcessor'}` (VideoLlama3Config model) - **video_llava** -- `{'pil': 'VideoLlavaImageProcessor'}` (VideoLlavaConfig model) - **videomae** -- `{'pil': 'VideoMAEImageProcessorPil', 'torchvision': 'VideoMAEImageProcessor'}` (VideoMAEConfig model) - **vilt** -- `{'pil': 'ViltImageProcessorPil', 'torchvision': 'ViltImageProcessor'}` (ViltConfig model) - **vipllava** -- `{'torchvision': 'CLIPImageProcessor', 'pil': 'CLIPImageProcessorPil'}` (VipLlavaConfig model) - **vit** -- `{'pil': 'ViTImageProcessorPil', 'torchvision': 'ViTImageProcessor'}` (ViTConfig model) - **vit_mae** -- `{'torchvision': 'ViTImageProcessor', 'pil': 'ViTImageProcessorPil'}` (ViTMAEConfig model) - **vit_msn** -- `{'torchvision': 'ViTImageProcessor', 'pil': 'ViTImageProcessorPil'}` (ViTMSNConfig model) - **vitmatte** -- `{'pil': 'VitMatteImageProcessorPil', 'torchvision': 'VitMatteImageProcessor'}` (VitMatteConfig model) - **vitpose** -- `{'pil': 'VitPoseImageProcessorPil', 'torchvision': 'VitPoseImageProcessor'}` (VitPoseConfig model) - **vivit** -- `{'torchvision': 'VivitImageProcessor'}` (VivitConfig model) - **xclip** -- `{'torchvision': 'CLIPImageProcessor', 'pil': 'CLIPImageProcessorPil'}` (XCLIPConfig model) - **yolos** -- `{'pil': 'YolosImageProcessorPil', 'torchvision': 'YolosImageProcessor'}` (YolosConfig model) - **zoedepth** -- `{'pil': 'ZoeDepthImageProcessorPil', 'torchvision': 'ZoeDepthImageProcessor'}` (ZoeDepthConfig model) Passing `token=True` is required when you want to use a private model. Examples: ```python >>> from transformers import AutoImageProcessor >>> # Download image processor from huggingface.co and cache. >>> image_processor = AutoImageProcessor.from_pretrained("google/vit-base-patch16-224-in21k") >>> # If image processor files are in a directory (e.g. image processor was saved using *save_pretrained('./test/saved_model/')*) >>> # image_processor = AutoImageProcessor.from_pretrained("./test/saved_model/") ``` **Parameters:** pretrained_model_name_or_path (`str` or `os.PathLike`) : This can be either: - a string, the *model id* of a pretrained image_processor hosted inside a model repo on huggingface.co. - a path to a *directory* containing a image processor file saved using the [save_pretrained()](/docs/transformers/v5.6.0/ja/main_classes/image_processor#transformers.ImageProcessingMixin.save_pretrained) method, e.g., `./my_model_directory/`. - a path to a saved image processor JSON *file*, e.g., `./my_model_directory/preprocessor_config.json`. cache_dir (`str` or `os.PathLike`, *optional*) : Path to a directory in which a downloaded pretrained model image processor should be cached if the standard cache should not be used. force_download (`bool`, *optional*, defaults to `False`) : Whether or not to force to (re-)download the image processor files and override the cached versions if they exist. proxies (`dict[str, str]`, *optional*) : A dictionary of proxy servers to use by protocol or endpoint, e.g., `{'http': 'foo.bar:3128', 'http://hostname': 'foo.bar:4012'}.` The proxies are used on each request. token (`str` or *bool*, *optional*) : The token to use as HTTP bearer authorization for remote files. If `True`, will use the token generated when running `hf auth login` (stored in `~/.huggingface`). revision (`str`, *optional*, defaults to `"main"`) : The specific model version to use. It can be a branch name, a tag name, or a commit id, since we use a git-based system for storing models and other artifacts on huggingface.co, so `revision` can be any identifier allowed by git. use_fast (`bool`, *optional*, defaults to `False`) : **Deprecated**: Use `backend="torchvision"` instead. This parameter is kept for backward compatibility. Use a fast torchvision-based image processor if it is supported for a given model. If a fast image processor is not available for a given model, a normal numpy-based image processor is returned instead. backend (`str`, *optional*, defaults to `None`) : The backend to use for image processing. Can be: - `None`: Automatically select the best available backend (torchvision if available, otherwise pil) - `"torchvision"`: Use Torchvision backend (GPU-accelerated, faster) - `"pil"`: Use PIL backend (portable, CPU-only) - Any custom backend name registered via `register()` method return_unused_kwargs (`bool`, *optional*, defaults to `False`) : If `False`, then this function returns just the final image processor object. If `True`, then this functions returns a `Tuple(image_processor, unused_kwargs)` where *unused_kwargs* is a dictionary consisting of the key/value pairs whose keys are not image processor attributes: i.e., the part of `kwargs` which has not been used to update `image_processor` and is otherwise ignored. trust_remote_code (`bool`, *optional*, defaults to `False`) : Whether or not to allow for custom models defined on the Hub in their own modeling files. This option should only be set to `True` for repositories you trust and in which you have read the code, as it will execute code present on the Hub on your local machine. image_processor_filename (`str`, *optional*, defaults to `"config.json"`) : The name of the file in the model directory to use for the image processor config. kwargs (`dict[str, Any]`, *optional*) : The values in kwargs of any keys which are image processor attributes will be used to override the loaded values. Behavior concerning key/value pairs whose keys are *not* image processor attributes is controlled by the `return_unused_kwargs` keyword parameter. #### register[[transformers.AutoImageProcessor.register]] [Source](https://github.com/huggingface/transformers/blob/v5.6.0/src/transformers/models/auto/image_processing_auto.py#L646) Register a new image processor for this class. **Parameters:** config_class ([PreTrainedConfig](/docs/transformers/v5.6.0/ja/main_classes/configuration#transformers.PreTrainedConfig)) : The configuration corresponding to the model to register. slow_image_processor_class (`type`, *optional*) : The PIL backend image processor class (deprecated, use `image_processor_classes={"pil": ...}`). fast_image_processor_class (`type`, *optional*) : The Torchvision backend image processor class (deprecated, use `image_processor_classes={"torchvision": ...}`). image_processor_classes (`dict[str, type]`, *optional*) : Dictionary mapping backend names to image processor classes. Allows registering custom backends. Example: `{"pil": MyPilProcessor, "torchvision": MyTorchvisionProcessor, "custom": MyCustomProcessor}` exist_ok (`bool`, *optional*, defaults to `False`) : If `True`, allow overwriting existing registrations. ## AutoProcessor[[transformers.AutoProcessor]] #### transformers.AutoProcessor[[transformers.AutoProcessor]] [Source](https://github.com/huggingface/transformers/blob/v5.6.0/src/transformers/models/auto/processing_auto.py#L217) This is a generic processor class that will be instantiated as one of the processor classes of the library when created with the [AutoProcessor.from_pretrained()](/docs/transformers/v5.6.0/ja/model_doc/auto#transformers.AutoProcessor.from_pretrained) class method. This class cannot be instantiated directly using `__init__()` (throws an error). from_pretrainedtransformers.AutoProcessor.from_pretrainedhttps://github.com/huggingface/transformers/blob/v5.6.0/src/transformers/models/auto/processing_auto.py#L231[{"name": "pretrained_model_name_or_path", "val": ""}, {"name": "**kwargs", "val": ""}]- **pretrained_model_name_or_path** (`str` or `os.PathLike`) -- This can be either: - a string, the *model id* of a pretrained feature_extractor hosted inside a model repo on huggingface.co. - a path to a *directory* containing a processor files saved using the `save_pretrained()` method, e.g., `./my_model_directory/`. - **cache_dir** (`str` or `os.PathLike`, *optional*) -- Path to a directory in which a downloaded pretrained model feature extractor should be cached if the standard cache should not be used. - **force_download** (`bool`, *optional*, defaults to `False`) -- Whether or not to force to (re-)download the feature extractor files and override the cached versions if they exist. - **proxies** (`dict[str, str]`, *optional*) -- A dictionary of proxy servers to use by protocol or endpoint, e.g., `{'http': 'foo.bar:3128', 'http://hostname': 'foo.bar:4012'}.` The proxies are used on each request. - **token** (`str` or *bool*, *optional*) -- The token to use as HTTP bearer authorization for remote files. If `True`, will use the token generated when running `hf auth login` (stored in `~/.huggingface`). - **revision** (`str`, *optional*, defaults to `"main"`) -- The specific model version to use. It can be a branch name, a tag name, or a commit id, since we use a git-based system for storing models and other artifacts on huggingface.co, so `revision` can be any identifier allowed by git. - **return_unused_kwargs** (`bool`, *optional*, defaults to `False`) -- If `False`, then this function returns just the final feature extractor object. If `True`, then this functions returns a `Tuple(feature_extractor, unused_kwargs)` where *unused_kwargs* is a dictionary consisting of the key/value pairs whose keys are not feature extractor attributes: i.e., the part of `kwargs` which has not been used to update `feature_extractor` and is otherwise ignored. - **trust_remote_code** (`bool`, *optional*, defaults to `False`) -- Whether or not to allow for custom models defined on the Hub in their own modeling files. This option should only be set to `True` for repositories you trust and in which you have read the code, as it will execute code present on the Hub on your local machine. - **kwargs** (`dict[str, Any]`, *optional*) -- The values in kwargs of any keys which are feature extractor attributes will be used to override the loaded values. Behavior concerning key/value pairs whose keys are *not* feature extractor attributes is controlled by the `return_unused_kwargs` keyword parameter.0 Instantiate one of the processor classes of the library from a pretrained model vocabulary. The processor class to instantiate is selected based on the `model_type` property of the config object (either passed as an argument or loaded from `pretrained_model_name_or_path` if possible): - **aimv2** -- [CLIPProcessor](/docs/transformers/v5.6.0/ja/model_doc/clip#transformers.CLIPProcessor) (Aimv2Config model) - **align** -- [AlignProcessor](/docs/transformers/v5.6.0/ja/model_doc/align#transformers.AlignProcessor) (AlignConfig model) - **altclip** -- [AltCLIPProcessor](/docs/transformers/v5.6.0/ja/model_doc/altclip#transformers.AltCLIPProcessor) (AltCLIPConfig model) - **aria** -- `AriaProcessor` (AriaConfig model) - **audioflamingo3** -- `AudioFlamingo3Processor` (AudioFlamingo3Config model) - **aya_vision** -- `AyaVisionProcessor` (AyaVisionConfig model) - **bark** -- [BarkProcessor](/docs/transformers/v5.6.0/ja/model_doc/bark#transformers.BarkProcessor) (BarkConfig model) - **blip** -- [BlipProcessor](/docs/transformers/v5.6.0/ja/model_doc/blip#transformers.BlipProcessor) (BlipConfig model) - **blip-2** -- [Blip2Processor](/docs/transformers/v5.6.0/ja/model_doc/blip-2#transformers.Blip2Processor) (Blip2Config model) - **bridgetower** -- [BridgeTowerProcessor](/docs/transformers/v5.6.0/ja/model_doc/bridgetower#transformers.BridgeTowerProcessor) (BridgeTowerConfig model) - **chameleon** -- `ChameleonProcessor` (ChameleonConfig model) - **chinese_clip** -- [ChineseCLIPProcessor](/docs/transformers/v5.6.0/ja/model_doc/chinese_clip#transformers.ChineseCLIPProcessor) (ChineseCLIPConfig model) - **clap** -- [ClapProcessor](/docs/transformers/v5.6.0/ja/model_doc/clap#transformers.ClapProcessor) (ClapConfig model) - **clip** -- [CLIPProcessor](/docs/transformers/v5.6.0/ja/model_doc/clip#transformers.CLIPProcessor) (CLIPConfig model) - **clipseg** -- [CLIPSegProcessor](/docs/transformers/v5.6.0/ja/model_doc/clipseg#transformers.CLIPSegProcessor) (CLIPSegConfig model) - **clvp** -- [ClvpProcessor](/docs/transformers/v5.6.0/ja/model_doc/clvp#transformers.ClvpProcessor) (ClvpConfig model) - **cohere2_vision** -- `Cohere2VisionProcessor` (Cohere2VisionConfig model) - **cohere_asr** -- `CohereAsrProcessor` (CohereAsrConfig model) - **colmodernvbert** -- `ColModernVBertProcessor` (ColModernVBertConfig model) - **colpali** -- `ColPaliProcessor` (ColPaliConfig model) - **colqwen2** -- `ColQwen2Processor` (ColQwen2Config model) - **deepseek_vl** -- `DeepseekVLProcessor` (DeepseekVLConfig model) - **deepseek_vl_hybrid** -- `DeepseekVLHybridProcessor` (DeepseekVLHybridConfig model) - **dia** -- `DiaProcessor` (DiaConfig model) - **edgetam** -- `Sam2Processor` (EdgeTamConfig model) - **emu3** -- `Emu3Processor` (Emu3Config model) - **ernie4_5_vl_moe** -- `Ernie4_5_VLMoeProcessor` (Ernie4_5_VLMoeConfig model) - **evolla** -- `EvollaProcessor` (EvollaConfig model) - **flava** -- `FlavaProcessor` (FlavaConfig model) - **florence2** -- `Florence2Processor` (Florence2Config model) - **fuyu** -- `FuyuProcessor` (FuyuConfig model) - **gemma3** -- `Gemma3Processor` (Gemma3Config model) - **gemma3n** -- `Gemma3nProcessor` (Gemma3nConfig model) - **gemma4** -- `Gemma4Processor` (Gemma4Config model) - **git** -- `GitProcessor` (GitConfig model) - **glm46v** -- `Glm46VProcessor` (Glm46VConfig model) - **glm4v** -- `Glm4vProcessor` (Glm4vConfig model) - **glm4v_moe** -- `Glm4vProcessor` (Glm4vMoeConfig model) - **glm_image** -- `Glm4vProcessor` (GlmImageConfig model) - **glmasr** -- `GlmAsrProcessor` (GlmAsrConfig model) - **got_ocr2** -- `GotOcr2Processor` (GotOcr2Config model) - **granite_speech** -- `GraniteSpeechProcessor` (GraniteSpeechConfig model) - **grounding-dino** -- `GroundingDinoProcessor` (GroundingDinoConfig model) - **groupvit** -- [CLIPProcessor](/docs/transformers/v5.6.0/ja/model_doc/clip#transformers.CLIPProcessor) (GroupViTConfig model) - **higgs_audio_v2** -- `HiggsAudioV2Processor` (HiggsAudioV2Config model) - **hubert** -- `Wav2Vec2Processor` (HubertConfig model) - **idefics** -- `IdeficsProcessor` (IdeficsConfig model) - **idefics2** -- `Idefics2Processor` (Idefics2Config model) - **idefics3** -- `Idefics3Processor` (Idefics3Config model) - **instructblip** -- `InstructBlipProcessor` (InstructBlipConfig model) - **instructblipvideo** -- `InstructBlipVideoProcessor` (InstructBlipVideoConfig model) - **internvl** -- `InternVLProcessor` (InternVLConfig model) - **janus** -- `JanusProcessor` (JanusConfig model) - **kosmos-2** -- `Kosmos2Processor` (Kosmos2Config model) - **kosmos-2.5** -- `Kosmos2_5Processor` (Kosmos2_5Config model) - **kyutai_speech_to_text** -- `KyutaiSpeechToTextProcessor` (KyutaiSpeechToTextConfig model) - **lasr_ctc** -- `LasrProcessor` (LasrCTCConfig model) - **lasr_encoder** -- `LasrProcessor` (LasrEncoderConfig model) - **layoutlmv2** -- `LayoutLMv2Processor` (LayoutLMv2Config model) - **layoutlmv3** -- `LayoutLMv3Processor` (LayoutLMv3Config model) - **layoutxlm** -- `LayoutXLMProcessor` (LayoutXLMConfig model) - **lfm2_vl** -- `Lfm2VlProcessor` (Lfm2VlConfig model) - **lighton_ocr** -- `LightOnOcrProcessor` (LightOnOcrConfig model) - **llama4** -- `Llama4Processor` (Llama4Config model) - **llava** -- `LlavaProcessor` (LlavaConfig model) - **llava_next** -- `LlavaNextProcessor` (LlavaNextConfig model) - **llava_next_video** -- `LlavaNextVideoProcessor` (LlavaNextVideoConfig model) - **llava_onevision** -- `LlavaOnevisionProcessor` (LlavaOnevisionConfig model) - **markuplm** -- `MarkupLMProcessor` (MarkupLMConfig model) - **metaclip_2** -- [CLIPProcessor](/docs/transformers/v5.6.0/ja/model_doc/clip#transformers.CLIPProcessor) (MetaClip2Config model) - **mgp-str** -- `MgpstrProcessor` (MgpstrConfig model) - **mistral3** -- `PixtralProcessor` (Mistral3Config model) - **mllama** -- `MllamaProcessor` (MllamaConfig model) - **mm-grounding-dino** -- `GroundingDinoProcessor` (MMGroundingDinoConfig model) - **modernvbert** -- `Idefics3Processor` (ModernVBertConfig model) - **moonshine** -- `Wav2Vec2Processor` (MoonshineConfig model) - **moonshine_streaming** -- `MoonshineStreamingProcessor` (MoonshineStreamingConfig model) - **musicflamingo** -- `MusicFlamingoProcessor` (MusicFlamingoConfig model) - **omdet-turbo** -- `OmDetTurboProcessor` (OmDetTurboConfig model) - **oneformer** -- `OneFormerProcessor` (OneFormerConfig model) - **ovis2** -- `Ovis2Processor` (Ovis2Config model) - **owlv2** -- `Owlv2Processor` (Owlv2Config model) - **owlvit** -- `OwlViTProcessor` (OwlViTConfig model) - **paddleocr_vl** -- `PaddleOCRVLProcessor` (PaddleOCRVLConfig model) - **paligemma** -- `PaliGemmaProcessor` (PaliGemmaConfig model) - **perception_lm** -- `PerceptionLMProcessor` (PerceptionLMConfig model) - **phi4_multimodal** -- `Phi4MultimodalProcessor` (Phi4MultimodalConfig model) - **pi0** -- `PI0Processor` (PI0Config model) - **pix2struct** -- `Pix2StructProcessor` (Pix2StructConfig model) - **pixtral** -- `PixtralProcessor` (PixtralVisionConfig model) - **pop2piano** -- `Pop2PianoProcessor` (Pop2PianoConfig model) - **pp_chart2table** -- `PPChart2TableProcessor` (PPChart2TableConfig model) - **qianfan_ocr** -- `QianfanOCRProcessor` (QianfanOCRConfig model) - **qwen2_5_omni** -- `Qwen2_5OmniProcessor` (Qwen2_5OmniConfig model) - **qwen2_5_vl** -- `Qwen2_5_VLProcessor` (Qwen2_5_VLConfig model) - **qwen2_audio** -- `Qwen2AudioProcessor` (Qwen2AudioConfig model) - **qwen2_vl** -- `Qwen2VLProcessor` (Qwen2VLConfig model) - **qwen3_5** -- `Qwen3VLProcessor` (Qwen3_5Config model) - **qwen3_5_moe** -- `Qwen3VLProcessor` (Qwen3_5MoeConfig model) - **qwen3_omni_moe** -- `Qwen3OmniMoeProcessor` (Qwen3OmniMoeConfig model) - **qwen3_vl** -- `Qwen3VLProcessor` (Qwen3VLConfig model) - **qwen3_vl_moe** -- `Qwen3VLProcessor` (Qwen3VLMoeConfig model) - **sam** -- `SamProcessor` (SamConfig model) - **sam2** -- `Sam2Processor` (Sam2Config model) - **sam3** -- `Sam3Processor` (Sam3Config model) - **sam3_lite_text** -- `Sam3Processor` (Sam3LiteTextConfig model) - **sam_hq** -- `SamHQProcessor` (SamHQConfig model) - **seamless_m4t** -- `SeamlessM4TProcessor` (SeamlessM4TConfig model) - **sew** -- `Wav2Vec2Processor` (SEWConfig model) - **sew-d** -- `Wav2Vec2Processor` (SEWDConfig model) - **shieldgemma2** -- `ShieldGemma2Processor` (ShieldGemma2Config model) - **siglip** -- `SiglipProcessor` (SiglipConfig model) - **siglip2** -- `Siglip2Processor` (Siglip2Config model) - **smolvlm** -- `SmolVLMProcessor` (SmolVLMConfig model) - **speech_to_text** -- `Speech2TextProcessor` (Speech2TextConfig model) - **speecht5** -- `SpeechT5Processor` (SpeechT5Config model) - **t5gemma2** -- `Gemma3Processor` (T5Gemma2Config model) - **t5gemma2_encoder** -- `Gemma3Processor` (T5Gemma2EncoderConfig model) - **trocr** -- `TrOCRProcessor` (TrOCRConfig model) - **tvp** -- `TvpProcessor` (TvpConfig model) - **udop** -- `UdopProcessor` (UdopConfig model) - **unispeech** -- `Wav2Vec2Processor` (UniSpeechConfig model) - **unispeech-sat** -- `Wav2Vec2Processor` (UniSpeechSatConfig model) - **vibevoice_asr** -- `VibeVoiceAsrProcessor` (VibeVoiceAsrConfig model) - **video_llava** -- `VideoLlavaProcessor` (VideoLlavaConfig model) - **vilt** -- `ViltProcessor` (ViltConfig model) - **vipllava** -- `LlavaProcessor` (VipLlavaConfig model) - **vision-text-dual-encoder** -- `VisionTextDualEncoderProcessor` (VisionTextDualEncoderConfig model) - **voxtral** -- `VoxtralProcessor` (VoxtralConfig model) - **voxtral_realtime** -- `VoxtralRealtimeProcessor` (VoxtralRealtimeConfig model) - **wav2vec2** -- `Wav2Vec2Processor` (Wav2Vec2Config model) - **wav2vec2-bert** -- `Wav2Vec2Processor` (Wav2Vec2BertConfig model) - **wav2vec2-conformer** -- `Wav2Vec2Processor` (Wav2Vec2ConformerConfig model) - **wavlm** -- `Wav2Vec2Processor` (WavLMConfig model) - **whisper** -- `WhisperProcessor` (WhisperConfig model) - **xclip** -- `XCLIPProcessor` (XCLIPConfig model) Passing `token=True` is required when you want to use a private model. Examples: ```python >>> from transformers import AutoProcessor >>> # Download processor from huggingface.co and cache. >>> processor = AutoProcessor.from_pretrained("facebook/wav2vec2-base-960h") >>> # If processor files are in a directory (e.g. processor was saved using *save_pretrained('./test/saved_model/')*) >>> # processor = AutoProcessor.from_pretrained("./test/saved_model/") ``` **Parameters:** pretrained_model_name_or_path (`str` or `os.PathLike`) : This can be either: - a string, the *model id* of a pretrained feature_extractor hosted inside a model repo on huggingface.co. - a path to a *directory* containing a processor files saved using the `save_pretrained()` method, e.g., `./my_model_directory/`. cache_dir (`str` or `os.PathLike`, *optional*) : Path to a directory in which a downloaded pretrained model feature extractor should be cached if the standard cache should not be used. force_download (`bool`, *optional*, defaults to `False`) : Whether or not to force to (re-)download the feature extractor files and override the cached versions if they exist. proxies (`dict[str, str]`, *optional*) : A dictionary of proxy servers to use by protocol or endpoint, e.g., `{'http': 'foo.bar:3128', 'http://hostname': 'foo.bar:4012'}.` The proxies are used on each request. token (`str` or *bool*, *optional*) : The token to use as HTTP bearer authorization for remote files. If `True`, will use the token generated when running `hf auth login` (stored in `~/.huggingface`). revision (`str`, *optional*, defaults to `"main"`) : The specific model version to use. It can be a branch name, a tag name, or a commit id, since we use a git-based system for storing models and other artifacts on huggingface.co, so `revision` can be any identifier allowed by git. return_unused_kwargs (`bool`, *optional*, defaults to `False`) : If `False`, then this function returns just the final feature extractor object. If `True`, then this functions returns a `Tuple(feature_extractor, unused_kwargs)` where *unused_kwargs* is a dictionary consisting of the key/value pairs whose keys are not feature extractor attributes: i.e., the part of `kwargs` which has not been used to update `feature_extractor` and is otherwise ignored. trust_remote_code (`bool`, *optional*, defaults to `False`) : Whether or not to allow for custom models defined on the Hub in their own modeling files. This option should only be set to `True` for repositories you trust and in which you have read the code, as it will execute code present on the Hub on your local machine. kwargs (`dict[str, Any]`, *optional*) : The values in kwargs of any keys which are feature extractor attributes will be used to override the loaded values. Behavior concerning key/value pairs whose keys are *not* feature extractor attributes is controlled by the `return_unused_kwargs` keyword parameter. #### register[[transformers.AutoProcessor.register]] [Source](https://github.com/huggingface/transformers/blob/v5.6.0/src/transformers/models/auto/processing_auto.py#L454) Register a new processor for this class. **Parameters:** config_class ([PreTrainedConfig](/docs/transformers/v5.6.0/ja/main_classes/configuration#transformers.PreTrainedConfig)) : The configuration corresponding to the model to register. processor_class ([ProcessorMixin](/docs/transformers/v5.6.0/ja/main_classes/processors#transformers.ProcessorMixin)) : The processor to register. ## Generic model classes 以下の自動クラスは、特定のヘッドを持たないベースモデルクラスをインスタンス化するために利用可能です。 ### AutoModel[[transformers.AutoModel]] #### transformers.AutoModel[[transformers.AutoModel]] [Source](https://github.com/huggingface/transformers/blob/v5.6.0/src/transformers/models/auto/modeling_auto.py#L1979) This is a generic model class that will be instantiated as one of the base model classes of the library when created with the [from_pretrained()](/docs/transformers/v5.6.0/ja/model_doc/auto#transformers.AutoModel.from_pretrained) class method or the [from_config()](/docs/transformers/v5.6.0/ja/model_doc/auto#transformers.AutoModel.from_config) class method. This class cannot be instantiated directly using `__init__()` (throws an error). from_configtransformers.AutoModel.from_confighttps://github.com/huggingface/transformers/blob/v5.6.0/src/transformers/models/auto/auto_factory.py#L206[{"name": "**kwargs", "val": ""}]- **config** ([PreTrainedConfig](/docs/transformers/v5.6.0/ja/main_classes/configuration#transformers.PreTrainedConfig)) -- The model class to instantiate is selected based on the configuration class: - [ASTConfig](/docs/transformers/v5.6.0/ja/model_doc/audio-spectrogram-transformer#transformers.ASTConfig) configuration class: [ASTModel](/docs/transformers/v5.6.0/ja/model_doc/audio-spectrogram-transformer#transformers.ASTModel) (ASTConfig model) - `AfmoeConfig` configuration class: `AfmoeModel` (AfmoeConfig model) - `Aimv2Config` configuration class: `Aimv2Model` (Aimv2Config model) - `Aimv2VisionConfig` configuration class: `Aimv2VisionModel` (Aimv2VisionConfig model) - [AlbertConfig](/docs/transformers/v5.6.0/ja/model_doc/albert#transformers.AlbertConfig) configuration class: [AlbertModel](/docs/transformers/v5.6.0/ja/model_doc/albert#transformers.AlbertModel) (AlbertConfig model) - [AlignConfig](/docs/transformers/v5.6.0/ja/model_doc/align#transformers.AlignConfig) configuration class: [AlignModel](/docs/transformers/v5.6.0/ja/model_doc/align#transformers.AlignModel) (AlignConfig model) - [AltCLIPConfig](/docs/transformers/v5.6.0/ja/model_doc/altclip#transformers.AltCLIPConfig) configuration class: [AltCLIPModel](/docs/transformers/v5.6.0/ja/model_doc/altclip#transformers.AltCLIPModel) (AltCLIPConfig model) - `ApertusConfig` configuration class: `ApertusModel` (ApertusConfig model) - `ArceeConfig` configuration class: `ArceeModel` (ArceeConfig model) - `AriaConfig` configuration class: `AriaModel` (AriaConfig model) - `AriaTextConfig` configuration class: `AriaTextModel` (AriaTextConfig model) - `AudioFlamingo3Config` configuration class: `AudioFlamingo3ForConditionalGeneration` (AudioFlamingo3Config model) - `AudioFlamingo3EncoderConfig` configuration class: `AudioFlamingo3Encoder` (AudioFlamingo3EncoderConfig model) - [AutoformerConfig](/docs/transformers/v5.6.0/ja/model_doc/autoformer#transformers.AutoformerConfig) configuration class: [AutoformerModel](/docs/transformers/v5.6.0/ja/model_doc/autoformer#transformers.AutoformerModel) (AutoformerConfig model) - `AyaVisionConfig` configuration class: `AyaVisionModel` (AyaVisionConfig model) - `BambaConfig` configuration class: `BambaModel` (BambaConfig model) - [BarkConfig](/docs/transformers/v5.6.0/ja/model_doc/bark#transformers.BarkConfig) configuration class: [BarkModel](/docs/transformers/v5.6.0/ja/model_doc/bark#transformers.BarkModel) (BarkConfig model) - [BartConfig](/docs/transformers/v5.6.0/ja/model_doc/bart#transformers.BartConfig) configuration class: [BartModel](/docs/transformers/v5.6.0/ja/model_doc/bart#transformers.BartModel) (BartConfig model) - [BeitConfig](/docs/transformers/v5.6.0/ja/model_doc/beit#transformers.BeitConfig) configuration class: [BeitModel](/docs/transformers/v5.6.0/ja/model_doc/beit#transformers.BeitModel) (BeitConfig model) - [BertConfig](/docs/transformers/v5.6.0/ja/model_doc/bert#transformers.BertConfig) configuration class: [BertModel](/docs/transformers/v5.6.0/ja/model_doc/bert#transformers.BertModel) (BertConfig model) - [BertGenerationConfig](/docs/transformers/v5.6.0/ja/model_doc/bert-generation#transformers.BertGenerationConfig) configuration class: [BertGenerationEncoder](/docs/transformers/v5.6.0/ja/model_doc/bert-generation#transformers.BertGenerationEncoder) (BertGenerationConfig model) - [BigBirdConfig](/docs/transformers/v5.6.0/ja/model_doc/big_bird#transformers.BigBirdConfig) configuration class: [BigBirdModel](/docs/transformers/v5.6.0/ja/model_doc/big_bird#transformers.BigBirdModel) (BigBirdConfig model) - [BigBirdPegasusConfig](/docs/transformers/v5.6.0/ja/model_doc/bigbird_pegasus#transformers.BigBirdPegasusConfig) configuration class: [BigBirdPegasusModel](/docs/transformers/v5.6.0/ja/model_doc/bigbird_pegasus#transformers.BigBirdPegasusModel) (BigBirdPegasusConfig model) - [BioGptConfig](/docs/transformers/v5.6.0/ja/model_doc/biogpt#transformers.BioGptConfig) configuration class: [BioGptModel](/docs/transformers/v5.6.0/ja/model_doc/biogpt#transformers.BioGptModel) (BioGptConfig model) - [BitConfig](/docs/transformers/v5.6.0/ja/model_doc/bit#transformers.BitConfig) configuration class: [BitModel](/docs/transformers/v5.6.0/ja/model_doc/bit#transformers.BitModel) (BitConfig model) - `BitNetConfig` configuration class: `BitNetModel` (BitNetConfig model) - [BlenderbotConfig](/docs/transformers/v5.6.0/ja/model_doc/blenderbot#transformers.BlenderbotConfig) configuration class: [BlenderbotModel](/docs/transformers/v5.6.0/ja/model_doc/blenderbot#transformers.BlenderbotModel) (BlenderbotConfig model) - [BlenderbotSmallConfig](/docs/transformers/v5.6.0/ja/model_doc/blenderbot-small#transformers.BlenderbotSmallConfig) configuration class: [BlenderbotSmallModel](/docs/transformers/v5.6.0/ja/model_doc/blenderbot-small#transformers.BlenderbotSmallModel) (BlenderbotSmallConfig model) - [Blip2Config](/docs/transformers/v5.6.0/ja/model_doc/blip-2#transformers.Blip2Config) configuration class: [Blip2Model](/docs/transformers/v5.6.0/ja/model_doc/blip-2#transformers.Blip2Model) (Blip2Config model) - [Blip2QFormerConfig](/docs/transformers/v5.6.0/ja/model_doc/blip-2#transformers.Blip2QFormerConfig) configuration class: [Blip2QFormerModel](/docs/transformers/v5.6.0/ja/model_doc/blip-2#transformers.Blip2QFormerModel) (Blip2QFormerConfig model) - [BlipConfig](/docs/transformers/v5.6.0/ja/model_doc/blip#transformers.BlipConfig) configuration class: [BlipModel](/docs/transformers/v5.6.0/ja/model_doc/blip#transformers.BlipModel) (BlipConfig model) - [BloomConfig](/docs/transformers/v5.6.0/ja/model_doc/bloom#transformers.BloomConfig) configuration class: [BloomModel](/docs/transformers/v5.6.0/ja/model_doc/bloom#transformers.BloomModel) (BloomConfig model) - `BltConfig` configuration class: `BltModel` (BltConfig model) - [BridgeTowerConfig](/docs/transformers/v5.6.0/ja/model_doc/bridgetower#transformers.BridgeTowerConfig) configuration class: [BridgeTowerModel](/docs/transformers/v5.6.0/ja/model_doc/bridgetower#transformers.BridgeTowerModel) (BridgeTowerConfig model) - [BrosConfig](/docs/transformers/v5.6.0/ja/model_doc/bros#transformers.BrosConfig) configuration class: [BrosModel](/docs/transformers/v5.6.0/ja/model_doc/bros#transformers.BrosModel) (BrosConfig model) - [CLIPConfig](/docs/transformers/v5.6.0/ja/model_doc/clip#transformers.CLIPConfig) configuration class: [CLIPModel](/docs/transformers/v5.6.0/ja/model_doc/clip#transformers.CLIPModel) (CLIPConfig model) - [CLIPSegConfig](/docs/transformers/v5.6.0/ja/model_doc/clipseg#transformers.CLIPSegConfig) configuration class: [CLIPSegModel](/docs/transformers/v5.6.0/ja/model_doc/clipseg#transformers.CLIPSegModel) (CLIPSegConfig model) - [CLIPTextConfig](/docs/transformers/v5.6.0/ja/model_doc/clip#transformers.CLIPTextConfig) configuration class: [CLIPTextModel](/docs/transformers/v5.6.0/ja/model_doc/clip#transformers.CLIPTextModel) (CLIPTextConfig model) - [CLIPVisionConfig](/docs/transformers/v5.6.0/ja/model_doc/clip#transformers.CLIPVisionConfig) configuration class: [CLIPVisionModel](/docs/transformers/v5.6.0/ja/model_doc/clip#transformers.CLIPVisionModel) (CLIPVisionConfig model) - [CTRLConfig](/docs/transformers/v5.6.0/ja/model_doc/ctrl#transformers.CTRLConfig) configuration class: [CTRLModel](/docs/transformers/v5.6.0/ja/model_doc/ctrl#transformers.CTRLModel) (CTRLConfig model) - [CamembertConfig](/docs/transformers/v5.6.0/ja/model_doc/camembert#transformers.CamembertConfig) configuration class: [CamembertModel](/docs/transformers/v5.6.0/ja/model_doc/camembert#transformers.CamembertModel) (CamembertConfig model) - [CanineConfig](/docs/transformers/v5.6.0/ja/model_doc/canine#transformers.CanineConfig) configuration class: [CanineModel](/docs/transformers/v5.6.0/ja/model_doc/canine#transformers.CanineModel) (CanineConfig model) - `ChameleonConfig` configuration class: `ChameleonModel` (ChameleonConfig model) - [ChineseCLIPConfig](/docs/transformers/v5.6.0/ja/model_doc/chinese_clip#transformers.ChineseCLIPConfig) configuration class: [ChineseCLIPModel](/docs/transformers/v5.6.0/ja/model_doc/chinese_clip#transformers.ChineseCLIPModel) (ChineseCLIPConfig model) - [ChineseCLIPVisionConfig](/docs/transformers/v5.6.0/ja/model_doc/chinese_clip#transformers.ChineseCLIPVisionConfig) configuration class: [ChineseCLIPVisionModel](/docs/transformers/v5.6.0/ja/model_doc/chinese_clip#transformers.ChineseCLIPVisionModel) (ChineseCLIPVisionConfig model) - [ClapConfig](/docs/transformers/v5.6.0/ja/model_doc/clap#transformers.ClapConfig) configuration class: [ClapModel](/docs/transformers/v5.6.0/ja/model_doc/clap#transformers.ClapModel) (ClapConfig model) - [ClvpConfig](/docs/transformers/v5.6.0/ja/model_doc/clvp#transformers.ClvpConfig) configuration class: [ClvpModelForConditionalGeneration](/docs/transformers/v5.6.0/ja/model_doc/clvp#transformers.ClvpModelForConditionalGeneration) (ClvpConfig model) - [CodeGenConfig](/docs/transformers/v5.6.0/ja/model_doc/codegen#transformers.CodeGenConfig) configuration class: [CodeGenModel](/docs/transformers/v5.6.0/ja/model_doc/codegen#transformers.CodeGenModel) (CodeGenConfig model) - `Cohere2Config` configuration class: `Cohere2Model` (Cohere2Config model) - `Cohere2VisionConfig` configuration class: `Cohere2VisionModel` (Cohere2VisionConfig model) - `CohereAsrConfig` configuration class: `CohereAsrModel` (CohereAsrConfig model) - `CohereConfig` configuration class: `CohereModel` (CohereConfig model) - [ConditionalDetrConfig](/docs/transformers/v5.6.0/ja/model_doc/conditional_detr#transformers.ConditionalDetrConfig) configuration class: [ConditionalDetrModel](/docs/transformers/v5.6.0/ja/model_doc/conditional_detr#transformers.ConditionalDetrModel) (ConditionalDetrConfig model) - [ConvBertConfig](/docs/transformers/v5.6.0/ja/model_doc/convbert#transformers.ConvBertConfig) configuration class: [ConvBertModel](/docs/transformers/v5.6.0/ja/model_doc/convbert#transformers.ConvBertModel) (ConvBertConfig model) - [ConvNextConfig](/docs/transformers/v5.6.0/ja/model_doc/convnext#transformers.ConvNextConfig) configuration class: [ConvNextModel](/docs/transformers/v5.6.0/ja/model_doc/convnext#transformers.ConvNextModel) (ConvNextConfig model) - [ConvNextV2Config](/docs/transformers/v5.6.0/ja/model_doc/convnextv2#transformers.ConvNextV2Config) configuration class: [ConvNextV2Model](/docs/transformers/v5.6.0/ja/model_doc/convnextv2#transformers.ConvNextV2Model) (ConvNextV2Config model) - [CpmAntConfig](/docs/transformers/v5.6.0/ja/model_doc/cpmant#transformers.CpmAntConfig) configuration class: [CpmAntModel](/docs/transformers/v5.6.0/ja/model_doc/cpmant#transformers.CpmAntModel) (CpmAntConfig model) - `CsmConfig` configuration class: `CsmForConditionalGeneration` (CsmConfig model) - [CvtConfig](/docs/transformers/v5.6.0/ja/model_doc/cvt#transformers.CvtConfig) configuration class: [CvtModel](/docs/transformers/v5.6.0/ja/model_doc/cvt#transformers.CvtModel) (CvtConfig model) - `CwmConfig` configuration class: `CwmModel` (CwmConfig model) - `DFineConfig` configuration class: `DFineModel` (DFineConfig model) - `DINOv3ConvNextConfig` configuration class: `DINOv3ConvNextModel` (DINOv3ConvNextConfig model) - `DINOv3ViTConfig` configuration class: `DINOv3ViTModel` (DINOv3ViTConfig model) - `DPRConfig` configuration class: `DPRQuestionEncoder` (DPRConfig model) - `DPTConfig` configuration class: `DPTModel` (DPTConfig model) - `DabDetrConfig` configuration class: `DabDetrModel` (DabDetrConfig model) - `DacConfig` configuration class: `DacModel` (DacConfig model) - [Data2VecAudioConfig](/docs/transformers/v5.6.0/ja/model_doc/data2vec#transformers.Data2VecAudioConfig) configuration class: [Data2VecAudioModel](/docs/transformers/v5.6.0/ja/model_doc/data2vec#transformers.Data2VecAudioModel) (Data2VecAudioConfig model) - [Data2VecTextConfig](/docs/transformers/v5.6.0/ja/model_doc/data2vec#transformers.Data2VecTextConfig) configuration class: [Data2VecTextModel](/docs/transformers/v5.6.0/ja/model_doc/data2vec#transformers.Data2VecTextModel) (Data2VecTextConfig model) - [Data2VecVisionConfig](/docs/transformers/v5.6.0/ja/model_doc/data2vec#transformers.Data2VecVisionConfig) configuration class: [Data2VecVisionModel](/docs/transformers/v5.6.0/ja/model_doc/data2vec#transformers.Data2VecVisionModel) (Data2VecVisionConfig model) - `DbrxConfig` configuration class: `DbrxModel` (DbrxConfig model) - [DebertaConfig](/docs/transformers/v5.6.0/ja/model_doc/deberta#transformers.DebertaConfig) configuration class: [DebertaModel](/docs/transformers/v5.6.0/ja/model_doc/deberta#transformers.DebertaModel) (DebertaConfig model) - [DebertaV2Config](/docs/transformers/v5.6.0/ja/model_doc/deberta-v2#transformers.DebertaV2Config) configuration class: [DebertaV2Model](/docs/transformers/v5.6.0/ja/model_doc/deberta-v2#transformers.DebertaV2Model) (DebertaV2Config model) - [DecisionTransformerConfig](/docs/transformers/v5.6.0/ja/model_doc/decision_transformer#transformers.DecisionTransformerConfig) configuration class: `DecisionTransformerModel` (DecisionTransformerConfig model) - `DeepseekV2Config` configuration class: `DeepseekV2Model` (DeepseekV2Config model) - `DeepseekV3Config` configuration class: `DeepseekV3Model` (DeepseekV3Config model) - `DeepseekVLConfig` configuration class: `DeepseekVLModel` (DeepseekVLConfig model) - `DeepseekVLHybridConfig` configuration class: `DeepseekVLHybridModel` (DeepseekVLHybridConfig model) - [DeformableDetrConfig](/docs/transformers/v5.6.0/ja/model_doc/deformable_detr#transformers.DeformableDetrConfig) configuration class: [DeformableDetrModel](/docs/transformers/v5.6.0/ja/model_doc/deformable_detr#transformers.DeformableDetrModel) (DeformableDetrConfig model) - [DeiTConfig](/docs/transformers/v5.6.0/ja/model_doc/deit#transformers.DeiTConfig) configuration class: [DeiTModel](/docs/transformers/v5.6.0/ja/model_doc/deit#transformers.DeiTModel) (DeiTConfig model) - `DepthProConfig` configuration class: `DepthProModel` (DepthProConfig model) - [DetrConfig](/docs/transformers/v5.6.0/ja/model_doc/detr#transformers.DetrConfig) configuration class: [DetrModel](/docs/transformers/v5.6.0/ja/model_doc/detr#transformers.DetrModel) (DetrConfig model) - `DiaConfig` configuration class: `DiaModel` (DiaConfig model) - `DiffLlamaConfig` configuration class: `DiffLlamaModel` (DiffLlamaConfig model) - [DinatConfig](/docs/transformers/v5.6.0/ja/model_doc/dinat#transformers.DinatConfig) configuration class: [DinatModel](/docs/transformers/v5.6.0/ja/model_doc/dinat#transformers.DinatModel) (DinatConfig model) - `Dinov2Config` configuration class: `Dinov2Model` (Dinov2Config model) - `Dinov2WithRegistersConfig` configuration class: `Dinov2WithRegistersModel` (Dinov2WithRegistersConfig model) - `DistilBertConfig` configuration class: `DistilBertModel` (DistilBertConfig model) - `DogeConfig` configuration class: `DogeModel` (DogeConfig model) - `DonutSwinConfig` configuration class: `DonutSwinModel` (DonutSwinConfig model) - `Dots1Config` configuration class: `Dots1Model` (Dots1Config model) - `EdgeTamConfig` configuration class: `EdgeTamModel` (EdgeTamConfig model) - `EdgeTamVideoConfig` configuration class: `EdgeTamVideoModel` (EdgeTamVideoConfig model) - `EdgeTamVisionConfig` configuration class: `EdgeTamVisionModel` (EdgeTamVisionConfig model) - `EfficientLoFTRConfig` configuration class: `EfficientLoFTRModel` (EfficientLoFTRConfig model) - `EfficientNetConfig` configuration class: `EfficientNetModel` (EfficientNetConfig model) - `ElectraConfig` configuration class: `ElectraModel` (ElectraConfig model) - `Emu3Config` configuration class: `Emu3Model` (Emu3Config model) - `EncodecConfig` configuration class: `EncodecModel` (EncodecConfig model) - `Ernie4_5Config` configuration class: `Ernie4_5Model` (Ernie4_5Config model) - `Ernie4_5_MoeConfig` configuration class: `Ernie4_5_MoeModel` (Ernie4_5_MoeConfig model) - `Ernie4_5_VLMoeConfig` configuration class: `Ernie4_5_VLMoeModel` (Ernie4_5_VLMoeConfig model) - `ErnieConfig` configuration class: `ErnieModel` (ErnieConfig model) - `EsmConfig` configuration class: `EsmModel` (EsmConfig model) - `EuroBertConfig` configuration class: `EuroBertModel` (EuroBertConfig model) - `EvollaConfig` configuration class: `EvollaModel` (EvollaConfig model) - `Exaone4Config` configuration class: `Exaone4Model` (Exaone4Config model) - `ExaoneMoeConfig` configuration class: `ExaoneMoeModel` (ExaoneMoeConfig model) - `FNetConfig` configuration class: `FNetModel` (FNetConfig model) - `FSMTConfig` configuration class: `FSMTModel` (FSMTConfig model) - `FalconConfig` configuration class: `FalconModel` (FalconConfig model) - `FalconH1Config` configuration class: `FalconH1Model` (FalconH1Config model) - `FalconMambaConfig` configuration class: `FalconMambaModel` (FalconMambaConfig model) - `FastSpeech2ConformerConfig` configuration class: `FastSpeech2ConformerModel` (FastSpeech2ConformerConfig model) - `FastSpeech2ConformerWithHifiGanConfig` configuration class: `FastSpeech2ConformerWithHifiGan` (FastSpeech2ConformerWithHifiGanConfig model) - `FastVlmConfig` configuration class: `FastVlmModel` (FastVlmConfig model) - `FlaubertConfig` configuration class: `FlaubertModel` (FlaubertConfig model) - `FlavaConfig` configuration class: `FlavaModel` (FlavaConfig model) - `FlexOlmoConfig` configuration class: `FlexOlmoModel` (FlexOlmoConfig model) - `Florence2Config` configuration class: `Florence2Model` (Florence2Config model) - `FocalNetConfig` configuration class: `FocalNetModel` (FocalNetConfig model) - `FunnelConfig` configuration class: `FunnelModel` or `FunnelBaseModel` (FunnelConfig model) - `FuyuConfig` configuration class: `FuyuModel` (FuyuConfig model) - `GLPNConfig` configuration class: `GLPNModel` (GLPNConfig model) - `GPT2Config` configuration class: `GPT2Model` (GPT2Config model) - `GPTBigCodeConfig` configuration class: `GPTBigCodeModel` (GPTBigCodeConfig model) - `GPTJConfig` configuration class: `GPTJModel` (GPTJConfig model) - `GPTNeoConfig` configuration class: `GPTNeoModel` (GPTNeoConfig model) - `GPTNeoXConfig` configuration class: `GPTNeoXModel` (GPTNeoXConfig model) - `GPTNeoXJapaneseConfig` configuration class: `GPTNeoXJapaneseModel` (GPTNeoXJapaneseConfig model) - `Gemma2Config` configuration class: `Gemma2Model` (Gemma2Config model) - `Gemma3Config` configuration class: `Gemma3Model` (Gemma3Config model) - `Gemma3TextConfig` configuration class: `Gemma3TextModel` (Gemma3TextConfig model) - `Gemma3nAudioConfig` configuration class: `Gemma3nAudioEncoder` (Gemma3nAudioConfig model) - `Gemma3nConfig` configuration class: `Gemma3nModel` (Gemma3nConfig model) - `Gemma3nTextConfig` configuration class: `Gemma3nTextModel` (Gemma3nTextConfig model) - `Gemma3nVisionConfig` configuration class: `TimmWrapperModel` (Gemma3nVisionConfig model) - `Gemma4AudioConfig` configuration class: `Gemma4AudioModel` (Gemma4AudioConfig model) - `Gemma4Config` configuration class: `Gemma4Model` (Gemma4Config model) - `Gemma4TextConfig` configuration class: `Gemma4TextModel` (Gemma4TextConfig model) - `Gemma4VisionConfig` configuration class: `Gemma4VisionModel` (Gemma4VisionConfig model) - `GemmaConfig` configuration class: `GemmaModel` (GemmaConfig model) - `GitConfig` configuration class: `GitModel` (GitConfig model) - `Glm46VConfig` configuration class: `Glm46VModel` (Glm46VConfig model) - `Glm4Config` configuration class: `Glm4Model` (Glm4Config model) - `Glm4MoeConfig` configuration class: `Glm4MoeModel` (Glm4MoeConfig model) - `Glm4MoeLiteConfig` configuration class: `Glm4MoeLiteModel` (Glm4MoeLiteConfig model) - `Glm4vConfig` configuration class: `Glm4vModel` (Glm4vConfig model) - `Glm4vMoeConfig` configuration class: `Glm4vMoeModel` (Glm4vMoeConfig model) - `Glm4vMoeTextConfig` configuration class: `Glm4vMoeTextModel` (Glm4vMoeTextConfig model) - `Glm4vMoeVisionConfig` configuration class: `Glm4vMoeVisionModel` (Glm4vMoeVisionConfig model) - `Glm4vTextConfig` configuration class: `Glm4vTextModel` (Glm4vTextConfig model) - `Glm4vVisionConfig` configuration class: `Glm4vVisionModel` (Glm4vVisionConfig model) - `GlmAsrConfig` configuration class: `GlmAsrForConditionalGeneration` (GlmAsrConfig model) - `GlmAsrEncoderConfig` configuration class: `GlmAsrEncoder` (GlmAsrEncoderConfig model) - `GlmConfig` configuration class: `GlmModel` (GlmConfig model) - `GlmImageConfig` configuration class: `GlmImageModel` (GlmImageConfig model) - `GlmImageTextConfig` configuration class: `GlmImageTextModel` (GlmImageTextConfig model) - `GlmImageVQVAEConfig` configuration class: `GlmImageVQVAE` (GlmImageVQVAEConfig model) - `GlmImageVisionConfig` configuration class: `GlmImageVisionModel` (GlmImageVisionConfig model) - `GlmMoeDsaConfig` configuration class: `GlmMoeDsaModel` (GlmMoeDsaConfig model) - `GlmOcrConfig` configuration class: `GlmOcrModel` (GlmOcrConfig model) - `GlmOcrTextConfig` configuration class: `GlmOcrTextModel` (GlmOcrTextConfig model) - `GlmOcrVisionConfig` configuration class: `GlmOcrVisionModel` (GlmOcrVisionConfig model) - `GotOcr2Config` configuration class: `GotOcr2Model` (GotOcr2Config model) - `GptOssConfig` configuration class: `GptOssModel` (GptOssConfig model) - `GraniteConfig` configuration class: `GraniteModel` (GraniteConfig model) - `GraniteMoeConfig` configuration class: `GraniteMoeModel` (GraniteMoeConfig model) - `GraniteMoeHybridConfig` configuration class: `GraniteMoeHybridModel` (GraniteMoeHybridConfig model) - `GraniteMoeSharedConfig` configuration class: `GraniteMoeSharedModel` (GraniteMoeSharedConfig model) - `GraniteSpeechConfig` configuration class: `GraniteSpeechForConditionalGeneration` (GraniteSpeechConfig model) - `GroundingDinoConfig` configuration class: `GroundingDinoModel` (GroundingDinoConfig model) - `GroupViTConfig` configuration class: `GroupViTModel` (GroupViTConfig model) - `HGNetV2Config` configuration class: `HGNetV2Backbone` (HGNetV2Config model) - `HYV3Config` configuration class: `HYV3Model` (HYV3Config model) - `HeliumConfig` configuration class: `HeliumModel` (HeliumConfig model) - `HieraConfig` configuration class: `HieraModel` (HieraConfig model) - `HiggsAudioV2Config` configuration class: `HiggsAudioV2ForConditionalGeneration` (HiggsAudioV2Config model) - `HiggsAudioV2TokenizerConfig` configuration class: `HiggsAudioV2TokenizerModel` (HiggsAudioV2TokenizerConfig model) - `HubertConfig` configuration class: `HubertModel` (HubertConfig model) - `HunYuanDenseV1Config` configuration class: `HunYuanDenseV1Model` (HunYuanDenseV1Config model) - `HunYuanMoEV1Config` configuration class: `HunYuanMoEV1Model` (HunYuanMoEV1Config model) - `IBertConfig` configuration class: `IBertModel` (IBertConfig model) - `IJepaConfig` configuration class: `IJepaModel` (IJepaConfig model) - `Idefics2Config` configuration class: `Idefics2Model` (Idefics2Config model) - `Idefics3Config` configuration class: `Idefics3Model` (Idefics3Config model) - `Idefics3VisionConfig` configuration class: `Idefics3VisionTransformer` (Idefics3VisionConfig model) - `IdeficsConfig` configuration class: `IdeficsModel` (IdeficsConfig model) - `ImageGPTConfig` configuration class: `ImageGPTModel` (ImageGPTConfig model) - `InformerConfig` configuration class: `InformerModel` (InformerConfig model) - `InstructBlipConfig` configuration class: `InstructBlipModel` (InstructBlipConfig model) - `InstructBlipVideoConfig` configuration class: `InstructBlipVideoModel` (InstructBlipVideoConfig model) - `InternVLConfig` configuration class: `InternVLModel` (InternVLConfig model) - `InternVLVisionConfig` configuration class: `InternVLVisionModel` (InternVLVisionConfig model) - `Jais2Config` configuration class: `Jais2Model` (Jais2Config model) - `JambaConfig` configuration class: `JambaModel` (JambaConfig model) - `JanusConfig` configuration class: `JanusModel` (JanusConfig model) - `JetMoeConfig` configuration class: `JetMoeModel` (JetMoeConfig model) - `JinaEmbeddingsV3Config` configuration class: `JinaEmbeddingsV3Model` (JinaEmbeddingsV3Config model) - `Kosmos2Config` configuration class: `Kosmos2Model` (Kosmos2Config model) - `Kosmos2_5Config` configuration class: `Kosmos2_5Model` (Kosmos2_5Config model) - `KyutaiSpeechToTextConfig` configuration class: `KyutaiSpeechToTextModel` (KyutaiSpeechToTextConfig model) - `LEDConfig` configuration class: `LEDModel` (LEDConfig model) - `LasrCTCConfig` configuration class: `LasrForCTC` (LasrCTCConfig model) - `LasrEncoderConfig` configuration class: `LasrEncoder` (LasrEncoderConfig model) - `LayoutLMConfig` configuration class: `LayoutLMModel` (LayoutLMConfig model) - `LayoutLMv2Config` configuration class: `LayoutLMv2Model` (LayoutLMv2Config model) - `LayoutLMv3Config` configuration class: `LayoutLMv3Model` (LayoutLMv3Config model) - `LevitConfig` configuration class: `LevitModel` (LevitConfig model) - `Lfm2Config` configuration class: `Lfm2Model` (Lfm2Config model) - `Lfm2MoeConfig` configuration class: `Lfm2MoeModel` (Lfm2MoeConfig model) - `Lfm2VlConfig` configuration class: `Lfm2VlModel` (Lfm2VlConfig model) - `LightGlueConfig` configuration class: `LightGlueForKeypointMatching` (LightGlueConfig model) - `LightOnOcrConfig` configuration class: `LightOnOcrModel` (LightOnOcrConfig model) - `LiltConfig` configuration class: `LiltModel` (LiltConfig model) - `Llama4Config` configuration class: `Llama4ForConditionalGeneration` (Llama4Config model) - `Llama4TextConfig` configuration class: `Llama4TextModel` (Llama4TextConfig model) - `LlamaConfig` configuration class: `LlamaModel` (LlamaConfig model) - `LlavaConfig` configuration class: `LlavaModel` (LlavaConfig model) - `LlavaNextConfig` configuration class: `LlavaNextModel` (LlavaNextConfig model) - `LlavaNextVideoConfig` configuration class: `LlavaNextVideoModel` (LlavaNextVideoConfig model) - `LlavaOnevisionConfig` configuration class: `LlavaOnevisionModel` (LlavaOnevisionConfig model) - `LongT5Config` configuration class: `LongT5Model` (LongT5Config model) - `LongcatFlashConfig` configuration class: `LongcatFlashModel` (LongcatFlashConfig model) - `LongformerConfig` configuration class: `LongformerModel` (LongformerConfig model) - `LukeConfig` configuration class: `LukeModel` (LukeConfig model) - `LwDetrConfig` configuration class: `LwDetrModel` (LwDetrConfig model) - `LxmertConfig` configuration class: `LxmertModel` (LxmertConfig model) - `M2M100Config` configuration class: `M2M100Model` (M2M100Config model) - `MBartConfig` configuration class: `MBartModel` (MBartConfig model) - `MLCDVisionConfig` configuration class: `MLCDVisionModel` (MLCDVisionConfig model) - `MMGroundingDinoConfig` configuration class: `MMGroundingDinoModel` (MMGroundingDinoConfig model) - `MPNetConfig` configuration class: `MPNetModel` (MPNetConfig model) - `MT5Config` configuration class: `MT5Model` (MT5Config model) - `Mamba2Config` configuration class: `Mamba2Model` (Mamba2Config model) - `MambaConfig` configuration class: `MambaModel` (MambaConfig model) - `MarianConfig` configuration class: `MarianModel` (MarianConfig model) - `MarkupLMConfig` configuration class: `MarkupLMModel` (MarkupLMConfig model) - `Mask2FormerConfig` configuration class: `Mask2FormerModel` (Mask2FormerConfig model) - `MaskFormerConfig` configuration class: `MaskFormerModel` (MaskFormerConfig model) - `MaskFormerSwinConfig` configuration class: `MaskFormerSwinModel` (MaskFormerSwinConfig model) - `MegatronBertConfig` configuration class: `MegatronBertModel` (MegatronBertConfig model) - `MetaClip2Config` configuration class: `MetaClip2Model` (MetaClip2Config model) - `MgpstrConfig` configuration class: `MgpstrForSceneTextRecognition` (MgpstrConfig model) - `MimiConfig` configuration class: `MimiModel` (MimiConfig model) - `MiniMaxConfig` configuration class: `MiniMaxModel` (MiniMaxConfig model) - `MiniMaxM2Config` configuration class: `MiniMaxM2Model` (MiniMaxM2Config model) - `Ministral3Config` configuration class: `Ministral3Model` (Ministral3Config model) - `MinistralConfig` configuration class: `MinistralModel` (MinistralConfig model) - `Mistral3Config` configuration class: `Mistral3Model` (Mistral3Config model) - `Mistral4Config` configuration class: `Mistral4Model` (Mistral4Config model) - `MistralConfig` configuration class: `MistralModel` (MistralConfig model) - `MixtralConfig` configuration class: `MixtralModel` (MixtralConfig model) - `MllamaConfig` configuration class: `MllamaModel` (MllamaConfig model) - `MobileBertConfig` configuration class: `MobileBertModel` (MobileBertConfig model) - `MobileNetV1Config` configuration class: `MobileNetV1Model` (MobileNetV1Config model) - `MobileNetV2Config` configuration class: `MobileNetV2Model` (MobileNetV2Config model) - `MobileViTConfig` configuration class: `MobileViTModel` (MobileViTConfig model) - `MobileViTV2Config` configuration class: `MobileViTV2Model` (MobileViTV2Config model) - `ModernBertConfig` configuration class: `ModernBertModel` (ModernBertConfig model) - `ModernBertDecoderConfig` configuration class: `ModernBertDecoderModel` (ModernBertDecoderConfig model) - `ModernVBertConfig` configuration class: `ModernVBertModel` (ModernVBertConfig model) - `MoonshineConfig` configuration class: `MoonshineModel` (MoonshineConfig model) - `MoonshineStreamingConfig` configuration class: `MoonshineStreamingModel` (MoonshineStreamingConfig model) - `MoshiConfig` configuration class: `MoshiModel` (MoshiConfig model) - `MptConfig` configuration class: `MptModel` (MptConfig model) - `MraConfig` configuration class: `MraModel` (MraConfig model) - `MusicFlamingoConfig` configuration class: `MusicFlamingoForConditionalGeneration` (MusicFlamingoConfig model) - `MusicgenConfig` configuration class: `MusicgenModel` (MusicgenConfig model) - `MusicgenMelodyConfig` configuration class: `MusicgenMelodyModel` (MusicgenMelodyConfig model) - `MvpConfig` configuration class: `MvpModel` (MvpConfig model) - `NanoChatConfig` configuration class: `NanoChatModel` (NanoChatConfig model) - `NemotronConfig` configuration class: `NemotronModel` (NemotronConfig model) - `NemotronHConfig` configuration class: `NemotronHModel` (NemotronHConfig model) - `NllbMoeConfig` configuration class: `NllbMoeModel` (NllbMoeConfig model) - `NomicBertConfig` configuration class: `NomicBertModel` (NomicBertConfig model) - `NystromformerConfig` configuration class: `NystromformerModel` (NystromformerConfig model) - `OPTConfig` configuration class: `OPTModel` (OPTConfig model) - `Olmo2Config` configuration class: `Olmo2Model` (Olmo2Config model) - `Olmo3Config` configuration class: `Olmo3Model` (Olmo3Config model) - `OlmoConfig` configuration class: `OlmoModel` (OlmoConfig model) - `OlmoHybridConfig` configuration class: `OlmoHybridModel` (OlmoHybridConfig model) - `OlmoeConfig` configuration class: `OlmoeModel` (OlmoeConfig model) - `OmDetTurboConfig` configuration class: `OmDetTurboForObjectDetection` (OmDetTurboConfig model) - `OneFormerConfig` configuration class: `OneFormerModel` (OneFormerConfig model) - `OpenAIGPTConfig` configuration class: `OpenAIGPTModel` (OpenAIGPTConfig model) - `OpenAIPrivacyFilterConfig` configuration class: `OpenAIPrivacyFilterModel` (OpenAIPrivacyFilterConfig model) - `Ovis2Config` configuration class: `Ovis2Model` (Ovis2Config model) - `OwlViTConfig` configuration class: `OwlViTModel` (OwlViTConfig model) - `Owlv2Config` configuration class: `Owlv2Model` (Owlv2Config model) - `PI0Config` configuration class: `PI0Model` (PI0Config model) - `PLBartConfig` configuration class: `PLBartModel` (PLBartConfig model) - `PPDocLayoutV3Config` configuration class: `PPDocLayoutV3Model` (PPDocLayoutV3Config model) - `PPOCRV5MobileRecConfig` configuration class: `PPOCRV5MobileRecModel` (PPOCRV5MobileRecConfig model) - `PPOCRV5ServerRecConfig` configuration class: `PPOCRV5ServerRecModel` (PPOCRV5ServerRecConfig model) - `PaliGemmaConfig` configuration class: `PaliGemmaModel` (PaliGemmaConfig model) - `ParakeetCTCConfig` configuration class: `ParakeetForCTC` (ParakeetCTCConfig model) - `ParakeetEncoderConfig` configuration class: `ParakeetEncoder` (ParakeetEncoderConfig model) - `PatchTSMixerConfig` configuration class: `PatchTSMixerModel` (PatchTSMixerConfig model) - `PatchTSTConfig` configuration class: `PatchTSTModel` (PatchTSTConfig model) - `PeAudioConfig` configuration class: `PeAudioModel` (PeAudioConfig model) - `PeAudioEncoderConfig` configuration class: `PeAudioEncoder` (PeAudioEncoderConfig model) - `PeAudioVideoConfig` configuration class: `PeAudioVideoModel` (PeAudioVideoConfig model) - `PeAudioVideoEncoderConfig` configuration class: `PeAudioVideoEncoder` (PeAudioVideoEncoderConfig model) - `PeVideoConfig` configuration class: `PeVideoModel` (PeVideoConfig model) - `PeVideoEncoderConfig` configuration class: `PeVideoEncoder` (PeVideoEncoderConfig model) - `PegasusConfig` configuration class: `PegasusModel` (PegasusConfig model) - `PegasusXConfig` configuration class: `PegasusXModel` (PegasusXConfig model) - `PerceiverConfig` configuration class: `PerceiverModel` (PerceiverConfig model) - `PerceptionLMConfig` configuration class: `PerceptionLMModel` (PerceptionLMConfig model) - `PersimmonConfig` configuration class: `PersimmonModel` (PersimmonConfig model) - `Phi3Config` configuration class: `Phi3Model` (Phi3Config model) - `Phi4MultimodalConfig` configuration class: `Phi4MultimodalModel` (Phi4MultimodalConfig model) - `PhiConfig` configuration class: `PhiModel` (PhiConfig model) - `PhimoeConfig` configuration class: `PhimoeModel` (PhimoeConfig model) - `PixioConfig` configuration class: `PixioModel` (PixioConfig model) - `PixtralVisionConfig` configuration class: `PixtralVisionModel` (PixtralVisionConfig model) - `PoolFormerConfig` configuration class: `PoolFormerModel` (PoolFormerConfig model) - `ProphetNetConfig` configuration class: `ProphetNetModel` (ProphetNetConfig model) - `PvtConfig` configuration class: `PvtModel` (PvtConfig model) - `PvtV2Config` configuration class: `PvtV2Model` (PvtV2Config model) - `QianfanOCRConfig` configuration class: `QianfanOCRModel` (QianfanOCRConfig model) - `QianfanOCRVisionConfig` configuration class: `QianfanOCRVisionModel` (QianfanOCRVisionConfig model) - `Qwen2AudioEncoderConfig` configuration class: `Qwen2AudioEncoder` (Qwen2AudioEncoderConfig model) - `Qwen2Config` configuration class: `Qwen2Model` (Qwen2Config model) - `Qwen2MoeConfig` configuration class: `Qwen2MoeModel` (Qwen2MoeConfig model) - `Qwen2VLConfig` configuration class: `Qwen2VLModel` (Qwen2VLConfig model) - `Qwen2VLTextConfig` configuration class: `Qwen2VLTextModel` (Qwen2VLTextConfig model) - `Qwen2_5_VLConfig` configuration class: `Qwen2_5_VLModel` (Qwen2_5_VLConfig model) - `Qwen2_5_VLTextConfig` configuration class: `Qwen2_5_VLTextModel` (Qwen2_5_VLTextConfig model) - `Qwen3Config` configuration class: `Qwen3Model` (Qwen3Config model) - `Qwen3MoeConfig` configuration class: `Qwen3MoeModel` (Qwen3MoeConfig model) - `Qwen3NextConfig` configuration class: `Qwen3NextModel` (Qwen3NextConfig model) - `Qwen3VLConfig` configuration class: `Qwen3VLModel` (Qwen3VLConfig model) - `Qwen3VLMoeConfig` configuration class: `Qwen3VLMoeModel` (Qwen3VLMoeConfig model) - `Qwen3VLMoeTextConfig` configuration class: `Qwen3VLMoeTextModel` (Qwen3VLMoeTextConfig model) - `Qwen3VLTextConfig` configuration class: `Qwen3VLTextModel` (Qwen3VLTextConfig model) - `Qwen3_5Config` configuration class: `Qwen3_5Model` (Qwen3_5Config model) - `Qwen3_5MoeConfig` configuration class: `Qwen3_5MoeModel` (Qwen3_5MoeConfig model) - `Qwen3_5MoeTextConfig` configuration class: `Qwen3_5MoeTextModel` (Qwen3_5MoeTextConfig model) - `Qwen3_5TextConfig` configuration class: `Qwen3_5TextModel` (Qwen3_5TextConfig model) - `RTDetrConfig` configuration class: `RTDetrModel` (RTDetrConfig model) - `RTDetrV2Config` configuration class: `RTDetrV2Model` (RTDetrV2Config model) - `RecurrentGemmaConfig` configuration class: `RecurrentGemmaModel` (RecurrentGemmaConfig model) - `ReformerConfig` configuration class: `ReformerModel` (ReformerConfig model) - `RegNetConfig` configuration class: `RegNetModel` (RegNetConfig model) - `RemBertConfig` configuration class: `RemBertModel` (RemBertConfig model) - `ResNetConfig` configuration class: `ResNetModel` (ResNetConfig model) - `RoCBertConfig` configuration class: `RoCBertModel` (RoCBertConfig model) - `RoFormerConfig` configuration class: `RoFormerModel` (RoFormerConfig model) - `RobertaConfig` configuration class: `RobertaModel` (RobertaConfig model) - `RobertaPreLayerNormConfig` configuration class: `RobertaPreLayerNormModel` (RobertaPreLayerNormConfig model) - `RwkvConfig` configuration class: `RwkvModel` (RwkvConfig model) - `SEWConfig` configuration class: `SEWModel` (SEWConfig model) - `SEWDConfig` configuration class: `SEWDModel` (SEWDConfig model) - `Sam2Config` configuration class: `Sam2Model` (Sam2Config model) - `Sam2HieraDetConfig` configuration class: `Sam2HieraDetModel` (Sam2HieraDetConfig model) - `Sam2VideoConfig` configuration class: `Sam2VideoModel` (Sam2VideoConfig model) - `Sam2VisionConfig` configuration class: `Sam2VisionModel` (Sam2VisionConfig model) - `Sam3Config` configuration class: `Sam3Model` (Sam3Config model) - `Sam3LiteTextConfig` configuration class: `Sam3LiteTextModel` (Sam3LiteTextConfig model) - `Sam3LiteTextTextConfig` configuration class: `Sam3LiteTextTextModel` (Sam3LiteTextTextConfig model) - `Sam3TrackerConfig` configuration class: `Sam3TrackerModel` (Sam3TrackerConfig model) - `Sam3TrackerVideoConfig` configuration class: `Sam3TrackerVideoModel` (Sam3TrackerVideoConfig model) - `Sam3ViTConfig` configuration class: `Sam3ViTModel` (Sam3ViTConfig model) - `Sam3VideoConfig` configuration class: `Sam3VideoModel` (Sam3VideoConfig model) - `Sam3VisionConfig` configuration class: `Sam3VisionModel` (Sam3VisionConfig model) - `SamConfig` configuration class: `SamModel` (SamConfig model) - `SamHQConfig` configuration class: `SamHQModel` (SamHQConfig model) - `SamHQVisionConfig` configuration class: `SamHQVisionModel` (SamHQVisionConfig model) - `SamVisionConfig` configuration class: `SamVisionModel` (SamVisionConfig model) - `SeamlessM4TConfig` configuration class: `SeamlessM4TModel` (SeamlessM4TConfig model) - `SeamlessM4Tv2Config` configuration class: `SeamlessM4Tv2Model` (SeamlessM4Tv2Config model) - `SeedOssConfig` configuration class: `SeedOssModel` (SeedOssConfig model) - `SegGptConfig` configuration class: `SegGptModel` (SegGptConfig model) - `SegformerConfig` configuration class: `SegformerModel` (SegformerConfig model) - `Siglip2Config` configuration class: `Siglip2Model` (Siglip2Config model) - `Siglip2VisionConfig` configuration class: `Siglip2VisionModel` (Siglip2VisionConfig model) - `SiglipConfig` configuration class: `SiglipModel` (SiglipConfig model) - `SiglipVisionConfig` configuration class: `SiglipVisionModel` (SiglipVisionConfig model) - `SmolLM3Config` configuration class: `SmolLM3Model` (SmolLM3Config model) - `SmolVLMConfig` configuration class: `SmolVLMModel` (SmolVLMConfig model) - `SmolVLMVisionConfig` configuration class: `SmolVLMVisionTransformer` (SmolVLMVisionConfig model) - `SolarOpenConfig` configuration class: `SolarOpenModel` (SolarOpenConfig model) - `Speech2TextConfig` configuration class: `Speech2TextModel` (Speech2TextConfig model) - `SpeechT5Config` configuration class: `SpeechT5Model` (SpeechT5Config model) - `SplinterConfig` configuration class: `SplinterModel` (SplinterConfig model) - `SqueezeBertConfig` configuration class: `SqueezeBertModel` (SqueezeBertConfig model) - `StableLmConfig` configuration class: `StableLmModel` (StableLmConfig model) - `Starcoder2Config` configuration class: `Starcoder2Model` (Starcoder2Config model) - `SwiftFormerConfig` configuration class: `SwiftFormerModel` (SwiftFormerConfig model) - `Swin2SRConfig` configuration class: `Swin2SRModel` (Swin2SRConfig model) - `SwinConfig` configuration class: `SwinModel` (SwinConfig model) - `Swinv2Config` configuration class: `Swinv2Model` (Swinv2Config model) - `SwitchTransformersConfig` configuration class: `SwitchTransformersModel` (SwitchTransformersConfig model) - `T5Config` configuration class: `T5Model` (T5Config model) - `T5Gemma2Config` configuration class: `T5Gemma2Model` (T5Gemma2Config model) - `T5Gemma2EncoderConfig` configuration class: `T5Gemma2Encoder` (T5Gemma2EncoderConfig model) - `T5GemmaConfig` configuration class: `T5GemmaModel` (T5GemmaConfig model) - `TableTransformerConfig` configuration class: `TableTransformerModel` (TableTransformerConfig model) - `TapasConfig` configuration class: `TapasModel` (TapasConfig model) - `TextNetConfig` configuration class: `TextNetModel` (TextNetConfig model) - `TimeSeriesTransformerConfig` configuration class: `TimeSeriesTransformerModel` (TimeSeriesTransformerConfig model) - `TimesFm2_5Config` configuration class: `TimesFm2_5Model` (TimesFm2_5Config model) - `TimesFmConfig` configuration class: `TimesFmModel` (TimesFmConfig model) - `TimesformerConfig` configuration class: `TimesformerModel` (TimesformerConfig model) - `TimmBackboneConfig` configuration class: `TimmBackbone` (TimmBackboneConfig model) - `TimmWrapperConfig` configuration class: `TimmWrapperModel` (TimmWrapperConfig model) - `TvpConfig` configuration class: `TvpModel` (TvpConfig model) - `UMT5Config` configuration class: `UMT5Model` (UMT5Config model) - `UVDocConfig` configuration class: `UVDocModel` (UVDocConfig model) - `UdopConfig` configuration class: `UdopModel` (UdopConfig model) - `UniSpeechConfig` configuration class: `UniSpeechModel` (UniSpeechConfig model) - `UniSpeechSatConfig` configuration class: `UniSpeechSatModel` (UniSpeechSatConfig model) - `UnivNetConfig` configuration class: `UnivNetModel` (UnivNetConfig model) - `VJEPA2Config` configuration class: `VJEPA2Model` (VJEPA2Config model) - `VaultGemmaConfig` configuration class: `VaultGemmaModel` (VaultGemmaConfig model) - `ViTConfig` configuration class: `ViTModel` (ViTConfig model) - `ViTMAEConfig` configuration class: `ViTMAEModel` (ViTMAEConfig model) - `ViTMSNConfig` configuration class: `ViTMSNModel` (ViTMSNConfig model) - `VibeVoiceAcousticTokenizerConfig` configuration class: `VibeVoiceAcousticTokenizerModel` (VibeVoiceAcousticTokenizerConfig model) - `VibeVoiceAcousticTokenizerDecoderConfig` configuration class: `VibeVoiceAcousticTokenizerDecoderModel` (VibeVoiceAcousticTokenizerDecoderConfig model) - `VibeVoiceAcousticTokenizerEncoderConfig` configuration class: `VibeVoiceAcousticTokenizerEncoderModel` (VibeVoiceAcousticTokenizerEncoderConfig model) - `VibeVoiceAsrConfig` configuration class: `VibeVoiceAsrForConditionalGeneration` (VibeVoiceAsrConfig model) - `VideoLlama3Config` configuration class: `VideoLlama3Model` (VideoLlama3Config model) - `VideoLlama3VisionConfig` configuration class: `VideoLlama3VisionModel` (VideoLlama3VisionConfig model) - `VideoLlavaConfig` configuration class: `VideoLlavaModel` (VideoLlavaConfig model) - `VideoMAEConfig` configuration class: `VideoMAEModel` (VideoMAEConfig model) - `ViltConfig` configuration class: `ViltModel` (ViltConfig model) - `VipLlavaConfig` configuration class: `VipLlavaModel` (VipLlavaConfig model) - `VisionTextDualEncoderConfig` configuration class: `VisionTextDualEncoderModel` (VisionTextDualEncoderConfig model) - `VisualBertConfig` configuration class: `VisualBertModel` (VisualBertConfig model) - `VitDetConfig` configuration class: `VitDetModel` (VitDetConfig model) - `VitsConfig` configuration class: `VitsModel` (VitsConfig model) - `VivitConfig` configuration class: `VivitModel` (VivitConfig model) - `VoxtralConfig` configuration class: `VoxtralForConditionalGeneration` (VoxtralConfig model) - `VoxtralEncoderConfig` configuration class: `VoxtralEncoder` (VoxtralEncoderConfig model) - `VoxtralRealtimeConfig` configuration class: `VoxtralRealtimeForConditionalGeneration` (VoxtralRealtimeConfig model) - `VoxtralRealtimeEncoderConfig` configuration class: `VoxtralRealtimeEncoder` (VoxtralRealtimeEncoderConfig model) - `VoxtralRealtimeTextConfig` configuration class: `VoxtralRealtimeTextModel` (VoxtralRealtimeTextConfig model) - `Wav2Vec2BertConfig` configuration class: `Wav2Vec2BertModel` (Wav2Vec2BertConfig model) - `Wav2Vec2Config` configuration class: `Wav2Vec2Model` (Wav2Vec2Config model) - `Wav2Vec2ConformerConfig` configuration class: `Wav2Vec2ConformerModel` (Wav2Vec2ConformerConfig model) - `WavLMConfig` configuration class: `WavLMModel` (WavLMConfig model) - `WhisperConfig` configuration class: `WhisperModel` (WhisperConfig model) - `XCLIPConfig` configuration class: `XCLIPModel` (XCLIPConfig model) - `XGLMConfig` configuration class: `XGLMModel` (XGLMConfig model) - `XLMConfig` configuration class: `XLMModel` (XLMConfig model) - `XLMRobertaConfig` configuration class: `XLMRobertaModel` (XLMRobertaConfig model) - `XLMRobertaXLConfig` configuration class: `XLMRobertaXLModel` (XLMRobertaXLConfig model) - `XLNetConfig` configuration class: `XLNetModel` (XLNetConfig model) - `XcodecConfig` configuration class: `XcodecModel` (XcodecConfig model) - `XmodConfig` configuration class: `XmodModel` (XmodConfig model) - `YolosConfig` configuration class: `YolosModel` (YolosConfig model) - `YosoConfig` configuration class: `YosoModel` (YosoConfig model) - `YoutuConfig` configuration class: `YoutuModel` (YoutuConfig model) - `Zamba2Config` configuration class: `Zamba2Model` (Zamba2Config model) - `ZambaConfig` configuration class: `ZambaModel` (ZambaConfig model) - `xLSTMConfig` configuration class: `xLSTMModel` (xLSTMConfig model) - **attn_implementation** (`str`, *optional*) -- The attention implementation to use in the model (if relevant). Can be any of `"eager"` (manual implementation of the attention), `"sdpa"` (using [`F.scaled_dot_product_attention`](https://pytorch.org/docs/master/generated/torch.nn.functional.scaled_dot_product_attention.html)), `"flash_attention_2"` (using [Dao-AILab/flash-attention](https://github.com/Dao-AILab/flash-attention)), or `"flash_attention_3"` (using [Dao-AILab/flash-attention/hopper](https://github.com/Dao-AILab/flash-attention/tree/main/hopper)). By default, if available, SDPA will be used for torch>=2.1.1. The default is otherwise the manual `"eager"` implementation.0 Instantiates one of the base model classes of the library from a configuration. Note: Loading a model from its configuration file does **not** load the model weights. It only affects the model's configuration. Use [from_pretrained()](/docs/transformers/v5.6.0/ja/model_doc/auto#transformers.AutoModel.from_pretrained) to load the model weights. Examples: ```python >>> from transformers import AutoConfig, AutoModel >>> # Download configuration from huggingface.co and cache. >>> config = AutoConfig.from_pretrained("google-bert/bert-base-cased") >>> model = AutoModel.from_config(config) ``` **Parameters:** config ([PreTrainedConfig](/docs/transformers/v5.6.0/ja/main_classes/configuration#transformers.PreTrainedConfig)) : The model class to instantiate is selected based on the configuration class: - [ASTConfig](/docs/transformers/v5.6.0/ja/model_doc/audio-spectrogram-transformer#transformers.ASTConfig) configuration class: [ASTModel](/docs/transformers/v5.6.0/ja/model_doc/audio-spectrogram-transformer#transformers.ASTModel) (ASTConfig model) - `AfmoeConfig` configuration class: `AfmoeModel` (AfmoeConfig model) - `Aimv2Config` configuration class: `Aimv2Model` (Aimv2Config model) - `Aimv2VisionConfig` configuration class: `Aimv2VisionModel` (Aimv2VisionConfig model) - [AlbertConfig](/docs/transformers/v5.6.0/ja/model_doc/albert#transformers.AlbertConfig) configuration class: [AlbertModel](/docs/transformers/v5.6.0/ja/model_doc/albert#transformers.AlbertModel) (AlbertConfig model) - [AlignConfig](/docs/transformers/v5.6.0/ja/model_doc/align#transformers.AlignConfig) configuration class: [AlignModel](/docs/transformers/v5.6.0/ja/model_doc/align#transformers.AlignModel) (AlignConfig model) - [AltCLIPConfig](/docs/transformers/v5.6.0/ja/model_doc/altclip#transformers.AltCLIPConfig) configuration class: [AltCLIPModel](/docs/transformers/v5.6.0/ja/model_doc/altclip#transformers.AltCLIPModel) (AltCLIPConfig model) - `ApertusConfig` configuration class: `ApertusModel` (ApertusConfig model) - `ArceeConfig` configuration class: `ArceeModel` (ArceeConfig model) - `AriaConfig` configuration class: `AriaModel` (AriaConfig model) - `AriaTextConfig` configuration class: `AriaTextModel` (AriaTextConfig model) - `AudioFlamingo3Config` configuration class: `AudioFlamingo3ForConditionalGeneration` (AudioFlamingo3Config model) - `AudioFlamingo3EncoderConfig` configuration class: `AudioFlamingo3Encoder` (AudioFlamingo3EncoderConfig model) - [AutoformerConfig](/docs/transformers/v5.6.0/ja/model_doc/autoformer#transformers.AutoformerConfig) configuration class: [AutoformerModel](/docs/transformers/v5.6.0/ja/model_doc/autoformer#transformers.AutoformerModel) (AutoformerConfig model) - `AyaVisionConfig` configuration class: `AyaVisionModel` (AyaVisionConfig model) - `BambaConfig` configuration class: `BambaModel` (BambaConfig model) - [BarkConfig](/docs/transformers/v5.6.0/ja/model_doc/bark#transformers.BarkConfig) configuration class: [BarkModel](/docs/transformers/v5.6.0/ja/model_doc/bark#transformers.BarkModel) (BarkConfig model) - [BartConfig](/docs/transformers/v5.6.0/ja/model_doc/bart#transformers.BartConfig) configuration class: [BartModel](/docs/transformers/v5.6.0/ja/model_doc/bart#transformers.BartModel) (BartConfig model) - [BeitConfig](/docs/transformers/v5.6.0/ja/model_doc/beit#transformers.BeitConfig) configuration class: [BeitModel](/docs/transformers/v5.6.0/ja/model_doc/beit#transformers.BeitModel) (BeitConfig model) - [BertConfig](/docs/transformers/v5.6.0/ja/model_doc/bert#transformers.BertConfig) configuration class: [BertModel](/docs/transformers/v5.6.0/ja/model_doc/bert#transformers.BertModel) (BertConfig model) - [BertGenerationConfig](/docs/transformers/v5.6.0/ja/model_doc/bert-generation#transformers.BertGenerationConfig) configuration class: [BertGenerationEncoder](/docs/transformers/v5.6.0/ja/model_doc/bert-generation#transformers.BertGenerationEncoder) (BertGenerationConfig model) - [BigBirdConfig](/docs/transformers/v5.6.0/ja/model_doc/big_bird#transformers.BigBirdConfig) configuration class: [BigBirdModel](/docs/transformers/v5.6.0/ja/model_doc/big_bird#transformers.BigBirdModel) (BigBirdConfig model) - [BigBirdPegasusConfig](/docs/transformers/v5.6.0/ja/model_doc/bigbird_pegasus#transformers.BigBirdPegasusConfig) configuration class: [BigBirdPegasusModel](/docs/transformers/v5.6.0/ja/model_doc/bigbird_pegasus#transformers.BigBirdPegasusModel) (BigBirdPegasusConfig model) - [BioGptConfig](/docs/transformers/v5.6.0/ja/model_doc/biogpt#transformers.BioGptConfig) configuration class: [BioGptModel](/docs/transformers/v5.6.0/ja/model_doc/biogpt#transformers.BioGptModel) (BioGptConfig model) - [BitConfig](/docs/transformers/v5.6.0/ja/model_doc/bit#transformers.BitConfig) configuration class: [BitModel](/docs/transformers/v5.6.0/ja/model_doc/bit#transformers.BitModel) (BitConfig model) - `BitNetConfig` configuration class: `BitNetModel` (BitNetConfig model) - [BlenderbotConfig](/docs/transformers/v5.6.0/ja/model_doc/blenderbot#transformers.BlenderbotConfig) configuration class: [BlenderbotModel](/docs/transformers/v5.6.0/ja/model_doc/blenderbot#transformers.BlenderbotModel) (BlenderbotConfig model) - [BlenderbotSmallConfig](/docs/transformers/v5.6.0/ja/model_doc/blenderbot-small#transformers.BlenderbotSmallConfig) configuration class: [BlenderbotSmallModel](/docs/transformers/v5.6.0/ja/model_doc/blenderbot-small#transformers.BlenderbotSmallModel) (BlenderbotSmallConfig model) - [Blip2Config](/docs/transformers/v5.6.0/ja/model_doc/blip-2#transformers.Blip2Config) configuration class: [Blip2Model](/docs/transformers/v5.6.0/ja/model_doc/blip-2#transformers.Blip2Model) (Blip2Config model) - [Blip2QFormerConfig](/docs/transformers/v5.6.0/ja/model_doc/blip-2#transformers.Blip2QFormerConfig) configuration class: [Blip2QFormerModel](/docs/transformers/v5.6.0/ja/model_doc/blip-2#transformers.Blip2QFormerModel) (Blip2QFormerConfig model) - [BlipConfig](/docs/transformers/v5.6.0/ja/model_doc/blip#transformers.BlipConfig) configuration class: [BlipModel](/docs/transformers/v5.6.0/ja/model_doc/blip#transformers.BlipModel) (BlipConfig model) - [BloomConfig](/docs/transformers/v5.6.0/ja/model_doc/bloom#transformers.BloomConfig) configuration class: [BloomModel](/docs/transformers/v5.6.0/ja/model_doc/bloom#transformers.BloomModel) (BloomConfig model) - `BltConfig` configuration class: `BltModel` (BltConfig model) - [BridgeTowerConfig](/docs/transformers/v5.6.0/ja/model_doc/bridgetower#transformers.BridgeTowerConfig) configuration class: [BridgeTowerModel](/docs/transformers/v5.6.0/ja/model_doc/bridgetower#transformers.BridgeTowerModel) (BridgeTowerConfig model) - [BrosConfig](/docs/transformers/v5.6.0/ja/model_doc/bros#transformers.BrosConfig) configuration class: [BrosModel](/docs/transformers/v5.6.0/ja/model_doc/bros#transformers.BrosModel) (BrosConfig model) - [CLIPConfig](/docs/transformers/v5.6.0/ja/model_doc/clip#transformers.CLIPConfig) configuration class: [CLIPModel](/docs/transformers/v5.6.0/ja/model_doc/clip#transformers.CLIPModel) (CLIPConfig model) - [CLIPSegConfig](/docs/transformers/v5.6.0/ja/model_doc/clipseg#transformers.CLIPSegConfig) configuration class: [CLIPSegModel](/docs/transformers/v5.6.0/ja/model_doc/clipseg#transformers.CLIPSegModel) (CLIPSegConfig model) - [CLIPTextConfig](/docs/transformers/v5.6.0/ja/model_doc/clip#transformers.CLIPTextConfig) configuration class: [CLIPTextModel](/docs/transformers/v5.6.0/ja/model_doc/clip#transformers.CLIPTextModel) (CLIPTextConfig model) - [CLIPVisionConfig](/docs/transformers/v5.6.0/ja/model_doc/clip#transformers.CLIPVisionConfig) configuration class: [CLIPVisionModel](/docs/transformers/v5.6.0/ja/model_doc/clip#transformers.CLIPVisionModel) (CLIPVisionConfig model) - [CTRLConfig](/docs/transformers/v5.6.0/ja/model_doc/ctrl#transformers.CTRLConfig) configuration class: [CTRLModel](/docs/transformers/v5.6.0/ja/model_doc/ctrl#transformers.CTRLModel) (CTRLConfig model) - [CamembertConfig](/docs/transformers/v5.6.0/ja/model_doc/camembert#transformers.CamembertConfig) configuration class: [CamembertModel](/docs/transformers/v5.6.0/ja/model_doc/camembert#transformers.CamembertModel) (CamembertConfig model) - [CanineConfig](/docs/transformers/v5.6.0/ja/model_doc/canine#transformers.CanineConfig) configuration class: [CanineModel](/docs/transformers/v5.6.0/ja/model_doc/canine#transformers.CanineModel) (CanineConfig model) - `ChameleonConfig` configuration class: `ChameleonModel` (ChameleonConfig model) - [ChineseCLIPConfig](/docs/transformers/v5.6.0/ja/model_doc/chinese_clip#transformers.ChineseCLIPConfig) configuration class: [ChineseCLIPModel](/docs/transformers/v5.6.0/ja/model_doc/chinese_clip#transformers.ChineseCLIPModel) (ChineseCLIPConfig model) - [ChineseCLIPVisionConfig](/docs/transformers/v5.6.0/ja/model_doc/chinese_clip#transformers.ChineseCLIPVisionConfig) configuration class: [ChineseCLIPVisionModel](/docs/transformers/v5.6.0/ja/model_doc/chinese_clip#transformers.ChineseCLIPVisionModel) (ChineseCLIPVisionConfig model) - [ClapConfig](/docs/transformers/v5.6.0/ja/model_doc/clap#transformers.ClapConfig) configuration class: [ClapModel](/docs/transformers/v5.6.0/ja/model_doc/clap#transformers.ClapModel) (ClapConfig model) - [ClvpConfig](/docs/transformers/v5.6.0/ja/model_doc/clvp#transformers.ClvpConfig) configuration class: [ClvpModelForConditionalGeneration](/docs/transformers/v5.6.0/ja/model_doc/clvp#transformers.ClvpModelForConditionalGeneration) (ClvpConfig model) - [CodeGenConfig](/docs/transformers/v5.6.0/ja/model_doc/codegen#transformers.CodeGenConfig) configuration class: [CodeGenModel](/docs/transformers/v5.6.0/ja/model_doc/codegen#transformers.CodeGenModel) (CodeGenConfig model) - `Cohere2Config` configuration class: `Cohere2Model` (Cohere2Config model) - `Cohere2VisionConfig` configuration class: `Cohere2VisionModel` (Cohere2VisionConfig model) - `CohereAsrConfig` configuration class: `CohereAsrModel` (CohereAsrConfig model) - `CohereConfig` configuration class: `CohereModel` (CohereConfig model) - [ConditionalDetrConfig](/docs/transformers/v5.6.0/ja/model_doc/conditional_detr#transformers.ConditionalDetrConfig) configuration class: [ConditionalDetrModel](/docs/transformers/v5.6.0/ja/model_doc/conditional_detr#transformers.ConditionalDetrModel) (ConditionalDetrConfig model) - [ConvBertConfig](/docs/transformers/v5.6.0/ja/model_doc/convbert#transformers.ConvBertConfig) configuration class: [ConvBertModel](/docs/transformers/v5.6.0/ja/model_doc/convbert#transformers.ConvBertModel) (ConvBertConfig model) - [ConvNextConfig](/docs/transformers/v5.6.0/ja/model_doc/convnext#transformers.ConvNextConfig) configuration class: [ConvNextModel](/docs/transformers/v5.6.0/ja/model_doc/convnext#transformers.ConvNextModel) (ConvNextConfig model) - [ConvNextV2Config](/docs/transformers/v5.6.0/ja/model_doc/convnextv2#transformers.ConvNextV2Config) configuration class: [ConvNextV2Model](/docs/transformers/v5.6.0/ja/model_doc/convnextv2#transformers.ConvNextV2Model) (ConvNextV2Config model) - [CpmAntConfig](/docs/transformers/v5.6.0/ja/model_doc/cpmant#transformers.CpmAntConfig) configuration class: [CpmAntModel](/docs/transformers/v5.6.0/ja/model_doc/cpmant#transformers.CpmAntModel) (CpmAntConfig model) - `CsmConfig` configuration class: `CsmForConditionalGeneration` (CsmConfig model) - [CvtConfig](/docs/transformers/v5.6.0/ja/model_doc/cvt#transformers.CvtConfig) configuration class: [CvtModel](/docs/transformers/v5.6.0/ja/model_doc/cvt#transformers.CvtModel) (CvtConfig model) - `CwmConfig` configuration class: `CwmModel` (CwmConfig model) - `DFineConfig` configuration class: `DFineModel` (DFineConfig model) - `DINOv3ConvNextConfig` configuration class: `DINOv3ConvNextModel` (DINOv3ConvNextConfig model) - `DINOv3ViTConfig` configuration class: `DINOv3ViTModel` (DINOv3ViTConfig model) - `DPRConfig` configuration class: `DPRQuestionEncoder` (DPRConfig model) - `DPTConfig` configuration class: `DPTModel` (DPTConfig model) - `DabDetrConfig` configuration class: `DabDetrModel` (DabDetrConfig model) - `DacConfig` configuration class: `DacModel` (DacConfig model) - [Data2VecAudioConfig](/docs/transformers/v5.6.0/ja/model_doc/data2vec#transformers.Data2VecAudioConfig) configuration class: [Data2VecAudioModel](/docs/transformers/v5.6.0/ja/model_doc/data2vec#transformers.Data2VecAudioModel) (Data2VecAudioConfig model) - [Data2VecTextConfig](/docs/transformers/v5.6.0/ja/model_doc/data2vec#transformers.Data2VecTextConfig) configuration class: [Data2VecTextModel](/docs/transformers/v5.6.0/ja/model_doc/data2vec#transformers.Data2VecTextModel) (Data2VecTextConfig model) - [Data2VecVisionConfig](/docs/transformers/v5.6.0/ja/model_doc/data2vec#transformers.Data2VecVisionConfig) configuration class: [Data2VecVisionModel](/docs/transformers/v5.6.0/ja/model_doc/data2vec#transformers.Data2VecVisionModel) (Data2VecVisionConfig model) - `DbrxConfig` configuration class: `DbrxModel` (DbrxConfig model) - [DebertaConfig](/docs/transformers/v5.6.0/ja/model_doc/deberta#transformers.DebertaConfig) configuration class: [DebertaModel](/docs/transformers/v5.6.0/ja/model_doc/deberta#transformers.DebertaModel) (DebertaConfig model) - [DebertaV2Config](/docs/transformers/v5.6.0/ja/model_doc/deberta-v2#transformers.DebertaV2Config) configuration class: [DebertaV2Model](/docs/transformers/v5.6.0/ja/model_doc/deberta-v2#transformers.DebertaV2Model) (DebertaV2Config model) - [DecisionTransformerConfig](/docs/transformers/v5.6.0/ja/model_doc/decision_transformer#transformers.DecisionTransformerConfig) configuration class: `DecisionTransformerModel` (DecisionTransformerConfig model) - `DeepseekV2Config` configuration class: `DeepseekV2Model` (DeepseekV2Config model) - `DeepseekV3Config` configuration class: `DeepseekV3Model` (DeepseekV3Config model) - `DeepseekVLConfig` configuration class: `DeepseekVLModel` (DeepseekVLConfig model) - `DeepseekVLHybridConfig` configuration class: `DeepseekVLHybridModel` (DeepseekVLHybridConfig model) - [DeformableDetrConfig](/docs/transformers/v5.6.0/ja/model_doc/deformable_detr#transformers.DeformableDetrConfig) configuration class: [DeformableDetrModel](/docs/transformers/v5.6.0/ja/model_doc/deformable_detr#transformers.DeformableDetrModel) (DeformableDetrConfig model) - [DeiTConfig](/docs/transformers/v5.6.0/ja/model_doc/deit#transformers.DeiTConfig) configuration class: [DeiTModel](/docs/transformers/v5.6.0/ja/model_doc/deit#transformers.DeiTModel) (DeiTConfig model) - `DepthProConfig` configuration class: `DepthProModel` (DepthProConfig model) - [DetrConfig](/docs/transformers/v5.6.0/ja/model_doc/detr#transformers.DetrConfig) configuration class: [DetrModel](/docs/transformers/v5.6.0/ja/model_doc/detr#transformers.DetrModel) (DetrConfig model) - `DiaConfig` configuration class: `DiaModel` (DiaConfig model) - `DiffLlamaConfig` configuration class: `DiffLlamaModel` (DiffLlamaConfig model) - [DinatConfig](/docs/transformers/v5.6.0/ja/model_doc/dinat#transformers.DinatConfig) configuration class: [DinatModel](/docs/transformers/v5.6.0/ja/model_doc/dinat#transformers.DinatModel) (DinatConfig model) - `Dinov2Config` configuration class: `Dinov2Model` (Dinov2Config model) - `Dinov2WithRegistersConfig` configuration class: `Dinov2WithRegistersModel` (Dinov2WithRegistersConfig model) - `DistilBertConfig` configuration class: `DistilBertModel` (DistilBertConfig model) - `DogeConfig` configuration class: `DogeModel` (DogeConfig model) - `DonutSwinConfig` configuration class: `DonutSwinModel` (DonutSwinConfig model) - `Dots1Config` configuration class: `Dots1Model` (Dots1Config model) - `EdgeTamConfig` configuration class: `EdgeTamModel` (EdgeTamConfig model) - `EdgeTamVideoConfig` configuration class: `EdgeTamVideoModel` (EdgeTamVideoConfig model) - `EdgeTamVisionConfig` configuration class: `EdgeTamVisionModel` (EdgeTamVisionConfig model) - `EfficientLoFTRConfig` configuration class: `EfficientLoFTRModel` (EfficientLoFTRConfig model) - `EfficientNetConfig` configuration class: `EfficientNetModel` (EfficientNetConfig model) - `ElectraConfig` configuration class: `ElectraModel` (ElectraConfig model) - `Emu3Config` configuration class: `Emu3Model` (Emu3Config model) - `EncodecConfig` configuration class: `EncodecModel` (EncodecConfig model) - `Ernie4_5Config` configuration class: `Ernie4_5Model` (Ernie4_5Config model) - `Ernie4_5_MoeConfig` configuration class: `Ernie4_5_MoeModel` (Ernie4_5_MoeConfig model) - `Ernie4_5_VLMoeConfig` configuration class: `Ernie4_5_VLMoeModel` (Ernie4_5_VLMoeConfig model) - `ErnieConfig` configuration class: `ErnieModel` (ErnieConfig model) - `EsmConfig` configuration class: `EsmModel` (EsmConfig model) - `EuroBertConfig` configuration class: `EuroBertModel` (EuroBertConfig model) - `EvollaConfig` configuration class: `EvollaModel` (EvollaConfig model) - `Exaone4Config` configuration class: `Exaone4Model` (Exaone4Config model) - `ExaoneMoeConfig` configuration class: `ExaoneMoeModel` (ExaoneMoeConfig model) - `FNetConfig` configuration class: `FNetModel` (FNetConfig model) - `FSMTConfig` configuration class: `FSMTModel` (FSMTConfig model) - `FalconConfig` configuration class: `FalconModel` (FalconConfig model) - `FalconH1Config` configuration class: `FalconH1Model` (FalconH1Config model) - `FalconMambaConfig` configuration class: `FalconMambaModel` (FalconMambaConfig model) - `FastSpeech2ConformerConfig` configuration class: `FastSpeech2ConformerModel` (FastSpeech2ConformerConfig model) - `FastSpeech2ConformerWithHifiGanConfig` configuration class: `FastSpeech2ConformerWithHifiGan` (FastSpeech2ConformerWithHifiGanConfig model) - `FastVlmConfig` configuration class: `FastVlmModel` (FastVlmConfig model) - `FlaubertConfig` configuration class: `FlaubertModel` (FlaubertConfig model) - `FlavaConfig` configuration class: `FlavaModel` (FlavaConfig model) - `FlexOlmoConfig` configuration class: `FlexOlmoModel` (FlexOlmoConfig model) - `Florence2Config` configuration class: `Florence2Model` (Florence2Config model) - `FocalNetConfig` configuration class: `FocalNetModel` (FocalNetConfig model) - `FunnelConfig` configuration class: `FunnelModel` or `FunnelBaseModel` (FunnelConfig model) - `FuyuConfig` configuration class: `FuyuModel` (FuyuConfig model) - `GLPNConfig` configuration class: `GLPNModel` (GLPNConfig model) - `GPT2Config` configuration class: `GPT2Model` (GPT2Config model) - `GPTBigCodeConfig` configuration class: `GPTBigCodeModel` (GPTBigCodeConfig model) - `GPTJConfig` configuration class: `GPTJModel` (GPTJConfig model) - `GPTNeoConfig` configuration class: `GPTNeoModel` (GPTNeoConfig model) - `GPTNeoXConfig` configuration class: `GPTNeoXModel` (GPTNeoXConfig model) - `GPTNeoXJapaneseConfig` configuration class: `GPTNeoXJapaneseModel` (GPTNeoXJapaneseConfig model) - `Gemma2Config` configuration class: `Gemma2Model` (Gemma2Config model) - `Gemma3Config` configuration class: `Gemma3Model` (Gemma3Config model) - `Gemma3TextConfig` configuration class: `Gemma3TextModel` (Gemma3TextConfig model) - `Gemma3nAudioConfig` configuration class: `Gemma3nAudioEncoder` (Gemma3nAudioConfig model) - `Gemma3nConfig` configuration class: `Gemma3nModel` (Gemma3nConfig model) - `Gemma3nTextConfig` configuration class: `Gemma3nTextModel` (Gemma3nTextConfig model) - `Gemma3nVisionConfig` configuration class: `TimmWrapperModel` (Gemma3nVisionConfig model) - `Gemma4AudioConfig` configuration class: `Gemma4AudioModel` (Gemma4AudioConfig model) - `Gemma4Config` configuration class: `Gemma4Model` (Gemma4Config model) - `Gemma4TextConfig` configuration class: `Gemma4TextModel` (Gemma4TextConfig model) - `Gemma4VisionConfig` configuration class: `Gemma4VisionModel` (Gemma4VisionConfig model) - `GemmaConfig` configuration class: `GemmaModel` (GemmaConfig model) - `GitConfig` configuration class: `GitModel` (GitConfig model) - `Glm46VConfig` configuration class: `Glm46VModel` (Glm46VConfig model) - `Glm4Config` configuration class: `Glm4Model` (Glm4Config model) - `Glm4MoeConfig` configuration class: `Glm4MoeModel` (Glm4MoeConfig model) - `Glm4MoeLiteConfig` configuration class: `Glm4MoeLiteModel` (Glm4MoeLiteConfig model) - `Glm4vConfig` configuration class: `Glm4vModel` (Glm4vConfig model) - `Glm4vMoeConfig` configuration class: `Glm4vMoeModel` (Glm4vMoeConfig model) - `Glm4vMoeTextConfig` configuration class: `Glm4vMoeTextModel` (Glm4vMoeTextConfig model) - `Glm4vMoeVisionConfig` configuration class: `Glm4vMoeVisionModel` (Glm4vMoeVisionConfig model) - `Glm4vTextConfig` configuration class: `Glm4vTextModel` (Glm4vTextConfig model) - `Glm4vVisionConfig` configuration class: `Glm4vVisionModel` (Glm4vVisionConfig model) - `GlmAsrConfig` configuration class: `GlmAsrForConditionalGeneration` (GlmAsrConfig model) - `GlmAsrEncoderConfig` configuration class: `GlmAsrEncoder` (GlmAsrEncoderConfig model) - `GlmConfig` configuration class: `GlmModel` (GlmConfig model) - `GlmImageConfig` configuration class: `GlmImageModel` (GlmImageConfig model) - `GlmImageTextConfig` configuration class: `GlmImageTextModel` (GlmImageTextConfig model) - `GlmImageVQVAEConfig` configuration class: `GlmImageVQVAE` (GlmImageVQVAEConfig model) - `GlmImageVisionConfig` configuration class: `GlmImageVisionModel` (GlmImageVisionConfig model) - `GlmMoeDsaConfig` configuration class: `GlmMoeDsaModel` (GlmMoeDsaConfig model) - `GlmOcrConfig` configuration class: `GlmOcrModel` (GlmOcrConfig model) - `GlmOcrTextConfig` configuration class: `GlmOcrTextModel` (GlmOcrTextConfig model) - `GlmOcrVisionConfig` configuration class: `GlmOcrVisionModel` (GlmOcrVisionConfig model) - `GotOcr2Config` configuration class: `GotOcr2Model` (GotOcr2Config model) - `GptOssConfig` configuration class: `GptOssModel` (GptOssConfig model) - `GraniteConfig` configuration class: `GraniteModel` (GraniteConfig model) - `GraniteMoeConfig` configuration class: `GraniteMoeModel` (GraniteMoeConfig model) - `GraniteMoeHybridConfig` configuration class: `GraniteMoeHybridModel` (GraniteMoeHybridConfig model) - `GraniteMoeSharedConfig` configuration class: `GraniteMoeSharedModel` (GraniteMoeSharedConfig model) - `GraniteSpeechConfig` configuration class: `GraniteSpeechForConditionalGeneration` (GraniteSpeechConfig model) - `GroundingDinoConfig` configuration class: `GroundingDinoModel` (GroundingDinoConfig model) - `GroupViTConfig` configuration class: `GroupViTModel` (GroupViTConfig model) - `HGNetV2Config` configuration class: `HGNetV2Backbone` (HGNetV2Config model) - `HYV3Config` configuration class: `HYV3Model` (HYV3Config model) - `HeliumConfig` configuration class: `HeliumModel` (HeliumConfig model) - `HieraConfig` configuration class: `HieraModel` (HieraConfig model) - `HiggsAudioV2Config` configuration class: `HiggsAudioV2ForConditionalGeneration` (HiggsAudioV2Config model) - `HiggsAudioV2TokenizerConfig` configuration class: `HiggsAudioV2TokenizerModel` (HiggsAudioV2TokenizerConfig model) - `HubertConfig` configuration class: `HubertModel` (HubertConfig model) - `HunYuanDenseV1Config` configuration class: `HunYuanDenseV1Model` (HunYuanDenseV1Config model) - `HunYuanMoEV1Config` configuration class: `HunYuanMoEV1Model` (HunYuanMoEV1Config model) - `IBertConfig` configuration class: `IBertModel` (IBertConfig model) - `IJepaConfig` configuration class: `IJepaModel` (IJepaConfig model) - `Idefics2Config` configuration class: `Idefics2Model` (Idefics2Config model) - `Idefics3Config` configuration class: `Idefics3Model` (Idefics3Config model) - `Idefics3VisionConfig` configuration class: `Idefics3VisionTransformer` (Idefics3VisionConfig model) - `IdeficsConfig` configuration class: `IdeficsModel` (IdeficsConfig model) - `ImageGPTConfig` configuration class: `ImageGPTModel` (ImageGPTConfig model) - `InformerConfig` configuration class: `InformerModel` (InformerConfig model) - `InstructBlipConfig` configuration class: `InstructBlipModel` (InstructBlipConfig model) - `InstructBlipVideoConfig` configuration class: `InstructBlipVideoModel` (InstructBlipVideoConfig model) - `InternVLConfig` configuration class: `InternVLModel` (InternVLConfig model) - `InternVLVisionConfig` configuration class: `InternVLVisionModel` (InternVLVisionConfig model) - `Jais2Config` configuration class: `Jais2Model` (Jais2Config model) - `JambaConfig` configuration class: `JambaModel` (JambaConfig model) - `JanusConfig` configuration class: `JanusModel` (JanusConfig model) - `JetMoeConfig` configuration class: `JetMoeModel` (JetMoeConfig model) - `JinaEmbeddingsV3Config` configuration class: `JinaEmbeddingsV3Model` (JinaEmbeddingsV3Config model) - `Kosmos2Config` configuration class: `Kosmos2Model` (Kosmos2Config model) - `Kosmos2_5Config` configuration class: `Kosmos2_5Model` (Kosmos2_5Config model) - `KyutaiSpeechToTextConfig` configuration class: `KyutaiSpeechToTextModel` (KyutaiSpeechToTextConfig model) - `LEDConfig` configuration class: `LEDModel` (LEDConfig model) - `LasrCTCConfig` configuration class: `LasrForCTC` (LasrCTCConfig model) - `LasrEncoderConfig` configuration class: `LasrEncoder` (LasrEncoderConfig model) - `LayoutLMConfig` configuration class: `LayoutLMModel` (LayoutLMConfig model) - `LayoutLMv2Config` configuration class: `LayoutLMv2Model` (LayoutLMv2Config model) - `LayoutLMv3Config` configuration class: `LayoutLMv3Model` (LayoutLMv3Config model) - `LevitConfig` configuration class: `LevitModel` (LevitConfig model) - `Lfm2Config` configuration class: `Lfm2Model` (Lfm2Config model) - `Lfm2MoeConfig` configuration class: `Lfm2MoeModel` (Lfm2MoeConfig model) - `Lfm2VlConfig` configuration class: `Lfm2VlModel` (Lfm2VlConfig model) - `LightGlueConfig` configuration class: `LightGlueForKeypointMatching` (LightGlueConfig model) - `LightOnOcrConfig` configuration class: `LightOnOcrModel` (LightOnOcrConfig model) - `LiltConfig` configuration class: `LiltModel` (LiltConfig model) - `Llama4Config` configuration class: `Llama4ForConditionalGeneration` (Llama4Config model) - `Llama4TextConfig` configuration class: `Llama4TextModel` (Llama4TextConfig model) - `LlamaConfig` configuration class: `LlamaModel` (LlamaConfig model) - `LlavaConfig` configuration class: `LlavaModel` (LlavaConfig model) - `LlavaNextConfig` configuration class: `LlavaNextModel` (LlavaNextConfig model) - `LlavaNextVideoConfig` configuration class: `LlavaNextVideoModel` (LlavaNextVideoConfig model) - `LlavaOnevisionConfig` configuration class: `LlavaOnevisionModel` (LlavaOnevisionConfig model) - `LongT5Config` configuration class: `LongT5Model` (LongT5Config model) - `LongcatFlashConfig` configuration class: `LongcatFlashModel` (LongcatFlashConfig model) - `LongformerConfig` configuration class: `LongformerModel` (LongformerConfig model) - `LukeConfig` configuration class: `LukeModel` (LukeConfig model) - `LwDetrConfig` configuration class: `LwDetrModel` (LwDetrConfig model) - `LxmertConfig` configuration class: `LxmertModel` (LxmertConfig model) - `M2M100Config` configuration class: `M2M100Model` (M2M100Config model) - `MBartConfig` configuration class: `MBartModel` (MBartConfig model) - `MLCDVisionConfig` configuration class: `MLCDVisionModel` (MLCDVisionConfig model) - `MMGroundingDinoConfig` configuration class: `MMGroundingDinoModel` (MMGroundingDinoConfig model) - `MPNetConfig` configuration class: `MPNetModel` (MPNetConfig model) - `MT5Config` configuration class: `MT5Model` (MT5Config model) - `Mamba2Config` configuration class: `Mamba2Model` (Mamba2Config model) - `MambaConfig` configuration class: `MambaModel` (MambaConfig model) - `MarianConfig` configuration class: `MarianModel` (MarianConfig model) - `MarkupLMConfig` configuration class: `MarkupLMModel` (MarkupLMConfig model) - `Mask2FormerConfig` configuration class: `Mask2FormerModel` (Mask2FormerConfig model) - `MaskFormerConfig` configuration class: `MaskFormerModel` (MaskFormerConfig model) - `MaskFormerSwinConfig` configuration class: `MaskFormerSwinModel` (MaskFormerSwinConfig model) - `MegatronBertConfig` configuration class: `MegatronBertModel` (MegatronBertConfig model) - `MetaClip2Config` configuration class: `MetaClip2Model` (MetaClip2Config model) - `MgpstrConfig` configuration class: `MgpstrForSceneTextRecognition` (MgpstrConfig model) - `MimiConfig` configuration class: `MimiModel` (MimiConfig model) - `MiniMaxConfig` configuration class: `MiniMaxModel` (MiniMaxConfig model) - `MiniMaxM2Config` configuration class: `MiniMaxM2Model` (MiniMaxM2Config model) - `Ministral3Config` configuration class: `Ministral3Model` (Ministral3Config model) - `MinistralConfig` configuration class: `MinistralModel` (MinistralConfig model) - `Mistral3Config` configuration class: `Mistral3Model` (Mistral3Config model) - `Mistral4Config` configuration class: `Mistral4Model` (Mistral4Config model) - `MistralConfig` configuration class: `MistralModel` (MistralConfig model) - `MixtralConfig` configuration class: `MixtralModel` (MixtralConfig model) - `MllamaConfig` configuration class: `MllamaModel` (MllamaConfig model) - `MobileBertConfig` configuration class: `MobileBertModel` (MobileBertConfig model) - `MobileNetV1Config` configuration class: `MobileNetV1Model` (MobileNetV1Config model) - `MobileNetV2Config` configuration class: `MobileNetV2Model` (MobileNetV2Config model) - `MobileViTConfig` configuration class: `MobileViTModel` (MobileViTConfig model) - `MobileViTV2Config` configuration class: `MobileViTV2Model` (MobileViTV2Config model) - `ModernBertConfig` configuration class: `ModernBertModel` (ModernBertConfig model) - `ModernBertDecoderConfig` configuration class: `ModernBertDecoderModel` (ModernBertDecoderConfig model) - `ModernVBertConfig` configuration class: `ModernVBertModel` (ModernVBertConfig model) - `MoonshineConfig` configuration class: `MoonshineModel` (MoonshineConfig model) - `MoonshineStreamingConfig` configuration class: `MoonshineStreamingModel` (MoonshineStreamingConfig model) - `MoshiConfig` configuration class: `MoshiModel` (MoshiConfig model) - `MptConfig` configuration class: `MptModel` (MptConfig model) - `MraConfig` configuration class: `MraModel` (MraConfig model) - `MusicFlamingoConfig` configuration class: `MusicFlamingoForConditionalGeneration` (MusicFlamingoConfig model) - `MusicgenConfig` configuration class: `MusicgenModel` (MusicgenConfig model) - `MusicgenMelodyConfig` configuration class: `MusicgenMelodyModel` (MusicgenMelodyConfig model) - `MvpConfig` configuration class: `MvpModel` (MvpConfig model) - `NanoChatConfig` configuration class: `NanoChatModel` (NanoChatConfig model) - `NemotronConfig` configuration class: `NemotronModel` (NemotronConfig model) - `NemotronHConfig` configuration class: `NemotronHModel` (NemotronHConfig model) - `NllbMoeConfig` configuration class: `NllbMoeModel` (NllbMoeConfig model) - `NomicBertConfig` configuration class: `NomicBertModel` (NomicBertConfig model) - `NystromformerConfig` configuration class: `NystromformerModel` (NystromformerConfig model) - `OPTConfig` configuration class: `OPTModel` (OPTConfig model) - `Olmo2Config` configuration class: `Olmo2Model` (Olmo2Config model) - `Olmo3Config` configuration class: `Olmo3Model` (Olmo3Config model) - `OlmoConfig` configuration class: `OlmoModel` (OlmoConfig model) - `OlmoHybridConfig` configuration class: `OlmoHybridModel` (OlmoHybridConfig model) - `OlmoeConfig` configuration class: `OlmoeModel` (OlmoeConfig model) - `OmDetTurboConfig` configuration class: `OmDetTurboForObjectDetection` (OmDetTurboConfig model) - `OneFormerConfig` configuration class: `OneFormerModel` (OneFormerConfig model) - `OpenAIGPTConfig` configuration class: `OpenAIGPTModel` (OpenAIGPTConfig model) - `OpenAIPrivacyFilterConfig` configuration class: `OpenAIPrivacyFilterModel` (OpenAIPrivacyFilterConfig model) - `Ovis2Config` configuration class: `Ovis2Model` (Ovis2Config model) - `OwlViTConfig` configuration class: `OwlViTModel` (OwlViTConfig model) - `Owlv2Config` configuration class: `Owlv2Model` (Owlv2Config model) - `PI0Config` configuration class: `PI0Model` (PI0Config model) - `PLBartConfig` configuration class: `PLBartModel` (PLBartConfig model) - `PPDocLayoutV3Config` configuration class: `PPDocLayoutV3Model` (PPDocLayoutV3Config model) - `PPOCRV5MobileRecConfig` configuration class: `PPOCRV5MobileRecModel` (PPOCRV5MobileRecConfig model) - `PPOCRV5ServerRecConfig` configuration class: `PPOCRV5ServerRecModel` (PPOCRV5ServerRecConfig model) - `PaliGemmaConfig` configuration class: `PaliGemmaModel` (PaliGemmaConfig model) - `ParakeetCTCConfig` configuration class: `ParakeetForCTC` (ParakeetCTCConfig model) - `ParakeetEncoderConfig` configuration class: `ParakeetEncoder` (ParakeetEncoderConfig model) - `PatchTSMixerConfig` configuration class: `PatchTSMixerModel` (PatchTSMixerConfig model) - `PatchTSTConfig` configuration class: `PatchTSTModel` (PatchTSTConfig model) - `PeAudioConfig` configuration class: `PeAudioModel` (PeAudioConfig model) - `PeAudioEncoderConfig` configuration class: `PeAudioEncoder` (PeAudioEncoderConfig model) - `PeAudioVideoConfig` configuration class: `PeAudioVideoModel` (PeAudioVideoConfig model) - `PeAudioVideoEncoderConfig` configuration class: `PeAudioVideoEncoder` (PeAudioVideoEncoderConfig model) - `PeVideoConfig` configuration class: `PeVideoModel` (PeVideoConfig model) - `PeVideoEncoderConfig` configuration class: `PeVideoEncoder` (PeVideoEncoderConfig model) - `PegasusConfig` configuration class: `PegasusModel` (PegasusConfig model) - `PegasusXConfig` configuration class: `PegasusXModel` (PegasusXConfig model) - `PerceiverConfig` configuration class: `PerceiverModel` (PerceiverConfig model) - `PerceptionLMConfig` configuration class: `PerceptionLMModel` (PerceptionLMConfig model) - `PersimmonConfig` configuration class: `PersimmonModel` (PersimmonConfig model) - `Phi3Config` configuration class: `Phi3Model` (Phi3Config model) - `Phi4MultimodalConfig` configuration class: `Phi4MultimodalModel` (Phi4MultimodalConfig model) - `PhiConfig` configuration class: `PhiModel` (PhiConfig model) - `PhimoeConfig` configuration class: `PhimoeModel` (PhimoeConfig model) - `PixioConfig` configuration class: `PixioModel` (PixioConfig model) - `PixtralVisionConfig` configuration class: `PixtralVisionModel` (PixtralVisionConfig model) - `PoolFormerConfig` configuration class: `PoolFormerModel` (PoolFormerConfig model) - `ProphetNetConfig` configuration class: `ProphetNetModel` (ProphetNetConfig model) - `PvtConfig` configuration class: `PvtModel` (PvtConfig model) - `PvtV2Config` configuration class: `PvtV2Model` (PvtV2Config model) - `QianfanOCRConfig` configuration class: `QianfanOCRModel` (QianfanOCRConfig model) - `QianfanOCRVisionConfig` configuration class: `QianfanOCRVisionModel` (QianfanOCRVisionConfig model) - `Qwen2AudioEncoderConfig` configuration class: `Qwen2AudioEncoder` (Qwen2AudioEncoderConfig model) - `Qwen2Config` configuration class: `Qwen2Model` (Qwen2Config model) - `Qwen2MoeConfig` configuration class: `Qwen2MoeModel` (Qwen2MoeConfig model) - `Qwen2VLConfig` configuration class: `Qwen2VLModel` (Qwen2VLConfig model) - `Qwen2VLTextConfig` configuration class: `Qwen2VLTextModel` (Qwen2VLTextConfig model) - `Qwen2_5_VLConfig` configuration class: `Qwen2_5_VLModel` (Qwen2_5_VLConfig model) - `Qwen2_5_VLTextConfig` configuration class: `Qwen2_5_VLTextModel` (Qwen2_5_VLTextConfig model) - `Qwen3Config` configuration class: `Qwen3Model` (Qwen3Config model) - `Qwen3MoeConfig` configuration class: `Qwen3MoeModel` (Qwen3MoeConfig model) - `Qwen3NextConfig` configuration class: `Qwen3NextModel` (Qwen3NextConfig model) - `Qwen3VLConfig` configuration class: `Qwen3VLModel` (Qwen3VLConfig model) - `Qwen3VLMoeConfig` configuration class: `Qwen3VLMoeModel` (Qwen3VLMoeConfig model) - `Qwen3VLMoeTextConfig` configuration class: `Qwen3VLMoeTextModel` (Qwen3VLMoeTextConfig model) - `Qwen3VLTextConfig` configuration class: `Qwen3VLTextModel` (Qwen3VLTextConfig model) - `Qwen3_5Config` configuration class: `Qwen3_5Model` (Qwen3_5Config model) - `Qwen3_5MoeConfig` configuration class: `Qwen3_5MoeModel` (Qwen3_5MoeConfig model) - `Qwen3_5MoeTextConfig` configuration class: `Qwen3_5MoeTextModel` (Qwen3_5MoeTextConfig model) - `Qwen3_5TextConfig` configuration class: `Qwen3_5TextModel` (Qwen3_5TextConfig model) - `RTDetrConfig` configuration class: `RTDetrModel` (RTDetrConfig model) - `RTDetrV2Config` configuration class: `RTDetrV2Model` (RTDetrV2Config model) - `RecurrentGemmaConfig` configuration class: `RecurrentGemmaModel` (RecurrentGemmaConfig model) - `ReformerConfig` configuration class: `ReformerModel` (ReformerConfig model) - `RegNetConfig` configuration class: `RegNetModel` (RegNetConfig model) - `RemBertConfig` configuration class: `RemBertModel` (RemBertConfig model) - `ResNetConfig` configuration class: `ResNetModel` (ResNetConfig model) - `RoCBertConfig` configuration class: `RoCBertModel` (RoCBertConfig model) - `RoFormerConfig` configuration class: `RoFormerModel` (RoFormerConfig model) - `RobertaConfig` configuration class: `RobertaModel` (RobertaConfig model) - `RobertaPreLayerNormConfig` configuration class: `RobertaPreLayerNormModel` (RobertaPreLayerNormConfig model) - `RwkvConfig` configuration class: `RwkvModel` (RwkvConfig model) - `SEWConfig` configuration class: `SEWModel` (SEWConfig model) - `SEWDConfig` configuration class: `SEWDModel` (SEWDConfig model) - `Sam2Config` configuration class: `Sam2Model` (Sam2Config model) - `Sam2HieraDetConfig` configuration class: `Sam2HieraDetModel` (Sam2HieraDetConfig model) - `Sam2VideoConfig` configuration class: `Sam2VideoModel` (Sam2VideoConfig model) - `Sam2VisionConfig` configuration class: `Sam2VisionModel` (Sam2VisionConfig model) - `Sam3Config` configuration class: `Sam3Model` (Sam3Config model) - `Sam3LiteTextConfig` configuration class: `Sam3LiteTextModel` (Sam3LiteTextConfig model) - `Sam3LiteTextTextConfig` configuration class: `Sam3LiteTextTextModel` (Sam3LiteTextTextConfig model) - `Sam3TrackerConfig` configuration class: `Sam3TrackerModel` (Sam3TrackerConfig model) - `Sam3TrackerVideoConfig` configuration class: `Sam3TrackerVideoModel` (Sam3TrackerVideoConfig model) - `Sam3ViTConfig` configuration class: `Sam3ViTModel` (Sam3ViTConfig model) - `Sam3VideoConfig` configuration class: `Sam3VideoModel` (Sam3VideoConfig model) - `Sam3VisionConfig` configuration class: `Sam3VisionModel` (Sam3VisionConfig model) - `SamConfig` configuration class: `SamModel` (SamConfig model) - `SamHQConfig` configuration class: `SamHQModel` (SamHQConfig model) - `SamHQVisionConfig` configuration class: `SamHQVisionModel` (SamHQVisionConfig model) - `SamVisionConfig` configuration class: `SamVisionModel` (SamVisionConfig model) - `SeamlessM4TConfig` configuration class: `SeamlessM4TModel` (SeamlessM4TConfig model) - `SeamlessM4Tv2Config` configuration class: `SeamlessM4Tv2Model` (SeamlessM4Tv2Config model) - `SeedOssConfig` configuration class: `SeedOssModel` (SeedOssConfig model) - `SegGptConfig` configuration class: `SegGptModel` (SegGptConfig model) - `SegformerConfig` configuration class: `SegformerModel` (SegformerConfig model) - `Siglip2Config` configuration class: `Siglip2Model` (Siglip2Config model) - `Siglip2VisionConfig` configuration class: `Siglip2VisionModel` (Siglip2VisionConfig model) - `SiglipConfig` configuration class: `SiglipModel` (SiglipConfig model) - `SiglipVisionConfig` configuration class: `SiglipVisionModel` (SiglipVisionConfig model) - `SmolLM3Config` configuration class: `SmolLM3Model` (SmolLM3Config model) - `SmolVLMConfig` configuration class: `SmolVLMModel` (SmolVLMConfig model) - `SmolVLMVisionConfig` configuration class: `SmolVLMVisionTransformer` (SmolVLMVisionConfig model) - `SolarOpenConfig` configuration class: `SolarOpenModel` (SolarOpenConfig model) - `Speech2TextConfig` configuration class: `Speech2TextModel` (Speech2TextConfig model) - `SpeechT5Config` configuration class: `SpeechT5Model` (SpeechT5Config model) - `SplinterConfig` configuration class: `SplinterModel` (SplinterConfig model) - `SqueezeBertConfig` configuration class: `SqueezeBertModel` (SqueezeBertConfig model) - `StableLmConfig` configuration class: `StableLmModel` (StableLmConfig model) - `Starcoder2Config` configuration class: `Starcoder2Model` (Starcoder2Config model) - `SwiftFormerConfig` configuration class: `SwiftFormerModel` (SwiftFormerConfig model) - `Swin2SRConfig` configuration class: `Swin2SRModel` (Swin2SRConfig model) - `SwinConfig` configuration class: `SwinModel` (SwinConfig model) - `Swinv2Config` configuration class: `Swinv2Model` (Swinv2Config model) - `SwitchTransformersConfig` configuration class: `SwitchTransformersModel` (SwitchTransformersConfig model) - `T5Config` configuration class: `T5Model` (T5Config model) - `T5Gemma2Config` configuration class: `T5Gemma2Model` (T5Gemma2Config model) - `T5Gemma2EncoderConfig` configuration class: `T5Gemma2Encoder` (T5Gemma2EncoderConfig model) - `T5GemmaConfig` configuration class: `T5GemmaModel` (T5GemmaConfig model) - `TableTransformerConfig` configuration class: `TableTransformerModel` (TableTransformerConfig model) - `TapasConfig` configuration class: `TapasModel` (TapasConfig model) - `TextNetConfig` configuration class: `TextNetModel` (TextNetConfig model) - `TimeSeriesTransformerConfig` configuration class: `TimeSeriesTransformerModel` (TimeSeriesTransformerConfig model) - `TimesFm2_5Config` configuration class: `TimesFm2_5Model` (TimesFm2_5Config model) - `TimesFmConfig` configuration class: `TimesFmModel` (TimesFmConfig model) - `TimesformerConfig` configuration class: `TimesformerModel` (TimesformerConfig model) - `TimmBackboneConfig` configuration class: `TimmBackbone` (TimmBackboneConfig model) - `TimmWrapperConfig` configuration class: `TimmWrapperModel` (TimmWrapperConfig model) - `TvpConfig` configuration class: `TvpModel` (TvpConfig model) - `UMT5Config` configuration class: `UMT5Model` (UMT5Config model) - `UVDocConfig` configuration class: `UVDocModel` (UVDocConfig model) - `UdopConfig` configuration class: `UdopModel` (UdopConfig model) - `UniSpeechConfig` configuration class: `UniSpeechModel` (UniSpeechConfig model) - `UniSpeechSatConfig` configuration class: `UniSpeechSatModel` (UniSpeechSatConfig model) - `UnivNetConfig` configuration class: `UnivNetModel` (UnivNetConfig model) - `VJEPA2Config` configuration class: `VJEPA2Model` (VJEPA2Config model) - `VaultGemmaConfig` configuration class: `VaultGemmaModel` (VaultGemmaConfig model) - `ViTConfig` configuration class: `ViTModel` (ViTConfig model) - `ViTMAEConfig` configuration class: `ViTMAEModel` (ViTMAEConfig model) - `ViTMSNConfig` configuration class: `ViTMSNModel` (ViTMSNConfig model) - `VibeVoiceAcousticTokenizerConfig` configuration class: `VibeVoiceAcousticTokenizerModel` (VibeVoiceAcousticTokenizerConfig model) - `VibeVoiceAcousticTokenizerDecoderConfig` configuration class: `VibeVoiceAcousticTokenizerDecoderModel` (VibeVoiceAcousticTokenizerDecoderConfig model) - `VibeVoiceAcousticTokenizerEncoderConfig` configuration class: `VibeVoiceAcousticTokenizerEncoderModel` (VibeVoiceAcousticTokenizerEncoderConfig model) - `VibeVoiceAsrConfig` configuration class: `VibeVoiceAsrForConditionalGeneration` (VibeVoiceAsrConfig model) - `VideoLlama3Config` configuration class: `VideoLlama3Model` (VideoLlama3Config model) - `VideoLlama3VisionConfig` configuration class: `VideoLlama3VisionModel` (VideoLlama3VisionConfig model) - `VideoLlavaConfig` configuration class: `VideoLlavaModel` (VideoLlavaConfig model) - `VideoMAEConfig` configuration class: `VideoMAEModel` (VideoMAEConfig model) - `ViltConfig` configuration class: `ViltModel` (ViltConfig model) - `VipLlavaConfig` configuration class: `VipLlavaModel` (VipLlavaConfig model) - `VisionTextDualEncoderConfig` configuration class: `VisionTextDualEncoderModel` (VisionTextDualEncoderConfig model) - `VisualBertConfig` configuration class: `VisualBertModel` (VisualBertConfig model) - `VitDetConfig` configuration class: `VitDetModel` (VitDetConfig model) - `VitsConfig` configuration class: `VitsModel` (VitsConfig model) - `VivitConfig` configuration class: `VivitModel` (VivitConfig model) - `VoxtralConfig` configuration class: `VoxtralForConditionalGeneration` (VoxtralConfig model) - `VoxtralEncoderConfig` configuration class: `VoxtralEncoder` (VoxtralEncoderConfig model) - `VoxtralRealtimeConfig` configuration class: `VoxtralRealtimeForConditionalGeneration` (VoxtralRealtimeConfig model) - `VoxtralRealtimeEncoderConfig` configuration class: `VoxtralRealtimeEncoder` (VoxtralRealtimeEncoderConfig model) - `VoxtralRealtimeTextConfig` configuration class: `VoxtralRealtimeTextModel` (VoxtralRealtimeTextConfig model) - `Wav2Vec2BertConfig` configuration class: `Wav2Vec2BertModel` (Wav2Vec2BertConfig model) - `Wav2Vec2Config` configuration class: `Wav2Vec2Model` (Wav2Vec2Config model) - `Wav2Vec2ConformerConfig` configuration class: `Wav2Vec2ConformerModel` (Wav2Vec2ConformerConfig model) - `WavLMConfig` configuration class: `WavLMModel` (WavLMConfig model) - `WhisperConfig` configuration class: `WhisperModel` (WhisperConfig model) - `XCLIPConfig` configuration class: `XCLIPModel` (XCLIPConfig model) - `XGLMConfig` configuration class: `XGLMModel` (XGLMConfig model) - `XLMConfig` configuration class: `XLMModel` (XLMConfig model) - `XLMRobertaConfig` configuration class: `XLMRobertaModel` (XLMRobertaConfig model) - `XLMRobertaXLConfig` configuration class: `XLMRobertaXLModel` (XLMRobertaXLConfig model) - `XLNetConfig` configuration class: `XLNetModel` (XLNetConfig model) - `XcodecConfig` configuration class: `XcodecModel` (XcodecConfig model) - `XmodConfig` configuration class: `XmodModel` (XmodConfig model) - `YolosConfig` configuration class: `YolosModel` (YolosConfig model) - `YosoConfig` configuration class: `YosoModel` (YosoConfig model) - `YoutuConfig` configuration class: `YoutuModel` (YoutuConfig model) - `Zamba2Config` configuration class: `Zamba2Model` (Zamba2Config model) - `ZambaConfig` configuration class: `ZambaModel` (ZambaConfig model) - `xLSTMConfig` configuration class: `xLSTMModel` (xLSTMConfig model) attn_implementation (`str`, *optional*) : The attention implementation to use in the model (if relevant). Can be any of `"eager"` (manual implementation of the attention), `"sdpa"` (using [`F.scaled_dot_product_attention`](https://pytorch.org/docs/master/generated/torch.nn.functional.scaled_dot_product_attention.html)), `"flash_attention_2"` (using [Dao-AILab/flash-attention](https://github.com/Dao-AILab/flash-attention)), or `"flash_attention_3"` (using [Dao-AILab/flash-attention/hopper](https://github.com/Dao-AILab/flash-attention/tree/main/hopper)). By default, if available, SDPA will be used for torch>=2.1.1. The default is otherwise the manual `"eager"` implementation. #### from_pretrained[[transformers.AutoModel.from_pretrained]] [Source](https://github.com/huggingface/transformers/blob/v5.6.0/src/transformers/models/auto/auto_factory.py#L253) Instantiate one of the base model classes of the library from a pretrained model. The model class to instantiate is selected based on the `model_type` property of the config object (either passed as an argument or loaded from `pretrained_model_name_or_path` if possible), or when it's missing, by falling back to using pattern matching on `pretrained_model_name_or_path`: - **afmoe** -- `AfmoeModel` (AfmoeConfig model) - **aimv2** -- `Aimv2Model` (Aimv2Config model) - **aimv2_vision_model** -- `Aimv2VisionModel` (Aimv2VisionConfig model) - **albert** -- [AlbertModel](/docs/transformers/v5.6.0/ja/model_doc/albert#transformers.AlbertModel) (AlbertConfig model) - **align** -- [AlignModel](/docs/transformers/v5.6.0/ja/model_doc/align#transformers.AlignModel) (AlignConfig model) - **altclip** -- [AltCLIPModel](/docs/transformers/v5.6.0/ja/model_doc/altclip#transformers.AltCLIPModel) (AltCLIPConfig model) - **apertus** -- `ApertusModel` (ApertusConfig model) - **arcee** -- `ArceeModel` (ArceeConfig model) - **aria** -- `AriaModel` (AriaConfig model) - **aria_text** -- `AriaTextModel` (AriaTextConfig model) - **audio-spectrogram-transformer** -- [ASTModel](/docs/transformers/v5.6.0/ja/model_doc/audio-spectrogram-transformer#transformers.ASTModel) (ASTConfig model) - **audioflamingo3** -- `AudioFlamingo3ForConditionalGeneration` (AudioFlamingo3Config model) - **audioflamingo3_encoder** -- `AudioFlamingo3Encoder` (AudioFlamingo3EncoderConfig model) - **autoformer** -- [AutoformerModel](/docs/transformers/v5.6.0/ja/model_doc/autoformer#transformers.AutoformerModel) (AutoformerConfig model) - **aya_vision** -- `AyaVisionModel` (AyaVisionConfig model) - **bamba** -- `BambaModel` (BambaConfig model) - **bark** -- [BarkModel](/docs/transformers/v5.6.0/ja/model_doc/bark#transformers.BarkModel) (BarkConfig model) - **bart** -- [BartModel](/docs/transformers/v5.6.0/ja/model_doc/bart#transformers.BartModel) (BartConfig model) - **beit** -- [BeitModel](/docs/transformers/v5.6.0/ja/model_doc/beit#transformers.BeitModel) (BeitConfig model) - **bert** -- [BertModel](/docs/transformers/v5.6.0/ja/model_doc/bert#transformers.BertModel) (BertConfig model) - **bert-generation** -- [BertGenerationEncoder](/docs/transformers/v5.6.0/ja/model_doc/bert-generation#transformers.BertGenerationEncoder) (BertGenerationConfig model) - **big_bird** -- [BigBirdModel](/docs/transformers/v5.6.0/ja/model_doc/big_bird#transformers.BigBirdModel) (BigBirdConfig model) - **bigbird_pegasus** -- [BigBirdPegasusModel](/docs/transformers/v5.6.0/ja/model_doc/bigbird_pegasus#transformers.BigBirdPegasusModel) (BigBirdPegasusConfig model) - **biogpt** -- [BioGptModel](/docs/transformers/v5.6.0/ja/model_doc/biogpt#transformers.BioGptModel) (BioGptConfig model) - **bit** -- [BitModel](/docs/transformers/v5.6.0/ja/model_doc/bit#transformers.BitModel) (BitConfig model) - **bitnet** -- `BitNetModel` (BitNetConfig model) - **blenderbot** -- [BlenderbotModel](/docs/transformers/v5.6.0/ja/model_doc/blenderbot#transformers.BlenderbotModel) (BlenderbotConfig model) - **blenderbot-small** -- [BlenderbotSmallModel](/docs/transformers/v5.6.0/ja/model_doc/blenderbot-small#transformers.BlenderbotSmallModel) (BlenderbotSmallConfig model) - **blip** -- [BlipModel](/docs/transformers/v5.6.0/ja/model_doc/blip#transformers.BlipModel) (BlipConfig model) - **blip-2** -- [Blip2Model](/docs/transformers/v5.6.0/ja/model_doc/blip-2#transformers.Blip2Model) (Blip2Config model) - **blip_2_qformer** -- [Blip2QFormerModel](/docs/transformers/v5.6.0/ja/model_doc/blip-2#transformers.Blip2QFormerModel) (Blip2QFormerConfig model) - **bloom** -- [BloomModel](/docs/transformers/v5.6.0/ja/model_doc/bloom#transformers.BloomModel) (BloomConfig model) - **blt** -- `BltModel` (BltConfig model) - **bridgetower** -- [BridgeTowerModel](/docs/transformers/v5.6.0/ja/model_doc/bridgetower#transformers.BridgeTowerModel) (BridgeTowerConfig model) - **bros** -- [BrosModel](/docs/transformers/v5.6.0/ja/model_doc/bros#transformers.BrosModel) (BrosConfig model) - **camembert** -- [CamembertModel](/docs/transformers/v5.6.0/ja/model_doc/camembert#transformers.CamembertModel) (CamembertConfig model) - **canine** -- [CanineModel](/docs/transformers/v5.6.0/ja/model_doc/canine#transformers.CanineModel) (CanineConfig model) - **chameleon** -- `ChameleonModel` (ChameleonConfig model) - **chinese_clip** -- [ChineseCLIPModel](/docs/transformers/v5.6.0/ja/model_doc/chinese_clip#transformers.ChineseCLIPModel) (ChineseCLIPConfig model) - **chinese_clip_vision_model** -- [ChineseCLIPVisionModel](/docs/transformers/v5.6.0/ja/model_doc/chinese_clip#transformers.ChineseCLIPVisionModel) (ChineseCLIPVisionConfig model) - **clap** -- [ClapModel](/docs/transformers/v5.6.0/ja/model_doc/clap#transformers.ClapModel) (ClapConfig model) - **clip** -- [CLIPModel](/docs/transformers/v5.6.0/ja/model_doc/clip#transformers.CLIPModel) (CLIPConfig model) - **clip_text_model** -- [CLIPTextModel](/docs/transformers/v5.6.0/ja/model_doc/clip#transformers.CLIPTextModel) (CLIPTextConfig model) - **clip_vision_model** -- [CLIPVisionModel](/docs/transformers/v5.6.0/ja/model_doc/clip#transformers.CLIPVisionModel) (CLIPVisionConfig model) - **clipseg** -- [CLIPSegModel](/docs/transformers/v5.6.0/ja/model_doc/clipseg#transformers.CLIPSegModel) (CLIPSegConfig model) - **clvp** -- [ClvpModelForConditionalGeneration](/docs/transformers/v5.6.0/ja/model_doc/clvp#transformers.ClvpModelForConditionalGeneration) (ClvpConfig model) - **codegen** -- [CodeGenModel](/docs/transformers/v5.6.0/ja/model_doc/codegen#transformers.CodeGenModel) (CodeGenConfig model) - **cohere** -- `CohereModel` (CohereConfig model) - **cohere2** -- `Cohere2Model` (Cohere2Config model) - **cohere2_vision** -- `Cohere2VisionModel` (Cohere2VisionConfig model) - **cohere_asr** -- `CohereAsrModel` (CohereAsrConfig model) - **conditional_detr** -- [ConditionalDetrModel](/docs/transformers/v5.6.0/ja/model_doc/conditional_detr#transformers.ConditionalDetrModel) (ConditionalDetrConfig model) - **convbert** -- [ConvBertModel](/docs/transformers/v5.6.0/ja/model_doc/convbert#transformers.ConvBertModel) (ConvBertConfig model) - **convnext** -- [ConvNextModel](/docs/transformers/v5.6.0/ja/model_doc/convnext#transformers.ConvNextModel) (ConvNextConfig model) - **convnextv2** -- [ConvNextV2Model](/docs/transformers/v5.6.0/ja/model_doc/convnextv2#transformers.ConvNextV2Model) (ConvNextV2Config model) - **cpmant** -- [CpmAntModel](/docs/transformers/v5.6.0/ja/model_doc/cpmant#transformers.CpmAntModel) (CpmAntConfig model) - **csm** -- `CsmForConditionalGeneration` (CsmConfig model) - **ctrl** -- [CTRLModel](/docs/transformers/v5.6.0/ja/model_doc/ctrl#transformers.CTRLModel) (CTRLConfig model) - **cvt** -- [CvtModel](/docs/transformers/v5.6.0/ja/model_doc/cvt#transformers.CvtModel) (CvtConfig model) - **cwm** -- `CwmModel` (CwmConfig model) - **d_fine** -- `DFineModel` (DFineConfig model) - **dab-detr** -- `DabDetrModel` (DabDetrConfig model) - **dac** -- `DacModel` (DacConfig model) - **data2vec-audio** -- [Data2VecAudioModel](/docs/transformers/v5.6.0/ja/model_doc/data2vec#transformers.Data2VecAudioModel) (Data2VecAudioConfig model) - **data2vec-text** -- [Data2VecTextModel](/docs/transformers/v5.6.0/ja/model_doc/data2vec#transformers.Data2VecTextModel) (Data2VecTextConfig model) - **data2vec-vision** -- [Data2VecVisionModel](/docs/transformers/v5.6.0/ja/model_doc/data2vec#transformers.Data2VecVisionModel) (Data2VecVisionConfig model) - **dbrx** -- `DbrxModel` (DbrxConfig model) - **deberta** -- [DebertaModel](/docs/transformers/v5.6.0/ja/model_doc/deberta#transformers.DebertaModel) (DebertaConfig model) - **deberta-v2** -- [DebertaV2Model](/docs/transformers/v5.6.0/ja/model_doc/deberta-v2#transformers.DebertaV2Model) (DebertaV2Config model) - **decision_transformer** -- `DecisionTransformerModel` (DecisionTransformerConfig model) - **deepseek_v2** -- `DeepseekV2Model` (DeepseekV2Config model) - **deepseek_v3** -- `DeepseekV3Model` (DeepseekV3Config model) - **deepseek_vl** -- `DeepseekVLModel` (DeepseekVLConfig model) - **deepseek_vl_hybrid** -- `DeepseekVLHybridModel` (DeepseekVLHybridConfig model) - **deformable_detr** -- [DeformableDetrModel](/docs/transformers/v5.6.0/ja/model_doc/deformable_detr#transformers.DeformableDetrModel) (DeformableDetrConfig model) - **deit** -- [DeiTModel](/docs/transformers/v5.6.0/ja/model_doc/deit#transformers.DeiTModel) (DeiTConfig model) - **depth_pro** -- `DepthProModel` (DepthProConfig model) - **detr** -- [DetrModel](/docs/transformers/v5.6.0/ja/model_doc/detr#transformers.DetrModel) (DetrConfig model) - **dia** -- `DiaModel` (DiaConfig model) - **diffllama** -- `DiffLlamaModel` (DiffLlamaConfig model) - **dinat** -- [DinatModel](/docs/transformers/v5.6.0/ja/model_doc/dinat#transformers.DinatModel) (DinatConfig model) - **dinov2** -- `Dinov2Model` (Dinov2Config model) - **dinov2_with_registers** -- `Dinov2WithRegistersModel` (Dinov2WithRegistersConfig model) - **dinov3_convnext** -- `DINOv3ConvNextModel` (DINOv3ConvNextConfig model) - **dinov3_vit** -- `DINOv3ViTModel` (DINOv3ViTConfig model) - **distilbert** -- `DistilBertModel` (DistilBertConfig model) - **doge** -- `DogeModel` (DogeConfig model) - **donut-swin** -- `DonutSwinModel` (DonutSwinConfig model) - **dots1** -- `Dots1Model` (Dots1Config model) - **dpr** -- `DPRQuestionEncoder` (DPRConfig model) - **dpt** -- `DPTModel` (DPTConfig model) - **edgetam** -- `EdgeTamModel` (EdgeTamConfig model) - **edgetam_video** -- `EdgeTamVideoModel` (EdgeTamVideoConfig model) - **edgetam_vision_model** -- `EdgeTamVisionModel` (EdgeTamVisionConfig model) - **efficientloftr** -- `EfficientLoFTRModel` (EfficientLoFTRConfig model) - **efficientnet** -- `EfficientNetModel` (EfficientNetConfig model) - **electra** -- `ElectraModel` (ElectraConfig model) - **emu3** -- `Emu3Model` (Emu3Config model) - **encodec** -- `EncodecModel` (EncodecConfig model) - **ernie** -- `ErnieModel` (ErnieConfig model) - **ernie4_5** -- `Ernie4_5Model` (Ernie4_5Config model) - **ernie4_5_moe** -- `Ernie4_5_MoeModel` (Ernie4_5_MoeConfig model) - **ernie4_5_vl_moe** -- `Ernie4_5_VLMoeModel` (Ernie4_5_VLMoeConfig model) - **esm** -- `EsmModel` (EsmConfig model) - **eurobert** -- `EuroBertModel` (EuroBertConfig model) - **evolla** -- `EvollaModel` (EvollaConfig model) - **exaone4** -- `Exaone4Model` (Exaone4Config model) - **exaone_moe** -- `ExaoneMoeModel` (ExaoneMoeConfig model) - **falcon** -- `FalconModel` (FalconConfig model) - **falcon_h1** -- `FalconH1Model` (FalconH1Config model) - **falcon_mamba** -- `FalconMambaModel` (FalconMambaConfig model) - **fast_vlm** -- `FastVlmModel` (FastVlmConfig model) - **fastspeech2_conformer** -- `FastSpeech2ConformerModel` (FastSpeech2ConformerConfig model) - **fastspeech2_conformer_with_hifigan** -- `FastSpeech2ConformerWithHifiGan` (FastSpeech2ConformerWithHifiGanConfig model) - **flaubert** -- `FlaubertModel` (FlaubertConfig model) - **flava** -- `FlavaModel` (FlavaConfig model) - **flex_olmo** -- `FlexOlmoModel` (FlexOlmoConfig model) - **florence2** -- `Florence2Model` (Florence2Config model) - **fnet** -- `FNetModel` (FNetConfig model) - **focalnet** -- `FocalNetModel` (FocalNetConfig model) - **fsmt** -- `FSMTModel` (FSMTConfig model) - **funnel** -- `FunnelModel` or `FunnelBaseModel` (FunnelConfig model) - **fuyu** -- `FuyuModel` (FuyuConfig model) - **gemma** -- `GemmaModel` (GemmaConfig model) - **gemma2** -- `Gemma2Model` (Gemma2Config model) - **gemma3** -- `Gemma3Model` (Gemma3Config model) - **gemma3_text** -- `Gemma3TextModel` (Gemma3TextConfig model) - **gemma3n** -- `Gemma3nModel` (Gemma3nConfig model) - **gemma3n_audio** -- `Gemma3nAudioEncoder` (Gemma3nAudioConfig model) - **gemma3n_text** -- `Gemma3nTextModel` (Gemma3nTextConfig model) - **gemma3n_vision** -- `TimmWrapperModel` (Gemma3nVisionConfig model) - **gemma4** -- `Gemma4Model` (Gemma4Config model) - **gemma4_audio** -- `Gemma4AudioModel` (Gemma4AudioConfig model) - **gemma4_text** -- `Gemma4TextModel` (Gemma4TextConfig model) - **gemma4_vision** -- `Gemma4VisionModel` (Gemma4VisionConfig model) - **git** -- `GitModel` (GitConfig model) - **glm** -- `GlmModel` (GlmConfig model) - **glm4** -- `Glm4Model` (Glm4Config model) - **glm46v** -- `Glm46VModel` (Glm46VConfig model) - **glm4_moe** -- `Glm4MoeModel` (Glm4MoeConfig model) - **glm4_moe_lite** -- `Glm4MoeLiteModel` (Glm4MoeLiteConfig model) - **glm4v** -- `Glm4vModel` (Glm4vConfig model) - **glm4v_moe** -- `Glm4vMoeModel` (Glm4vMoeConfig model) - **glm4v_moe_text** -- `Glm4vMoeTextModel` (Glm4vMoeTextConfig model) - **glm4v_moe_vision** -- `Glm4vMoeVisionModel` (Glm4vMoeVisionConfig model) - **glm4v_text** -- `Glm4vTextModel` (Glm4vTextConfig model) - **glm4v_vision** -- `Glm4vVisionModel` (Glm4vVisionConfig model) - **glm_image** -- `GlmImageModel` (GlmImageConfig model) - **glm_image_text** -- `GlmImageTextModel` (GlmImageTextConfig model) - **glm_image_vision** -- `GlmImageVisionModel` (GlmImageVisionConfig model) - **glm_image_vqmodel** -- `GlmImageVQVAE` (GlmImageVQVAEConfig model) - **glm_moe_dsa** -- `GlmMoeDsaModel` (GlmMoeDsaConfig model) - **glm_ocr** -- `GlmOcrModel` (GlmOcrConfig model) - **glm_ocr_text** -- `GlmOcrTextModel` (GlmOcrTextConfig model) - **glm_ocr_vision** -- `GlmOcrVisionModel` (GlmOcrVisionConfig model) - **glmasr** -- `GlmAsrForConditionalGeneration` (GlmAsrConfig model) - **glmasr_encoder** -- `GlmAsrEncoder` (GlmAsrEncoderConfig model) - **glpn** -- `GLPNModel` (GLPNConfig model) - **got_ocr2** -- `GotOcr2Model` (GotOcr2Config model) - **gpt-sw3** -- `GPT2Model` (GPT2Config model) - **gpt2** -- `GPT2Model` (GPT2Config model) - **gpt_bigcode** -- `GPTBigCodeModel` (GPTBigCodeConfig model) - **gpt_neo** -- `GPTNeoModel` (GPTNeoConfig model) - **gpt_neox** -- `GPTNeoXModel` (GPTNeoXConfig model) - **gpt_neox_japanese** -- `GPTNeoXJapaneseModel` (GPTNeoXJapaneseConfig model) - **gpt_oss** -- `GptOssModel` (GptOssConfig model) - **gptj** -- `GPTJModel` (GPTJConfig model) - **granite** -- `GraniteModel` (GraniteConfig model) - **granite_speech** -- `GraniteSpeechForConditionalGeneration` (GraniteSpeechConfig model) - **granitemoe** -- `GraniteMoeModel` (GraniteMoeConfig model) - **granitemoehybrid** -- `GraniteMoeHybridModel` (GraniteMoeHybridConfig model) - **granitemoeshared** -- `GraniteMoeSharedModel` (GraniteMoeSharedConfig model) - **grounding-dino** -- `GroundingDinoModel` (GroundingDinoConfig model) - **groupvit** -- `GroupViTModel` (GroupViTConfig model) - **helium** -- `HeliumModel` (HeliumConfig model) - **hgnet_v2** -- `HGNetV2Backbone` (HGNetV2Config model) - **hiera** -- `HieraModel` (HieraConfig model) - **higgs_audio_v2** -- `HiggsAudioV2ForConditionalGeneration` (HiggsAudioV2Config model) - **higgs_audio_v2_tokenizer** -- `HiggsAudioV2TokenizerModel` (HiggsAudioV2TokenizerConfig model) - **hubert** -- `HubertModel` (HubertConfig model) - **hunyuan_v1_dense** -- `HunYuanDenseV1Model` (HunYuanDenseV1Config model) - **hunyuan_v1_moe** -- `HunYuanMoEV1Model` (HunYuanMoEV1Config model) - **hy_v3** -- `HYV3Model` (HYV3Config model) - **ibert** -- `IBertModel` (IBertConfig model) - **idefics** -- `IdeficsModel` (IdeficsConfig model) - **idefics2** -- `Idefics2Model` (Idefics2Config model) - **idefics3** -- `Idefics3Model` (Idefics3Config model) - **idefics3_vision** -- `Idefics3VisionTransformer` (Idefics3VisionConfig model) - **ijepa** -- `IJepaModel` (IJepaConfig model) - **imagegpt** -- `ImageGPTModel` (ImageGPTConfig model) - **informer** -- `InformerModel` (InformerConfig model) - **instructblip** -- `InstructBlipModel` (InstructBlipConfig model) - **instructblipvideo** -- `InstructBlipVideoModel` (InstructBlipVideoConfig model) - **internvl** -- `InternVLModel` (InternVLConfig model) - **internvl_vision** -- `InternVLVisionModel` (InternVLVisionConfig model) - **jais2** -- `Jais2Model` (Jais2Config model) - **jamba** -- `JambaModel` (JambaConfig model) - **janus** -- `JanusModel` (JanusConfig model) - **jetmoe** -- `JetMoeModel` (JetMoeConfig model) - **jina_embeddings_v3** -- `JinaEmbeddingsV3Model` (JinaEmbeddingsV3Config model) - **kosmos-2** -- `Kosmos2Model` (Kosmos2Config model) - **kosmos-2.5** -- `Kosmos2_5Model` (Kosmos2_5Config model) - **kyutai_speech_to_text** -- `KyutaiSpeechToTextModel` (KyutaiSpeechToTextConfig model) - **lasr_ctc** -- `LasrForCTC` (LasrCTCConfig model) - **lasr_encoder** -- `LasrEncoder` (LasrEncoderConfig model) - **layoutlm** -- `LayoutLMModel` (LayoutLMConfig model) - **layoutlmv2** -- `LayoutLMv2Model` (LayoutLMv2Config model) - **layoutlmv3** -- `LayoutLMv3Model` (LayoutLMv3Config model) - **led** -- `LEDModel` (LEDConfig model) - **levit** -- `LevitModel` (LevitConfig model) - **lfm2** -- `Lfm2Model` (Lfm2Config model) - **lfm2_moe** -- `Lfm2MoeModel` (Lfm2MoeConfig model) - **lfm2_vl** -- `Lfm2VlModel` (Lfm2VlConfig model) - **lightglue** -- `LightGlueForKeypointMatching` (LightGlueConfig model) - **lighton_ocr** -- `LightOnOcrModel` (LightOnOcrConfig model) - **lilt** -- `LiltModel` (LiltConfig model) - **llama** -- `LlamaModel` (LlamaConfig model) - **llama4** -- `Llama4ForConditionalGeneration` (Llama4Config model) - **llama4_text** -- `Llama4TextModel` (Llama4TextConfig model) - **llava** -- `LlavaModel` (LlavaConfig model) - **llava_next** -- `LlavaNextModel` (LlavaNextConfig model) - **llava_next_video** -- `LlavaNextVideoModel` (LlavaNextVideoConfig model) - **llava_onevision** -- `LlavaOnevisionModel` (LlavaOnevisionConfig model) - **longcat_flash** -- `LongcatFlashModel` (LongcatFlashConfig model) - **longformer** -- `LongformerModel` (LongformerConfig model) - **longt5** -- `LongT5Model` (LongT5Config model) - **luke** -- `LukeModel` (LukeConfig model) - **lw_detr** -- `LwDetrModel` (LwDetrConfig model) - **lxmert** -- `LxmertModel` (LxmertConfig model) - **m2m_100** -- `M2M100Model` (M2M100Config model) - **mamba** -- `MambaModel` (MambaConfig model) - **mamba2** -- `Mamba2Model` (Mamba2Config model) - **marian** -- `MarianModel` (MarianConfig model) - **markuplm** -- `MarkupLMModel` (MarkupLMConfig model) - **mask2former** -- `Mask2FormerModel` (Mask2FormerConfig model) - **maskformer** -- `MaskFormerModel` (MaskFormerConfig model) - **maskformer-swin** -- `MaskFormerSwinModel` (MaskFormerSwinConfig model) - **mbart** -- `MBartModel` (MBartConfig model) - **megatron-bert** -- `MegatronBertModel` (MegatronBertConfig model) - **metaclip_2** -- `MetaClip2Model` (MetaClip2Config model) - **mgp-str** -- `MgpstrForSceneTextRecognition` (MgpstrConfig model) - **mimi** -- `MimiModel` (MimiConfig model) - **minimax** -- `MiniMaxModel` (MiniMaxConfig model) - **minimax_m2** -- `MiniMaxM2Model` (MiniMaxM2Config model) - **ministral** -- `MinistralModel` (MinistralConfig model) - **ministral3** -- `Ministral3Model` (Ministral3Config model) - **mistral** -- `MistralModel` (MistralConfig model) - **mistral3** -- `Mistral3Model` (Mistral3Config model) - **mistral4** -- `Mistral4Model` (Mistral4Config model) - **mixtral** -- `MixtralModel` (MixtralConfig model) - **mlcd** -- `MLCDVisionModel` (MLCDVisionConfig model) - **mlcd_vision_model** -- `MLCDVisionModel` (MLCDVisionConfig model) - **mllama** -- `MllamaModel` (MllamaConfig model) - **mm-grounding-dino** -- `MMGroundingDinoModel` (MMGroundingDinoConfig model) - **mobilebert** -- `MobileBertModel` (MobileBertConfig model) - **mobilenet_v1** -- `MobileNetV1Model` (MobileNetV1Config model) - **mobilenet_v2** -- `MobileNetV2Model` (MobileNetV2Config model) - **mobilevit** -- `MobileViTModel` (MobileViTConfig model) - **mobilevitv2** -- `MobileViTV2Model` (MobileViTV2Config model) - **modernbert** -- `ModernBertModel` (ModernBertConfig model) - **modernbert-decoder** -- `ModernBertDecoderModel` (ModernBertDecoderConfig model) - **modernvbert** -- `ModernVBertModel` (ModernVBertConfig model) - **moonshine** -- `MoonshineModel` (MoonshineConfig model) - **moonshine_streaming** -- `MoonshineStreamingModel` (MoonshineStreamingConfig model) - **moshi** -- `MoshiModel` (MoshiConfig model) - **mpnet** -- `MPNetModel` (MPNetConfig model) - **mpt** -- `MptModel` (MptConfig model) - **mra** -- `MraModel` (MraConfig model) - **mt5** -- `MT5Model` (MT5Config model) - **musicflamingo** -- `MusicFlamingoForConditionalGeneration` (MusicFlamingoConfig model) - **musicgen** -- `MusicgenModel` (MusicgenConfig model) - **musicgen_melody** -- `MusicgenMelodyModel` (MusicgenMelodyConfig model) - **mvp** -- `MvpModel` (MvpConfig model) - **nanochat** -- `NanoChatModel` (NanoChatConfig model) - **nemotron** -- `NemotronModel` (NemotronConfig model) - **nemotron_h** -- `NemotronHModel` (NemotronHConfig model) - **nllb-moe** -- `NllbMoeModel` (NllbMoeConfig model) - **nomic_bert** -- `NomicBertModel` (NomicBertConfig model) - **nystromformer** -- `NystromformerModel` (NystromformerConfig model) - **olmo** -- `OlmoModel` (OlmoConfig model) - **olmo2** -- `Olmo2Model` (Olmo2Config model) - **olmo3** -- `Olmo3Model` (Olmo3Config model) - **olmo_hybrid** -- `OlmoHybridModel` (OlmoHybridConfig model) - **olmoe** -- `OlmoeModel` (OlmoeConfig model) - **omdet-turbo** -- `OmDetTurboForObjectDetection` (OmDetTurboConfig model) - **oneformer** -- `OneFormerModel` (OneFormerConfig model) - **openai-gpt** -- `OpenAIGPTModel` (OpenAIGPTConfig model) - **openai_privacy_filter** -- `OpenAIPrivacyFilterModel` (OpenAIPrivacyFilterConfig model) - **opt** -- `OPTModel` (OPTConfig model) - **ovis2** -- `Ovis2Model` (Ovis2Config model) - **owlv2** -- `Owlv2Model` (Owlv2Config model) - **owlvit** -- `OwlViTModel` (OwlViTConfig model) - **paligemma** -- `PaliGemmaModel` (PaliGemmaConfig model) - **parakeet_ctc** -- `ParakeetForCTC` (ParakeetCTCConfig model) - **parakeet_encoder** -- `ParakeetEncoder` (ParakeetEncoderConfig model) - **patchtsmixer** -- `PatchTSMixerModel` (PatchTSMixerConfig model) - **patchtst** -- `PatchTSTModel` (PatchTSTConfig model) - **pe_audio** -- `PeAudioModel` (PeAudioConfig model) - **pe_audio_encoder** -- `PeAudioEncoder` (PeAudioEncoderConfig model) - **pe_audio_video** -- `PeAudioVideoModel` (PeAudioVideoConfig model) - **pe_audio_video_encoder** -- `PeAudioVideoEncoder` (PeAudioVideoEncoderConfig model) - **pe_video** -- `PeVideoModel` (PeVideoConfig model) - **pe_video_encoder** -- `PeVideoEncoder` (PeVideoEncoderConfig model) - **pegasus** -- `PegasusModel` (PegasusConfig model) - **pegasus_x** -- `PegasusXModel` (PegasusXConfig model) - **perceiver** -- `PerceiverModel` (PerceiverConfig model) - **perception_lm** -- `PerceptionLMModel` (PerceptionLMConfig model) - **persimmon** -- `PersimmonModel` (PersimmonConfig model) - **phi** -- `PhiModel` (PhiConfig model) - **phi3** -- `Phi3Model` (Phi3Config model) - **phi4_multimodal** -- `Phi4MultimodalModel` (Phi4MultimodalConfig model) - **phimoe** -- `PhimoeModel` (PhimoeConfig model) - **pi0** -- `PI0Model` (PI0Config model) - **pixio** -- `PixioModel` (PixioConfig model) - **pixtral** -- `PixtralVisionModel` (PixtralVisionConfig model) - **plbart** -- `PLBartModel` (PLBartConfig model) - **poolformer** -- `PoolFormerModel` (PoolFormerConfig model) - **pp_doclayout_v3** -- `PPDocLayoutV3Model` (PPDocLayoutV3Config model) - **pp_ocrv5_mobile_rec** -- `PPOCRV5MobileRecModel` (PPOCRV5MobileRecConfig model) - **pp_ocrv5_server_rec** -- `PPOCRV5ServerRecModel` (PPOCRV5ServerRecConfig model) - **prophetnet** -- `ProphetNetModel` (ProphetNetConfig model) - **pvt** -- `PvtModel` (PvtConfig model) - **pvt_v2** -- `PvtV2Model` (PvtV2Config model) - **qianfan_ocr** -- `QianfanOCRModel` (QianfanOCRConfig model) - **qianfan_ocr_vision** -- `QianfanOCRVisionModel` (QianfanOCRVisionConfig model) - **qwen2** -- `Qwen2Model` (Qwen2Config model) - **qwen2_5_vl** -- `Qwen2_5_VLModel` (Qwen2_5_VLConfig model) - **qwen2_5_vl_text** -- `Qwen2_5_VLTextModel` (Qwen2_5_VLTextConfig model) - **qwen2_audio_encoder** -- `Qwen2AudioEncoder` (Qwen2AudioEncoderConfig model) - **qwen2_moe** -- `Qwen2MoeModel` (Qwen2MoeConfig model) - **qwen2_vl** -- `Qwen2VLModel` (Qwen2VLConfig model) - **qwen2_vl_text** -- `Qwen2VLTextModel` (Qwen2VLTextConfig model) - **qwen3** -- `Qwen3Model` (Qwen3Config model) - **qwen3_5** -- `Qwen3_5Model` (Qwen3_5Config model) - **qwen3_5_moe** -- `Qwen3_5MoeModel` (Qwen3_5MoeConfig model) - **qwen3_5_moe_text** -- `Qwen3_5MoeTextModel` (Qwen3_5MoeTextConfig model) - **qwen3_5_text** -- `Qwen3_5TextModel` (Qwen3_5TextConfig model) - **qwen3_moe** -- `Qwen3MoeModel` (Qwen3MoeConfig model) - **qwen3_next** -- `Qwen3NextModel` (Qwen3NextConfig model) - **qwen3_vl** -- `Qwen3VLModel` (Qwen3VLConfig model) - **qwen3_vl_moe** -- `Qwen3VLMoeModel` (Qwen3VLMoeConfig model) - **qwen3_vl_moe_text** -- `Qwen3VLMoeTextModel` (Qwen3VLMoeTextConfig model) - **qwen3_vl_text** -- `Qwen3VLTextModel` (Qwen3VLTextConfig model) - **recurrent_gemma** -- `RecurrentGemmaModel` (RecurrentGemmaConfig model) - **reformer** -- `ReformerModel` (ReformerConfig model) - **regnet** -- `RegNetModel` (RegNetConfig model) - **rembert** -- `RemBertModel` (RemBertConfig model) - **resnet** -- `ResNetModel` (ResNetConfig model) - **roberta** -- `RobertaModel` (RobertaConfig model) - **roberta-prelayernorm** -- `RobertaPreLayerNormModel` (RobertaPreLayerNormConfig model) - **roc_bert** -- `RoCBertModel` (RoCBertConfig model) - **roformer** -- `RoFormerModel` (RoFormerConfig model) - **rt_detr** -- `RTDetrModel` (RTDetrConfig model) - **rt_detr_v2** -- `RTDetrV2Model` (RTDetrV2Config model) - **rwkv** -- `RwkvModel` (RwkvConfig model) - **sam** -- `SamModel` (SamConfig model) - **sam2** -- `Sam2Model` (Sam2Config model) - **sam2_hiera_det_model** -- `Sam2HieraDetModel` (Sam2HieraDetConfig model) - **sam2_video** -- `Sam2VideoModel` (Sam2VideoConfig model) - **sam2_vision_model** -- `Sam2VisionModel` (Sam2VisionConfig model) - **sam3** -- `Sam3Model` (Sam3Config model) - **sam3_lite_text** -- `Sam3LiteTextModel` (Sam3LiteTextConfig model) - **sam3_lite_text_text_model** -- `Sam3LiteTextTextModel` (Sam3LiteTextTextConfig model) - **sam3_tracker** -- `Sam3TrackerModel` (Sam3TrackerConfig model) - **sam3_tracker_video** -- `Sam3TrackerVideoModel` (Sam3TrackerVideoConfig model) - **sam3_video** -- `Sam3VideoModel` (Sam3VideoConfig model) - **sam3_vision_model** -- `Sam3VisionModel` (Sam3VisionConfig model) - **sam3_vit_model** -- `Sam3ViTModel` (Sam3ViTConfig model) - **sam_hq** -- `SamHQModel` (SamHQConfig model) - **sam_hq_vision_model** -- `SamHQVisionModel` (SamHQVisionConfig model) - **sam_vision_model** -- `SamVisionModel` (SamVisionConfig model) - **seamless_m4t** -- `SeamlessM4TModel` (SeamlessM4TConfig model) - **seamless_m4t_v2** -- `SeamlessM4Tv2Model` (SeamlessM4Tv2Config model) - **seed_oss** -- `SeedOssModel` (SeedOssConfig model) - **segformer** -- `SegformerModel` (SegformerConfig model) - **seggpt** -- `SegGptModel` (SegGptConfig model) - **sew** -- `SEWModel` (SEWConfig model) - **sew-d** -- `SEWDModel` (SEWDConfig model) - **siglip** -- `SiglipModel` (SiglipConfig model) - **siglip2** -- `Siglip2Model` (Siglip2Config model) - **siglip2_vision_model** -- `Siglip2VisionModel` (Siglip2VisionConfig model) - **siglip_vision_model** -- `SiglipVisionModel` (SiglipVisionConfig model) - **smollm3** -- `SmolLM3Model` (SmolLM3Config model) - **smolvlm** -- `SmolVLMModel` (SmolVLMConfig model) - **smolvlm_vision** -- `SmolVLMVisionTransformer` (SmolVLMVisionConfig model) - **solar_open** -- `SolarOpenModel` (SolarOpenConfig model) - **speech_to_text** -- `Speech2TextModel` (Speech2TextConfig model) - **speecht5** -- `SpeechT5Model` (SpeechT5Config model) - **splinter** -- `SplinterModel` (SplinterConfig model) - **squeezebert** -- `SqueezeBertModel` (SqueezeBertConfig model) - **stablelm** -- `StableLmModel` (StableLmConfig model) - **starcoder2** -- `Starcoder2Model` (Starcoder2Config model) - **swiftformer** -- `SwiftFormerModel` (SwiftFormerConfig model) - **swin** -- `SwinModel` (SwinConfig model) - **swin2sr** -- `Swin2SRModel` (Swin2SRConfig model) - **swinv2** -- `Swinv2Model` (Swinv2Config model) - **switch_transformers** -- `SwitchTransformersModel` (SwitchTransformersConfig model) - **t5** -- `T5Model` (T5Config model) - **t5gemma** -- `T5GemmaModel` (T5GemmaConfig model) - **t5gemma2** -- `T5Gemma2Model` (T5Gemma2Config model) - **t5gemma2_encoder** -- `T5Gemma2Encoder` (T5Gemma2EncoderConfig model) - **table-transformer** -- `TableTransformerModel` (TableTransformerConfig model) - **tapas** -- `TapasModel` (TapasConfig model) - **textnet** -- `TextNetModel` (TextNetConfig model) - **time_series_transformer** -- `TimeSeriesTransformerModel` (TimeSeriesTransformerConfig model) - **timesfm** -- `TimesFmModel` (TimesFmConfig model) - **timesfm2_5** -- `TimesFm2_5Model` (TimesFm2_5Config model) - **timesformer** -- `TimesformerModel` (TimesformerConfig model) - **timm_backbone** -- `TimmBackbone` (TimmBackboneConfig model) - **timm_wrapper** -- `TimmWrapperModel` (TimmWrapperConfig model) - **tvp** -- `TvpModel` (TvpConfig model) - **udop** -- `UdopModel` (UdopConfig model) - **umt5** -- `UMT5Model` (UMT5Config model) - **unispeech** -- `UniSpeechModel` (UniSpeechConfig model) - **unispeech-sat** -- `UniSpeechSatModel` (UniSpeechSatConfig model) - **univnet** -- `UnivNetModel` (UnivNetConfig model) - **uvdoc** -- `UVDocModel` (UVDocConfig model) - **vaultgemma** -- `VaultGemmaModel` (VaultGemmaConfig model) - **vibevoice_acoustic_tokenizer** -- `VibeVoiceAcousticTokenizerModel` (VibeVoiceAcousticTokenizerConfig model) - **vibevoice_acoustic_tokenizer_decoder** -- `VibeVoiceAcousticTokenizerDecoderModel` (VibeVoiceAcousticTokenizerDecoderConfig model) - **vibevoice_acoustic_tokenizer_encoder** -- `VibeVoiceAcousticTokenizerEncoderModel` (VibeVoiceAcousticTokenizerEncoderConfig model) - **vibevoice_asr** -- `VibeVoiceAsrForConditionalGeneration` (VibeVoiceAsrConfig model) - **video_llama_3** -- `VideoLlama3Model` (VideoLlama3Config model) - **video_llama_3_vision** -- `VideoLlama3VisionModel` (VideoLlama3VisionConfig model) - **video_llava** -- `VideoLlavaModel` (VideoLlavaConfig model) - **videomae** -- `VideoMAEModel` (VideoMAEConfig model) - **vilt** -- `ViltModel` (ViltConfig model) - **vipllava** -- `VipLlavaModel` (VipLlavaConfig model) - **vision-text-dual-encoder** -- `VisionTextDualEncoderModel` (VisionTextDualEncoderConfig model) - **visual_bert** -- `VisualBertModel` (VisualBertConfig model) - **vit** -- `ViTModel` (ViTConfig model) - **vit_mae** -- `ViTMAEModel` (ViTMAEConfig model) - **vit_msn** -- `ViTMSNModel` (ViTMSNConfig model) - **vitdet** -- `VitDetModel` (VitDetConfig model) - **vits** -- `VitsModel` (VitsConfig model) - **vivit** -- `VivitModel` (VivitConfig model) - **vjepa2** -- `VJEPA2Model` (VJEPA2Config model) - **voxtral** -- `VoxtralForConditionalGeneration` (VoxtralConfig model) - **voxtral_encoder** -- `VoxtralEncoder` (VoxtralEncoderConfig model) - **voxtral_realtime** -- `VoxtralRealtimeForConditionalGeneration` (VoxtralRealtimeConfig model) - **voxtral_realtime_encoder** -- `VoxtralRealtimeEncoder` (VoxtralRealtimeEncoderConfig model) - **voxtral_realtime_text** -- `VoxtralRealtimeTextModel` (VoxtralRealtimeTextConfig model) - **wav2vec2** -- `Wav2Vec2Model` (Wav2Vec2Config model) - **wav2vec2-bert** -- `Wav2Vec2BertModel` (Wav2Vec2BertConfig model) - **wav2vec2-conformer** -- `Wav2Vec2ConformerModel` (Wav2Vec2ConformerConfig model) - **wavlm** -- `WavLMModel` (WavLMConfig model) - **whisper** -- `WhisperModel` (WhisperConfig model) - **xclip** -- `XCLIPModel` (XCLIPConfig model) - **xcodec** -- `XcodecModel` (XcodecConfig model) - **xglm** -- `XGLMModel` (XGLMConfig model) - **xlm** -- `XLMModel` (XLMConfig model) - **xlm-roberta** -- `XLMRobertaModel` (XLMRobertaConfig model) - **xlm-roberta-xl** -- `XLMRobertaXLModel` (XLMRobertaXLConfig model) - **xlnet** -- `XLNetModel` (XLNetConfig model) - **xlstm** -- `xLSTMModel` (xLSTMConfig model) - **xmod** -- `XmodModel` (XmodConfig model) - **yolos** -- `YolosModel` (YolosConfig model) - **yoso** -- `YosoModel` (YosoConfig model) - **youtu** -- `YoutuModel` (YoutuConfig model) - **zamba** -- `ZambaModel` (ZambaConfig model) - **zamba2** -- `Zamba2Model` (Zamba2Config model) The model is set in evaluation mode by default using `model.eval()` (so for instance, dropout modules are deactivated). To train the model, you should first set it back in training mode with `model.train()` Examples: ```python >>> from transformers import AutoConfig, AutoModel >>> # Download model and configuration from huggingface.co and cache. >>> model = AutoModel.from_pretrained("google-bert/bert-base-cased") >>> # Update configuration during loading >>> model = AutoModel.from_pretrained("google-bert/bert-base-cased", output_attentions=True) >>> model.config.output_attentions True ``` **Parameters:** pretrained_model_name_or_path (`str` or `os.PathLike`) : Can be either: - A string, the *model id* of a pretrained model hosted inside a model repo on huggingface.co. - A path to a *directory* containing model weights saved using [save_pretrained()](/docs/transformers/v5.6.0/ja/main_classes/model#transformers.PreTrainedModel.save_pretrained), e.g., `./my_model_directory/`. model_args (additional positional arguments, *optional*) : Will be passed along to the underlying model `__init__()` method. config ([PreTrainedConfig](/docs/transformers/v5.6.0/ja/main_classes/configuration#transformers.PreTrainedConfig), *optional*) : Configuration for the model to use instead of an automatically loaded configuration. Configuration can be automatically loaded when: - The model is a model provided by the library (loaded with the *model id* string of a pretrained model). - The model was saved using [save_pretrained()](/docs/transformers/v5.6.0/ja/main_classes/model#transformers.PreTrainedModel.save_pretrained) and is reloaded by supplying the save directory. - The model is loaded by supplying a local directory as `pretrained_model_name_or_path` and a configuration JSON file named *config.json* is found in the directory. state_dict (*dict[str, torch.Tensor]*, *optional*) : A state dictionary to use instead of a state dictionary loaded from saved weights file. This option can be used if you want to create a model from a pretrained configuration but load your own weights. In this case though, you should check if using [save_pretrained()](/docs/transformers/v5.6.0/ja/main_classes/model#transformers.PreTrainedModel.save_pretrained) and [from_pretrained()](/docs/transformers/v5.6.0/ja/main_classes/model#transformers.PreTrainedModel.from_pretrained) is not a simpler option. cache_dir (`str` or `os.PathLike`, *optional*) : Path to a directory in which a downloaded pretrained model configuration should be cached if the standard cache should not be used. force_download (`bool`, *optional*, defaults to `False`) : Whether or not to force the (re-)download of the model weights and configuration files, overriding the cached versions if they exist. proxies (`dict[str, str]`, *optional*) : A dictionary of proxy servers to use by protocol or endpoint, e.g., `{'http': 'foo.bar:3128', 'http://hostname': 'foo.bar:4012'}`. The proxies are used on each request. output_loading_info(`bool`, *optional*, defaults to `False`) : Whether or not to also return a dictionary containing missing keys, unexpected keys and error messages. local_files_only(`bool`, *optional*, defaults to `False`) : Whether or not to only look at local files (e.g., not try downloading the model). revision (`str`, *optional*, defaults to `"main"`) : The specific model version to use. It can be a branch name, a tag name, or a commit id, since we use a git-based system for storing models and other artifacts on huggingface.co, so `revision` can be any identifier allowed by git. trust_remote_code (`bool`, *optional*, defaults to `False`) : Whether or not to allow for custom models defined on the Hub in their own modeling files. This option should only be set to `True` for repositories you trust and in which you have read the code, as it will execute code present on the Hub on your local machine. code_revision (`str`, *optional*, defaults to `"main"`) : The specific revision to use for the code on the Hub, if the code leaves in a different repository than the rest of the model. It can be a branch name, a tag name, or a commit id, since we use a git-based system for storing models and other artifacts on huggingface.co, so `revision` can be any identifier allowed by git. kwargs (additional keyword arguments, *optional*) : Can be used to update the configuration object (after it being loaded) and initiate the model (e.g., `output_attentions=True`). Behaves differently depending on whether a `config` is provided or automatically loaded: - If a configuration is provided with `config`, `**kwargs` will be directly passed to the underlying model's `__init__` method (we assume all relevant updates to the configuration have already been done) - If a configuration is not provided, `kwargs` will be first passed to the configuration class initialization function ([from_pretrained()](/docs/transformers/v5.6.0/ja/main_classes/configuration#transformers.PreTrainedConfig.from_pretrained)). Each key of `kwargs` that corresponds to a configuration attribute will be used to override said attribute with the supplied `kwargs` value. Remaining keys that do not correspond to any configuration attribute will be passed to the underlying model's `__init__` function. ## Generic pretraining classes 以下の自動クラスは、事前学習ヘッドを持つモデルをインスタンス化するために利用可能です。 ### AutoModelForPreTraining[[transformers.AutoModelForPreTraining]] #### transformers.AutoModelForPreTraining[[transformers.AutoModelForPreTraining]] [Source](https://github.com/huggingface/transformers/blob/v5.6.0/src/transformers/models/auto/modeling_auto.py#L1986) This is a generic model class that will be instantiated as one of the model classes of the library (with a pretraining head) when created with the [from_pretrained()](/docs/transformers/v5.6.0/ja/model_doc/auto#transformers.AutoModel.from_pretrained) class method or the [from_config()](/docs/transformers/v5.6.0/ja/model_doc/auto#transformers.AutoModel.from_config) class method. This class cannot be instantiated directly using `__init__()` (throws an error). from_configtransformers.AutoModelForPreTraining.from_confighttps://github.com/huggingface/transformers/blob/v5.6.0/src/transformers/models/auto/auto_factory.py#L206[{"name": "**kwargs", "val": ""}]- **config** ([PreTrainedConfig](/docs/transformers/v5.6.0/ja/main_classes/configuration#transformers.PreTrainedConfig)) -- The model class to instantiate is selected based on the configuration class: - [AlbertConfig](/docs/transformers/v5.6.0/ja/model_doc/albert#transformers.AlbertConfig) configuration class: [AlbertForPreTraining](/docs/transformers/v5.6.0/ja/model_doc/albert#transformers.AlbertForPreTraining) (AlbertConfig model) - `AudioFlamingo3Config` configuration class: `AudioFlamingo3ForConditionalGeneration` (AudioFlamingo3Config model) - [BartConfig](/docs/transformers/v5.6.0/ja/model_doc/bart#transformers.BartConfig) configuration class: [BartForConditionalGeneration](/docs/transformers/v5.6.0/ja/model_doc/bart#transformers.BartForConditionalGeneration) (BartConfig model) - [BertConfig](/docs/transformers/v5.6.0/ja/model_doc/bert#transformers.BertConfig) configuration class: [BertForPreTraining](/docs/transformers/v5.6.0/ja/model_doc/bert#transformers.BertForPreTraining) (BertConfig model) - [BigBirdConfig](/docs/transformers/v5.6.0/ja/model_doc/big_bird#transformers.BigBirdConfig) configuration class: [BigBirdForPreTraining](/docs/transformers/v5.6.0/ja/model_doc/big_bird#transformers.BigBirdForPreTraining) (BigBirdConfig model) - [BloomConfig](/docs/transformers/v5.6.0/ja/model_doc/bloom#transformers.BloomConfig) configuration class: [BloomForCausalLM](/docs/transformers/v5.6.0/ja/model_doc/bloom#transformers.BloomForCausalLM) (BloomConfig model) - [CTRLConfig](/docs/transformers/v5.6.0/ja/model_doc/ctrl#transformers.CTRLConfig) configuration class: [CTRLLMHeadModel](/docs/transformers/v5.6.0/ja/model_doc/ctrl#transformers.CTRLLMHeadModel) (CTRLConfig model) - [CamembertConfig](/docs/transformers/v5.6.0/ja/model_doc/camembert#transformers.CamembertConfig) configuration class: [CamembertForMaskedLM](/docs/transformers/v5.6.0/ja/model_doc/camembert#transformers.CamembertForMaskedLM) (CamembertConfig model) - `ColModernVBertConfig` configuration class: `ColModernVBertForRetrieval` (ColModernVBertConfig model) - `ColPaliConfig` configuration class: `ColPaliForRetrieval` (ColPaliConfig model) - `ColQwen2Config` configuration class: `ColQwen2ForRetrieval` (ColQwen2Config model) - [Data2VecTextConfig](/docs/transformers/v5.6.0/ja/model_doc/data2vec#transformers.Data2VecTextConfig) configuration class: [Data2VecTextForMaskedLM](/docs/transformers/v5.6.0/ja/model_doc/data2vec#transformers.Data2VecTextForMaskedLM) (Data2VecTextConfig model) - [DebertaConfig](/docs/transformers/v5.6.0/ja/model_doc/deberta#transformers.DebertaConfig) configuration class: [DebertaForMaskedLM](/docs/transformers/v5.6.0/ja/model_doc/deberta#transformers.DebertaForMaskedLM) (DebertaConfig model) - [DebertaV2Config](/docs/transformers/v5.6.0/ja/model_doc/deberta-v2#transformers.DebertaV2Config) configuration class: [DebertaV2ForMaskedLM](/docs/transformers/v5.6.0/ja/model_doc/deberta-v2#transformers.DebertaV2ForMaskedLM) (DebertaV2Config model) - `DistilBertConfig` configuration class: `DistilBertForMaskedLM` (DistilBertConfig model) - `ElectraConfig` configuration class: `ElectraForPreTraining` (ElectraConfig model) - `ErnieConfig` configuration class: `ErnieForPreTraining` (ErnieConfig model) - `EvollaConfig` configuration class: `EvollaForProteinText2Text` (EvollaConfig model) - `Exaone4Config` configuration class: `Exaone4ForCausalLM` (Exaone4Config model) - `ExaoneMoeConfig` configuration class: `ExaoneMoeForCausalLM` (ExaoneMoeConfig model) - `FNetConfig` configuration class: `FNetForPreTraining` (FNetConfig model) - `FSMTConfig` configuration class: `FSMTForConditionalGeneration` (FSMTConfig model) - `FalconMambaConfig` configuration class: `FalconMambaForCausalLM` (FalconMambaConfig model) - `FlaubertConfig` configuration class: `FlaubertWithLMHeadModel` (FlaubertConfig model) - `FlavaConfig` configuration class: `FlavaForPreTraining` (FlavaConfig model) - `Florence2Config` configuration class: `Florence2ForConditionalGeneration` (Florence2Config model) - `FunnelConfig` configuration class: `FunnelForPreTraining` (FunnelConfig model) - `GPT2Config` configuration class: `GPT2LMHeadModel` (GPT2Config model) - `GPTBigCodeConfig` configuration class: `GPTBigCodeForCausalLM` (GPTBigCodeConfig model) - `Gemma3Config` configuration class: `Gemma3ForConditionalGeneration` (Gemma3Config model) - `Gemma4Config` configuration class: `Gemma4ForConditionalGeneration` (Gemma4Config model) - `GlmAsrConfig` configuration class: `GlmAsrForConditionalGeneration` (GlmAsrConfig model) - `HieraConfig` configuration class: `HieraForPreTraining` (HieraConfig model) - `IBertConfig` configuration class: `IBertForMaskedLM` (IBertConfig model) - `Idefics2Config` configuration class: `Idefics2ForConditionalGeneration` (Idefics2Config model) - `Idefics3Config` configuration class: `Idefics3ForConditionalGeneration` (Idefics3Config model) - `IdeficsConfig` configuration class: `IdeficsForVisionText2Text` (IdeficsConfig model) - `JanusConfig` configuration class: `JanusForConditionalGeneration` (JanusConfig model) - `LayoutLMConfig` configuration class: `LayoutLMForMaskedLM` (LayoutLMConfig model) - `LlavaConfig` configuration class: `LlavaForConditionalGeneration` (LlavaConfig model) - `LlavaNextConfig` configuration class: `LlavaNextForConditionalGeneration` (LlavaNextConfig model) - `LlavaNextVideoConfig` configuration class: `LlavaNextVideoForConditionalGeneration` (LlavaNextVideoConfig model) - `LlavaOnevisionConfig` configuration class: `LlavaOnevisionForConditionalGeneration` (LlavaOnevisionConfig model) - `LongformerConfig` configuration class: `LongformerForMaskedLM` (LongformerConfig model) - `LukeConfig` configuration class: `LukeForMaskedLM` (LukeConfig model) - `LxmertConfig` configuration class: `LxmertForPreTraining` (LxmertConfig model) - `MPNetConfig` configuration class: `MPNetForMaskedLM` (MPNetConfig model) - `Mamba2Config` configuration class: `Mamba2ForCausalLM` (Mamba2Config model) - `MambaConfig` configuration class: `MambaForCausalLM` (MambaConfig model) - `MegatronBertConfig` configuration class: `MegatronBertForPreTraining` (MegatronBertConfig model) - `Mistral3Config` configuration class: `Mistral3ForConditionalGeneration` (Mistral3Config model) - `Mistral4Config` configuration class: `Mistral4ForCausalLM` (Mistral4Config model) - `MllamaConfig` configuration class: `MllamaForConditionalGeneration` (MllamaConfig model) - `MobileBertConfig` configuration class: `MobileBertForPreTraining` (MobileBertConfig model) - `MptConfig` configuration class: `MptForCausalLM` (MptConfig model) - `MraConfig` configuration class: `MraForMaskedLM` (MraConfig model) - `MusicFlamingoConfig` configuration class: `MusicFlamingoForConditionalGeneration` (MusicFlamingoConfig model) - `MvpConfig` configuration class: `MvpForConditionalGeneration` (MvpConfig model) - `NanoChatConfig` configuration class: `NanoChatForCausalLM` (NanoChatConfig model) - `NllbMoeConfig` configuration class: `NllbMoeForConditionalGeneration` (NllbMoeConfig model) - `OpenAIGPTConfig` configuration class: `OpenAIGPTLMHeadModel` (OpenAIGPTConfig model) - `PaliGemmaConfig` configuration class: `PaliGemmaForConditionalGeneration` (PaliGemmaConfig model) - `Qwen2AudioConfig` configuration class: `Qwen2AudioForConditionalGeneration` (Qwen2AudioConfig model) - `RoCBertConfig` configuration class: `RoCBertForPreTraining` (RoCBertConfig model) - `RobertaConfig` configuration class: `RobertaForMaskedLM` (RobertaConfig model) - `RobertaPreLayerNormConfig` configuration class: `RobertaPreLayerNormForMaskedLM` (RobertaPreLayerNormConfig model) - `RwkvConfig` configuration class: `RwkvForCausalLM` (RwkvConfig model) - `SplinterConfig` configuration class: `SplinterForPreTraining` (SplinterConfig model) - `SqueezeBertConfig` configuration class: `SqueezeBertForMaskedLM` (SqueezeBertConfig model) - `SwitchTransformersConfig` configuration class: `SwitchTransformersForConditionalGeneration` (SwitchTransformersConfig model) - `T5Config` configuration class: `T5ForConditionalGeneration` (T5Config model) - `T5Gemma2Config` configuration class: `T5Gemma2ForConditionalGeneration` (T5Gemma2Config model) - `T5GemmaConfig` configuration class: `T5GemmaForConditionalGeneration` (T5GemmaConfig model) - `TapasConfig` configuration class: `TapasForMaskedLM` (TapasConfig model) - `UniSpeechConfig` configuration class: `UniSpeechForPreTraining` (UniSpeechConfig model) - `UniSpeechSatConfig` configuration class: `UniSpeechSatForPreTraining` (UniSpeechSatConfig model) - `ViTMAEConfig` configuration class: `ViTMAEForPreTraining` (ViTMAEConfig model) - `VibeVoiceAsrConfig` configuration class: `VibeVoiceAsrForConditionalGeneration` (VibeVoiceAsrConfig model) - `VideoLlavaConfig` configuration class: `VideoLlavaForConditionalGeneration` (VideoLlavaConfig model) - `VideoMAEConfig` configuration class: `VideoMAEForPreTraining` (VideoMAEConfig model) - `VipLlavaConfig` configuration class: `VipLlavaForConditionalGeneration` (VipLlavaConfig model) - `VisualBertConfig` configuration class: `VisualBertForPreTraining` (VisualBertConfig model) - `VoxtralConfig` configuration class: `VoxtralForConditionalGeneration` (VoxtralConfig model) - `VoxtralRealtimeConfig` configuration class: `VoxtralRealtimeForConditionalGeneration` (VoxtralRealtimeConfig model) - `Wav2Vec2Config` configuration class: `Wav2Vec2ForPreTraining` (Wav2Vec2Config model) - `Wav2Vec2ConformerConfig` configuration class: `Wav2Vec2ConformerForPreTraining` (Wav2Vec2ConformerConfig model) - `XLMConfig` configuration class: `XLMWithLMHeadModel` (XLMConfig model) - `XLMRobertaConfig` configuration class: `XLMRobertaForMaskedLM` (XLMRobertaConfig model) - `XLMRobertaXLConfig` configuration class: `XLMRobertaXLForMaskedLM` (XLMRobertaXLConfig model) - `XLNetConfig` configuration class: `XLNetLMHeadModel` (XLNetConfig model) - `XmodConfig` configuration class: `XmodForMaskedLM` (XmodConfig model) - `xLSTMConfig` configuration class: `xLSTMForCausalLM` (xLSTMConfig model) - **attn_implementation** (`str`, *optional*) -- The attention implementation to use in the model (if relevant). Can be any of `"eager"` (manual implementation of the attention), `"sdpa"` (using [`F.scaled_dot_product_attention`](https://pytorch.org/docs/master/generated/torch.nn.functional.scaled_dot_product_attention.html)), `"flash_attention_2"` (using [Dao-AILab/flash-attention](https://github.com/Dao-AILab/flash-attention)), or `"flash_attention_3"` (using [Dao-AILab/flash-attention/hopper](https://github.com/Dao-AILab/flash-attention/tree/main/hopper)). By default, if available, SDPA will be used for torch>=2.1.1. The default is otherwise the manual `"eager"` implementation.0 Instantiates one of the model classes of the library (with a pretraining head) from a configuration. Note: Loading a model from its configuration file does **not** load the model weights. It only affects the model's configuration. Use [from_pretrained()](/docs/transformers/v5.6.0/ja/model_doc/auto#transformers.AutoModel.from_pretrained) to load the model weights. Examples: ```python >>> from transformers import AutoConfig, AutoModelForPreTraining >>> # Download configuration from huggingface.co and cache. >>> config = AutoConfig.from_pretrained("google-bert/bert-base-cased") >>> model = AutoModelForPreTraining.from_config(config) ``` **Parameters:** config ([PreTrainedConfig](/docs/transformers/v5.6.0/ja/main_classes/configuration#transformers.PreTrainedConfig)) : The model class to instantiate is selected based on the configuration class: - [AlbertConfig](/docs/transformers/v5.6.0/ja/model_doc/albert#transformers.AlbertConfig) configuration class: [AlbertForPreTraining](/docs/transformers/v5.6.0/ja/model_doc/albert#transformers.AlbertForPreTraining) (AlbertConfig model) - `AudioFlamingo3Config` configuration class: `AudioFlamingo3ForConditionalGeneration` (AudioFlamingo3Config model) - [BartConfig](/docs/transformers/v5.6.0/ja/model_doc/bart#transformers.BartConfig) configuration class: [BartForConditionalGeneration](/docs/transformers/v5.6.0/ja/model_doc/bart#transformers.BartForConditionalGeneration) (BartConfig model) - [BertConfig](/docs/transformers/v5.6.0/ja/model_doc/bert#transformers.BertConfig) configuration class: [BertForPreTraining](/docs/transformers/v5.6.0/ja/model_doc/bert#transformers.BertForPreTraining) (BertConfig model) - [BigBirdConfig](/docs/transformers/v5.6.0/ja/model_doc/big_bird#transformers.BigBirdConfig) configuration class: [BigBirdForPreTraining](/docs/transformers/v5.6.0/ja/model_doc/big_bird#transformers.BigBirdForPreTraining) (BigBirdConfig model) - [BloomConfig](/docs/transformers/v5.6.0/ja/model_doc/bloom#transformers.BloomConfig) configuration class: [BloomForCausalLM](/docs/transformers/v5.6.0/ja/model_doc/bloom#transformers.BloomForCausalLM) (BloomConfig model) - [CTRLConfig](/docs/transformers/v5.6.0/ja/model_doc/ctrl#transformers.CTRLConfig) configuration class: [CTRLLMHeadModel](/docs/transformers/v5.6.0/ja/model_doc/ctrl#transformers.CTRLLMHeadModel) (CTRLConfig model) - [CamembertConfig](/docs/transformers/v5.6.0/ja/model_doc/camembert#transformers.CamembertConfig) configuration class: [CamembertForMaskedLM](/docs/transformers/v5.6.0/ja/model_doc/camembert#transformers.CamembertForMaskedLM) (CamembertConfig model) - `ColModernVBertConfig` configuration class: `ColModernVBertForRetrieval` (ColModernVBertConfig model) - `ColPaliConfig` configuration class: `ColPaliForRetrieval` (ColPaliConfig model) - `ColQwen2Config` configuration class: `ColQwen2ForRetrieval` (ColQwen2Config model) - [Data2VecTextConfig](/docs/transformers/v5.6.0/ja/model_doc/data2vec#transformers.Data2VecTextConfig) configuration class: [Data2VecTextForMaskedLM](/docs/transformers/v5.6.0/ja/model_doc/data2vec#transformers.Data2VecTextForMaskedLM) (Data2VecTextConfig model) - [DebertaConfig](/docs/transformers/v5.6.0/ja/model_doc/deberta#transformers.DebertaConfig) configuration class: [DebertaForMaskedLM](/docs/transformers/v5.6.0/ja/model_doc/deberta#transformers.DebertaForMaskedLM) (DebertaConfig model) - [DebertaV2Config](/docs/transformers/v5.6.0/ja/model_doc/deberta-v2#transformers.DebertaV2Config) configuration class: [DebertaV2ForMaskedLM](/docs/transformers/v5.6.0/ja/model_doc/deberta-v2#transformers.DebertaV2ForMaskedLM) (DebertaV2Config model) - `DistilBertConfig` configuration class: `DistilBertForMaskedLM` (DistilBertConfig model) - `ElectraConfig` configuration class: `ElectraForPreTraining` (ElectraConfig model) - `ErnieConfig` configuration class: `ErnieForPreTraining` (ErnieConfig model) - `EvollaConfig` configuration class: `EvollaForProteinText2Text` (EvollaConfig model) - `Exaone4Config` configuration class: `Exaone4ForCausalLM` (Exaone4Config model) - `ExaoneMoeConfig` configuration class: `ExaoneMoeForCausalLM` (ExaoneMoeConfig model) - `FNetConfig` configuration class: `FNetForPreTraining` (FNetConfig model) - `FSMTConfig` configuration class: `FSMTForConditionalGeneration` (FSMTConfig model) - `FalconMambaConfig` configuration class: `FalconMambaForCausalLM` (FalconMambaConfig model) - `FlaubertConfig` configuration class: `FlaubertWithLMHeadModel` (FlaubertConfig model) - `FlavaConfig` configuration class: `FlavaForPreTraining` (FlavaConfig model) - `Florence2Config` configuration class: `Florence2ForConditionalGeneration` (Florence2Config model) - `FunnelConfig` configuration class: `FunnelForPreTraining` (FunnelConfig model) - `GPT2Config` configuration class: `GPT2LMHeadModel` (GPT2Config model) - `GPTBigCodeConfig` configuration class: `GPTBigCodeForCausalLM` (GPTBigCodeConfig model) - `Gemma3Config` configuration class: `Gemma3ForConditionalGeneration` (Gemma3Config model) - `Gemma4Config` configuration class: `Gemma4ForConditionalGeneration` (Gemma4Config model) - `GlmAsrConfig` configuration class: `GlmAsrForConditionalGeneration` (GlmAsrConfig model) - `HieraConfig` configuration class: `HieraForPreTraining` (HieraConfig model) - `IBertConfig` configuration class: `IBertForMaskedLM` (IBertConfig model) - `Idefics2Config` configuration class: `Idefics2ForConditionalGeneration` (Idefics2Config model) - `Idefics3Config` configuration class: `Idefics3ForConditionalGeneration` (Idefics3Config model) - `IdeficsConfig` configuration class: `IdeficsForVisionText2Text` (IdeficsConfig model) - `JanusConfig` configuration class: `JanusForConditionalGeneration` (JanusConfig model) - `LayoutLMConfig` configuration class: `LayoutLMForMaskedLM` (LayoutLMConfig model) - `LlavaConfig` configuration class: `LlavaForConditionalGeneration` (LlavaConfig model) - `LlavaNextConfig` configuration class: `LlavaNextForConditionalGeneration` (LlavaNextConfig model) - `LlavaNextVideoConfig` configuration class: `LlavaNextVideoForConditionalGeneration` (LlavaNextVideoConfig model) - `LlavaOnevisionConfig` configuration class: `LlavaOnevisionForConditionalGeneration` (LlavaOnevisionConfig model) - `LongformerConfig` configuration class: `LongformerForMaskedLM` (LongformerConfig model) - `LukeConfig` configuration class: `LukeForMaskedLM` (LukeConfig model) - `LxmertConfig` configuration class: `LxmertForPreTraining` (LxmertConfig model) - `MPNetConfig` configuration class: `MPNetForMaskedLM` (MPNetConfig model) - `Mamba2Config` configuration class: `Mamba2ForCausalLM` (Mamba2Config model) - `MambaConfig` configuration class: `MambaForCausalLM` (MambaConfig model) - `MegatronBertConfig` configuration class: `MegatronBertForPreTraining` (MegatronBertConfig model) - `Mistral3Config` configuration class: `Mistral3ForConditionalGeneration` (Mistral3Config model) - `Mistral4Config` configuration class: `Mistral4ForCausalLM` (Mistral4Config model) - `MllamaConfig` configuration class: `MllamaForConditionalGeneration` (MllamaConfig model) - `MobileBertConfig` configuration class: `MobileBertForPreTraining` (MobileBertConfig model) - `MptConfig` configuration class: `MptForCausalLM` (MptConfig model) - `MraConfig` configuration class: `MraForMaskedLM` (MraConfig model) - `MusicFlamingoConfig` configuration class: `MusicFlamingoForConditionalGeneration` (MusicFlamingoConfig model) - `MvpConfig` configuration class: `MvpForConditionalGeneration` (MvpConfig model) - `NanoChatConfig` configuration class: `NanoChatForCausalLM` (NanoChatConfig model) - `NllbMoeConfig` configuration class: `NllbMoeForConditionalGeneration` (NllbMoeConfig model) - `OpenAIGPTConfig` configuration class: `OpenAIGPTLMHeadModel` (OpenAIGPTConfig model) - `PaliGemmaConfig` configuration class: `PaliGemmaForConditionalGeneration` (PaliGemmaConfig model) - `Qwen2AudioConfig` configuration class: `Qwen2AudioForConditionalGeneration` (Qwen2AudioConfig model) - `RoCBertConfig` configuration class: `RoCBertForPreTraining` (RoCBertConfig model) - `RobertaConfig` configuration class: `RobertaForMaskedLM` (RobertaConfig model) - `RobertaPreLayerNormConfig` configuration class: `RobertaPreLayerNormForMaskedLM` (RobertaPreLayerNormConfig model) - `RwkvConfig` configuration class: `RwkvForCausalLM` (RwkvConfig model) - `SplinterConfig` configuration class: `SplinterForPreTraining` (SplinterConfig model) - `SqueezeBertConfig` configuration class: `SqueezeBertForMaskedLM` (SqueezeBertConfig model) - `SwitchTransformersConfig` configuration class: `SwitchTransformersForConditionalGeneration` (SwitchTransformersConfig model) - `T5Config` configuration class: `T5ForConditionalGeneration` (T5Config model) - `T5Gemma2Config` configuration class: `T5Gemma2ForConditionalGeneration` (T5Gemma2Config model) - `T5GemmaConfig` configuration class: `T5GemmaForConditionalGeneration` (T5GemmaConfig model) - `TapasConfig` configuration class: `TapasForMaskedLM` (TapasConfig model) - `UniSpeechConfig` configuration class: `UniSpeechForPreTraining` (UniSpeechConfig model) - `UniSpeechSatConfig` configuration class: `UniSpeechSatForPreTraining` (UniSpeechSatConfig model) - `ViTMAEConfig` configuration class: `ViTMAEForPreTraining` (ViTMAEConfig model) - `VibeVoiceAsrConfig` configuration class: `VibeVoiceAsrForConditionalGeneration` (VibeVoiceAsrConfig model) - `VideoLlavaConfig` configuration class: `VideoLlavaForConditionalGeneration` (VideoLlavaConfig model) - `VideoMAEConfig` configuration class: `VideoMAEForPreTraining` (VideoMAEConfig model) - `VipLlavaConfig` configuration class: `VipLlavaForConditionalGeneration` (VipLlavaConfig model) - `VisualBertConfig` configuration class: `VisualBertForPreTraining` (VisualBertConfig model) - `VoxtralConfig` configuration class: `VoxtralForConditionalGeneration` (VoxtralConfig model) - `VoxtralRealtimeConfig` configuration class: `VoxtralRealtimeForConditionalGeneration` (VoxtralRealtimeConfig model) - `Wav2Vec2Config` configuration class: `Wav2Vec2ForPreTraining` (Wav2Vec2Config model) - `Wav2Vec2ConformerConfig` configuration class: `Wav2Vec2ConformerForPreTraining` (Wav2Vec2ConformerConfig model) - `XLMConfig` configuration class: `XLMWithLMHeadModel` (XLMConfig model) - `XLMRobertaConfig` configuration class: `XLMRobertaForMaskedLM` (XLMRobertaConfig model) - `XLMRobertaXLConfig` configuration class: `XLMRobertaXLForMaskedLM` (XLMRobertaXLConfig model) - `XLNetConfig` configuration class: `XLNetLMHeadModel` (XLNetConfig model) - `XmodConfig` configuration class: `XmodForMaskedLM` (XmodConfig model) - `xLSTMConfig` configuration class: `xLSTMForCausalLM` (xLSTMConfig model) attn_implementation (`str`, *optional*) : The attention implementation to use in the model (if relevant). Can be any of `"eager"` (manual implementation of the attention), `"sdpa"` (using [`F.scaled_dot_product_attention`](https://pytorch.org/docs/master/generated/torch.nn.functional.scaled_dot_product_attention.html)), `"flash_attention_2"` (using [Dao-AILab/flash-attention](https://github.com/Dao-AILab/flash-attention)), or `"flash_attention_3"` (using [Dao-AILab/flash-attention/hopper](https://github.com/Dao-AILab/flash-attention/tree/main/hopper)). By default, if available, SDPA will be used for torch>=2.1.1. The default is otherwise the manual `"eager"` implementation. #### from_pretrained[[transformers.AutoModelForPreTraining.from_pretrained]] [Source](https://github.com/huggingface/transformers/blob/v5.6.0/src/transformers/models/auto/auto_factory.py#L253) Instantiate one of the model classes of the library (with a pretraining head) from a pretrained model. The model class to instantiate is selected based on the `model_type` property of the config object (either passed as an argument or loaded from `pretrained_model_name_or_path` if possible), or when it's missing, by falling back to using pattern matching on `pretrained_model_name_or_path`: - **albert** -- [AlbertForPreTraining](/docs/transformers/v5.6.0/ja/model_doc/albert#transformers.AlbertForPreTraining) (AlbertConfig model) - **audioflamingo3** -- `AudioFlamingo3ForConditionalGeneration` (AudioFlamingo3Config model) - **bart** -- [BartForConditionalGeneration](/docs/transformers/v5.6.0/ja/model_doc/bart#transformers.BartForConditionalGeneration) (BartConfig model) - **bert** -- [BertForPreTraining](/docs/transformers/v5.6.0/ja/model_doc/bert#transformers.BertForPreTraining) (BertConfig model) - **big_bird** -- [BigBirdForPreTraining](/docs/transformers/v5.6.0/ja/model_doc/big_bird#transformers.BigBirdForPreTraining) (BigBirdConfig model) - **bloom** -- [BloomForCausalLM](/docs/transformers/v5.6.0/ja/model_doc/bloom#transformers.BloomForCausalLM) (BloomConfig model) - **camembert** -- [CamembertForMaskedLM](/docs/transformers/v5.6.0/ja/model_doc/camembert#transformers.CamembertForMaskedLM) (CamembertConfig model) - **colmodernvbert** -- `ColModernVBertForRetrieval` (ColModernVBertConfig model) - **colpali** -- `ColPaliForRetrieval` (ColPaliConfig model) - **colqwen2** -- `ColQwen2ForRetrieval` (ColQwen2Config model) - **ctrl** -- [CTRLLMHeadModel](/docs/transformers/v5.6.0/ja/model_doc/ctrl#transformers.CTRLLMHeadModel) (CTRLConfig model) - **data2vec-text** -- [Data2VecTextForMaskedLM](/docs/transformers/v5.6.0/ja/model_doc/data2vec#transformers.Data2VecTextForMaskedLM) (Data2VecTextConfig model) - **deberta** -- [DebertaForMaskedLM](/docs/transformers/v5.6.0/ja/model_doc/deberta#transformers.DebertaForMaskedLM) (DebertaConfig model) - **deberta-v2** -- [DebertaV2ForMaskedLM](/docs/transformers/v5.6.0/ja/model_doc/deberta-v2#transformers.DebertaV2ForMaskedLM) (DebertaV2Config model) - **distilbert** -- `DistilBertForMaskedLM` (DistilBertConfig model) - **electra** -- `ElectraForPreTraining` (ElectraConfig model) - **ernie** -- `ErnieForPreTraining` (ErnieConfig model) - **evolla** -- `EvollaForProteinText2Text` (EvollaConfig model) - **exaone4** -- `Exaone4ForCausalLM` (Exaone4Config model) - **exaone_moe** -- `ExaoneMoeForCausalLM` (ExaoneMoeConfig model) - **falcon_mamba** -- `FalconMambaForCausalLM` (FalconMambaConfig model) - **flaubert** -- `FlaubertWithLMHeadModel` (FlaubertConfig model) - **flava** -- `FlavaForPreTraining` (FlavaConfig model) - **florence2** -- `Florence2ForConditionalGeneration` (Florence2Config model) - **fnet** -- `FNetForPreTraining` (FNetConfig model) - **fsmt** -- `FSMTForConditionalGeneration` (FSMTConfig model) - **funnel** -- `FunnelForPreTraining` (FunnelConfig model) - **gemma3** -- `Gemma3ForConditionalGeneration` (Gemma3Config model) - **gemma4** -- `Gemma4ForConditionalGeneration` (Gemma4Config model) - **glmasr** -- `GlmAsrForConditionalGeneration` (GlmAsrConfig model) - **gpt-sw3** -- `GPT2LMHeadModel` (GPT2Config model) - **gpt2** -- `GPT2LMHeadModel` (GPT2Config model) - **gpt_bigcode** -- `GPTBigCodeForCausalLM` (GPTBigCodeConfig model) - **hiera** -- `HieraForPreTraining` (HieraConfig model) - **ibert** -- `IBertForMaskedLM` (IBertConfig model) - **idefics** -- `IdeficsForVisionText2Text` (IdeficsConfig model) - **idefics2** -- `Idefics2ForConditionalGeneration` (Idefics2Config model) - **idefics3** -- `Idefics3ForConditionalGeneration` (Idefics3Config model) - **janus** -- `JanusForConditionalGeneration` (JanusConfig model) - **layoutlm** -- `LayoutLMForMaskedLM` (LayoutLMConfig model) - **llava** -- `LlavaForConditionalGeneration` (LlavaConfig model) - **llava_next** -- `LlavaNextForConditionalGeneration` (LlavaNextConfig model) - **llava_next_video** -- `LlavaNextVideoForConditionalGeneration` (LlavaNextVideoConfig model) - **llava_onevision** -- `LlavaOnevisionForConditionalGeneration` (LlavaOnevisionConfig model) - **longformer** -- `LongformerForMaskedLM` (LongformerConfig model) - **luke** -- `LukeForMaskedLM` (LukeConfig model) - **lxmert** -- `LxmertForPreTraining` (LxmertConfig model) - **mamba** -- `MambaForCausalLM` (MambaConfig model) - **mamba2** -- `Mamba2ForCausalLM` (Mamba2Config model) - **megatron-bert** -- `MegatronBertForPreTraining` (MegatronBertConfig model) - **mistral3** -- `Mistral3ForConditionalGeneration` (Mistral3Config model) - **mistral4** -- `Mistral4ForCausalLM` (Mistral4Config model) - **mllama** -- `MllamaForConditionalGeneration` (MllamaConfig model) - **mobilebert** -- `MobileBertForPreTraining` (MobileBertConfig model) - **mpnet** -- `MPNetForMaskedLM` (MPNetConfig model) - **mpt** -- `MptForCausalLM` (MptConfig model) - **mra** -- `MraForMaskedLM` (MraConfig model) - **musicflamingo** -- `MusicFlamingoForConditionalGeneration` (MusicFlamingoConfig model) - **mvp** -- `MvpForConditionalGeneration` (MvpConfig model) - **nanochat** -- `NanoChatForCausalLM` (NanoChatConfig model) - **nllb-moe** -- `NllbMoeForConditionalGeneration` (NllbMoeConfig model) - **openai-gpt** -- `OpenAIGPTLMHeadModel` (OpenAIGPTConfig model) - **paligemma** -- `PaliGemmaForConditionalGeneration` (PaliGemmaConfig model) - **qwen2_audio** -- `Qwen2AudioForConditionalGeneration` (Qwen2AudioConfig model) - **roberta** -- `RobertaForMaskedLM` (RobertaConfig model) - **roberta-prelayernorm** -- `RobertaPreLayerNormForMaskedLM` (RobertaPreLayerNormConfig model) - **roc_bert** -- `RoCBertForPreTraining` (RoCBertConfig model) - **rwkv** -- `RwkvForCausalLM` (RwkvConfig model) - **splinter** -- `SplinterForPreTraining` (SplinterConfig model) - **squeezebert** -- `SqueezeBertForMaskedLM` (SqueezeBertConfig model) - **switch_transformers** -- `SwitchTransformersForConditionalGeneration` (SwitchTransformersConfig model) - **t5** -- `T5ForConditionalGeneration` (T5Config model) - **t5gemma** -- `T5GemmaForConditionalGeneration` (T5GemmaConfig model) - **t5gemma2** -- `T5Gemma2ForConditionalGeneration` (T5Gemma2Config model) - **tapas** -- `TapasForMaskedLM` (TapasConfig model) - **unispeech** -- `UniSpeechForPreTraining` (UniSpeechConfig model) - **unispeech-sat** -- `UniSpeechSatForPreTraining` (UniSpeechSatConfig model) - **vibevoice_asr** -- `VibeVoiceAsrForConditionalGeneration` (VibeVoiceAsrConfig model) - **video_llava** -- `VideoLlavaForConditionalGeneration` (VideoLlavaConfig model) - **videomae** -- `VideoMAEForPreTraining` (VideoMAEConfig model) - **vipllava** -- `VipLlavaForConditionalGeneration` (VipLlavaConfig model) - **visual_bert** -- `VisualBertForPreTraining` (VisualBertConfig model) - **vit_mae** -- `ViTMAEForPreTraining` (ViTMAEConfig model) - **voxtral** -- `VoxtralForConditionalGeneration` (VoxtralConfig model) - **voxtral_realtime** -- `VoxtralRealtimeForConditionalGeneration` (VoxtralRealtimeConfig model) - **wav2vec2** -- `Wav2Vec2ForPreTraining` (Wav2Vec2Config model) - **wav2vec2-conformer** -- `Wav2Vec2ConformerForPreTraining` (Wav2Vec2ConformerConfig model) - **xlm** -- `XLMWithLMHeadModel` (XLMConfig model) - **xlm-roberta** -- `XLMRobertaForMaskedLM` (XLMRobertaConfig model) - **xlm-roberta-xl** -- `XLMRobertaXLForMaskedLM` (XLMRobertaXLConfig model) - **xlnet** -- `XLNetLMHeadModel` (XLNetConfig model) - **xlstm** -- `xLSTMForCausalLM` (xLSTMConfig model) - **xmod** -- `XmodForMaskedLM` (XmodConfig model) The model is set in evaluation mode by default using `model.eval()` (so for instance, dropout modules are deactivated). To train the model, you should first set it back in training mode with `model.train()` Examples: ```python >>> from transformers import AutoConfig, AutoModelForPreTraining >>> # Download model and configuration from huggingface.co and cache. >>> model = AutoModelForPreTraining.from_pretrained("google-bert/bert-base-cased") >>> # Update configuration during loading >>> model = AutoModelForPreTraining.from_pretrained("google-bert/bert-base-cased", output_attentions=True) >>> model.config.output_attentions True ``` **Parameters:** pretrained_model_name_or_path (`str` or `os.PathLike`) : Can be either: - A string, the *model id* of a pretrained model hosted inside a model repo on huggingface.co. - A path to a *directory* containing model weights saved using [save_pretrained()](/docs/transformers/v5.6.0/ja/main_classes/model#transformers.PreTrainedModel.save_pretrained), e.g., `./my_model_directory/`. model_args (additional positional arguments, *optional*) : Will be passed along to the underlying model `__init__()` method. config ([PreTrainedConfig](/docs/transformers/v5.6.0/ja/main_classes/configuration#transformers.PreTrainedConfig), *optional*) : Configuration for the model to use instead of an automatically loaded configuration. Configuration can be automatically loaded when: - The model is a model provided by the library (loaded with the *model id* string of a pretrained model). - The model was saved using [save_pretrained()](/docs/transformers/v5.6.0/ja/main_classes/model#transformers.PreTrainedModel.save_pretrained) and is reloaded by supplying the save directory. - The model is loaded by supplying a local directory as `pretrained_model_name_or_path` and a configuration JSON file named *config.json* is found in the directory. state_dict (*dict[str, torch.Tensor]*, *optional*) : A state dictionary to use instead of a state dictionary loaded from saved weights file. This option can be used if you want to create a model from a pretrained configuration but load your own weights. In this case though, you should check if using [save_pretrained()](/docs/transformers/v5.6.0/ja/main_classes/model#transformers.PreTrainedModel.save_pretrained) and [from_pretrained()](/docs/transformers/v5.6.0/ja/main_classes/model#transformers.PreTrainedModel.from_pretrained) is not a simpler option. cache_dir (`str` or `os.PathLike`, *optional*) : Path to a directory in which a downloaded pretrained model configuration should be cached if the standard cache should not be used. force_download (`bool`, *optional*, defaults to `False`) : Whether or not to force the (re-)download of the model weights and configuration files, overriding the cached versions if they exist. proxies (`dict[str, str]`, *optional*) : A dictionary of proxy servers to use by protocol or endpoint, e.g., `{'http': 'foo.bar:3128', 'http://hostname': 'foo.bar:4012'}`. The proxies are used on each request. output_loading_info(`bool`, *optional*, defaults to `False`) : Whether or not to also return a dictionary containing missing keys, unexpected keys and error messages. local_files_only(`bool`, *optional*, defaults to `False`) : Whether or not to only look at local files (e.g., not try downloading the model). revision (`str`, *optional*, defaults to `"main"`) : The specific model version to use. It can be a branch name, a tag name, or a commit id, since we use a git-based system for storing models and other artifacts on huggingface.co, so `revision` can be any identifier allowed by git. trust_remote_code (`bool`, *optional*, defaults to `False`) : Whether or not to allow for custom models defined on the Hub in their own modeling files. This option should only be set to `True` for repositories you trust and in which you have read the code, as it will execute code present on the Hub on your local machine. code_revision (`str`, *optional*, defaults to `"main"`) : The specific revision to use for the code on the Hub, if the code leaves in a different repository than the rest of the model. It can be a branch name, a tag name, or a commit id, since we use a git-based system for storing models and other artifacts on huggingface.co, so `revision` can be any identifier allowed by git. kwargs (additional keyword arguments, *optional*) : Can be used to update the configuration object (after it being loaded) and initiate the model (e.g., `output_attentions=True`). Behaves differently depending on whether a `config` is provided or automatically loaded: - If a configuration is provided with `config`, `**kwargs` will be directly passed to the underlying model's `__init__` method (we assume all relevant updates to the configuration have already been done) - If a configuration is not provided, `kwargs` will be first passed to the configuration class initialization function ([from_pretrained()](/docs/transformers/v5.6.0/ja/main_classes/configuration#transformers.PreTrainedConfig.from_pretrained)). Each key of `kwargs` that corresponds to a configuration attribute will be used to override said attribute with the supplied `kwargs` value. Remaining keys that do not correspond to any configuration attribute will be passed to the underlying model's `__init__` function. ## Natural Language Processing 以下の自動クラスは、次の自然言語処理タスクに利用可能です。 ### AutoModelForCausalLM[[transformers.AutoModelForCausalLM]] #### transformers.AutoModelForCausalLM[[transformers.AutoModelForCausalLM]] [Source](https://github.com/huggingface/transformers/blob/v5.6.0/src/transformers/models/auto/modeling_auto.py#L1993) This is a generic model class that will be instantiated as one of the model classes of the library (with a causal language modeling head) when created with the [from_pretrained()](/docs/transformers/v5.6.0/ja/model_doc/auto#transformers.AutoModel.from_pretrained) class method or the [from_config()](/docs/transformers/v5.6.0/ja/model_doc/auto#transformers.AutoModel.from_config) class method. This class cannot be instantiated directly using `__init__()` (throws an error). from_configtransformers.AutoModelForCausalLM.from_confighttps://github.com/huggingface/transformers/blob/v5.6.0/src/transformers/models/auto/auto_factory.py#L206[{"name": "**kwargs", "val": ""}]- **config** ([PreTrainedConfig](/docs/transformers/v5.6.0/ja/main_classes/configuration#transformers.PreTrainedConfig)) -- The model class to instantiate is selected based on the configuration class: - `AfmoeConfig` configuration class: `AfmoeForCausalLM` (AfmoeConfig model) - `ApertusConfig` configuration class: `ApertusForCausalLM` (ApertusConfig model) - `ArceeConfig` configuration class: `ArceeForCausalLM` (ArceeConfig model) - `AriaTextConfig` configuration class: `AriaTextForCausalLM` (AriaTextConfig model) - `BambaConfig` configuration class: `BambaForCausalLM` (BambaConfig model) - [BartConfig](/docs/transformers/v5.6.0/ja/model_doc/bart#transformers.BartConfig) configuration class: [BartForCausalLM](/docs/transformers/v5.6.0/ja/model_doc/bart#transformers.BartForCausalLM) (BartConfig model) - [BertConfig](/docs/transformers/v5.6.0/ja/model_doc/bert#transformers.BertConfig) configuration class: [BertLMHeadModel](/docs/transformers/v5.6.0/ja/model_doc/bert#transformers.BertLMHeadModel) (BertConfig model) - [BertGenerationConfig](/docs/transformers/v5.6.0/ja/model_doc/bert-generation#transformers.BertGenerationConfig) configuration class: [BertGenerationDecoder](/docs/transformers/v5.6.0/ja/model_doc/bert-generation#transformers.BertGenerationDecoder) (BertGenerationConfig model) - [BigBirdConfig](/docs/transformers/v5.6.0/ja/model_doc/big_bird#transformers.BigBirdConfig) configuration class: [BigBirdForCausalLM](/docs/transformers/v5.6.0/ja/model_doc/big_bird#transformers.BigBirdForCausalLM) (BigBirdConfig model) - [BigBirdPegasusConfig](/docs/transformers/v5.6.0/ja/model_doc/bigbird_pegasus#transformers.BigBirdPegasusConfig) configuration class: [BigBirdPegasusForCausalLM](/docs/transformers/v5.6.0/ja/model_doc/bigbird_pegasus#transformers.BigBirdPegasusForCausalLM) (BigBirdPegasusConfig model) - [BioGptConfig](/docs/transformers/v5.6.0/ja/model_doc/biogpt#transformers.BioGptConfig) configuration class: [BioGptForCausalLM](/docs/transformers/v5.6.0/ja/model_doc/biogpt#transformers.BioGptForCausalLM) (BioGptConfig model) - `BitNetConfig` configuration class: `BitNetForCausalLM` (BitNetConfig model) - [BlenderbotConfig](/docs/transformers/v5.6.0/ja/model_doc/blenderbot#transformers.BlenderbotConfig) configuration class: [BlenderbotForCausalLM](/docs/transformers/v5.6.0/ja/model_doc/blenderbot#transformers.BlenderbotForCausalLM) (BlenderbotConfig model) - [BlenderbotSmallConfig](/docs/transformers/v5.6.0/ja/model_doc/blenderbot-small#transformers.BlenderbotSmallConfig) configuration class: [BlenderbotSmallForCausalLM](/docs/transformers/v5.6.0/ja/model_doc/blenderbot-small#transformers.BlenderbotSmallForCausalLM) (BlenderbotSmallConfig model) - [BloomConfig](/docs/transformers/v5.6.0/ja/model_doc/bloom#transformers.BloomConfig) configuration class: [BloomForCausalLM](/docs/transformers/v5.6.0/ja/model_doc/bloom#transformers.BloomForCausalLM) (BloomConfig model) - `BltConfig` configuration class: `BltForCausalLM` (BltConfig model) - [CTRLConfig](/docs/transformers/v5.6.0/ja/model_doc/ctrl#transformers.CTRLConfig) configuration class: [CTRLLMHeadModel](/docs/transformers/v5.6.0/ja/model_doc/ctrl#transformers.CTRLLMHeadModel) (CTRLConfig model) - [CamembertConfig](/docs/transformers/v5.6.0/ja/model_doc/camembert#transformers.CamembertConfig) configuration class: [CamembertForCausalLM](/docs/transformers/v5.6.0/ja/model_doc/camembert#transformers.CamembertForCausalLM) (CamembertConfig model) - [CodeGenConfig](/docs/transformers/v5.6.0/ja/model_doc/codegen#transformers.CodeGenConfig) configuration class: [CodeGenForCausalLM](/docs/transformers/v5.6.0/ja/model_doc/codegen#transformers.CodeGenForCausalLM) (CodeGenConfig model) - `Cohere2Config` configuration class: `Cohere2ForCausalLM` (Cohere2Config model) - `CohereConfig` configuration class: `CohereForCausalLM` (CohereConfig model) - [CpmAntConfig](/docs/transformers/v5.6.0/ja/model_doc/cpmant#transformers.CpmAntConfig) configuration class: [CpmAntForCausalLM](/docs/transformers/v5.6.0/ja/model_doc/cpmant#transformers.CpmAntForCausalLM) (CpmAntConfig model) - `CwmConfig` configuration class: `CwmForCausalLM` (CwmConfig model) - [Data2VecTextConfig](/docs/transformers/v5.6.0/ja/model_doc/data2vec#transformers.Data2VecTextConfig) configuration class: [Data2VecTextForCausalLM](/docs/transformers/v5.6.0/ja/model_doc/data2vec#transformers.Data2VecTextForCausalLM) (Data2VecTextConfig model) - `DbrxConfig` configuration class: `DbrxForCausalLM` (DbrxConfig model) - `DeepseekV2Config` configuration class: `DeepseekV2ForCausalLM` (DeepseekV2Config model) - `DeepseekV3Config` configuration class: `DeepseekV3ForCausalLM` (DeepseekV3Config model) - `DiffLlamaConfig` configuration class: `DiffLlamaForCausalLM` (DiffLlamaConfig model) - `DogeConfig` configuration class: `DogeForCausalLM` (DogeConfig model) - `Dots1Config` configuration class: `Dots1ForCausalLM` (Dots1Config model) - `ElectraConfig` configuration class: `ElectraForCausalLM` (ElectraConfig model) - `Emu3Config` configuration class: `Emu3ForCausalLM` (Emu3Config model) - `Ernie4_5Config` configuration class: `Ernie4_5ForCausalLM` (Ernie4_5Config model) - `Ernie4_5_MoeConfig` configuration class: `Ernie4_5_MoeForCausalLM` (Ernie4_5_MoeConfig model) - `ErnieConfig` configuration class: `ErnieForCausalLM` (ErnieConfig model) - `Exaone4Config` configuration class: `Exaone4ForCausalLM` (Exaone4Config model) - `ExaoneMoeConfig` configuration class: `ExaoneMoeForCausalLM` (ExaoneMoeConfig model) - `FalconConfig` configuration class: `FalconForCausalLM` (FalconConfig model) - `FalconH1Config` configuration class: `FalconH1ForCausalLM` (FalconH1Config model) - `FalconMambaConfig` configuration class: `FalconMambaForCausalLM` (FalconMambaConfig model) - `FlexOlmoConfig` configuration class: `FlexOlmoForCausalLM` (FlexOlmoConfig model) - `FuyuConfig` configuration class: `FuyuForCausalLM` (FuyuConfig model) - `GPT2Config` configuration class: `GPT2LMHeadModel` (GPT2Config model) - `GPTBigCodeConfig` configuration class: `GPTBigCodeForCausalLM` (GPTBigCodeConfig model) - `GPTJConfig` configuration class: `GPTJForCausalLM` (GPTJConfig model) - `GPTNeoConfig` configuration class: `GPTNeoForCausalLM` (GPTNeoConfig model) - `GPTNeoXConfig` configuration class: `GPTNeoXForCausalLM` (GPTNeoXConfig model) - `GPTNeoXJapaneseConfig` configuration class: `GPTNeoXJapaneseForCausalLM` (GPTNeoXJapaneseConfig model) - `Gemma2Config` configuration class: `Gemma2ForCausalLM` (Gemma2Config model) - `Gemma3Config` configuration class: `Gemma3ForConditionalGeneration` (Gemma3Config model) - `Gemma3TextConfig` configuration class: `Gemma3ForCausalLM` (Gemma3TextConfig model) - `Gemma3nConfig` configuration class: `Gemma3nForConditionalGeneration` (Gemma3nConfig model) - `Gemma3nTextConfig` configuration class: `Gemma3nForCausalLM` (Gemma3nTextConfig model) - `Gemma4Config` configuration class: `Gemma4ForConditionalGeneration` (Gemma4Config model) - `Gemma4TextConfig` configuration class: `Gemma4ForCausalLM` (Gemma4TextConfig model) - `GemmaConfig` configuration class: `GemmaForCausalLM` (GemmaConfig model) - `GitConfig` configuration class: `GitForCausalLM` (GitConfig model) - `Glm4Config` configuration class: `Glm4ForCausalLM` (Glm4Config model) - `Glm4MoeConfig` configuration class: `Glm4MoeForCausalLM` (Glm4MoeConfig model) - `Glm4MoeLiteConfig` configuration class: `Glm4MoeLiteForCausalLM` (Glm4MoeLiteConfig model) - `GlmConfig` configuration class: `GlmForCausalLM` (GlmConfig model) - `GlmMoeDsaConfig` configuration class: `GlmMoeDsaForCausalLM` (GlmMoeDsaConfig model) - `GotOcr2Config` configuration class: `GotOcr2ForConditionalGeneration` (GotOcr2Config model) - `GptOssConfig` configuration class: `GptOssForCausalLM` (GptOssConfig model) - `GraniteConfig` configuration class: `GraniteForCausalLM` (GraniteConfig model) - `GraniteMoeConfig` configuration class: `GraniteMoeForCausalLM` (GraniteMoeConfig model) - `GraniteMoeHybridConfig` configuration class: `GraniteMoeHybridForCausalLM` (GraniteMoeHybridConfig model) - `GraniteMoeSharedConfig` configuration class: `GraniteMoeSharedForCausalLM` (GraniteMoeSharedConfig model) - `HYV3Config` configuration class: `HYV3ForCausalLM` (HYV3Config model) - `HeliumConfig` configuration class: `HeliumForCausalLM` (HeliumConfig model) - `HunYuanDenseV1Config` configuration class: `HunYuanDenseV1ForCausalLM` (HunYuanDenseV1Config model) - `HunYuanMoEV1Config` configuration class: `HunYuanMoEV1ForCausalLM` (HunYuanMoEV1Config model) - `Jais2Config` configuration class: `Jais2ForCausalLM` (Jais2Config model) - `JambaConfig` configuration class: `JambaForCausalLM` (JambaConfig model) - `JetMoeConfig` configuration class: `JetMoeForCausalLM` (JetMoeConfig model) - `Lfm2Config` configuration class: `Lfm2ForCausalLM` (Lfm2Config model) - `Lfm2MoeConfig` configuration class: `Lfm2MoeForCausalLM` (Lfm2MoeConfig model) - `Llama4Config` configuration class: `Llama4ForCausalLM` (Llama4Config model) - `Llama4TextConfig` configuration class: `Llama4ForCausalLM` (Llama4TextConfig model) - `LlamaConfig` configuration class: `LlamaForCausalLM` (LlamaConfig model) - `LongcatFlashConfig` configuration class: `LongcatFlashForCausalLM` (LongcatFlashConfig model) - `MBartConfig` configuration class: `MBartForCausalLM` (MBartConfig model) - `Mamba2Config` configuration class: `Mamba2ForCausalLM` (Mamba2Config model) - `MambaConfig` configuration class: `MambaForCausalLM` (MambaConfig model) - `MarianConfig` configuration class: `MarianForCausalLM` (MarianConfig model) - `MegatronBertConfig` configuration class: `MegatronBertForCausalLM` (MegatronBertConfig model) - `MiniMaxConfig` configuration class: `MiniMaxForCausalLM` (MiniMaxConfig model) - `MiniMaxM2Config` configuration class: `MiniMaxM2ForCausalLM` (MiniMaxM2Config model) - `Ministral3Config` configuration class: `Ministral3ForCausalLM` (Ministral3Config model) - `MinistralConfig` configuration class: `MinistralForCausalLM` (MinistralConfig model) - `MistralConfig` configuration class: `MistralForCausalLM` (MistralConfig model) - `MixtralConfig` configuration class: `MixtralForCausalLM` (MixtralConfig model) - `MllamaConfig` configuration class: `MllamaForCausalLM` (MllamaConfig model) - `ModernBertDecoderConfig` configuration class: `ModernBertDecoderForCausalLM` (ModernBertDecoderConfig model) - `MoshiConfig` configuration class: `MoshiForCausalLM` (MoshiConfig model) - `MptConfig` configuration class: `MptForCausalLM` (MptConfig model) - `MusicgenConfig` configuration class: `MusicgenForCausalLM` (MusicgenConfig model) - `MusicgenMelodyConfig` configuration class: `MusicgenMelodyForCausalLM` (MusicgenMelodyConfig model) - `MvpConfig` configuration class: `MvpForCausalLM` (MvpConfig model) - `NanoChatConfig` configuration class: `NanoChatForCausalLM` (NanoChatConfig model) - `NemotronConfig` configuration class: `NemotronForCausalLM` (NemotronConfig model) - `NemotronHConfig` configuration class: `NemotronHForCausalLM` (NemotronHConfig model) - `OPTConfig` configuration class: `OPTForCausalLM` (OPTConfig model) - `Olmo2Config` configuration class: `Olmo2ForCausalLM` (Olmo2Config model) - `Olmo3Config` configuration class: `Olmo3ForCausalLM` (Olmo3Config model) - `OlmoConfig` configuration class: `OlmoForCausalLM` (OlmoConfig model) - `OlmoHybridConfig` configuration class: `OlmoHybridForCausalLM` (OlmoHybridConfig model) - `OlmoeConfig` configuration class: `OlmoeForCausalLM` (OlmoeConfig model) - `OpenAIGPTConfig` configuration class: `OpenAIGPTLMHeadModel` (OpenAIGPTConfig model) - `PLBartConfig` configuration class: `PLBartForCausalLM` (PLBartConfig model) - `PegasusConfig` configuration class: `PegasusForCausalLM` (PegasusConfig model) - `PersimmonConfig` configuration class: `PersimmonForCausalLM` (PersimmonConfig model) - `Phi3Config` configuration class: `Phi3ForCausalLM` (Phi3Config model) - `Phi4MultimodalConfig` configuration class: `Phi4MultimodalForCausalLM` (Phi4MultimodalConfig model) - `PhiConfig` configuration class: `PhiForCausalLM` (PhiConfig model) - `PhimoeConfig` configuration class: `PhimoeForCausalLM` (PhimoeConfig model) - `ProphetNetConfig` configuration class: `ProphetNetForCausalLM` (ProphetNetConfig model) - `Qwen2Config` configuration class: `Qwen2ForCausalLM` (Qwen2Config model) - `Qwen2MoeConfig` configuration class: `Qwen2MoeForCausalLM` (Qwen2MoeConfig model) - `Qwen3Config` configuration class: `Qwen3ForCausalLM` (Qwen3Config model) - `Qwen3MoeConfig` configuration class: `Qwen3MoeForCausalLM` (Qwen3MoeConfig model) - `Qwen3NextConfig` configuration class: `Qwen3NextForCausalLM` (Qwen3NextConfig model) - `Qwen3_5Config` configuration class: `Qwen3_5ForCausalLM` (Qwen3_5Config model) - `Qwen3_5MoeConfig` configuration class: `Qwen3_5MoeForCausalLM` (Qwen3_5MoeConfig model) - `Qwen3_5MoeTextConfig` configuration class: `Qwen3_5MoeForCausalLM` (Qwen3_5MoeTextConfig model) - `Qwen3_5TextConfig` configuration class: `Qwen3_5ForCausalLM` (Qwen3_5TextConfig model) - `RecurrentGemmaConfig` configuration class: `RecurrentGemmaForCausalLM` (RecurrentGemmaConfig model) - `ReformerConfig` configuration class: `ReformerModelWithLMHead` (ReformerConfig model) - `RemBertConfig` configuration class: `RemBertForCausalLM` (RemBertConfig model) - `RoCBertConfig` configuration class: `RoCBertForCausalLM` (RoCBertConfig model) - `RoFormerConfig` configuration class: `RoFormerForCausalLM` (RoFormerConfig model) - `RobertaConfig` configuration class: `RobertaForCausalLM` (RobertaConfig model) - `RobertaPreLayerNormConfig` configuration class: `RobertaPreLayerNormForCausalLM` (RobertaPreLayerNormConfig model) - `RwkvConfig` configuration class: `RwkvForCausalLM` (RwkvConfig model) - `SeedOssConfig` configuration class: `SeedOssForCausalLM` (SeedOssConfig model) - `SmolLM3Config` configuration class: `SmolLM3ForCausalLM` (SmolLM3Config model) - `SolarOpenConfig` configuration class: `SolarOpenForCausalLM` (SolarOpenConfig model) - `StableLmConfig` configuration class: `StableLmForCausalLM` (StableLmConfig model) - `Starcoder2Config` configuration class: `Starcoder2ForCausalLM` (Starcoder2Config model) - `TrOCRConfig` configuration class: `TrOCRForCausalLM` (TrOCRConfig model) - `VaultGemmaConfig` configuration class: `VaultGemmaForCausalLM` (VaultGemmaConfig model) - `WhisperConfig` configuration class: `WhisperForCausalLM` (WhisperConfig model) - `XGLMConfig` configuration class: `XGLMForCausalLM` (XGLMConfig model) - `XLMConfig` configuration class: `XLMWithLMHeadModel` (XLMConfig model) - `XLMRobertaConfig` configuration class: `XLMRobertaForCausalLM` (XLMRobertaConfig model) - `XLMRobertaXLConfig` configuration class: `XLMRobertaXLForCausalLM` (XLMRobertaXLConfig model) - `XLNetConfig` configuration class: `XLNetLMHeadModel` (XLNetConfig model) - `XmodConfig` configuration class: `XmodForCausalLM` (XmodConfig model) - `YoutuConfig` configuration class: `YoutuForCausalLM` (YoutuConfig model) - `Zamba2Config` configuration class: `Zamba2ForCausalLM` (Zamba2Config model) - `ZambaConfig` configuration class: `ZambaForCausalLM` (ZambaConfig model) - `xLSTMConfig` configuration class: `xLSTMForCausalLM` (xLSTMConfig model) - **attn_implementation** (`str`, *optional*) -- The attention implementation to use in the model (if relevant). Can be any of `"eager"` (manual implementation of the attention), `"sdpa"` (using [`F.scaled_dot_product_attention`](https://pytorch.org/docs/master/generated/torch.nn.functional.scaled_dot_product_attention.html)), `"flash_attention_2"` (using [Dao-AILab/flash-attention](https://github.com/Dao-AILab/flash-attention)), or `"flash_attention_3"` (using [Dao-AILab/flash-attention/hopper](https://github.com/Dao-AILab/flash-attention/tree/main/hopper)). By default, if available, SDPA will be used for torch>=2.1.1. The default is otherwise the manual `"eager"` implementation.0 Instantiates one of the model classes of the library (with a causal language modeling head) from a configuration. Note: Loading a model from its configuration file does **not** load the model weights. It only affects the model's configuration. Use [from_pretrained()](/docs/transformers/v5.6.0/ja/model_doc/auto#transformers.AutoModel.from_pretrained) to load the model weights. Examples: ```python >>> from transformers import AutoConfig, AutoModelForCausalLM >>> # Download configuration from huggingface.co and cache. >>> config = AutoConfig.from_pretrained("google-bert/bert-base-cased") >>> model = AutoModelForCausalLM.from_config(config) ``` **Parameters:** config ([PreTrainedConfig](/docs/transformers/v5.6.0/ja/main_classes/configuration#transformers.PreTrainedConfig)) : The model class to instantiate is selected based on the configuration class: - `AfmoeConfig` configuration class: `AfmoeForCausalLM` (AfmoeConfig model) - `ApertusConfig` configuration class: `ApertusForCausalLM` (ApertusConfig model) - `ArceeConfig` configuration class: `ArceeForCausalLM` (ArceeConfig model) - `AriaTextConfig` configuration class: `AriaTextForCausalLM` (AriaTextConfig model) - `BambaConfig` configuration class: `BambaForCausalLM` (BambaConfig model) - [BartConfig](/docs/transformers/v5.6.0/ja/model_doc/bart#transformers.BartConfig) configuration class: [BartForCausalLM](/docs/transformers/v5.6.0/ja/model_doc/bart#transformers.BartForCausalLM) (BartConfig model) - [BertConfig](/docs/transformers/v5.6.0/ja/model_doc/bert#transformers.BertConfig) configuration class: [BertLMHeadModel](/docs/transformers/v5.6.0/ja/model_doc/bert#transformers.BertLMHeadModel) (BertConfig model) - [BertGenerationConfig](/docs/transformers/v5.6.0/ja/model_doc/bert-generation#transformers.BertGenerationConfig) configuration class: [BertGenerationDecoder](/docs/transformers/v5.6.0/ja/model_doc/bert-generation#transformers.BertGenerationDecoder) (BertGenerationConfig model) - [BigBirdConfig](/docs/transformers/v5.6.0/ja/model_doc/big_bird#transformers.BigBirdConfig) configuration class: [BigBirdForCausalLM](/docs/transformers/v5.6.0/ja/model_doc/big_bird#transformers.BigBirdForCausalLM) (BigBirdConfig model) - [BigBirdPegasusConfig](/docs/transformers/v5.6.0/ja/model_doc/bigbird_pegasus#transformers.BigBirdPegasusConfig) configuration class: [BigBirdPegasusForCausalLM](/docs/transformers/v5.6.0/ja/model_doc/bigbird_pegasus#transformers.BigBirdPegasusForCausalLM) (BigBirdPegasusConfig model) - [BioGptConfig](/docs/transformers/v5.6.0/ja/model_doc/biogpt#transformers.BioGptConfig) configuration class: [BioGptForCausalLM](/docs/transformers/v5.6.0/ja/model_doc/biogpt#transformers.BioGptForCausalLM) (BioGptConfig model) - `BitNetConfig` configuration class: `BitNetForCausalLM` (BitNetConfig model) - [BlenderbotConfig](/docs/transformers/v5.6.0/ja/model_doc/blenderbot#transformers.BlenderbotConfig) configuration class: [BlenderbotForCausalLM](/docs/transformers/v5.6.0/ja/model_doc/blenderbot#transformers.BlenderbotForCausalLM) (BlenderbotConfig model) - [BlenderbotSmallConfig](/docs/transformers/v5.6.0/ja/model_doc/blenderbot-small#transformers.BlenderbotSmallConfig) configuration class: [BlenderbotSmallForCausalLM](/docs/transformers/v5.6.0/ja/model_doc/blenderbot-small#transformers.BlenderbotSmallForCausalLM) (BlenderbotSmallConfig model) - [BloomConfig](/docs/transformers/v5.6.0/ja/model_doc/bloom#transformers.BloomConfig) configuration class: [BloomForCausalLM](/docs/transformers/v5.6.0/ja/model_doc/bloom#transformers.BloomForCausalLM) (BloomConfig model) - `BltConfig` configuration class: `BltForCausalLM` (BltConfig model) - [CTRLConfig](/docs/transformers/v5.6.0/ja/model_doc/ctrl#transformers.CTRLConfig) configuration class: [CTRLLMHeadModel](/docs/transformers/v5.6.0/ja/model_doc/ctrl#transformers.CTRLLMHeadModel) (CTRLConfig model) - [CamembertConfig](/docs/transformers/v5.6.0/ja/model_doc/camembert#transformers.CamembertConfig) configuration class: [CamembertForCausalLM](/docs/transformers/v5.6.0/ja/model_doc/camembert#transformers.CamembertForCausalLM) (CamembertConfig model) - [CodeGenConfig](/docs/transformers/v5.6.0/ja/model_doc/codegen#transformers.CodeGenConfig) configuration class: [CodeGenForCausalLM](/docs/transformers/v5.6.0/ja/model_doc/codegen#transformers.CodeGenForCausalLM) (CodeGenConfig model) - `Cohere2Config` configuration class: `Cohere2ForCausalLM` (Cohere2Config model) - `CohereConfig` configuration class: `CohereForCausalLM` (CohereConfig model) - [CpmAntConfig](/docs/transformers/v5.6.0/ja/model_doc/cpmant#transformers.CpmAntConfig) configuration class: [CpmAntForCausalLM](/docs/transformers/v5.6.0/ja/model_doc/cpmant#transformers.CpmAntForCausalLM) (CpmAntConfig model) - `CwmConfig` configuration class: `CwmForCausalLM` (CwmConfig model) - [Data2VecTextConfig](/docs/transformers/v5.6.0/ja/model_doc/data2vec#transformers.Data2VecTextConfig) configuration class: [Data2VecTextForCausalLM](/docs/transformers/v5.6.0/ja/model_doc/data2vec#transformers.Data2VecTextForCausalLM) (Data2VecTextConfig model) - `DbrxConfig` configuration class: `DbrxForCausalLM` (DbrxConfig model) - `DeepseekV2Config` configuration class: `DeepseekV2ForCausalLM` (DeepseekV2Config model) - `DeepseekV3Config` configuration class: `DeepseekV3ForCausalLM` (DeepseekV3Config model) - `DiffLlamaConfig` configuration class: `DiffLlamaForCausalLM` (DiffLlamaConfig model) - `DogeConfig` configuration class: `DogeForCausalLM` (DogeConfig model) - `Dots1Config` configuration class: `Dots1ForCausalLM` (Dots1Config model) - `ElectraConfig` configuration class: `ElectraForCausalLM` (ElectraConfig model) - `Emu3Config` configuration class: `Emu3ForCausalLM` (Emu3Config model) - `Ernie4_5Config` configuration class: `Ernie4_5ForCausalLM` (Ernie4_5Config model) - `Ernie4_5_MoeConfig` configuration class: `Ernie4_5_MoeForCausalLM` (Ernie4_5_MoeConfig model) - `ErnieConfig` configuration class: `ErnieForCausalLM` (ErnieConfig model) - `Exaone4Config` configuration class: `Exaone4ForCausalLM` (Exaone4Config model) - `ExaoneMoeConfig` configuration class: `ExaoneMoeForCausalLM` (ExaoneMoeConfig model) - `FalconConfig` configuration class: `FalconForCausalLM` (FalconConfig model) - `FalconH1Config` configuration class: `FalconH1ForCausalLM` (FalconH1Config model) - `FalconMambaConfig` configuration class: `FalconMambaForCausalLM` (FalconMambaConfig model) - `FlexOlmoConfig` configuration class: `FlexOlmoForCausalLM` (FlexOlmoConfig model) - `FuyuConfig` configuration class: `FuyuForCausalLM` (FuyuConfig model) - `GPT2Config` configuration class: `GPT2LMHeadModel` (GPT2Config model) - `GPTBigCodeConfig` configuration class: `GPTBigCodeForCausalLM` (GPTBigCodeConfig model) - `GPTJConfig` configuration class: `GPTJForCausalLM` (GPTJConfig model) - `GPTNeoConfig` configuration class: `GPTNeoForCausalLM` (GPTNeoConfig model) - `GPTNeoXConfig` configuration class: `GPTNeoXForCausalLM` (GPTNeoXConfig model) - `GPTNeoXJapaneseConfig` configuration class: `GPTNeoXJapaneseForCausalLM` (GPTNeoXJapaneseConfig model) - `Gemma2Config` configuration class: `Gemma2ForCausalLM` (Gemma2Config model) - `Gemma3Config` configuration class: `Gemma3ForConditionalGeneration` (Gemma3Config model) - `Gemma3TextConfig` configuration class: `Gemma3ForCausalLM` (Gemma3TextConfig model) - `Gemma3nConfig` configuration class: `Gemma3nForConditionalGeneration` (Gemma3nConfig model) - `Gemma3nTextConfig` configuration class: `Gemma3nForCausalLM` (Gemma3nTextConfig model) - `Gemma4Config` configuration class: `Gemma4ForConditionalGeneration` (Gemma4Config model) - `Gemma4TextConfig` configuration class: `Gemma4ForCausalLM` (Gemma4TextConfig model) - `GemmaConfig` configuration class: `GemmaForCausalLM` (GemmaConfig model) - `GitConfig` configuration class: `GitForCausalLM` (GitConfig model) - `Glm4Config` configuration class: `Glm4ForCausalLM` (Glm4Config model) - `Glm4MoeConfig` configuration class: `Glm4MoeForCausalLM` (Glm4MoeConfig model) - `Glm4MoeLiteConfig` configuration class: `Glm4MoeLiteForCausalLM` (Glm4MoeLiteConfig model) - `GlmConfig` configuration class: `GlmForCausalLM` (GlmConfig model) - `GlmMoeDsaConfig` configuration class: `GlmMoeDsaForCausalLM` (GlmMoeDsaConfig model) - `GotOcr2Config` configuration class: `GotOcr2ForConditionalGeneration` (GotOcr2Config model) - `GptOssConfig` configuration class: `GptOssForCausalLM` (GptOssConfig model) - `GraniteConfig` configuration class: `GraniteForCausalLM` (GraniteConfig model) - `GraniteMoeConfig` configuration class: `GraniteMoeForCausalLM` (GraniteMoeConfig model) - `GraniteMoeHybridConfig` configuration class: `GraniteMoeHybridForCausalLM` (GraniteMoeHybridConfig model) - `GraniteMoeSharedConfig` configuration class: `GraniteMoeSharedForCausalLM` (GraniteMoeSharedConfig model) - `HYV3Config` configuration class: `HYV3ForCausalLM` (HYV3Config model) - `HeliumConfig` configuration class: `HeliumForCausalLM` (HeliumConfig model) - `HunYuanDenseV1Config` configuration class: `HunYuanDenseV1ForCausalLM` (HunYuanDenseV1Config model) - `HunYuanMoEV1Config` configuration class: `HunYuanMoEV1ForCausalLM` (HunYuanMoEV1Config model) - `Jais2Config` configuration class: `Jais2ForCausalLM` (Jais2Config model) - `JambaConfig` configuration class: `JambaForCausalLM` (JambaConfig model) - `JetMoeConfig` configuration class: `JetMoeForCausalLM` (JetMoeConfig model) - `Lfm2Config` configuration class: `Lfm2ForCausalLM` (Lfm2Config model) - `Lfm2MoeConfig` configuration class: `Lfm2MoeForCausalLM` (Lfm2MoeConfig model) - `Llama4Config` configuration class: `Llama4ForCausalLM` (Llama4Config model) - `Llama4TextConfig` configuration class: `Llama4ForCausalLM` (Llama4TextConfig model) - `LlamaConfig` configuration class: `LlamaForCausalLM` (LlamaConfig model) - `LongcatFlashConfig` configuration class: `LongcatFlashForCausalLM` (LongcatFlashConfig model) - `MBartConfig` configuration class: `MBartForCausalLM` (MBartConfig model) - `Mamba2Config` configuration class: `Mamba2ForCausalLM` (Mamba2Config model) - `MambaConfig` configuration class: `MambaForCausalLM` (MambaConfig model) - `MarianConfig` configuration class: `MarianForCausalLM` (MarianConfig model) - `MegatronBertConfig` configuration class: `MegatronBertForCausalLM` (MegatronBertConfig model) - `MiniMaxConfig` configuration class: `MiniMaxForCausalLM` (MiniMaxConfig model) - `MiniMaxM2Config` configuration class: `MiniMaxM2ForCausalLM` (MiniMaxM2Config model) - `Ministral3Config` configuration class: `Ministral3ForCausalLM` (Ministral3Config model) - `MinistralConfig` configuration class: `MinistralForCausalLM` (MinistralConfig model) - `MistralConfig` configuration class: `MistralForCausalLM` (MistralConfig model) - `MixtralConfig` configuration class: `MixtralForCausalLM` (MixtralConfig model) - `MllamaConfig` configuration class: `MllamaForCausalLM` (MllamaConfig model) - `ModernBertDecoderConfig` configuration class: `ModernBertDecoderForCausalLM` (ModernBertDecoderConfig model) - `MoshiConfig` configuration class: `MoshiForCausalLM` (MoshiConfig model) - `MptConfig` configuration class: `MptForCausalLM` (MptConfig model) - `MusicgenConfig` configuration class: `MusicgenForCausalLM` (MusicgenConfig model) - `MusicgenMelodyConfig` configuration class: `MusicgenMelodyForCausalLM` (MusicgenMelodyConfig model) - `MvpConfig` configuration class: `MvpForCausalLM` (MvpConfig model) - `NanoChatConfig` configuration class: `NanoChatForCausalLM` (NanoChatConfig model) - `NemotronConfig` configuration class: `NemotronForCausalLM` (NemotronConfig model) - `NemotronHConfig` configuration class: `NemotronHForCausalLM` (NemotronHConfig model) - `OPTConfig` configuration class: `OPTForCausalLM` (OPTConfig model) - `Olmo2Config` configuration class: `Olmo2ForCausalLM` (Olmo2Config model) - `Olmo3Config` configuration class: `Olmo3ForCausalLM` (Olmo3Config model) - `OlmoConfig` configuration class: `OlmoForCausalLM` (OlmoConfig model) - `OlmoHybridConfig` configuration class: `OlmoHybridForCausalLM` (OlmoHybridConfig model) - `OlmoeConfig` configuration class: `OlmoeForCausalLM` (OlmoeConfig model) - `OpenAIGPTConfig` configuration class: `OpenAIGPTLMHeadModel` (OpenAIGPTConfig model) - `PLBartConfig` configuration class: `PLBartForCausalLM` (PLBartConfig model) - `PegasusConfig` configuration class: `PegasusForCausalLM` (PegasusConfig model) - `PersimmonConfig` configuration class: `PersimmonForCausalLM` (PersimmonConfig model) - `Phi3Config` configuration class: `Phi3ForCausalLM` (Phi3Config model) - `Phi4MultimodalConfig` configuration class: `Phi4MultimodalForCausalLM` (Phi4MultimodalConfig model) - `PhiConfig` configuration class: `PhiForCausalLM` (PhiConfig model) - `PhimoeConfig` configuration class: `PhimoeForCausalLM` (PhimoeConfig model) - `ProphetNetConfig` configuration class: `ProphetNetForCausalLM` (ProphetNetConfig model) - `Qwen2Config` configuration class: `Qwen2ForCausalLM` (Qwen2Config model) - `Qwen2MoeConfig` configuration class: `Qwen2MoeForCausalLM` (Qwen2MoeConfig model) - `Qwen3Config` configuration class: `Qwen3ForCausalLM` (Qwen3Config model) - `Qwen3MoeConfig` configuration class: `Qwen3MoeForCausalLM` (Qwen3MoeConfig model) - `Qwen3NextConfig` configuration class: `Qwen3NextForCausalLM` (Qwen3NextConfig model) - `Qwen3_5Config` configuration class: `Qwen3_5ForCausalLM` (Qwen3_5Config model) - `Qwen3_5MoeConfig` configuration class: `Qwen3_5MoeForCausalLM` (Qwen3_5MoeConfig model) - `Qwen3_5MoeTextConfig` configuration class: `Qwen3_5MoeForCausalLM` (Qwen3_5MoeTextConfig model) - `Qwen3_5TextConfig` configuration class: `Qwen3_5ForCausalLM` (Qwen3_5TextConfig model) - `RecurrentGemmaConfig` configuration class: `RecurrentGemmaForCausalLM` (RecurrentGemmaConfig model) - `ReformerConfig` configuration class: `ReformerModelWithLMHead` (ReformerConfig model) - `RemBertConfig` configuration class: `RemBertForCausalLM` (RemBertConfig model) - `RoCBertConfig` configuration class: `RoCBertForCausalLM` (RoCBertConfig model) - `RoFormerConfig` configuration class: `RoFormerForCausalLM` (RoFormerConfig model) - `RobertaConfig` configuration class: `RobertaForCausalLM` (RobertaConfig model) - `RobertaPreLayerNormConfig` configuration class: `RobertaPreLayerNormForCausalLM` (RobertaPreLayerNormConfig model) - `RwkvConfig` configuration class: `RwkvForCausalLM` (RwkvConfig model) - `SeedOssConfig` configuration class: `SeedOssForCausalLM` (SeedOssConfig model) - `SmolLM3Config` configuration class: `SmolLM3ForCausalLM` (SmolLM3Config model) - `SolarOpenConfig` configuration class: `SolarOpenForCausalLM` (SolarOpenConfig model) - `StableLmConfig` configuration class: `StableLmForCausalLM` (StableLmConfig model) - `Starcoder2Config` configuration class: `Starcoder2ForCausalLM` (Starcoder2Config model) - `TrOCRConfig` configuration class: `TrOCRForCausalLM` (TrOCRConfig model) - `VaultGemmaConfig` configuration class: `VaultGemmaForCausalLM` (VaultGemmaConfig model) - `WhisperConfig` configuration class: `WhisperForCausalLM` (WhisperConfig model) - `XGLMConfig` configuration class: `XGLMForCausalLM` (XGLMConfig model) - `XLMConfig` configuration class: `XLMWithLMHeadModel` (XLMConfig model) - `XLMRobertaConfig` configuration class: `XLMRobertaForCausalLM` (XLMRobertaConfig model) - `XLMRobertaXLConfig` configuration class: `XLMRobertaXLForCausalLM` (XLMRobertaXLConfig model) - `XLNetConfig` configuration class: `XLNetLMHeadModel` (XLNetConfig model) - `XmodConfig` configuration class: `XmodForCausalLM` (XmodConfig model) - `YoutuConfig` configuration class: `YoutuForCausalLM` (YoutuConfig model) - `Zamba2Config` configuration class: `Zamba2ForCausalLM` (Zamba2Config model) - `ZambaConfig` configuration class: `ZambaForCausalLM` (ZambaConfig model) - `xLSTMConfig` configuration class: `xLSTMForCausalLM` (xLSTMConfig model) attn_implementation (`str`, *optional*) : The attention implementation to use in the model (if relevant). Can be any of `"eager"` (manual implementation of the attention), `"sdpa"` (using [`F.scaled_dot_product_attention`](https://pytorch.org/docs/master/generated/torch.nn.functional.scaled_dot_product_attention.html)), `"flash_attention_2"` (using [Dao-AILab/flash-attention](https://github.com/Dao-AILab/flash-attention)), or `"flash_attention_3"` (using [Dao-AILab/flash-attention/hopper](https://github.com/Dao-AILab/flash-attention/tree/main/hopper)). By default, if available, SDPA will be used for torch>=2.1.1. The default is otherwise the manual `"eager"` implementation. #### from_pretrained[[transformers.AutoModelForCausalLM.from_pretrained]] [Source](https://github.com/huggingface/transformers/blob/v5.6.0/src/transformers/models/auto/auto_factory.py#L253) Instantiate one of the model classes of the library (with a causal language modeling head) from a pretrained model. The model class to instantiate is selected based on the `model_type` property of the config object (either passed as an argument or loaded from `pretrained_model_name_or_path` if possible), or when it's missing, by falling back to using pattern matching on `pretrained_model_name_or_path`: - **afmoe** -- `AfmoeForCausalLM` (AfmoeConfig model) - **apertus** -- `ApertusForCausalLM` (ApertusConfig model) - **arcee** -- `ArceeForCausalLM` (ArceeConfig model) - **aria_text** -- `AriaTextForCausalLM` (AriaTextConfig model) - **bamba** -- `BambaForCausalLM` (BambaConfig model) - **bart** -- [BartForCausalLM](/docs/transformers/v5.6.0/ja/model_doc/bart#transformers.BartForCausalLM) (BartConfig model) - **bert** -- [BertLMHeadModel](/docs/transformers/v5.6.0/ja/model_doc/bert#transformers.BertLMHeadModel) (BertConfig model) - **bert-generation** -- [BertGenerationDecoder](/docs/transformers/v5.6.0/ja/model_doc/bert-generation#transformers.BertGenerationDecoder) (BertGenerationConfig model) - **big_bird** -- [BigBirdForCausalLM](/docs/transformers/v5.6.0/ja/model_doc/big_bird#transformers.BigBirdForCausalLM) (BigBirdConfig model) - **bigbird_pegasus** -- [BigBirdPegasusForCausalLM](/docs/transformers/v5.6.0/ja/model_doc/bigbird_pegasus#transformers.BigBirdPegasusForCausalLM) (BigBirdPegasusConfig model) - **biogpt** -- [BioGptForCausalLM](/docs/transformers/v5.6.0/ja/model_doc/biogpt#transformers.BioGptForCausalLM) (BioGptConfig model) - **bitnet** -- `BitNetForCausalLM` (BitNetConfig model) - **blenderbot** -- [BlenderbotForCausalLM](/docs/transformers/v5.6.0/ja/model_doc/blenderbot#transformers.BlenderbotForCausalLM) (BlenderbotConfig model) - **blenderbot-small** -- [BlenderbotSmallForCausalLM](/docs/transformers/v5.6.0/ja/model_doc/blenderbot-small#transformers.BlenderbotSmallForCausalLM) (BlenderbotSmallConfig model) - **bloom** -- [BloomForCausalLM](/docs/transformers/v5.6.0/ja/model_doc/bloom#transformers.BloomForCausalLM) (BloomConfig model) - **blt** -- `BltForCausalLM` (BltConfig model) - **camembert** -- [CamembertForCausalLM](/docs/transformers/v5.6.0/ja/model_doc/camembert#transformers.CamembertForCausalLM) (CamembertConfig model) - **codegen** -- [CodeGenForCausalLM](/docs/transformers/v5.6.0/ja/model_doc/codegen#transformers.CodeGenForCausalLM) (CodeGenConfig model) - **cohere** -- `CohereForCausalLM` (CohereConfig model) - **cohere2** -- `Cohere2ForCausalLM` (Cohere2Config model) - **cpmant** -- [CpmAntForCausalLM](/docs/transformers/v5.6.0/ja/model_doc/cpmant#transformers.CpmAntForCausalLM) (CpmAntConfig model) - **ctrl** -- [CTRLLMHeadModel](/docs/transformers/v5.6.0/ja/model_doc/ctrl#transformers.CTRLLMHeadModel) (CTRLConfig model) - **cwm** -- `CwmForCausalLM` (CwmConfig model) - **data2vec-text** -- [Data2VecTextForCausalLM](/docs/transformers/v5.6.0/ja/model_doc/data2vec#transformers.Data2VecTextForCausalLM) (Data2VecTextConfig model) - **dbrx** -- `DbrxForCausalLM` (DbrxConfig model) - **deepseek_v2** -- `DeepseekV2ForCausalLM` (DeepseekV2Config model) - **deepseek_v3** -- `DeepseekV3ForCausalLM` (DeepseekV3Config model) - **diffllama** -- `DiffLlamaForCausalLM` (DiffLlamaConfig model) - **doge** -- `DogeForCausalLM` (DogeConfig model) - **dots1** -- `Dots1ForCausalLM` (Dots1Config model) - **electra** -- `ElectraForCausalLM` (ElectraConfig model) - **emu3** -- `Emu3ForCausalLM` (Emu3Config model) - **ernie** -- `ErnieForCausalLM` (ErnieConfig model) - **ernie4_5** -- `Ernie4_5ForCausalLM` (Ernie4_5Config model) - **ernie4_5_moe** -- `Ernie4_5_MoeForCausalLM` (Ernie4_5_MoeConfig model) - **exaone4** -- `Exaone4ForCausalLM` (Exaone4Config model) - **exaone_moe** -- `ExaoneMoeForCausalLM` (ExaoneMoeConfig model) - **falcon** -- `FalconForCausalLM` (FalconConfig model) - **falcon_h1** -- `FalconH1ForCausalLM` (FalconH1Config model) - **falcon_mamba** -- `FalconMambaForCausalLM` (FalconMambaConfig model) - **flex_olmo** -- `FlexOlmoForCausalLM` (FlexOlmoConfig model) - **fuyu** -- `FuyuForCausalLM` (FuyuConfig model) - **gemma** -- `GemmaForCausalLM` (GemmaConfig model) - **gemma2** -- `Gemma2ForCausalLM` (Gemma2Config model) - **gemma3** -- `Gemma3ForConditionalGeneration` (Gemma3Config model) - **gemma3_text** -- `Gemma3ForCausalLM` (Gemma3TextConfig model) - **gemma3n** -- `Gemma3nForConditionalGeneration` (Gemma3nConfig model) - **gemma3n_text** -- `Gemma3nForCausalLM` (Gemma3nTextConfig model) - **gemma4** -- `Gemma4ForConditionalGeneration` (Gemma4Config model) - **gemma4_text** -- `Gemma4ForCausalLM` (Gemma4TextConfig model) - **git** -- `GitForCausalLM` (GitConfig model) - **glm** -- `GlmForCausalLM` (GlmConfig model) - **glm4** -- `Glm4ForCausalLM` (Glm4Config model) - **glm4_moe** -- `Glm4MoeForCausalLM` (Glm4MoeConfig model) - **glm4_moe_lite** -- `Glm4MoeLiteForCausalLM` (Glm4MoeLiteConfig model) - **glm_moe_dsa** -- `GlmMoeDsaForCausalLM` (GlmMoeDsaConfig model) - **got_ocr2** -- `GotOcr2ForConditionalGeneration` (GotOcr2Config model) - **gpt-sw3** -- `GPT2LMHeadModel` (GPT2Config model) - **gpt2** -- `GPT2LMHeadModel` (GPT2Config model) - **gpt_bigcode** -- `GPTBigCodeForCausalLM` (GPTBigCodeConfig model) - **gpt_neo** -- `GPTNeoForCausalLM` (GPTNeoConfig model) - **gpt_neox** -- `GPTNeoXForCausalLM` (GPTNeoXConfig model) - **gpt_neox_japanese** -- `GPTNeoXJapaneseForCausalLM` (GPTNeoXJapaneseConfig model) - **gpt_oss** -- `GptOssForCausalLM` (GptOssConfig model) - **gptj** -- `GPTJForCausalLM` (GPTJConfig model) - **granite** -- `GraniteForCausalLM` (GraniteConfig model) - **granitemoe** -- `GraniteMoeForCausalLM` (GraniteMoeConfig model) - **granitemoehybrid** -- `GraniteMoeHybridForCausalLM` (GraniteMoeHybridConfig model) - **granitemoeshared** -- `GraniteMoeSharedForCausalLM` (GraniteMoeSharedConfig model) - **helium** -- `HeliumForCausalLM` (HeliumConfig model) - **hunyuan_v1_dense** -- `HunYuanDenseV1ForCausalLM` (HunYuanDenseV1Config model) - **hunyuan_v1_moe** -- `HunYuanMoEV1ForCausalLM` (HunYuanMoEV1Config model) - **hy_v3** -- `HYV3ForCausalLM` (HYV3Config model) - **jais2** -- `Jais2ForCausalLM` (Jais2Config model) - **jamba** -- `JambaForCausalLM` (JambaConfig model) - **jetmoe** -- `JetMoeForCausalLM` (JetMoeConfig model) - **lfm2** -- `Lfm2ForCausalLM` (Lfm2Config model) - **lfm2_moe** -- `Lfm2MoeForCausalLM` (Lfm2MoeConfig model) - **llama** -- `LlamaForCausalLM` (LlamaConfig model) - **llama4** -- `Llama4ForCausalLM` (Llama4Config model) - **llama4_text** -- `Llama4ForCausalLM` (Llama4TextConfig model) - **longcat_flash** -- `LongcatFlashForCausalLM` (LongcatFlashConfig model) - **mamba** -- `MambaForCausalLM` (MambaConfig model) - **mamba2** -- `Mamba2ForCausalLM` (Mamba2Config model) - **marian** -- `MarianForCausalLM` (MarianConfig model) - **mbart** -- `MBartForCausalLM` (MBartConfig model) - **megatron-bert** -- `MegatronBertForCausalLM` (MegatronBertConfig model) - **minimax** -- `MiniMaxForCausalLM` (MiniMaxConfig model) - **minimax_m2** -- `MiniMaxM2ForCausalLM` (MiniMaxM2Config model) - **ministral** -- `MinistralForCausalLM` (MinistralConfig model) - **ministral3** -- `Ministral3ForCausalLM` (Ministral3Config model) - **mistral** -- `MistralForCausalLM` (MistralConfig model) - **mixtral** -- `MixtralForCausalLM` (MixtralConfig model) - **mllama** -- `MllamaForCausalLM` (MllamaConfig model) - **modernbert-decoder** -- `ModernBertDecoderForCausalLM` (ModernBertDecoderConfig model) - **moshi** -- `MoshiForCausalLM` (MoshiConfig model) - **mpt** -- `MptForCausalLM` (MptConfig model) - **musicgen** -- `MusicgenForCausalLM` (MusicgenConfig model) - **musicgen_melody** -- `MusicgenMelodyForCausalLM` (MusicgenMelodyConfig model) - **mvp** -- `MvpForCausalLM` (MvpConfig model) - **nanochat** -- `NanoChatForCausalLM` (NanoChatConfig model) - **nemotron** -- `NemotronForCausalLM` (NemotronConfig model) - **nemotron_h** -- `NemotronHForCausalLM` (NemotronHConfig model) - **olmo** -- `OlmoForCausalLM` (OlmoConfig model) - **olmo2** -- `Olmo2ForCausalLM` (Olmo2Config model) - **olmo3** -- `Olmo3ForCausalLM` (Olmo3Config model) - **olmo_hybrid** -- `OlmoHybridForCausalLM` (OlmoHybridConfig model) - **olmoe** -- `OlmoeForCausalLM` (OlmoeConfig model) - **openai-gpt** -- `OpenAIGPTLMHeadModel` (OpenAIGPTConfig model) - **opt** -- `OPTForCausalLM` (OPTConfig model) - **pegasus** -- `PegasusForCausalLM` (PegasusConfig model) - **persimmon** -- `PersimmonForCausalLM` (PersimmonConfig model) - **phi** -- `PhiForCausalLM` (PhiConfig model) - **phi3** -- `Phi3ForCausalLM` (Phi3Config model) - **phi4_multimodal** -- `Phi4MultimodalForCausalLM` (Phi4MultimodalConfig model) - **phimoe** -- `PhimoeForCausalLM` (PhimoeConfig model) - **plbart** -- `PLBartForCausalLM` (PLBartConfig model) - **prophetnet** -- `ProphetNetForCausalLM` (ProphetNetConfig model) - **qwen2** -- `Qwen2ForCausalLM` (Qwen2Config model) - **qwen2_moe** -- `Qwen2MoeForCausalLM` (Qwen2MoeConfig model) - **qwen3** -- `Qwen3ForCausalLM` (Qwen3Config model) - **qwen3_5** -- `Qwen3_5ForCausalLM` (Qwen3_5Config model) - **qwen3_5_moe** -- `Qwen3_5MoeForCausalLM` (Qwen3_5MoeConfig model) - **qwen3_5_moe_text** -- `Qwen3_5MoeForCausalLM` (Qwen3_5MoeTextConfig model) - **qwen3_5_text** -- `Qwen3_5ForCausalLM` (Qwen3_5TextConfig model) - **qwen3_moe** -- `Qwen3MoeForCausalLM` (Qwen3MoeConfig model) - **qwen3_next** -- `Qwen3NextForCausalLM` (Qwen3NextConfig model) - **recurrent_gemma** -- `RecurrentGemmaForCausalLM` (RecurrentGemmaConfig model) - **reformer** -- `ReformerModelWithLMHead` (ReformerConfig model) - **rembert** -- `RemBertForCausalLM` (RemBertConfig model) - **roberta** -- `RobertaForCausalLM` (RobertaConfig model) - **roberta-prelayernorm** -- `RobertaPreLayerNormForCausalLM` (RobertaPreLayerNormConfig model) - **roc_bert** -- `RoCBertForCausalLM` (RoCBertConfig model) - **roformer** -- `RoFormerForCausalLM` (RoFormerConfig model) - **rwkv** -- `RwkvForCausalLM` (RwkvConfig model) - **seed_oss** -- `SeedOssForCausalLM` (SeedOssConfig model) - **smollm3** -- `SmolLM3ForCausalLM` (SmolLM3Config model) - **solar_open** -- `SolarOpenForCausalLM` (SolarOpenConfig model) - **stablelm** -- `StableLmForCausalLM` (StableLmConfig model) - **starcoder2** -- `Starcoder2ForCausalLM` (Starcoder2Config model) - **trocr** -- `TrOCRForCausalLM` (TrOCRConfig model) - **vaultgemma** -- `VaultGemmaForCausalLM` (VaultGemmaConfig model) - **whisper** -- `WhisperForCausalLM` (WhisperConfig model) - **xglm** -- `XGLMForCausalLM` (XGLMConfig model) - **xlm** -- `XLMWithLMHeadModel` (XLMConfig model) - **xlm-roberta** -- `XLMRobertaForCausalLM` (XLMRobertaConfig model) - **xlm-roberta-xl** -- `XLMRobertaXLForCausalLM` (XLMRobertaXLConfig model) - **xlnet** -- `XLNetLMHeadModel` (XLNetConfig model) - **xlstm** -- `xLSTMForCausalLM` (xLSTMConfig model) - **xmod** -- `XmodForCausalLM` (XmodConfig model) - **youtu** -- `YoutuForCausalLM` (YoutuConfig model) - **zamba** -- `ZambaForCausalLM` (ZambaConfig model) - **zamba2** -- `Zamba2ForCausalLM` (Zamba2Config model) The model is set in evaluation mode by default using `model.eval()` (so for instance, dropout modules are deactivated). To train the model, you should first set it back in training mode with `model.train()` Examples: ```python >>> from transformers import AutoConfig, AutoModelForCausalLM >>> # Download model and configuration from huggingface.co and cache. >>> model = AutoModelForCausalLM.from_pretrained("google-bert/bert-base-cased") >>> # Update configuration during loading >>> model = AutoModelForCausalLM.from_pretrained("google-bert/bert-base-cased", output_attentions=True) >>> model.config.output_attentions True ``` **Parameters:** pretrained_model_name_or_path (`str` or `os.PathLike`) : Can be either: - A string, the *model id* of a pretrained model hosted inside a model repo on huggingface.co. - A path to a *directory* containing model weights saved using [save_pretrained()](/docs/transformers/v5.6.0/ja/main_classes/model#transformers.PreTrainedModel.save_pretrained), e.g., `./my_model_directory/`. model_args (additional positional arguments, *optional*) : Will be passed along to the underlying model `__init__()` method. config ([PreTrainedConfig](/docs/transformers/v5.6.0/ja/main_classes/configuration#transformers.PreTrainedConfig), *optional*) : Configuration for the model to use instead of an automatically loaded configuration. Configuration can be automatically loaded when: - The model is a model provided by the library (loaded with the *model id* string of a pretrained model). - The model was saved using [save_pretrained()](/docs/transformers/v5.6.0/ja/main_classes/model#transformers.PreTrainedModel.save_pretrained) and is reloaded by supplying the save directory. - The model is loaded by supplying a local directory as `pretrained_model_name_or_path` and a configuration JSON file named *config.json* is found in the directory. state_dict (*dict[str, torch.Tensor]*, *optional*) : A state dictionary to use instead of a state dictionary loaded from saved weights file. This option can be used if you want to create a model from a pretrained configuration but load your own weights. In this case though, you should check if using [save_pretrained()](/docs/transformers/v5.6.0/ja/main_classes/model#transformers.PreTrainedModel.save_pretrained) and [from_pretrained()](/docs/transformers/v5.6.0/ja/main_classes/model#transformers.PreTrainedModel.from_pretrained) is not a simpler option. cache_dir (`str` or `os.PathLike`, *optional*) : Path to a directory in which a downloaded pretrained model configuration should be cached if the standard cache should not be used. force_download (`bool`, *optional*, defaults to `False`) : Whether or not to force the (re-)download of the model weights and configuration files, overriding the cached versions if they exist. proxies (`dict[str, str]`, *optional*) : A dictionary of proxy servers to use by protocol or endpoint, e.g., `{'http': 'foo.bar:3128', 'http://hostname': 'foo.bar:4012'}`. The proxies are used on each request. output_loading_info(`bool`, *optional*, defaults to `False`) : Whether or not to also return a dictionary containing missing keys, unexpected keys and error messages. local_files_only(`bool`, *optional*, defaults to `False`) : Whether or not to only look at local files (e.g., not try downloading the model). revision (`str`, *optional*, defaults to `"main"`) : The specific model version to use. It can be a branch name, a tag name, or a commit id, since we use a git-based system for storing models and other artifacts on huggingface.co, so `revision` can be any identifier allowed by git. trust_remote_code (`bool`, *optional*, defaults to `False`) : Whether or not to allow for custom models defined on the Hub in their own modeling files. This option should only be set to `True` for repositories you trust and in which you have read the code, as it will execute code present on the Hub on your local machine. code_revision (`str`, *optional*, defaults to `"main"`) : The specific revision to use for the code on the Hub, if the code leaves in a different repository than the rest of the model. It can be a branch name, a tag name, or a commit id, since we use a git-based system for storing models and other artifacts on huggingface.co, so `revision` can be any identifier allowed by git. kwargs (additional keyword arguments, *optional*) : Can be used to update the configuration object (after it being loaded) and initiate the model (e.g., `output_attentions=True`). Behaves differently depending on whether a `config` is provided or automatically loaded: - If a configuration is provided with `config`, `**kwargs` will be directly passed to the underlying model's `__init__` method (we assume all relevant updates to the configuration have already been done) - If a configuration is not provided, `kwargs` will be first passed to the configuration class initialization function ([from_pretrained()](/docs/transformers/v5.6.0/ja/main_classes/configuration#transformers.PreTrainedConfig.from_pretrained)). Each key of `kwargs` that corresponds to a configuration attribute will be used to override said attribute with the supplied `kwargs` value. Remaining keys that do not correspond to any configuration attribute will be passed to the underlying model's `__init__` function. ### AutoModelForMaskedLM[[transformers.AutoModelForMaskedLM]] #### transformers.AutoModelForMaskedLM[[transformers.AutoModelForMaskedLM]] [Source](https://github.com/huggingface/transformers/blob/v5.6.0/src/transformers/models/auto/modeling_auto.py#L2010) This is a generic model class that will be instantiated as one of the model classes of the library (with a masked language modeling head) when created with the [from_pretrained()](/docs/transformers/v5.6.0/ja/model_doc/auto#transformers.AutoModel.from_pretrained) class method or the [from_config()](/docs/transformers/v5.6.0/ja/model_doc/auto#transformers.AutoModel.from_config) class method. This class cannot be instantiated directly using `__init__()` (throws an error). from_configtransformers.AutoModelForMaskedLM.from_confighttps://github.com/huggingface/transformers/blob/v5.6.0/src/transformers/models/auto/auto_factory.py#L206[{"name": "**kwargs", "val": ""}]- **config** ([PreTrainedConfig](/docs/transformers/v5.6.0/ja/main_classes/configuration#transformers.PreTrainedConfig)) -- The model class to instantiate is selected based on the configuration class: - [AlbertConfig](/docs/transformers/v5.6.0/ja/model_doc/albert#transformers.AlbertConfig) configuration class: [AlbertForMaskedLM](/docs/transformers/v5.6.0/ja/model_doc/albert#transformers.AlbertForMaskedLM) (AlbertConfig model) - [BartConfig](/docs/transformers/v5.6.0/ja/model_doc/bart#transformers.BartConfig) configuration class: [BartForConditionalGeneration](/docs/transformers/v5.6.0/ja/model_doc/bart#transformers.BartForConditionalGeneration) (BartConfig model) - [BertConfig](/docs/transformers/v5.6.0/ja/model_doc/bert#transformers.BertConfig) configuration class: [BertForMaskedLM](/docs/transformers/v5.6.0/ja/model_doc/bert#transformers.BertForMaskedLM) (BertConfig model) - [BigBirdConfig](/docs/transformers/v5.6.0/ja/model_doc/big_bird#transformers.BigBirdConfig) configuration class: [BigBirdForMaskedLM](/docs/transformers/v5.6.0/ja/model_doc/big_bird#transformers.BigBirdForMaskedLM) (BigBirdConfig model) - [CamembertConfig](/docs/transformers/v5.6.0/ja/model_doc/camembert#transformers.CamembertConfig) configuration class: [CamembertForMaskedLM](/docs/transformers/v5.6.0/ja/model_doc/camembert#transformers.CamembertForMaskedLM) (CamembertConfig model) - [ConvBertConfig](/docs/transformers/v5.6.0/ja/model_doc/convbert#transformers.ConvBertConfig) configuration class: [ConvBertForMaskedLM](/docs/transformers/v5.6.0/ja/model_doc/convbert#transformers.ConvBertForMaskedLM) (ConvBertConfig model) - [Data2VecTextConfig](/docs/transformers/v5.6.0/ja/model_doc/data2vec#transformers.Data2VecTextConfig) configuration class: [Data2VecTextForMaskedLM](/docs/transformers/v5.6.0/ja/model_doc/data2vec#transformers.Data2VecTextForMaskedLM) (Data2VecTextConfig model) - [DebertaConfig](/docs/transformers/v5.6.0/ja/model_doc/deberta#transformers.DebertaConfig) configuration class: [DebertaForMaskedLM](/docs/transformers/v5.6.0/ja/model_doc/deberta#transformers.DebertaForMaskedLM) (DebertaConfig model) - [DebertaV2Config](/docs/transformers/v5.6.0/ja/model_doc/deberta-v2#transformers.DebertaV2Config) configuration class: [DebertaV2ForMaskedLM](/docs/transformers/v5.6.0/ja/model_doc/deberta-v2#transformers.DebertaV2ForMaskedLM) (DebertaV2Config model) - `DistilBertConfig` configuration class: `DistilBertForMaskedLM` (DistilBertConfig model) - `ElectraConfig` configuration class: `ElectraForMaskedLM` (ElectraConfig model) - `ErnieConfig` configuration class: `ErnieForMaskedLM` (ErnieConfig model) - `EsmConfig` configuration class: `EsmForMaskedLM` (EsmConfig model) - `EuroBertConfig` configuration class: `EuroBertForMaskedLM` (EuroBertConfig model) - `FNetConfig` configuration class: `FNetForMaskedLM` (FNetConfig model) - `FlaubertConfig` configuration class: `FlaubertWithLMHeadModel` (FlaubertConfig model) - `FunnelConfig` configuration class: `FunnelForMaskedLM` (FunnelConfig model) - `IBertConfig` configuration class: `IBertForMaskedLM` (IBertConfig model) - `JinaEmbeddingsV3Config` configuration class: `JinaEmbeddingsV3ForMaskedLM` (JinaEmbeddingsV3Config model) - `LayoutLMConfig` configuration class: `LayoutLMForMaskedLM` (LayoutLMConfig model) - `LongformerConfig` configuration class: `LongformerForMaskedLM` (LongformerConfig model) - `LukeConfig` configuration class: `LukeForMaskedLM` (LukeConfig model) - `MBartConfig` configuration class: `MBartForConditionalGeneration` (MBartConfig model) - `MPNetConfig` configuration class: `MPNetForMaskedLM` (MPNetConfig model) - `MegatronBertConfig` configuration class: `MegatronBertForMaskedLM` (MegatronBertConfig model) - `MobileBertConfig` configuration class: `MobileBertForMaskedLM` (MobileBertConfig model) - `ModernBertConfig` configuration class: `ModernBertForMaskedLM` (ModernBertConfig model) - `ModernVBertConfig` configuration class: `ModernVBertForMaskedLM` (ModernVBertConfig model) - `MraConfig` configuration class: `MraForMaskedLM` (MraConfig model) - `MvpConfig` configuration class: `MvpForConditionalGeneration` (MvpConfig model) - `NomicBertConfig` configuration class: `NomicBertForMaskedLM` (NomicBertConfig model) - `NystromformerConfig` configuration class: `NystromformerForMaskedLM` (NystromformerConfig model) - `PerceiverConfig` configuration class: `PerceiverForMaskedLM` (PerceiverConfig model) - `ReformerConfig` configuration class: `ReformerForMaskedLM` (ReformerConfig model) - `RemBertConfig` configuration class: `RemBertForMaskedLM` (RemBertConfig model) - `RoCBertConfig` configuration class: `RoCBertForMaskedLM` (RoCBertConfig model) - `RoFormerConfig` configuration class: `RoFormerForMaskedLM` (RoFormerConfig model) - `RobertaConfig` configuration class: `RobertaForMaskedLM` (RobertaConfig model) - `RobertaPreLayerNormConfig` configuration class: `RobertaPreLayerNormForMaskedLM` (RobertaPreLayerNormConfig model) - `SqueezeBertConfig` configuration class: `SqueezeBertForMaskedLM` (SqueezeBertConfig model) - `TapasConfig` configuration class: `TapasForMaskedLM` (TapasConfig model) - `XLMConfig` configuration class: `XLMWithLMHeadModel` (XLMConfig model) - `XLMRobertaConfig` configuration class: `XLMRobertaForMaskedLM` (XLMRobertaConfig model) - `XLMRobertaXLConfig` configuration class: `XLMRobertaXLForMaskedLM` (XLMRobertaXLConfig model) - `XmodConfig` configuration class: `XmodForMaskedLM` (XmodConfig model) - `YosoConfig` configuration class: `YosoForMaskedLM` (YosoConfig model) - **attn_implementation** (`str`, *optional*) -- The attention implementation to use in the model (if relevant). Can be any of `"eager"` (manual implementation of the attention), `"sdpa"` (using [`F.scaled_dot_product_attention`](https://pytorch.org/docs/master/generated/torch.nn.functional.scaled_dot_product_attention.html)), `"flash_attention_2"` (using [Dao-AILab/flash-attention](https://github.com/Dao-AILab/flash-attention)), or `"flash_attention_3"` (using [Dao-AILab/flash-attention/hopper](https://github.com/Dao-AILab/flash-attention/tree/main/hopper)). By default, if available, SDPA will be used for torch>=2.1.1. The default is otherwise the manual `"eager"` implementation.0 Instantiates one of the model classes of the library (with a masked language modeling head) from a configuration. Note: Loading a model from its configuration file does **not** load the model weights. It only affects the model's configuration. Use [from_pretrained()](/docs/transformers/v5.6.0/ja/model_doc/auto#transformers.AutoModel.from_pretrained) to load the model weights. Examples: ```python >>> from transformers import AutoConfig, AutoModelForMaskedLM >>> # Download configuration from huggingface.co and cache. >>> config = AutoConfig.from_pretrained("google-bert/bert-base-cased") >>> model = AutoModelForMaskedLM.from_config(config) ``` **Parameters:** config ([PreTrainedConfig](/docs/transformers/v5.6.0/ja/main_classes/configuration#transformers.PreTrainedConfig)) : The model class to instantiate is selected based on the configuration class: - [AlbertConfig](/docs/transformers/v5.6.0/ja/model_doc/albert#transformers.AlbertConfig) configuration class: [AlbertForMaskedLM](/docs/transformers/v5.6.0/ja/model_doc/albert#transformers.AlbertForMaskedLM) (AlbertConfig model) - [BartConfig](/docs/transformers/v5.6.0/ja/model_doc/bart#transformers.BartConfig) configuration class: [BartForConditionalGeneration](/docs/transformers/v5.6.0/ja/model_doc/bart#transformers.BartForConditionalGeneration) (BartConfig model) - [BertConfig](/docs/transformers/v5.6.0/ja/model_doc/bert#transformers.BertConfig) configuration class: [BertForMaskedLM](/docs/transformers/v5.6.0/ja/model_doc/bert#transformers.BertForMaskedLM) (BertConfig model) - [BigBirdConfig](/docs/transformers/v5.6.0/ja/model_doc/big_bird#transformers.BigBirdConfig) configuration class: [BigBirdForMaskedLM](/docs/transformers/v5.6.0/ja/model_doc/big_bird#transformers.BigBirdForMaskedLM) (BigBirdConfig model) - [CamembertConfig](/docs/transformers/v5.6.0/ja/model_doc/camembert#transformers.CamembertConfig) configuration class: [CamembertForMaskedLM](/docs/transformers/v5.6.0/ja/model_doc/camembert#transformers.CamembertForMaskedLM) (CamembertConfig model) - [ConvBertConfig](/docs/transformers/v5.6.0/ja/model_doc/convbert#transformers.ConvBertConfig) configuration class: [ConvBertForMaskedLM](/docs/transformers/v5.6.0/ja/model_doc/convbert#transformers.ConvBertForMaskedLM) (ConvBertConfig model) - [Data2VecTextConfig](/docs/transformers/v5.6.0/ja/model_doc/data2vec#transformers.Data2VecTextConfig) configuration class: [Data2VecTextForMaskedLM](/docs/transformers/v5.6.0/ja/model_doc/data2vec#transformers.Data2VecTextForMaskedLM) (Data2VecTextConfig model) - [DebertaConfig](/docs/transformers/v5.6.0/ja/model_doc/deberta#transformers.DebertaConfig) configuration class: [DebertaForMaskedLM](/docs/transformers/v5.6.0/ja/model_doc/deberta#transformers.DebertaForMaskedLM) (DebertaConfig model) - [DebertaV2Config](/docs/transformers/v5.6.0/ja/model_doc/deberta-v2#transformers.DebertaV2Config) configuration class: [DebertaV2ForMaskedLM](/docs/transformers/v5.6.0/ja/model_doc/deberta-v2#transformers.DebertaV2ForMaskedLM) (DebertaV2Config model) - `DistilBertConfig` configuration class: `DistilBertForMaskedLM` (DistilBertConfig model) - `ElectraConfig` configuration class: `ElectraForMaskedLM` (ElectraConfig model) - `ErnieConfig` configuration class: `ErnieForMaskedLM` (ErnieConfig model) - `EsmConfig` configuration class: `EsmForMaskedLM` (EsmConfig model) - `EuroBertConfig` configuration class: `EuroBertForMaskedLM` (EuroBertConfig model) - `FNetConfig` configuration class: `FNetForMaskedLM` (FNetConfig model) - `FlaubertConfig` configuration class: `FlaubertWithLMHeadModel` (FlaubertConfig model) - `FunnelConfig` configuration class: `FunnelForMaskedLM` (FunnelConfig model) - `IBertConfig` configuration class: `IBertForMaskedLM` (IBertConfig model) - `JinaEmbeddingsV3Config` configuration class: `JinaEmbeddingsV3ForMaskedLM` (JinaEmbeddingsV3Config model) - `LayoutLMConfig` configuration class: `LayoutLMForMaskedLM` (LayoutLMConfig model) - `LongformerConfig` configuration class: `LongformerForMaskedLM` (LongformerConfig model) - `LukeConfig` configuration class: `LukeForMaskedLM` (LukeConfig model) - `MBartConfig` configuration class: `MBartForConditionalGeneration` (MBartConfig model) - `MPNetConfig` configuration class: `MPNetForMaskedLM` (MPNetConfig model) - `MegatronBertConfig` configuration class: `MegatronBertForMaskedLM` (MegatronBertConfig model) - `MobileBertConfig` configuration class: `MobileBertForMaskedLM` (MobileBertConfig model) - `ModernBertConfig` configuration class: `ModernBertForMaskedLM` (ModernBertConfig model) - `ModernVBertConfig` configuration class: `ModernVBertForMaskedLM` (ModernVBertConfig model) - `MraConfig` configuration class: `MraForMaskedLM` (MraConfig model) - `MvpConfig` configuration class: `MvpForConditionalGeneration` (MvpConfig model) - `NomicBertConfig` configuration class: `NomicBertForMaskedLM` (NomicBertConfig model) - `NystromformerConfig` configuration class: `NystromformerForMaskedLM` (NystromformerConfig model) - `PerceiverConfig` configuration class: `PerceiverForMaskedLM` (PerceiverConfig model) - `ReformerConfig` configuration class: `ReformerForMaskedLM` (ReformerConfig model) - `RemBertConfig` configuration class: `RemBertForMaskedLM` (RemBertConfig model) - `RoCBertConfig` configuration class: `RoCBertForMaskedLM` (RoCBertConfig model) - `RoFormerConfig` configuration class: `RoFormerForMaskedLM` (RoFormerConfig model) - `RobertaConfig` configuration class: `RobertaForMaskedLM` (RobertaConfig model) - `RobertaPreLayerNormConfig` configuration class: `RobertaPreLayerNormForMaskedLM` (RobertaPreLayerNormConfig model) - `SqueezeBertConfig` configuration class: `SqueezeBertForMaskedLM` (SqueezeBertConfig model) - `TapasConfig` configuration class: `TapasForMaskedLM` (TapasConfig model) - `XLMConfig` configuration class: `XLMWithLMHeadModel` (XLMConfig model) - `XLMRobertaConfig` configuration class: `XLMRobertaForMaskedLM` (XLMRobertaConfig model) - `XLMRobertaXLConfig` configuration class: `XLMRobertaXLForMaskedLM` (XLMRobertaXLConfig model) - `XmodConfig` configuration class: `XmodForMaskedLM` (XmodConfig model) - `YosoConfig` configuration class: `YosoForMaskedLM` (YosoConfig model) attn_implementation (`str`, *optional*) : The attention implementation to use in the model (if relevant). Can be any of `"eager"` (manual implementation of the attention), `"sdpa"` (using [`F.scaled_dot_product_attention`](https://pytorch.org/docs/master/generated/torch.nn.functional.scaled_dot_product_attention.html)), `"flash_attention_2"` (using [Dao-AILab/flash-attention](https://github.com/Dao-AILab/flash-attention)), or `"flash_attention_3"` (using [Dao-AILab/flash-attention/hopper](https://github.com/Dao-AILab/flash-attention/tree/main/hopper)). By default, if available, SDPA will be used for torch>=2.1.1. The default is otherwise the manual `"eager"` implementation. #### from_pretrained[[transformers.AutoModelForMaskedLM.from_pretrained]] [Source](https://github.com/huggingface/transformers/blob/v5.6.0/src/transformers/models/auto/auto_factory.py#L253) Instantiate one of the model classes of the library (with a masked language modeling head) from a pretrained model. The model class to instantiate is selected based on the `model_type` property of the config object (either passed as an argument or loaded from `pretrained_model_name_or_path` if possible), or when it's missing, by falling back to using pattern matching on `pretrained_model_name_or_path`: - **albert** -- [AlbertForMaskedLM](/docs/transformers/v5.6.0/ja/model_doc/albert#transformers.AlbertForMaskedLM) (AlbertConfig model) - **bart** -- [BartForConditionalGeneration](/docs/transformers/v5.6.0/ja/model_doc/bart#transformers.BartForConditionalGeneration) (BartConfig model) - **bert** -- [BertForMaskedLM](/docs/transformers/v5.6.0/ja/model_doc/bert#transformers.BertForMaskedLM) (BertConfig model) - **big_bird** -- [BigBirdForMaskedLM](/docs/transformers/v5.6.0/ja/model_doc/big_bird#transformers.BigBirdForMaskedLM) (BigBirdConfig model) - **camembert** -- [CamembertForMaskedLM](/docs/transformers/v5.6.0/ja/model_doc/camembert#transformers.CamembertForMaskedLM) (CamembertConfig model) - **convbert** -- [ConvBertForMaskedLM](/docs/transformers/v5.6.0/ja/model_doc/convbert#transformers.ConvBertForMaskedLM) (ConvBertConfig model) - **data2vec-text** -- [Data2VecTextForMaskedLM](/docs/transformers/v5.6.0/ja/model_doc/data2vec#transformers.Data2VecTextForMaskedLM) (Data2VecTextConfig model) - **deberta** -- [DebertaForMaskedLM](/docs/transformers/v5.6.0/ja/model_doc/deberta#transformers.DebertaForMaskedLM) (DebertaConfig model) - **deberta-v2** -- [DebertaV2ForMaskedLM](/docs/transformers/v5.6.0/ja/model_doc/deberta-v2#transformers.DebertaV2ForMaskedLM) (DebertaV2Config model) - **distilbert** -- `DistilBertForMaskedLM` (DistilBertConfig model) - **electra** -- `ElectraForMaskedLM` (ElectraConfig model) - **ernie** -- `ErnieForMaskedLM` (ErnieConfig model) - **esm** -- `EsmForMaskedLM` (EsmConfig model) - **eurobert** -- `EuroBertForMaskedLM` (EuroBertConfig model) - **flaubert** -- `FlaubertWithLMHeadModel` (FlaubertConfig model) - **fnet** -- `FNetForMaskedLM` (FNetConfig model) - **funnel** -- `FunnelForMaskedLM` (FunnelConfig model) - **ibert** -- `IBertForMaskedLM` (IBertConfig model) - **jina_embeddings_v3** -- `JinaEmbeddingsV3ForMaskedLM` (JinaEmbeddingsV3Config model) - **layoutlm** -- `LayoutLMForMaskedLM` (LayoutLMConfig model) - **longformer** -- `LongformerForMaskedLM` (LongformerConfig model) - **luke** -- `LukeForMaskedLM` (LukeConfig model) - **mbart** -- `MBartForConditionalGeneration` (MBartConfig model) - **megatron-bert** -- `MegatronBertForMaskedLM` (MegatronBertConfig model) - **mobilebert** -- `MobileBertForMaskedLM` (MobileBertConfig model) - **modernbert** -- `ModernBertForMaskedLM` (ModernBertConfig model) - **modernvbert** -- `ModernVBertForMaskedLM` (ModernVBertConfig model) - **mpnet** -- `MPNetForMaskedLM` (MPNetConfig model) - **mra** -- `MraForMaskedLM` (MraConfig model) - **mvp** -- `MvpForConditionalGeneration` (MvpConfig model) - **nomic_bert** -- `NomicBertForMaskedLM` (NomicBertConfig model) - **nystromformer** -- `NystromformerForMaskedLM` (NystromformerConfig model) - **perceiver** -- `PerceiverForMaskedLM` (PerceiverConfig model) - **reformer** -- `ReformerForMaskedLM` (ReformerConfig model) - **rembert** -- `RemBertForMaskedLM` (RemBertConfig model) - **roberta** -- `RobertaForMaskedLM` (RobertaConfig model) - **roberta-prelayernorm** -- `RobertaPreLayerNormForMaskedLM` (RobertaPreLayerNormConfig model) - **roc_bert** -- `RoCBertForMaskedLM` (RoCBertConfig model) - **roformer** -- `RoFormerForMaskedLM` (RoFormerConfig model) - **squeezebert** -- `SqueezeBertForMaskedLM` (SqueezeBertConfig model) - **tapas** -- `TapasForMaskedLM` (TapasConfig model) - **xlm** -- `XLMWithLMHeadModel` (XLMConfig model) - **xlm-roberta** -- `XLMRobertaForMaskedLM` (XLMRobertaConfig model) - **xlm-roberta-xl** -- `XLMRobertaXLForMaskedLM` (XLMRobertaXLConfig model) - **xmod** -- `XmodForMaskedLM` (XmodConfig model) - **yoso** -- `YosoForMaskedLM` (YosoConfig model) The model is set in evaluation mode by default using `model.eval()` (so for instance, dropout modules are deactivated). To train the model, you should first set it back in training mode with `model.train()` Examples: ```python >>> from transformers import AutoConfig, AutoModelForMaskedLM >>> # Download model and configuration from huggingface.co and cache. >>> model = AutoModelForMaskedLM.from_pretrained("google-bert/bert-base-cased") >>> # Update configuration during loading >>> model = AutoModelForMaskedLM.from_pretrained("google-bert/bert-base-cased", output_attentions=True) >>> model.config.output_attentions True ``` **Parameters:** pretrained_model_name_or_path (`str` or `os.PathLike`) : Can be either: - A string, the *model id* of a pretrained model hosted inside a model repo on huggingface.co. - A path to a *directory* containing model weights saved using [save_pretrained()](/docs/transformers/v5.6.0/ja/main_classes/model#transformers.PreTrainedModel.save_pretrained), e.g., `./my_model_directory/`. model_args (additional positional arguments, *optional*) : Will be passed along to the underlying model `__init__()` method. config ([PreTrainedConfig](/docs/transformers/v5.6.0/ja/main_classes/configuration#transformers.PreTrainedConfig), *optional*) : Configuration for the model to use instead of an automatically loaded configuration. Configuration can be automatically loaded when: - The model is a model provided by the library (loaded with the *model id* string of a pretrained model). - The model was saved using [save_pretrained()](/docs/transformers/v5.6.0/ja/main_classes/model#transformers.PreTrainedModel.save_pretrained) and is reloaded by supplying the save directory. - The model is loaded by supplying a local directory as `pretrained_model_name_or_path` and a configuration JSON file named *config.json* is found in the directory. state_dict (*dict[str, torch.Tensor]*, *optional*) : A state dictionary to use instead of a state dictionary loaded from saved weights file. This option can be used if you want to create a model from a pretrained configuration but load your own weights. In this case though, you should check if using [save_pretrained()](/docs/transformers/v5.6.0/ja/main_classes/model#transformers.PreTrainedModel.save_pretrained) and [from_pretrained()](/docs/transformers/v5.6.0/ja/main_classes/model#transformers.PreTrainedModel.from_pretrained) is not a simpler option. cache_dir (`str` or `os.PathLike`, *optional*) : Path to a directory in which a downloaded pretrained model configuration should be cached if the standard cache should not be used. force_download (`bool`, *optional*, defaults to `False`) : Whether or not to force the (re-)download of the model weights and configuration files, overriding the cached versions if they exist. proxies (`dict[str, str]`, *optional*) : A dictionary of proxy servers to use by protocol or endpoint, e.g., `{'http': 'foo.bar:3128', 'http://hostname': 'foo.bar:4012'}`. The proxies are used on each request. output_loading_info(`bool`, *optional*, defaults to `False`) : Whether or not to also return a dictionary containing missing keys, unexpected keys and error messages. local_files_only(`bool`, *optional*, defaults to `False`) : Whether or not to only look at local files (e.g., not try downloading the model). revision (`str`, *optional*, defaults to `"main"`) : The specific model version to use. It can be a branch name, a tag name, or a commit id, since we use a git-based system for storing models and other artifacts on huggingface.co, so `revision` can be any identifier allowed by git. trust_remote_code (`bool`, *optional*, defaults to `False`) : Whether or not to allow for custom models defined on the Hub in their own modeling files. This option should only be set to `True` for repositories you trust and in which you have read the code, as it will execute code present on the Hub on your local machine. code_revision (`str`, *optional*, defaults to `"main"`) : The specific revision to use for the code on the Hub, if the code leaves in a different repository than the rest of the model. It can be a branch name, a tag name, or a commit id, since we use a git-based system for storing models and other artifacts on huggingface.co, so `revision` can be any identifier allowed by git. kwargs (additional keyword arguments, *optional*) : Can be used to update the configuration object (after it being loaded) and initiate the model (e.g., `output_attentions=True`). Behaves differently depending on whether a `config` is provided or automatically loaded: - If a configuration is provided with `config`, `**kwargs` will be directly passed to the underlying model's `__init__` method (we assume all relevant updates to the configuration have already been done) - If a configuration is not provided, `kwargs` will be first passed to the configuration class initialization function ([from_pretrained()](/docs/transformers/v5.6.0/ja/main_classes/configuration#transformers.PreTrainedConfig.from_pretrained)). Each key of `kwargs` that corresponds to a configuration attribute will be used to override said attribute with the supplied `kwargs` value. Remaining keys that do not correspond to any configuration attribute will be passed to the underlying model's `__init__` function. ### AutoModelForMaskGeneration[[transformers.AutoModelForMaskGeneration]] #### transformers.AutoModelForMaskGeneration[[transformers.AutoModelForMaskGeneration]] [Source](https://github.com/huggingface/transformers/blob/v5.6.0/src/transformers/models/auto/modeling_auto.py#L1959) ### AutoModelForSeq2SeqLM[[transformers.AutoModelForSeq2SeqLM]] #### transformers.AutoModelForSeq2SeqLM[[transformers.AutoModelForSeq2SeqLM]] [Source](https://github.com/huggingface/transformers/blob/v5.6.0/src/transformers/models/auto/modeling_auto.py#L2017) This is a generic model class that will be instantiated as one of the model classes of the library (with a sequence-to-sequence language modeling head) when created with the [from_pretrained()](/docs/transformers/v5.6.0/ja/model_doc/auto#transformers.AutoModel.from_pretrained) class method or the [from_config()](/docs/transformers/v5.6.0/ja/model_doc/auto#transformers.AutoModel.from_config) class method. This class cannot be instantiated directly using `__init__()` (throws an error). from_configtransformers.AutoModelForSeq2SeqLM.from_confighttps://github.com/huggingface/transformers/blob/v5.6.0/src/transformers/models/auto/auto_factory.py#L206[{"name": "**kwargs", "val": ""}]- **config** ([PreTrainedConfig](/docs/transformers/v5.6.0/ja/main_classes/configuration#transformers.PreTrainedConfig)) -- The model class to instantiate is selected based on the configuration class: - `AudioFlamingo3Config` configuration class: `AudioFlamingo3ForConditionalGeneration` (AudioFlamingo3Config model) - [BartConfig](/docs/transformers/v5.6.0/ja/model_doc/bart#transformers.BartConfig) configuration class: [BartForConditionalGeneration](/docs/transformers/v5.6.0/ja/model_doc/bart#transformers.BartForConditionalGeneration) (BartConfig model) - [BigBirdPegasusConfig](/docs/transformers/v5.6.0/ja/model_doc/bigbird_pegasus#transformers.BigBirdPegasusConfig) configuration class: [BigBirdPegasusForConditionalGeneration](/docs/transformers/v5.6.0/ja/model_doc/bigbird_pegasus#transformers.BigBirdPegasusForConditionalGeneration) (BigBirdPegasusConfig model) - [BlenderbotConfig](/docs/transformers/v5.6.0/ja/model_doc/blenderbot#transformers.BlenderbotConfig) configuration class: [BlenderbotForConditionalGeneration](/docs/transformers/v5.6.0/ja/model_doc/blenderbot#transformers.BlenderbotForConditionalGeneration) (BlenderbotConfig model) - [BlenderbotSmallConfig](/docs/transformers/v5.6.0/ja/model_doc/blenderbot-small#transformers.BlenderbotSmallConfig) configuration class: [BlenderbotSmallForConditionalGeneration](/docs/transformers/v5.6.0/ja/model_doc/blenderbot-small#transformers.BlenderbotSmallForConditionalGeneration) (BlenderbotSmallConfig model) - `EncoderDecoderConfig` configuration class: `EncoderDecoderModel` (EncoderDecoderConfig model) - `FSMTConfig` configuration class: `FSMTForConditionalGeneration` (FSMTConfig model) - `GlmAsrConfig` configuration class: `GlmAsrForConditionalGeneration` (GlmAsrConfig model) - `GraniteSpeechConfig` configuration class: `GraniteSpeechForConditionalGeneration` (GraniteSpeechConfig model) - `LEDConfig` configuration class: `LEDForConditionalGeneration` (LEDConfig model) - `LongT5Config` configuration class: `LongT5ForConditionalGeneration` (LongT5Config model) - `M2M100Config` configuration class: `M2M100ForConditionalGeneration` (M2M100Config model) - `MBartConfig` configuration class: `MBartForConditionalGeneration` (MBartConfig model) - `MT5Config` configuration class: `MT5ForConditionalGeneration` (MT5Config model) - `MarianConfig` configuration class: `MarianMTModel` (MarianConfig model) - `MusicFlamingoConfig` configuration class: `MusicFlamingoForConditionalGeneration` (MusicFlamingoConfig model) - `MvpConfig` configuration class: `MvpForConditionalGeneration` (MvpConfig model) - `NllbMoeConfig` configuration class: `NllbMoeForConditionalGeneration` (NllbMoeConfig model) - `PLBartConfig` configuration class: `PLBartForConditionalGeneration` (PLBartConfig model) - `PegasusConfig` configuration class: `PegasusForConditionalGeneration` (PegasusConfig model) - `PegasusXConfig` configuration class: `PegasusXForConditionalGeneration` (PegasusXConfig model) - `ProphetNetConfig` configuration class: `ProphetNetForConditionalGeneration` (ProphetNetConfig model) - `Qwen2AudioConfig` configuration class: `Qwen2AudioForConditionalGeneration` (Qwen2AudioConfig model) - `SeamlessM4TConfig` configuration class: `SeamlessM4TForTextToText` (SeamlessM4TConfig model) - `SeamlessM4Tv2Config` configuration class: `SeamlessM4Tv2ForTextToText` (SeamlessM4Tv2Config model) - `SwitchTransformersConfig` configuration class: `SwitchTransformersForConditionalGeneration` (SwitchTransformersConfig model) - `T5Config` configuration class: `T5ForConditionalGeneration` (T5Config model) - `T5Gemma2Config` configuration class: `T5Gemma2ForConditionalGeneration` (T5Gemma2Config model) - `T5GemmaConfig` configuration class: `T5GemmaForConditionalGeneration` (T5GemmaConfig model) - `UMT5Config` configuration class: `UMT5ForConditionalGeneration` (UMT5Config model) - `VibeVoiceAsrConfig` configuration class: `VibeVoiceAsrForConditionalGeneration` (VibeVoiceAsrConfig model) - `VoxtralConfig` configuration class: `VoxtralForConditionalGeneration` (VoxtralConfig model) - `VoxtralRealtimeConfig` configuration class: `VoxtralRealtimeForConditionalGeneration` (VoxtralRealtimeConfig model) - **attn_implementation** (`str`, *optional*) -- The attention implementation to use in the model (if relevant). Can be any of `"eager"` (manual implementation of the attention), `"sdpa"` (using [`F.scaled_dot_product_attention`](https://pytorch.org/docs/master/generated/torch.nn.functional.scaled_dot_product_attention.html)), `"flash_attention_2"` (using [Dao-AILab/flash-attention](https://github.com/Dao-AILab/flash-attention)), or `"flash_attention_3"` (using [Dao-AILab/flash-attention/hopper](https://github.com/Dao-AILab/flash-attention/tree/main/hopper)). By default, if available, SDPA will be used for torch>=2.1.1. The default is otherwise the manual `"eager"` implementation.0 Instantiates one of the model classes of the library (with a sequence-to-sequence language modeling head) from a configuration. Note: Loading a model from its configuration file does **not** load the model weights. It only affects the model's configuration. Use [from_pretrained()](/docs/transformers/v5.6.0/ja/model_doc/auto#transformers.AutoModel.from_pretrained) to load the model weights. Examples: ```python >>> from transformers import AutoConfig, AutoModelForSeq2SeqLM >>> # Download configuration from huggingface.co and cache. >>> config = AutoConfig.from_pretrained("google-t5/t5-base") >>> model = AutoModelForSeq2SeqLM.from_config(config) ``` **Parameters:** config ([PreTrainedConfig](/docs/transformers/v5.6.0/ja/main_classes/configuration#transformers.PreTrainedConfig)) : The model class to instantiate is selected based on the configuration class: - `AudioFlamingo3Config` configuration class: `AudioFlamingo3ForConditionalGeneration` (AudioFlamingo3Config model) - [BartConfig](/docs/transformers/v5.6.0/ja/model_doc/bart#transformers.BartConfig) configuration class: [BartForConditionalGeneration](/docs/transformers/v5.6.0/ja/model_doc/bart#transformers.BartForConditionalGeneration) (BartConfig model) - [BigBirdPegasusConfig](/docs/transformers/v5.6.0/ja/model_doc/bigbird_pegasus#transformers.BigBirdPegasusConfig) configuration class: [BigBirdPegasusForConditionalGeneration](/docs/transformers/v5.6.0/ja/model_doc/bigbird_pegasus#transformers.BigBirdPegasusForConditionalGeneration) (BigBirdPegasusConfig model) - [BlenderbotConfig](/docs/transformers/v5.6.0/ja/model_doc/blenderbot#transformers.BlenderbotConfig) configuration class: [BlenderbotForConditionalGeneration](/docs/transformers/v5.6.0/ja/model_doc/blenderbot#transformers.BlenderbotForConditionalGeneration) (BlenderbotConfig model) - [BlenderbotSmallConfig](/docs/transformers/v5.6.0/ja/model_doc/blenderbot-small#transformers.BlenderbotSmallConfig) configuration class: [BlenderbotSmallForConditionalGeneration](/docs/transformers/v5.6.0/ja/model_doc/blenderbot-small#transformers.BlenderbotSmallForConditionalGeneration) (BlenderbotSmallConfig model) - `EncoderDecoderConfig` configuration class: `EncoderDecoderModel` (EncoderDecoderConfig model) - `FSMTConfig` configuration class: `FSMTForConditionalGeneration` (FSMTConfig model) - `GlmAsrConfig` configuration class: `GlmAsrForConditionalGeneration` (GlmAsrConfig model) - `GraniteSpeechConfig` configuration class: `GraniteSpeechForConditionalGeneration` (GraniteSpeechConfig model) - `LEDConfig` configuration class: `LEDForConditionalGeneration` (LEDConfig model) - `LongT5Config` configuration class: `LongT5ForConditionalGeneration` (LongT5Config model) - `M2M100Config` configuration class: `M2M100ForConditionalGeneration` (M2M100Config model) - `MBartConfig` configuration class: `MBartForConditionalGeneration` (MBartConfig model) - `MT5Config` configuration class: `MT5ForConditionalGeneration` (MT5Config model) - `MarianConfig` configuration class: `MarianMTModel` (MarianConfig model) - `MusicFlamingoConfig` configuration class: `MusicFlamingoForConditionalGeneration` (MusicFlamingoConfig model) - `MvpConfig` configuration class: `MvpForConditionalGeneration` (MvpConfig model) - `NllbMoeConfig` configuration class: `NllbMoeForConditionalGeneration` (NllbMoeConfig model) - `PLBartConfig` configuration class: `PLBartForConditionalGeneration` (PLBartConfig model) - `PegasusConfig` configuration class: `PegasusForConditionalGeneration` (PegasusConfig model) - `PegasusXConfig` configuration class: `PegasusXForConditionalGeneration` (PegasusXConfig model) - `ProphetNetConfig` configuration class: `ProphetNetForConditionalGeneration` (ProphetNetConfig model) - `Qwen2AudioConfig` configuration class: `Qwen2AudioForConditionalGeneration` (Qwen2AudioConfig model) - `SeamlessM4TConfig` configuration class: `SeamlessM4TForTextToText` (SeamlessM4TConfig model) - `SeamlessM4Tv2Config` configuration class: `SeamlessM4Tv2ForTextToText` (SeamlessM4Tv2Config model) - `SwitchTransformersConfig` configuration class: `SwitchTransformersForConditionalGeneration` (SwitchTransformersConfig model) - `T5Config` configuration class: `T5ForConditionalGeneration` (T5Config model) - `T5Gemma2Config` configuration class: `T5Gemma2ForConditionalGeneration` (T5Gemma2Config model) - `T5GemmaConfig` configuration class: `T5GemmaForConditionalGeneration` (T5GemmaConfig model) - `UMT5Config` configuration class: `UMT5ForConditionalGeneration` (UMT5Config model) - `VibeVoiceAsrConfig` configuration class: `VibeVoiceAsrForConditionalGeneration` (VibeVoiceAsrConfig model) - `VoxtralConfig` configuration class: `VoxtralForConditionalGeneration` (VoxtralConfig model) - `VoxtralRealtimeConfig` configuration class: `VoxtralRealtimeForConditionalGeneration` (VoxtralRealtimeConfig model) attn_implementation (`str`, *optional*) : The attention implementation to use in the model (if relevant). Can be any of `"eager"` (manual implementation of the attention), `"sdpa"` (using [`F.scaled_dot_product_attention`](https://pytorch.org/docs/master/generated/torch.nn.functional.scaled_dot_product_attention.html)), `"flash_attention_2"` (using [Dao-AILab/flash-attention](https://github.com/Dao-AILab/flash-attention)), or `"flash_attention_3"` (using [Dao-AILab/flash-attention/hopper](https://github.com/Dao-AILab/flash-attention/tree/main/hopper)). By default, if available, SDPA will be used for torch>=2.1.1. The default is otherwise the manual `"eager"` implementation. #### from_pretrained[[transformers.AutoModelForSeq2SeqLM.from_pretrained]] [Source](https://github.com/huggingface/transformers/blob/v5.6.0/src/transformers/models/auto/auto_factory.py#L253) Instantiate one of the model classes of the library (with a sequence-to-sequence language modeling head) from a pretrained model. The model class to instantiate is selected based on the `model_type` property of the config object (either passed as an argument or loaded from `pretrained_model_name_or_path` if possible), or when it's missing, by falling back to using pattern matching on `pretrained_model_name_or_path`: - **audioflamingo3** -- `AudioFlamingo3ForConditionalGeneration` (AudioFlamingo3Config model) - **bart** -- [BartForConditionalGeneration](/docs/transformers/v5.6.0/ja/model_doc/bart#transformers.BartForConditionalGeneration) (BartConfig model) - **bigbird_pegasus** -- [BigBirdPegasusForConditionalGeneration](/docs/transformers/v5.6.0/ja/model_doc/bigbird_pegasus#transformers.BigBirdPegasusForConditionalGeneration) (BigBirdPegasusConfig model) - **blenderbot** -- [BlenderbotForConditionalGeneration](/docs/transformers/v5.6.0/ja/model_doc/blenderbot#transformers.BlenderbotForConditionalGeneration) (BlenderbotConfig model) - **blenderbot-small** -- [BlenderbotSmallForConditionalGeneration](/docs/transformers/v5.6.0/ja/model_doc/blenderbot-small#transformers.BlenderbotSmallForConditionalGeneration) (BlenderbotSmallConfig model) - **encoder-decoder** -- `EncoderDecoderModel` (EncoderDecoderConfig model) - **fsmt** -- `FSMTForConditionalGeneration` (FSMTConfig model) - **glmasr** -- `GlmAsrForConditionalGeneration` (GlmAsrConfig model) - **granite_speech** -- `GraniteSpeechForConditionalGeneration` (GraniteSpeechConfig model) - **led** -- `LEDForConditionalGeneration` (LEDConfig model) - **longt5** -- `LongT5ForConditionalGeneration` (LongT5Config model) - **m2m_100** -- `M2M100ForConditionalGeneration` (M2M100Config model) - **marian** -- `MarianMTModel` (MarianConfig model) - **mbart** -- `MBartForConditionalGeneration` (MBartConfig model) - **mt5** -- `MT5ForConditionalGeneration` (MT5Config model) - **musicflamingo** -- `MusicFlamingoForConditionalGeneration` (MusicFlamingoConfig model) - **mvp** -- `MvpForConditionalGeneration` (MvpConfig model) - **nllb-moe** -- `NllbMoeForConditionalGeneration` (NllbMoeConfig model) - **pegasus** -- `PegasusForConditionalGeneration` (PegasusConfig model) - **pegasus_x** -- `PegasusXForConditionalGeneration` (PegasusXConfig model) - **plbart** -- `PLBartForConditionalGeneration` (PLBartConfig model) - **prophetnet** -- `ProphetNetForConditionalGeneration` (ProphetNetConfig model) - **qwen2_audio** -- `Qwen2AudioForConditionalGeneration` (Qwen2AudioConfig model) - **seamless_m4t** -- `SeamlessM4TForTextToText` (SeamlessM4TConfig model) - **seamless_m4t_v2** -- `SeamlessM4Tv2ForTextToText` (SeamlessM4Tv2Config model) - **switch_transformers** -- `SwitchTransformersForConditionalGeneration` (SwitchTransformersConfig model) - **t5** -- `T5ForConditionalGeneration` (T5Config model) - **t5gemma** -- `T5GemmaForConditionalGeneration` (T5GemmaConfig model) - **t5gemma2** -- `T5Gemma2ForConditionalGeneration` (T5Gemma2Config model) - **umt5** -- `UMT5ForConditionalGeneration` (UMT5Config model) - **vibevoice_asr** -- `VibeVoiceAsrForConditionalGeneration` (VibeVoiceAsrConfig model) - **voxtral** -- `VoxtralForConditionalGeneration` (VoxtralConfig model) - **voxtral_realtime** -- `VoxtralRealtimeForConditionalGeneration` (VoxtralRealtimeConfig model) The model is set in evaluation mode by default using `model.eval()` (so for instance, dropout modules are deactivated). To train the model, you should first set it back in training mode with `model.train()` Examples: ```python >>> from transformers import AutoConfig, AutoModelForSeq2SeqLM >>> # Download model and configuration from huggingface.co and cache. >>> model = AutoModelForSeq2SeqLM.from_pretrained("google-t5/t5-base") >>> # Update configuration during loading >>> model = AutoModelForSeq2SeqLM.from_pretrained("google-t5/t5-base", output_attentions=True) >>> model.config.output_attentions True ``` **Parameters:** pretrained_model_name_or_path (`str` or `os.PathLike`) : Can be either: - A string, the *model id* of a pretrained model hosted inside a model repo on huggingface.co. - A path to a *directory* containing model weights saved using [save_pretrained()](/docs/transformers/v5.6.0/ja/main_classes/model#transformers.PreTrainedModel.save_pretrained), e.g., `./my_model_directory/`. model_args (additional positional arguments, *optional*) : Will be passed along to the underlying model `__init__()` method. config ([PreTrainedConfig](/docs/transformers/v5.6.0/ja/main_classes/configuration#transformers.PreTrainedConfig), *optional*) : Configuration for the model to use instead of an automatically loaded configuration. Configuration can be automatically loaded when: - The model is a model provided by the library (loaded with the *model id* string of a pretrained model). - The model was saved using [save_pretrained()](/docs/transformers/v5.6.0/ja/main_classes/model#transformers.PreTrainedModel.save_pretrained) and is reloaded by supplying the save directory. - The model is loaded by supplying a local directory as `pretrained_model_name_or_path` and a configuration JSON file named *config.json* is found in the directory. state_dict (*dict[str, torch.Tensor]*, *optional*) : A state dictionary to use instead of a state dictionary loaded from saved weights file. This option can be used if you want to create a model from a pretrained configuration but load your own weights. In this case though, you should check if using [save_pretrained()](/docs/transformers/v5.6.0/ja/main_classes/model#transformers.PreTrainedModel.save_pretrained) and [from_pretrained()](/docs/transformers/v5.6.0/ja/main_classes/model#transformers.PreTrainedModel.from_pretrained) is not a simpler option. cache_dir (`str` or `os.PathLike`, *optional*) : Path to a directory in which a downloaded pretrained model configuration should be cached if the standard cache should not be used. force_download (`bool`, *optional*, defaults to `False`) : Whether or not to force the (re-)download of the model weights and configuration files, overriding the cached versions if they exist. proxies (`dict[str, str]`, *optional*) : A dictionary of proxy servers to use by protocol or endpoint, e.g., `{'http': 'foo.bar:3128', 'http://hostname': 'foo.bar:4012'}`. The proxies are used on each request. output_loading_info(`bool`, *optional*, defaults to `False`) : Whether or not to also return a dictionary containing missing keys, unexpected keys and error messages. local_files_only(`bool`, *optional*, defaults to `False`) : Whether or not to only look at local files (e.g., not try downloading the model). revision (`str`, *optional*, defaults to `"main"`) : The specific model version to use. It can be a branch name, a tag name, or a commit id, since we use a git-based system for storing models and other artifacts on huggingface.co, so `revision` can be any identifier allowed by git. trust_remote_code (`bool`, *optional*, defaults to `False`) : Whether or not to allow for custom models defined on the Hub in their own modeling files. This option should only be set to `True` for repositories you trust and in which you have read the code, as it will execute code present on the Hub on your local machine. code_revision (`str`, *optional*, defaults to `"main"`) : The specific revision to use for the code on the Hub, if the code leaves in a different repository than the rest of the model. It can be a branch name, a tag name, or a commit id, since we use a git-based system for storing models and other artifacts on huggingface.co, so `revision` can be any identifier allowed by git. kwargs (additional keyword arguments, *optional*) : Can be used to update the configuration object (after it being loaded) and initiate the model (e.g., `output_attentions=True`). Behaves differently depending on whether a `config` is provided or automatically loaded: - If a configuration is provided with `config`, `**kwargs` will be directly passed to the underlying model's `__init__` method (we assume all relevant updates to the configuration have already been done) - If a configuration is not provided, `kwargs` will be first passed to the configuration class initialization function ([from_pretrained()](/docs/transformers/v5.6.0/ja/main_classes/configuration#transformers.PreTrainedConfig.from_pretrained)). Each key of `kwargs` that corresponds to a configuration attribute will be used to override said attribute with the supplied `kwargs` value. Remaining keys that do not correspond to any configuration attribute will be passed to the underlying model's `__init__` function. ### AutoModelForSequenceClassification[[transformers.AutoModelForSequenceClassification]] #### transformers.AutoModelForSequenceClassification[[transformers.AutoModelForSequenceClassification]] [Source](https://github.com/huggingface/transformers/blob/v5.6.0/src/transformers/models/auto/modeling_auto.py#L2028) This is a generic model class that will be instantiated as one of the model classes of the library (with a sequence classification head) when created with the [from_pretrained()](/docs/transformers/v5.6.0/ja/model_doc/auto#transformers.AutoModel.from_pretrained) class method or the [from_config()](/docs/transformers/v5.6.0/ja/model_doc/auto#transformers.AutoModel.from_config) class method. This class cannot be instantiated directly using `__init__()` (throws an error). from_configtransformers.AutoModelForSequenceClassification.from_confighttps://github.com/huggingface/transformers/blob/v5.6.0/src/transformers/models/auto/auto_factory.py#L206[{"name": "**kwargs", "val": ""}]- **config** ([PreTrainedConfig](/docs/transformers/v5.6.0/ja/main_classes/configuration#transformers.PreTrainedConfig)) -- The model class to instantiate is selected based on the configuration class: - [AlbertConfig](/docs/transformers/v5.6.0/ja/model_doc/albert#transformers.AlbertConfig) configuration class: [AlbertForSequenceClassification](/docs/transformers/v5.6.0/ja/model_doc/albert#transformers.AlbertForSequenceClassification) (AlbertConfig model) - `ArceeConfig` configuration class: `ArceeForSequenceClassification` (ArceeConfig model) - [BartConfig](/docs/transformers/v5.6.0/ja/model_doc/bart#transformers.BartConfig) configuration class: [BartForSequenceClassification](/docs/transformers/v5.6.0/ja/model_doc/bart#transformers.BartForSequenceClassification) (BartConfig model) - [BertConfig](/docs/transformers/v5.6.0/ja/model_doc/bert#transformers.BertConfig) configuration class: [BertForSequenceClassification](/docs/transformers/v5.6.0/ja/model_doc/bert#transformers.BertForSequenceClassification) (BertConfig model) - [BigBirdConfig](/docs/transformers/v5.6.0/ja/model_doc/big_bird#transformers.BigBirdConfig) configuration class: [BigBirdForSequenceClassification](/docs/transformers/v5.6.0/ja/model_doc/big_bird#transformers.BigBirdForSequenceClassification) (BigBirdConfig model) - [BigBirdPegasusConfig](/docs/transformers/v5.6.0/ja/model_doc/bigbird_pegasus#transformers.BigBirdPegasusConfig) configuration class: [BigBirdPegasusForSequenceClassification](/docs/transformers/v5.6.0/ja/model_doc/bigbird_pegasus#transformers.BigBirdPegasusForSequenceClassification) (BigBirdPegasusConfig model) - [BioGptConfig](/docs/transformers/v5.6.0/ja/model_doc/biogpt#transformers.BioGptConfig) configuration class: [BioGptForSequenceClassification](/docs/transformers/v5.6.0/ja/model_doc/biogpt#transformers.BioGptForSequenceClassification) (BioGptConfig model) - [BloomConfig](/docs/transformers/v5.6.0/ja/model_doc/bloom#transformers.BloomConfig) configuration class: [BloomForSequenceClassification](/docs/transformers/v5.6.0/ja/model_doc/bloom#transformers.BloomForSequenceClassification) (BloomConfig model) - [CTRLConfig](/docs/transformers/v5.6.0/ja/model_doc/ctrl#transformers.CTRLConfig) configuration class: [CTRLForSequenceClassification](/docs/transformers/v5.6.0/ja/model_doc/ctrl#transformers.CTRLForSequenceClassification) (CTRLConfig model) - [CamembertConfig](/docs/transformers/v5.6.0/ja/model_doc/camembert#transformers.CamembertConfig) configuration class: [CamembertForSequenceClassification](/docs/transformers/v5.6.0/ja/model_doc/camembert#transformers.CamembertForSequenceClassification) (CamembertConfig model) - [CanineConfig](/docs/transformers/v5.6.0/ja/model_doc/canine#transformers.CanineConfig) configuration class: [CanineForSequenceClassification](/docs/transformers/v5.6.0/ja/model_doc/canine#transformers.CanineForSequenceClassification) (CanineConfig model) - [ConvBertConfig](/docs/transformers/v5.6.0/ja/model_doc/convbert#transformers.ConvBertConfig) configuration class: [ConvBertForSequenceClassification](/docs/transformers/v5.6.0/ja/model_doc/convbert#transformers.ConvBertForSequenceClassification) (ConvBertConfig model) - [Data2VecTextConfig](/docs/transformers/v5.6.0/ja/model_doc/data2vec#transformers.Data2VecTextConfig) configuration class: [Data2VecTextForSequenceClassification](/docs/transformers/v5.6.0/ja/model_doc/data2vec#transformers.Data2VecTextForSequenceClassification) (Data2VecTextConfig model) - [DebertaConfig](/docs/transformers/v5.6.0/ja/model_doc/deberta#transformers.DebertaConfig) configuration class: [DebertaForSequenceClassification](/docs/transformers/v5.6.0/ja/model_doc/deberta#transformers.DebertaForSequenceClassification) (DebertaConfig model) - [DebertaV2Config](/docs/transformers/v5.6.0/ja/model_doc/deberta-v2#transformers.DebertaV2Config) configuration class: [DebertaV2ForSequenceClassification](/docs/transformers/v5.6.0/ja/model_doc/deberta-v2#transformers.DebertaV2ForSequenceClassification) (DebertaV2Config model) - `DeepseekV2Config` configuration class: `DeepseekV2ForSequenceClassification` (DeepseekV2Config model) - `DeepseekV3Config` configuration class: `DeepseekV3ForSequenceClassification` (DeepseekV3Config model) - `DiffLlamaConfig` configuration class: `DiffLlamaForSequenceClassification` (DiffLlamaConfig model) - `DistilBertConfig` configuration class: `DistilBertForSequenceClassification` (DistilBertConfig model) - `DogeConfig` configuration class: `DogeForSequenceClassification` (DogeConfig model) - `ElectraConfig` configuration class: `ElectraForSequenceClassification` (ElectraConfig model) - `ErnieConfig` configuration class: `ErnieForSequenceClassification` (ErnieConfig model) - `EsmConfig` configuration class: `EsmForSequenceClassification` (EsmConfig model) - `EuroBertConfig` configuration class: `EuroBertForSequenceClassification` (EuroBertConfig model) - `Exaone4Config` configuration class: `Exaone4ForSequenceClassification` (Exaone4Config model) - `FNetConfig` configuration class: `FNetForSequenceClassification` (FNetConfig model) - `FalconConfig` configuration class: `FalconForSequenceClassification` (FalconConfig model) - `FlaubertConfig` configuration class: `FlaubertForSequenceClassification` (FlaubertConfig model) - `FunnelConfig` configuration class: `FunnelForSequenceClassification` (FunnelConfig model) - `GPT2Config` configuration class: `GPT2ForSequenceClassification` (GPT2Config model) - `GPTBigCodeConfig` configuration class: `GPTBigCodeForSequenceClassification` (GPTBigCodeConfig model) - `GPTJConfig` configuration class: `GPTJForSequenceClassification` (GPTJConfig model) - `GPTNeoConfig` configuration class: `GPTNeoForSequenceClassification` (GPTNeoConfig model) - `GPTNeoXConfig` configuration class: `GPTNeoXForSequenceClassification` (GPTNeoXConfig model) - `Gemma2Config` configuration class: `Gemma2ForSequenceClassification` (Gemma2Config model) - `Gemma3Config` configuration class: `Gemma3ForSequenceClassification` (Gemma3Config model) - `Gemma3TextConfig` configuration class: `Gemma3TextForSequenceClassification` (Gemma3TextConfig model) - `GemmaConfig` configuration class: `GemmaForSequenceClassification` (GemmaConfig model) - `Glm4Config` configuration class: `Glm4ForSequenceClassification` (Glm4Config model) - `GlmConfig` configuration class: `GlmForSequenceClassification` (GlmConfig model) - `GptOssConfig` configuration class: `GptOssForSequenceClassification` (GptOssConfig model) - `HeliumConfig` configuration class: `HeliumForSequenceClassification` (HeliumConfig model) - `HunYuanDenseV1Config` configuration class: `HunYuanDenseV1ForSequenceClassification` (HunYuanDenseV1Config model) - `HunYuanMoEV1Config` configuration class: `HunYuanMoEV1ForSequenceClassification` (HunYuanMoEV1Config model) - `IBertConfig` configuration class: `IBertForSequenceClassification` (IBertConfig model) - `JambaConfig` configuration class: `JambaForSequenceClassification` (JambaConfig model) - `JetMoeConfig` configuration class: `JetMoeForSequenceClassification` (JetMoeConfig model) - `JinaEmbeddingsV3Config` configuration class: `JinaEmbeddingsV3ForSequenceClassification` (JinaEmbeddingsV3Config model) - `LayoutLMConfig` configuration class: `LayoutLMForSequenceClassification` (LayoutLMConfig model) - `LayoutLMv2Config` configuration class: `LayoutLMv2ForSequenceClassification` (LayoutLMv2Config model) - `LayoutLMv3Config` configuration class: `LayoutLMv3ForSequenceClassification` (LayoutLMv3Config model) - `LiltConfig` configuration class: `LiltForSequenceClassification` (LiltConfig model) - `LlamaConfig` configuration class: `LlamaForSequenceClassification` (LlamaConfig model) - `LongformerConfig` configuration class: `LongformerForSequenceClassification` (LongformerConfig model) - `LukeConfig` configuration class: `LukeForSequenceClassification` (LukeConfig model) - `MBartConfig` configuration class: `MBartForSequenceClassification` (MBartConfig model) - `MPNetConfig` configuration class: `MPNetForSequenceClassification` (MPNetConfig model) - `MT5Config` configuration class: `MT5ForSequenceClassification` (MT5Config model) - `MarkupLMConfig` configuration class: `MarkupLMForSequenceClassification` (MarkupLMConfig model) - `MegatronBertConfig` configuration class: `MegatronBertForSequenceClassification` (MegatronBertConfig model) - `MiniMaxConfig` configuration class: `MiniMaxForSequenceClassification` (MiniMaxConfig model) - `Ministral3Config` configuration class: `Ministral3ForSequenceClassification` (Ministral3Config model) - `MinistralConfig` configuration class: `MinistralForSequenceClassification` (MinistralConfig model) - `Mistral4Config` configuration class: `Mistral4ForSequenceClassification` (Mistral4Config model) - `MistralConfig` configuration class: `MistralForSequenceClassification` (MistralConfig model) - `MixtralConfig` configuration class: `MixtralForSequenceClassification` (MixtralConfig model) - `MobileBertConfig` configuration class: `MobileBertForSequenceClassification` (MobileBertConfig model) - `ModernBertConfig` configuration class: `ModernBertForSequenceClassification` (ModernBertConfig model) - `ModernBertDecoderConfig` configuration class: `ModernBertDecoderForSequenceClassification` (ModernBertDecoderConfig model) - `ModernVBertConfig` configuration class: `ModernVBertForSequenceClassification` (ModernVBertConfig model) - `MptConfig` configuration class: `MptForSequenceClassification` (MptConfig model) - `MraConfig` configuration class: `MraForSequenceClassification` (MraConfig model) - `MvpConfig` configuration class: `MvpForSequenceClassification` (MvpConfig model) - `NemotronConfig` configuration class: `NemotronForSequenceClassification` (NemotronConfig model) - `NomicBertConfig` configuration class: `NomicBertForSequenceClassification` (NomicBertConfig model) - `NystromformerConfig` configuration class: `NystromformerForSequenceClassification` (NystromformerConfig model) - `OPTConfig` configuration class: `OPTForSequenceClassification` (OPTConfig model) - `Olmo2Config` configuration class: `Olmo2ForSequenceClassification` (Olmo2Config model) - `Olmo3Config` configuration class: `Olmo3ForSequenceClassification` (Olmo3Config model) - `OlmoConfig` configuration class: `OlmoForSequenceClassification` (OlmoConfig model) - `OpenAIGPTConfig` configuration class: `OpenAIGPTForSequenceClassification` (OpenAIGPTConfig model) - `PLBartConfig` configuration class: `PLBartForSequenceClassification` (PLBartConfig model) - `PerceiverConfig` configuration class: `PerceiverForSequenceClassification` (PerceiverConfig model) - `PersimmonConfig` configuration class: `PersimmonForSequenceClassification` (PersimmonConfig model) - `Phi3Config` configuration class: `Phi3ForSequenceClassification` (Phi3Config model) - `PhiConfig` configuration class: `PhiForSequenceClassification` (PhiConfig model) - `PhimoeConfig` configuration class: `PhimoeForSequenceClassification` (PhimoeConfig model) - `Qwen2Config` configuration class: `Qwen2ForSequenceClassification` (Qwen2Config model) - `Qwen2MoeConfig` configuration class: `Qwen2MoeForSequenceClassification` (Qwen2MoeConfig model) - `Qwen3Config` configuration class: `Qwen3ForSequenceClassification` (Qwen3Config model) - `Qwen3MoeConfig` configuration class: `Qwen3MoeForSequenceClassification` (Qwen3MoeConfig model) - `Qwen3NextConfig` configuration class: `Qwen3NextForSequenceClassification` (Qwen3NextConfig model) - `Qwen3_5Config` configuration class: `Qwen3_5ForSequenceClassification` (Qwen3_5Config model) - `Qwen3_5TextConfig` configuration class: `Qwen3_5ForSequenceClassification` (Qwen3_5TextConfig model) - `ReformerConfig` configuration class: `ReformerForSequenceClassification` (ReformerConfig model) - `RemBertConfig` configuration class: `RemBertForSequenceClassification` (RemBertConfig model) - `RoCBertConfig` configuration class: `RoCBertForSequenceClassification` (RoCBertConfig model) - `RoFormerConfig` configuration class: `RoFormerForSequenceClassification` (RoFormerConfig model) - `RobertaConfig` configuration class: `RobertaForSequenceClassification` (RobertaConfig model) - `RobertaPreLayerNormConfig` configuration class: `RobertaPreLayerNormForSequenceClassification` (RobertaPreLayerNormConfig model) - `SeedOssConfig` configuration class: `SeedOssForSequenceClassification` (SeedOssConfig model) - `SmolLM3Config` configuration class: `SmolLM3ForSequenceClassification` (SmolLM3Config model) - `SqueezeBertConfig` configuration class: `SqueezeBertForSequenceClassification` (SqueezeBertConfig model) - `StableLmConfig` configuration class: `StableLmForSequenceClassification` (StableLmConfig model) - `Starcoder2Config` configuration class: `Starcoder2ForSequenceClassification` (Starcoder2Config model) - `T5Config` configuration class: `T5ForSequenceClassification` (T5Config model) - `T5Gemma2Config` configuration class: `T5Gemma2ForSequenceClassification` (T5Gemma2Config model) - `T5GemmaConfig` configuration class: `T5GemmaForSequenceClassification` (T5GemmaConfig model) - `TapasConfig` configuration class: `TapasForSequenceClassification` (TapasConfig model) - `UMT5Config` configuration class: `UMT5ForSequenceClassification` (UMT5Config model) - `XLMConfig` configuration class: `XLMForSequenceClassification` (XLMConfig model) - `XLMRobertaConfig` configuration class: `XLMRobertaForSequenceClassification` (XLMRobertaConfig model) - `XLMRobertaXLConfig` configuration class: `XLMRobertaXLForSequenceClassification` (XLMRobertaXLConfig model) - `XLNetConfig` configuration class: `XLNetForSequenceClassification` (XLNetConfig model) - `XmodConfig` configuration class: `XmodForSequenceClassification` (XmodConfig model) - `YosoConfig` configuration class: `YosoForSequenceClassification` (YosoConfig model) - `Zamba2Config` configuration class: `Zamba2ForSequenceClassification` (Zamba2Config model) - `ZambaConfig` configuration class: `ZambaForSequenceClassification` (ZambaConfig model) - **attn_implementation** (`str`, *optional*) -- The attention implementation to use in the model (if relevant). Can be any of `"eager"` (manual implementation of the attention), `"sdpa"` (using [`F.scaled_dot_product_attention`](https://pytorch.org/docs/master/generated/torch.nn.functional.scaled_dot_product_attention.html)), `"flash_attention_2"` (using [Dao-AILab/flash-attention](https://github.com/Dao-AILab/flash-attention)), or `"flash_attention_3"` (using [Dao-AILab/flash-attention/hopper](https://github.com/Dao-AILab/flash-attention/tree/main/hopper)). By default, if available, SDPA will be used for torch>=2.1.1. The default is otherwise the manual `"eager"` implementation.0 Instantiates one of the model classes of the library (with a sequence classification head) from a configuration. Note: Loading a model from its configuration file does **not** load the model weights. It only affects the model's configuration. Use [from_pretrained()](/docs/transformers/v5.6.0/ja/model_doc/auto#transformers.AutoModel.from_pretrained) to load the model weights. Examples: ```python >>> from transformers import AutoConfig, AutoModelForSequenceClassification >>> # Download configuration from huggingface.co and cache. >>> config = AutoConfig.from_pretrained("google-bert/bert-base-cased") >>> model = AutoModelForSequenceClassification.from_config(config) ``` **Parameters:** config ([PreTrainedConfig](/docs/transformers/v5.6.0/ja/main_classes/configuration#transformers.PreTrainedConfig)) : The model class to instantiate is selected based on the configuration class: - [AlbertConfig](/docs/transformers/v5.6.0/ja/model_doc/albert#transformers.AlbertConfig) configuration class: [AlbertForSequenceClassification](/docs/transformers/v5.6.0/ja/model_doc/albert#transformers.AlbertForSequenceClassification) (AlbertConfig model) - `ArceeConfig` configuration class: `ArceeForSequenceClassification` (ArceeConfig model) - [BartConfig](/docs/transformers/v5.6.0/ja/model_doc/bart#transformers.BartConfig) configuration class: [BartForSequenceClassification](/docs/transformers/v5.6.0/ja/model_doc/bart#transformers.BartForSequenceClassification) (BartConfig model) - [BertConfig](/docs/transformers/v5.6.0/ja/model_doc/bert#transformers.BertConfig) configuration class: [BertForSequenceClassification](/docs/transformers/v5.6.0/ja/model_doc/bert#transformers.BertForSequenceClassification) (BertConfig model) - [BigBirdConfig](/docs/transformers/v5.6.0/ja/model_doc/big_bird#transformers.BigBirdConfig) configuration class: [BigBirdForSequenceClassification](/docs/transformers/v5.6.0/ja/model_doc/big_bird#transformers.BigBirdForSequenceClassification) (BigBirdConfig model) - [BigBirdPegasusConfig](/docs/transformers/v5.6.0/ja/model_doc/bigbird_pegasus#transformers.BigBirdPegasusConfig) configuration class: [BigBirdPegasusForSequenceClassification](/docs/transformers/v5.6.0/ja/model_doc/bigbird_pegasus#transformers.BigBirdPegasusForSequenceClassification) (BigBirdPegasusConfig model) - [BioGptConfig](/docs/transformers/v5.6.0/ja/model_doc/biogpt#transformers.BioGptConfig) configuration class: [BioGptForSequenceClassification](/docs/transformers/v5.6.0/ja/model_doc/biogpt#transformers.BioGptForSequenceClassification) (BioGptConfig model) - [BloomConfig](/docs/transformers/v5.6.0/ja/model_doc/bloom#transformers.BloomConfig) configuration class: [BloomForSequenceClassification](/docs/transformers/v5.6.0/ja/model_doc/bloom#transformers.BloomForSequenceClassification) (BloomConfig model) - [CTRLConfig](/docs/transformers/v5.6.0/ja/model_doc/ctrl#transformers.CTRLConfig) configuration class: [CTRLForSequenceClassification](/docs/transformers/v5.6.0/ja/model_doc/ctrl#transformers.CTRLForSequenceClassification) (CTRLConfig model) - [CamembertConfig](/docs/transformers/v5.6.0/ja/model_doc/camembert#transformers.CamembertConfig) configuration class: [CamembertForSequenceClassification](/docs/transformers/v5.6.0/ja/model_doc/camembert#transformers.CamembertForSequenceClassification) (CamembertConfig model) - [CanineConfig](/docs/transformers/v5.6.0/ja/model_doc/canine#transformers.CanineConfig) configuration class: [CanineForSequenceClassification](/docs/transformers/v5.6.0/ja/model_doc/canine#transformers.CanineForSequenceClassification) (CanineConfig model) - [ConvBertConfig](/docs/transformers/v5.6.0/ja/model_doc/convbert#transformers.ConvBertConfig) configuration class: [ConvBertForSequenceClassification](/docs/transformers/v5.6.0/ja/model_doc/convbert#transformers.ConvBertForSequenceClassification) (ConvBertConfig model) - [Data2VecTextConfig](/docs/transformers/v5.6.0/ja/model_doc/data2vec#transformers.Data2VecTextConfig) configuration class: [Data2VecTextForSequenceClassification](/docs/transformers/v5.6.0/ja/model_doc/data2vec#transformers.Data2VecTextForSequenceClassification) (Data2VecTextConfig model) - [DebertaConfig](/docs/transformers/v5.6.0/ja/model_doc/deberta#transformers.DebertaConfig) configuration class: [DebertaForSequenceClassification](/docs/transformers/v5.6.0/ja/model_doc/deberta#transformers.DebertaForSequenceClassification) (DebertaConfig model) - [DebertaV2Config](/docs/transformers/v5.6.0/ja/model_doc/deberta-v2#transformers.DebertaV2Config) configuration class: [DebertaV2ForSequenceClassification](/docs/transformers/v5.6.0/ja/model_doc/deberta-v2#transformers.DebertaV2ForSequenceClassification) (DebertaV2Config model) - `DeepseekV2Config` configuration class: `DeepseekV2ForSequenceClassification` (DeepseekV2Config model) - `DeepseekV3Config` configuration class: `DeepseekV3ForSequenceClassification` (DeepseekV3Config model) - `DiffLlamaConfig` configuration class: `DiffLlamaForSequenceClassification` (DiffLlamaConfig model) - `DistilBertConfig` configuration class: `DistilBertForSequenceClassification` (DistilBertConfig model) - `DogeConfig` configuration class: `DogeForSequenceClassification` (DogeConfig model) - `ElectraConfig` configuration class: `ElectraForSequenceClassification` (ElectraConfig model) - `ErnieConfig` configuration class: `ErnieForSequenceClassification` (ErnieConfig model) - `EsmConfig` configuration class: `EsmForSequenceClassification` (EsmConfig model) - `EuroBertConfig` configuration class: `EuroBertForSequenceClassification` (EuroBertConfig model) - `Exaone4Config` configuration class: `Exaone4ForSequenceClassification` (Exaone4Config model) - `FNetConfig` configuration class: `FNetForSequenceClassification` (FNetConfig model) - `FalconConfig` configuration class: `FalconForSequenceClassification` (FalconConfig model) - `FlaubertConfig` configuration class: `FlaubertForSequenceClassification` (FlaubertConfig model) - `FunnelConfig` configuration class: `FunnelForSequenceClassification` (FunnelConfig model) - `GPT2Config` configuration class: `GPT2ForSequenceClassification` (GPT2Config model) - `GPTBigCodeConfig` configuration class: `GPTBigCodeForSequenceClassification` (GPTBigCodeConfig model) - `GPTJConfig` configuration class: `GPTJForSequenceClassification` (GPTJConfig model) - `GPTNeoConfig` configuration class: `GPTNeoForSequenceClassification` (GPTNeoConfig model) - `GPTNeoXConfig` configuration class: `GPTNeoXForSequenceClassification` (GPTNeoXConfig model) - `Gemma2Config` configuration class: `Gemma2ForSequenceClassification` (Gemma2Config model) - `Gemma3Config` configuration class: `Gemma3ForSequenceClassification` (Gemma3Config model) - `Gemma3TextConfig` configuration class: `Gemma3TextForSequenceClassification` (Gemma3TextConfig model) - `GemmaConfig` configuration class: `GemmaForSequenceClassification` (GemmaConfig model) - `Glm4Config` configuration class: `Glm4ForSequenceClassification` (Glm4Config model) - `GlmConfig` configuration class: `GlmForSequenceClassification` (GlmConfig model) - `GptOssConfig` configuration class: `GptOssForSequenceClassification` (GptOssConfig model) - `HeliumConfig` configuration class: `HeliumForSequenceClassification` (HeliumConfig model) - `HunYuanDenseV1Config` configuration class: `HunYuanDenseV1ForSequenceClassification` (HunYuanDenseV1Config model) - `HunYuanMoEV1Config` configuration class: `HunYuanMoEV1ForSequenceClassification` (HunYuanMoEV1Config model) - `IBertConfig` configuration class: `IBertForSequenceClassification` (IBertConfig model) - `JambaConfig` configuration class: `JambaForSequenceClassification` (JambaConfig model) - `JetMoeConfig` configuration class: `JetMoeForSequenceClassification` (JetMoeConfig model) - `JinaEmbeddingsV3Config` configuration class: `JinaEmbeddingsV3ForSequenceClassification` (JinaEmbeddingsV3Config model) - `LayoutLMConfig` configuration class: `LayoutLMForSequenceClassification` (LayoutLMConfig model) - `LayoutLMv2Config` configuration class: `LayoutLMv2ForSequenceClassification` (LayoutLMv2Config model) - `LayoutLMv3Config` configuration class: `LayoutLMv3ForSequenceClassification` (LayoutLMv3Config model) - `LiltConfig` configuration class: `LiltForSequenceClassification` (LiltConfig model) - `LlamaConfig` configuration class: `LlamaForSequenceClassification` (LlamaConfig model) - `LongformerConfig` configuration class: `LongformerForSequenceClassification` (LongformerConfig model) - `LukeConfig` configuration class: `LukeForSequenceClassification` (LukeConfig model) - `MBartConfig` configuration class: `MBartForSequenceClassification` (MBartConfig model) - `MPNetConfig` configuration class: `MPNetForSequenceClassification` (MPNetConfig model) - `MT5Config` configuration class: `MT5ForSequenceClassification` (MT5Config model) - `MarkupLMConfig` configuration class: `MarkupLMForSequenceClassification` (MarkupLMConfig model) - `MegatronBertConfig` configuration class: `MegatronBertForSequenceClassification` (MegatronBertConfig model) - `MiniMaxConfig` configuration class: `MiniMaxForSequenceClassification` (MiniMaxConfig model) - `Ministral3Config` configuration class: `Ministral3ForSequenceClassification` (Ministral3Config model) - `MinistralConfig` configuration class: `MinistralForSequenceClassification` (MinistralConfig model) - `Mistral4Config` configuration class: `Mistral4ForSequenceClassification` (Mistral4Config model) - `MistralConfig` configuration class: `MistralForSequenceClassification` (MistralConfig model) - `MixtralConfig` configuration class: `MixtralForSequenceClassification` (MixtralConfig model) - `MobileBertConfig` configuration class: `MobileBertForSequenceClassification` (MobileBertConfig model) - `ModernBertConfig` configuration class: `ModernBertForSequenceClassification` (ModernBertConfig model) - `ModernBertDecoderConfig` configuration class: `ModernBertDecoderForSequenceClassification` (ModernBertDecoderConfig model) - `ModernVBertConfig` configuration class: `ModernVBertForSequenceClassification` (ModernVBertConfig model) - `MptConfig` configuration class: `MptForSequenceClassification` (MptConfig model) - `MraConfig` configuration class: `MraForSequenceClassification` (MraConfig model) - `MvpConfig` configuration class: `MvpForSequenceClassification` (MvpConfig model) - `NemotronConfig` configuration class: `NemotronForSequenceClassification` (NemotronConfig model) - `NomicBertConfig` configuration class: `NomicBertForSequenceClassification` (NomicBertConfig model) - `NystromformerConfig` configuration class: `NystromformerForSequenceClassification` (NystromformerConfig model) - `OPTConfig` configuration class: `OPTForSequenceClassification` (OPTConfig model) - `Olmo2Config` configuration class: `Olmo2ForSequenceClassification` (Olmo2Config model) - `Olmo3Config` configuration class: `Olmo3ForSequenceClassification` (Olmo3Config model) - `OlmoConfig` configuration class: `OlmoForSequenceClassification` (OlmoConfig model) - `OpenAIGPTConfig` configuration class: `OpenAIGPTForSequenceClassification` (OpenAIGPTConfig model) - `PLBartConfig` configuration class: `PLBartForSequenceClassification` (PLBartConfig model) - `PerceiverConfig` configuration class: `PerceiverForSequenceClassification` (PerceiverConfig model) - `PersimmonConfig` configuration class: `PersimmonForSequenceClassification` (PersimmonConfig model) - `Phi3Config` configuration class: `Phi3ForSequenceClassification` (Phi3Config model) - `PhiConfig` configuration class: `PhiForSequenceClassification` (PhiConfig model) - `PhimoeConfig` configuration class: `PhimoeForSequenceClassification` (PhimoeConfig model) - `Qwen2Config` configuration class: `Qwen2ForSequenceClassification` (Qwen2Config model) - `Qwen2MoeConfig` configuration class: `Qwen2MoeForSequenceClassification` (Qwen2MoeConfig model) - `Qwen3Config` configuration class: `Qwen3ForSequenceClassification` (Qwen3Config model) - `Qwen3MoeConfig` configuration class: `Qwen3MoeForSequenceClassification` (Qwen3MoeConfig model) - `Qwen3NextConfig` configuration class: `Qwen3NextForSequenceClassification` (Qwen3NextConfig model) - `Qwen3_5Config` configuration class: `Qwen3_5ForSequenceClassification` (Qwen3_5Config model) - `Qwen3_5TextConfig` configuration class: `Qwen3_5ForSequenceClassification` (Qwen3_5TextConfig model) - `ReformerConfig` configuration class: `ReformerForSequenceClassification` (ReformerConfig model) - `RemBertConfig` configuration class: `RemBertForSequenceClassification` (RemBertConfig model) - `RoCBertConfig` configuration class: `RoCBertForSequenceClassification` (RoCBertConfig model) - `RoFormerConfig` configuration class: `RoFormerForSequenceClassification` (RoFormerConfig model) - `RobertaConfig` configuration class: `RobertaForSequenceClassification` (RobertaConfig model) - `RobertaPreLayerNormConfig` configuration class: `RobertaPreLayerNormForSequenceClassification` (RobertaPreLayerNormConfig model) - `SeedOssConfig` configuration class: `SeedOssForSequenceClassification` (SeedOssConfig model) - `SmolLM3Config` configuration class: `SmolLM3ForSequenceClassification` (SmolLM3Config model) - `SqueezeBertConfig` configuration class: `SqueezeBertForSequenceClassification` (SqueezeBertConfig model) - `StableLmConfig` configuration class: `StableLmForSequenceClassification` (StableLmConfig model) - `Starcoder2Config` configuration class: `Starcoder2ForSequenceClassification` (Starcoder2Config model) - `T5Config` configuration class: `T5ForSequenceClassification` (T5Config model) - `T5Gemma2Config` configuration class: `T5Gemma2ForSequenceClassification` (T5Gemma2Config model) - `T5GemmaConfig` configuration class: `T5GemmaForSequenceClassification` (T5GemmaConfig model) - `TapasConfig` configuration class: `TapasForSequenceClassification` (TapasConfig model) - `UMT5Config` configuration class: `UMT5ForSequenceClassification` (UMT5Config model) - `XLMConfig` configuration class: `XLMForSequenceClassification` (XLMConfig model) - `XLMRobertaConfig` configuration class: `XLMRobertaForSequenceClassification` (XLMRobertaConfig model) - `XLMRobertaXLConfig` configuration class: `XLMRobertaXLForSequenceClassification` (XLMRobertaXLConfig model) - `XLNetConfig` configuration class: `XLNetForSequenceClassification` (XLNetConfig model) - `XmodConfig` configuration class: `XmodForSequenceClassification` (XmodConfig model) - `YosoConfig` configuration class: `YosoForSequenceClassification` (YosoConfig model) - `Zamba2Config` configuration class: `Zamba2ForSequenceClassification` (Zamba2Config model) - `ZambaConfig` configuration class: `ZambaForSequenceClassification` (ZambaConfig model) attn_implementation (`str`, *optional*) : The attention implementation to use in the model (if relevant). Can be any of `"eager"` (manual implementation of the attention), `"sdpa"` (using [`F.scaled_dot_product_attention`](https://pytorch.org/docs/master/generated/torch.nn.functional.scaled_dot_product_attention.html)), `"flash_attention_2"` (using [Dao-AILab/flash-attention](https://github.com/Dao-AILab/flash-attention)), or `"flash_attention_3"` (using [Dao-AILab/flash-attention/hopper](https://github.com/Dao-AILab/flash-attention/tree/main/hopper)). By default, if available, SDPA will be used for torch>=2.1.1. The default is otherwise the manual `"eager"` implementation. #### from_pretrained[[transformers.AutoModelForSequenceClassification.from_pretrained]] [Source](https://github.com/huggingface/transformers/blob/v5.6.0/src/transformers/models/auto/auto_factory.py#L253) Instantiate one of the model classes of the library (with a sequence classification head) from a pretrained model. The model class to instantiate is selected based on the `model_type` property of the config object (either passed as an argument or loaded from `pretrained_model_name_or_path` if possible), or when it's missing, by falling back to using pattern matching on `pretrained_model_name_or_path`: - **albert** -- [AlbertForSequenceClassification](/docs/transformers/v5.6.0/ja/model_doc/albert#transformers.AlbertForSequenceClassification) (AlbertConfig model) - **arcee** -- `ArceeForSequenceClassification` (ArceeConfig model) - **bart** -- [BartForSequenceClassification](/docs/transformers/v5.6.0/ja/model_doc/bart#transformers.BartForSequenceClassification) (BartConfig model) - **bert** -- [BertForSequenceClassification](/docs/transformers/v5.6.0/ja/model_doc/bert#transformers.BertForSequenceClassification) (BertConfig model) - **big_bird** -- [BigBirdForSequenceClassification](/docs/transformers/v5.6.0/ja/model_doc/big_bird#transformers.BigBirdForSequenceClassification) (BigBirdConfig model) - **bigbird_pegasus** -- [BigBirdPegasusForSequenceClassification](/docs/transformers/v5.6.0/ja/model_doc/bigbird_pegasus#transformers.BigBirdPegasusForSequenceClassification) (BigBirdPegasusConfig model) - **biogpt** -- [BioGptForSequenceClassification](/docs/transformers/v5.6.0/ja/model_doc/biogpt#transformers.BioGptForSequenceClassification) (BioGptConfig model) - **bloom** -- [BloomForSequenceClassification](/docs/transformers/v5.6.0/ja/model_doc/bloom#transformers.BloomForSequenceClassification) (BloomConfig model) - **camembert** -- [CamembertForSequenceClassification](/docs/transformers/v5.6.0/ja/model_doc/camembert#transformers.CamembertForSequenceClassification) (CamembertConfig model) - **canine** -- [CanineForSequenceClassification](/docs/transformers/v5.6.0/ja/model_doc/canine#transformers.CanineForSequenceClassification) (CanineConfig model) - **convbert** -- [ConvBertForSequenceClassification](/docs/transformers/v5.6.0/ja/model_doc/convbert#transformers.ConvBertForSequenceClassification) (ConvBertConfig model) - **ctrl** -- [CTRLForSequenceClassification](/docs/transformers/v5.6.0/ja/model_doc/ctrl#transformers.CTRLForSequenceClassification) (CTRLConfig model) - **data2vec-text** -- [Data2VecTextForSequenceClassification](/docs/transformers/v5.6.0/ja/model_doc/data2vec#transformers.Data2VecTextForSequenceClassification) (Data2VecTextConfig model) - **deberta** -- [DebertaForSequenceClassification](/docs/transformers/v5.6.0/ja/model_doc/deberta#transformers.DebertaForSequenceClassification) (DebertaConfig model) - **deberta-v2** -- [DebertaV2ForSequenceClassification](/docs/transformers/v5.6.0/ja/model_doc/deberta-v2#transformers.DebertaV2ForSequenceClassification) (DebertaV2Config model) - **deepseek_v2** -- `DeepseekV2ForSequenceClassification` (DeepseekV2Config model) - **deepseek_v3** -- `DeepseekV3ForSequenceClassification` (DeepseekV3Config model) - **diffllama** -- `DiffLlamaForSequenceClassification` (DiffLlamaConfig model) - **distilbert** -- `DistilBertForSequenceClassification` (DistilBertConfig model) - **doge** -- `DogeForSequenceClassification` (DogeConfig model) - **electra** -- `ElectraForSequenceClassification` (ElectraConfig model) - **ernie** -- `ErnieForSequenceClassification` (ErnieConfig model) - **esm** -- `EsmForSequenceClassification` (EsmConfig model) - **eurobert** -- `EuroBertForSequenceClassification` (EuroBertConfig model) - **exaone4** -- `Exaone4ForSequenceClassification` (Exaone4Config model) - **falcon** -- `FalconForSequenceClassification` (FalconConfig model) - **flaubert** -- `FlaubertForSequenceClassification` (FlaubertConfig model) - **fnet** -- `FNetForSequenceClassification` (FNetConfig model) - **funnel** -- `FunnelForSequenceClassification` (FunnelConfig model) - **gemma** -- `GemmaForSequenceClassification` (GemmaConfig model) - **gemma2** -- `Gemma2ForSequenceClassification` (Gemma2Config model) - **gemma3** -- `Gemma3ForSequenceClassification` (Gemma3Config model) - **gemma3_text** -- `Gemma3TextForSequenceClassification` (Gemma3TextConfig model) - **glm** -- `GlmForSequenceClassification` (GlmConfig model) - **glm4** -- `Glm4ForSequenceClassification` (Glm4Config model) - **gpt-sw3** -- `GPT2ForSequenceClassification` (GPT2Config model) - **gpt2** -- `GPT2ForSequenceClassification` (GPT2Config model) - **gpt_bigcode** -- `GPTBigCodeForSequenceClassification` (GPTBigCodeConfig model) - **gpt_neo** -- `GPTNeoForSequenceClassification` (GPTNeoConfig model) - **gpt_neox** -- `GPTNeoXForSequenceClassification` (GPTNeoXConfig model) - **gpt_oss** -- `GptOssForSequenceClassification` (GptOssConfig model) - **gptj** -- `GPTJForSequenceClassification` (GPTJConfig model) - **helium** -- `HeliumForSequenceClassification` (HeliumConfig model) - **hunyuan_v1_dense** -- `HunYuanDenseV1ForSequenceClassification` (HunYuanDenseV1Config model) - **hunyuan_v1_moe** -- `HunYuanMoEV1ForSequenceClassification` (HunYuanMoEV1Config model) - **ibert** -- `IBertForSequenceClassification` (IBertConfig model) - **jamba** -- `JambaForSequenceClassification` (JambaConfig model) - **jetmoe** -- `JetMoeForSequenceClassification` (JetMoeConfig model) - **jina_embeddings_v3** -- `JinaEmbeddingsV3ForSequenceClassification` (JinaEmbeddingsV3Config model) - **layoutlm** -- `LayoutLMForSequenceClassification` (LayoutLMConfig model) - **layoutlmv2** -- `LayoutLMv2ForSequenceClassification` (LayoutLMv2Config model) - **layoutlmv3** -- `LayoutLMv3ForSequenceClassification` (LayoutLMv3Config model) - **lilt** -- `LiltForSequenceClassification` (LiltConfig model) - **llama** -- `LlamaForSequenceClassification` (LlamaConfig model) - **longformer** -- `LongformerForSequenceClassification` (LongformerConfig model) - **luke** -- `LukeForSequenceClassification` (LukeConfig model) - **markuplm** -- `MarkupLMForSequenceClassification` (MarkupLMConfig model) - **mbart** -- `MBartForSequenceClassification` (MBartConfig model) - **megatron-bert** -- `MegatronBertForSequenceClassification` (MegatronBertConfig model) - **minimax** -- `MiniMaxForSequenceClassification` (MiniMaxConfig model) - **ministral** -- `MinistralForSequenceClassification` (MinistralConfig model) - **ministral3** -- `Ministral3ForSequenceClassification` (Ministral3Config model) - **mistral** -- `MistralForSequenceClassification` (MistralConfig model) - **mistral4** -- `Mistral4ForSequenceClassification` (Mistral4Config model) - **mixtral** -- `MixtralForSequenceClassification` (MixtralConfig model) - **mobilebert** -- `MobileBertForSequenceClassification` (MobileBertConfig model) - **modernbert** -- `ModernBertForSequenceClassification` (ModernBertConfig model) - **modernbert-decoder** -- `ModernBertDecoderForSequenceClassification` (ModernBertDecoderConfig model) - **modernvbert** -- `ModernVBertForSequenceClassification` (ModernVBertConfig model) - **mpnet** -- `MPNetForSequenceClassification` (MPNetConfig model) - **mpt** -- `MptForSequenceClassification` (MptConfig model) - **mra** -- `MraForSequenceClassification` (MraConfig model) - **mt5** -- `MT5ForSequenceClassification` (MT5Config model) - **mvp** -- `MvpForSequenceClassification` (MvpConfig model) - **nemotron** -- `NemotronForSequenceClassification` (NemotronConfig model) - **nomic_bert** -- `NomicBertForSequenceClassification` (NomicBertConfig model) - **nystromformer** -- `NystromformerForSequenceClassification` (NystromformerConfig model) - **olmo** -- `OlmoForSequenceClassification` (OlmoConfig model) - **olmo2** -- `Olmo2ForSequenceClassification` (Olmo2Config model) - **olmo3** -- `Olmo3ForSequenceClassification` (Olmo3Config model) - **openai-gpt** -- `OpenAIGPTForSequenceClassification` (OpenAIGPTConfig model) - **opt** -- `OPTForSequenceClassification` (OPTConfig model) - **perceiver** -- `PerceiverForSequenceClassification` (PerceiverConfig model) - **persimmon** -- `PersimmonForSequenceClassification` (PersimmonConfig model) - **phi** -- `PhiForSequenceClassification` (PhiConfig model) - **phi3** -- `Phi3ForSequenceClassification` (Phi3Config model) - **phimoe** -- `PhimoeForSequenceClassification` (PhimoeConfig model) - **plbart** -- `PLBartForSequenceClassification` (PLBartConfig model) - **qwen2** -- `Qwen2ForSequenceClassification` (Qwen2Config model) - **qwen2_moe** -- `Qwen2MoeForSequenceClassification` (Qwen2MoeConfig model) - **qwen3** -- `Qwen3ForSequenceClassification` (Qwen3Config model) - **qwen3_5** -- `Qwen3_5ForSequenceClassification` (Qwen3_5Config model) - **qwen3_5_text** -- `Qwen3_5ForSequenceClassification` (Qwen3_5TextConfig model) - **qwen3_moe** -- `Qwen3MoeForSequenceClassification` (Qwen3MoeConfig model) - **qwen3_next** -- `Qwen3NextForSequenceClassification` (Qwen3NextConfig model) - **reformer** -- `ReformerForSequenceClassification` (ReformerConfig model) - **rembert** -- `RemBertForSequenceClassification` (RemBertConfig model) - **roberta** -- `RobertaForSequenceClassification` (RobertaConfig model) - **roberta-prelayernorm** -- `RobertaPreLayerNormForSequenceClassification` (RobertaPreLayerNormConfig model) - **roc_bert** -- `RoCBertForSequenceClassification` (RoCBertConfig model) - **roformer** -- `RoFormerForSequenceClassification` (RoFormerConfig model) - **seed_oss** -- `SeedOssForSequenceClassification` (SeedOssConfig model) - **smollm3** -- `SmolLM3ForSequenceClassification` (SmolLM3Config model) - **squeezebert** -- `SqueezeBertForSequenceClassification` (SqueezeBertConfig model) - **stablelm** -- `StableLmForSequenceClassification` (StableLmConfig model) - **starcoder2** -- `Starcoder2ForSequenceClassification` (Starcoder2Config model) - **t5** -- `T5ForSequenceClassification` (T5Config model) - **t5gemma** -- `T5GemmaForSequenceClassification` (T5GemmaConfig model) - **t5gemma2** -- `T5Gemma2ForSequenceClassification` (T5Gemma2Config model) - **tapas** -- `TapasForSequenceClassification` (TapasConfig model) - **umt5** -- `UMT5ForSequenceClassification` (UMT5Config model) - **xlm** -- `XLMForSequenceClassification` (XLMConfig model) - **xlm-roberta** -- `XLMRobertaForSequenceClassification` (XLMRobertaConfig model) - **xlm-roberta-xl** -- `XLMRobertaXLForSequenceClassification` (XLMRobertaXLConfig model) - **xlnet** -- `XLNetForSequenceClassification` (XLNetConfig model) - **xmod** -- `XmodForSequenceClassification` (XmodConfig model) - **yoso** -- `YosoForSequenceClassification` (YosoConfig model) - **zamba** -- `ZambaForSequenceClassification` (ZambaConfig model) - **zamba2** -- `Zamba2ForSequenceClassification` (Zamba2Config model) The model is set in evaluation mode by default using `model.eval()` (so for instance, dropout modules are deactivated). To train the model, you should first set it back in training mode with `model.train()` Examples: ```python >>> from transformers import AutoConfig, AutoModelForSequenceClassification >>> # Download model and configuration from huggingface.co and cache. >>> model = AutoModelForSequenceClassification.from_pretrained("google-bert/bert-base-cased") >>> # Update configuration during loading >>> model = AutoModelForSequenceClassification.from_pretrained("google-bert/bert-base-cased", output_attentions=True) >>> model.config.output_attentions True ``` **Parameters:** pretrained_model_name_or_path (`str` or `os.PathLike`) : Can be either: - A string, the *model id* of a pretrained model hosted inside a model repo on huggingface.co. - A path to a *directory* containing model weights saved using [save_pretrained()](/docs/transformers/v5.6.0/ja/main_classes/model#transformers.PreTrainedModel.save_pretrained), e.g., `./my_model_directory/`. model_args (additional positional arguments, *optional*) : Will be passed along to the underlying model `__init__()` method. config ([PreTrainedConfig](/docs/transformers/v5.6.0/ja/main_classes/configuration#transformers.PreTrainedConfig), *optional*) : Configuration for the model to use instead of an automatically loaded configuration. Configuration can be automatically loaded when: - The model is a model provided by the library (loaded with the *model id* string of a pretrained model). - The model was saved using [save_pretrained()](/docs/transformers/v5.6.0/ja/main_classes/model#transformers.PreTrainedModel.save_pretrained) and is reloaded by supplying the save directory. - The model is loaded by supplying a local directory as `pretrained_model_name_or_path` and a configuration JSON file named *config.json* is found in the directory. state_dict (*dict[str, torch.Tensor]*, *optional*) : A state dictionary to use instead of a state dictionary loaded from saved weights file. This option can be used if you want to create a model from a pretrained configuration but load your own weights. In this case though, you should check if using [save_pretrained()](/docs/transformers/v5.6.0/ja/main_classes/model#transformers.PreTrainedModel.save_pretrained) and [from_pretrained()](/docs/transformers/v5.6.0/ja/main_classes/model#transformers.PreTrainedModel.from_pretrained) is not a simpler option. cache_dir (`str` or `os.PathLike`, *optional*) : Path to a directory in which a downloaded pretrained model configuration should be cached if the standard cache should not be used. force_download (`bool`, *optional*, defaults to `False`) : Whether or not to force the (re-)download of the model weights and configuration files, overriding the cached versions if they exist. proxies (`dict[str, str]`, *optional*) : A dictionary of proxy servers to use by protocol or endpoint, e.g., `{'http': 'foo.bar:3128', 'http://hostname': 'foo.bar:4012'}`. The proxies are used on each request. output_loading_info(`bool`, *optional*, defaults to `False`) : Whether or not to also return a dictionary containing missing keys, unexpected keys and error messages. local_files_only(`bool`, *optional*, defaults to `False`) : Whether or not to only look at local files (e.g., not try downloading the model). revision (`str`, *optional*, defaults to `"main"`) : The specific model version to use. It can be a branch name, a tag name, or a commit id, since we use a git-based system for storing models and other artifacts on huggingface.co, so `revision` can be any identifier allowed by git. trust_remote_code (`bool`, *optional*, defaults to `False`) : Whether or not to allow for custom models defined on the Hub in their own modeling files. This option should only be set to `True` for repositories you trust and in which you have read the code, as it will execute code present on the Hub on your local machine. code_revision (`str`, *optional*, defaults to `"main"`) : The specific revision to use for the code on the Hub, if the code leaves in a different repository than the rest of the model. It can be a branch name, a tag name, or a commit id, since we use a git-based system for storing models and other artifacts on huggingface.co, so `revision` can be any identifier allowed by git. kwargs (additional keyword arguments, *optional*) : Can be used to update the configuration object (after it being loaded) and initiate the model (e.g., `output_attentions=True`). Behaves differently depending on whether a `config` is provided or automatically loaded: - If a configuration is provided with `config`, `**kwargs` will be directly passed to the underlying model's `__init__` method (we assume all relevant updates to the configuration have already been done) - If a configuration is not provided, `kwargs` will be first passed to the configuration class initialization function ([from_pretrained()](/docs/transformers/v5.6.0/ja/main_classes/configuration#transformers.PreTrainedConfig.from_pretrained)). Each key of `kwargs` that corresponds to a configuration attribute will be used to override said attribute with the supplied `kwargs` value. Remaining keys that do not correspond to any configuration attribute will be passed to the underlying model's `__init__` function. ### AutoModelForMultipleChoice[[transformers.AutoModelForMultipleChoice]] #### transformers.AutoModelForMultipleChoice[[transformers.AutoModelForMultipleChoice]] [Source](https://github.com/huggingface/transformers/blob/v5.6.0/src/transformers/models/auto/modeling_auto.py#L2084) This is a generic model class that will be instantiated as one of the model classes of the library (with a multiple choice head) when created with the [from_pretrained()](/docs/transformers/v5.6.0/ja/model_doc/auto#transformers.AutoModel.from_pretrained) class method or the [from_config()](/docs/transformers/v5.6.0/ja/model_doc/auto#transformers.AutoModel.from_config) class method. This class cannot be instantiated directly using `__init__()` (throws an error). from_configtransformers.AutoModelForMultipleChoice.from_confighttps://github.com/huggingface/transformers/blob/v5.6.0/src/transformers/models/auto/auto_factory.py#L206[{"name": "**kwargs", "val": ""}]- **config** ([PreTrainedConfig](/docs/transformers/v5.6.0/ja/main_classes/configuration#transformers.PreTrainedConfig)) -- The model class to instantiate is selected based on the configuration class: - [AlbertConfig](/docs/transformers/v5.6.0/ja/model_doc/albert#transformers.AlbertConfig) configuration class: [AlbertForMultipleChoice](/docs/transformers/v5.6.0/ja/model_doc/albert#transformers.AlbertForMultipleChoice) (AlbertConfig model) - [BertConfig](/docs/transformers/v5.6.0/ja/model_doc/bert#transformers.BertConfig) configuration class: [BertForMultipleChoice](/docs/transformers/v5.6.0/ja/model_doc/bert#transformers.BertForMultipleChoice) (BertConfig model) - [BigBirdConfig](/docs/transformers/v5.6.0/ja/model_doc/big_bird#transformers.BigBirdConfig) configuration class: [BigBirdForMultipleChoice](/docs/transformers/v5.6.0/ja/model_doc/big_bird#transformers.BigBirdForMultipleChoice) (BigBirdConfig model) - [CamembertConfig](/docs/transformers/v5.6.0/ja/model_doc/camembert#transformers.CamembertConfig) configuration class: [CamembertForMultipleChoice](/docs/transformers/v5.6.0/ja/model_doc/camembert#transformers.CamembertForMultipleChoice) (CamembertConfig model) - [CanineConfig](/docs/transformers/v5.6.0/ja/model_doc/canine#transformers.CanineConfig) configuration class: [CanineForMultipleChoice](/docs/transformers/v5.6.0/ja/model_doc/canine#transformers.CanineForMultipleChoice) (CanineConfig model) - [ConvBertConfig](/docs/transformers/v5.6.0/ja/model_doc/convbert#transformers.ConvBertConfig) configuration class: [ConvBertForMultipleChoice](/docs/transformers/v5.6.0/ja/model_doc/convbert#transformers.ConvBertForMultipleChoice) (ConvBertConfig model) - [Data2VecTextConfig](/docs/transformers/v5.6.0/ja/model_doc/data2vec#transformers.Data2VecTextConfig) configuration class: [Data2VecTextForMultipleChoice](/docs/transformers/v5.6.0/ja/model_doc/data2vec#transformers.Data2VecTextForMultipleChoice) (Data2VecTextConfig model) - [DebertaV2Config](/docs/transformers/v5.6.0/ja/model_doc/deberta-v2#transformers.DebertaV2Config) configuration class: [DebertaV2ForMultipleChoice](/docs/transformers/v5.6.0/ja/model_doc/deberta-v2#transformers.DebertaV2ForMultipleChoice) (DebertaV2Config model) - `DistilBertConfig` configuration class: `DistilBertForMultipleChoice` (DistilBertConfig model) - `ElectraConfig` configuration class: `ElectraForMultipleChoice` (ElectraConfig model) - `ErnieConfig` configuration class: `ErnieForMultipleChoice` (ErnieConfig model) - `FNetConfig` configuration class: `FNetForMultipleChoice` (FNetConfig model) - `FlaubertConfig` configuration class: `FlaubertForMultipleChoice` (FlaubertConfig model) - `FunnelConfig` configuration class: `FunnelForMultipleChoice` (FunnelConfig model) - `IBertConfig` configuration class: `IBertForMultipleChoice` (IBertConfig model) - `LongformerConfig` configuration class: `LongformerForMultipleChoice` (LongformerConfig model) - `LukeConfig` configuration class: `LukeForMultipleChoice` (LukeConfig model) - `MPNetConfig` configuration class: `MPNetForMultipleChoice` (MPNetConfig model) - `MegatronBertConfig` configuration class: `MegatronBertForMultipleChoice` (MegatronBertConfig model) - `MobileBertConfig` configuration class: `MobileBertForMultipleChoice` (MobileBertConfig model) - `ModernBertConfig` configuration class: `ModernBertForMultipleChoice` (ModernBertConfig model) - `MraConfig` configuration class: `MraForMultipleChoice` (MraConfig model) - `NystromformerConfig` configuration class: `NystromformerForMultipleChoice` (NystromformerConfig model) - `RemBertConfig` configuration class: `RemBertForMultipleChoice` (RemBertConfig model) - `RoCBertConfig` configuration class: `RoCBertForMultipleChoice` (RoCBertConfig model) - `RoFormerConfig` configuration class: `RoFormerForMultipleChoice` (RoFormerConfig model) - `RobertaConfig` configuration class: `RobertaForMultipleChoice` (RobertaConfig model) - `RobertaPreLayerNormConfig` configuration class: `RobertaPreLayerNormForMultipleChoice` (RobertaPreLayerNormConfig model) - `SqueezeBertConfig` configuration class: `SqueezeBertForMultipleChoice` (SqueezeBertConfig model) - `XLMConfig` configuration class: `XLMForMultipleChoice` (XLMConfig model) - `XLMRobertaConfig` configuration class: `XLMRobertaForMultipleChoice` (XLMRobertaConfig model) - `XLMRobertaXLConfig` configuration class: `XLMRobertaXLForMultipleChoice` (XLMRobertaXLConfig model) - `XLNetConfig` configuration class: `XLNetForMultipleChoice` (XLNetConfig model) - `XmodConfig` configuration class: `XmodForMultipleChoice` (XmodConfig model) - `YosoConfig` configuration class: `YosoForMultipleChoice` (YosoConfig model) - **attn_implementation** (`str`, *optional*) -- The attention implementation to use in the model (if relevant). Can be any of `"eager"` (manual implementation of the attention), `"sdpa"` (using [`F.scaled_dot_product_attention`](https://pytorch.org/docs/master/generated/torch.nn.functional.scaled_dot_product_attention.html)), `"flash_attention_2"` (using [Dao-AILab/flash-attention](https://github.com/Dao-AILab/flash-attention)), or `"flash_attention_3"` (using [Dao-AILab/flash-attention/hopper](https://github.com/Dao-AILab/flash-attention/tree/main/hopper)). By default, if available, SDPA will be used for torch>=2.1.1. The default is otherwise the manual `"eager"` implementation.0 Instantiates one of the model classes of the library (with a multiple choice head) from a configuration. Note: Loading a model from its configuration file does **not** load the model weights. It only affects the model's configuration. Use [from_pretrained()](/docs/transformers/v5.6.0/ja/model_doc/auto#transformers.AutoModel.from_pretrained) to load the model weights. Examples: ```python >>> from transformers import AutoConfig, AutoModelForMultipleChoice >>> # Download configuration from huggingface.co and cache. >>> config = AutoConfig.from_pretrained("google-bert/bert-base-cased") >>> model = AutoModelForMultipleChoice.from_config(config) ``` **Parameters:** config ([PreTrainedConfig](/docs/transformers/v5.6.0/ja/main_classes/configuration#transformers.PreTrainedConfig)) : The model class to instantiate is selected based on the configuration class: - [AlbertConfig](/docs/transformers/v5.6.0/ja/model_doc/albert#transformers.AlbertConfig) configuration class: [AlbertForMultipleChoice](/docs/transformers/v5.6.0/ja/model_doc/albert#transformers.AlbertForMultipleChoice) (AlbertConfig model) - [BertConfig](/docs/transformers/v5.6.0/ja/model_doc/bert#transformers.BertConfig) configuration class: [BertForMultipleChoice](/docs/transformers/v5.6.0/ja/model_doc/bert#transformers.BertForMultipleChoice) (BertConfig model) - [BigBirdConfig](/docs/transformers/v5.6.0/ja/model_doc/big_bird#transformers.BigBirdConfig) configuration class: [BigBirdForMultipleChoice](/docs/transformers/v5.6.0/ja/model_doc/big_bird#transformers.BigBirdForMultipleChoice) (BigBirdConfig model) - [CamembertConfig](/docs/transformers/v5.6.0/ja/model_doc/camembert#transformers.CamembertConfig) configuration class: [CamembertForMultipleChoice](/docs/transformers/v5.6.0/ja/model_doc/camembert#transformers.CamembertForMultipleChoice) (CamembertConfig model) - [CanineConfig](/docs/transformers/v5.6.0/ja/model_doc/canine#transformers.CanineConfig) configuration class: [CanineForMultipleChoice](/docs/transformers/v5.6.0/ja/model_doc/canine#transformers.CanineForMultipleChoice) (CanineConfig model) - [ConvBertConfig](/docs/transformers/v5.6.0/ja/model_doc/convbert#transformers.ConvBertConfig) configuration class: [ConvBertForMultipleChoice](/docs/transformers/v5.6.0/ja/model_doc/convbert#transformers.ConvBertForMultipleChoice) (ConvBertConfig model) - [Data2VecTextConfig](/docs/transformers/v5.6.0/ja/model_doc/data2vec#transformers.Data2VecTextConfig) configuration class: [Data2VecTextForMultipleChoice](/docs/transformers/v5.6.0/ja/model_doc/data2vec#transformers.Data2VecTextForMultipleChoice) (Data2VecTextConfig model) - [DebertaV2Config](/docs/transformers/v5.6.0/ja/model_doc/deberta-v2#transformers.DebertaV2Config) configuration class: [DebertaV2ForMultipleChoice](/docs/transformers/v5.6.0/ja/model_doc/deberta-v2#transformers.DebertaV2ForMultipleChoice) (DebertaV2Config model) - `DistilBertConfig` configuration class: `DistilBertForMultipleChoice` (DistilBertConfig model) - `ElectraConfig` configuration class: `ElectraForMultipleChoice` (ElectraConfig model) - `ErnieConfig` configuration class: `ErnieForMultipleChoice` (ErnieConfig model) - `FNetConfig` configuration class: `FNetForMultipleChoice` (FNetConfig model) - `FlaubertConfig` configuration class: `FlaubertForMultipleChoice` (FlaubertConfig model) - `FunnelConfig` configuration class: `FunnelForMultipleChoice` (FunnelConfig model) - `IBertConfig` configuration class: `IBertForMultipleChoice` (IBertConfig model) - `LongformerConfig` configuration class: `LongformerForMultipleChoice` (LongformerConfig model) - `LukeConfig` configuration class: `LukeForMultipleChoice` (LukeConfig model) - `MPNetConfig` configuration class: `MPNetForMultipleChoice` (MPNetConfig model) - `MegatronBertConfig` configuration class: `MegatronBertForMultipleChoice` (MegatronBertConfig model) - `MobileBertConfig` configuration class: `MobileBertForMultipleChoice` (MobileBertConfig model) - `ModernBertConfig` configuration class: `ModernBertForMultipleChoice` (ModernBertConfig model) - `MraConfig` configuration class: `MraForMultipleChoice` (MraConfig model) - `NystromformerConfig` configuration class: `NystromformerForMultipleChoice` (NystromformerConfig model) - `RemBertConfig` configuration class: `RemBertForMultipleChoice` (RemBertConfig model) - `RoCBertConfig` configuration class: `RoCBertForMultipleChoice` (RoCBertConfig model) - `RoFormerConfig` configuration class: `RoFormerForMultipleChoice` (RoFormerConfig model) - `RobertaConfig` configuration class: `RobertaForMultipleChoice` (RobertaConfig model) - `RobertaPreLayerNormConfig` configuration class: `RobertaPreLayerNormForMultipleChoice` (RobertaPreLayerNormConfig model) - `SqueezeBertConfig` configuration class: `SqueezeBertForMultipleChoice` (SqueezeBertConfig model) - `XLMConfig` configuration class: `XLMForMultipleChoice` (XLMConfig model) - `XLMRobertaConfig` configuration class: `XLMRobertaForMultipleChoice` (XLMRobertaConfig model) - `XLMRobertaXLConfig` configuration class: `XLMRobertaXLForMultipleChoice` (XLMRobertaXLConfig model) - `XLNetConfig` configuration class: `XLNetForMultipleChoice` (XLNetConfig model) - `XmodConfig` configuration class: `XmodForMultipleChoice` (XmodConfig model) - `YosoConfig` configuration class: `YosoForMultipleChoice` (YosoConfig model) attn_implementation (`str`, *optional*) : The attention implementation to use in the model (if relevant). Can be any of `"eager"` (manual implementation of the attention), `"sdpa"` (using [`F.scaled_dot_product_attention`](https://pytorch.org/docs/master/generated/torch.nn.functional.scaled_dot_product_attention.html)), `"flash_attention_2"` (using [Dao-AILab/flash-attention](https://github.com/Dao-AILab/flash-attention)), or `"flash_attention_3"` (using [Dao-AILab/flash-attention/hopper](https://github.com/Dao-AILab/flash-attention/tree/main/hopper)). By default, if available, SDPA will be used for torch>=2.1.1. The default is otherwise the manual `"eager"` implementation. #### from_pretrained[[transformers.AutoModelForMultipleChoice.from_pretrained]] [Source](https://github.com/huggingface/transformers/blob/v5.6.0/src/transformers/models/auto/auto_factory.py#L253) Instantiate one of the model classes of the library (with a multiple choice head) from a pretrained model. The model class to instantiate is selected based on the `model_type` property of the config object (either passed as an argument or loaded from `pretrained_model_name_or_path` if possible), or when it's missing, by falling back to using pattern matching on `pretrained_model_name_or_path`: - **albert** -- [AlbertForMultipleChoice](/docs/transformers/v5.6.0/ja/model_doc/albert#transformers.AlbertForMultipleChoice) (AlbertConfig model) - **bert** -- [BertForMultipleChoice](/docs/transformers/v5.6.0/ja/model_doc/bert#transformers.BertForMultipleChoice) (BertConfig model) - **big_bird** -- [BigBirdForMultipleChoice](/docs/transformers/v5.6.0/ja/model_doc/big_bird#transformers.BigBirdForMultipleChoice) (BigBirdConfig model) - **camembert** -- [CamembertForMultipleChoice](/docs/transformers/v5.6.0/ja/model_doc/camembert#transformers.CamembertForMultipleChoice) (CamembertConfig model) - **canine** -- [CanineForMultipleChoice](/docs/transformers/v5.6.0/ja/model_doc/canine#transformers.CanineForMultipleChoice) (CanineConfig model) - **convbert** -- [ConvBertForMultipleChoice](/docs/transformers/v5.6.0/ja/model_doc/convbert#transformers.ConvBertForMultipleChoice) (ConvBertConfig model) - **data2vec-text** -- [Data2VecTextForMultipleChoice](/docs/transformers/v5.6.0/ja/model_doc/data2vec#transformers.Data2VecTextForMultipleChoice) (Data2VecTextConfig model) - **deberta-v2** -- [DebertaV2ForMultipleChoice](/docs/transformers/v5.6.0/ja/model_doc/deberta-v2#transformers.DebertaV2ForMultipleChoice) (DebertaV2Config model) - **distilbert** -- `DistilBertForMultipleChoice` (DistilBertConfig model) - **electra** -- `ElectraForMultipleChoice` (ElectraConfig model) - **ernie** -- `ErnieForMultipleChoice` (ErnieConfig model) - **flaubert** -- `FlaubertForMultipleChoice` (FlaubertConfig model) - **fnet** -- `FNetForMultipleChoice` (FNetConfig model) - **funnel** -- `FunnelForMultipleChoice` (FunnelConfig model) - **ibert** -- `IBertForMultipleChoice` (IBertConfig model) - **longformer** -- `LongformerForMultipleChoice` (LongformerConfig model) - **luke** -- `LukeForMultipleChoice` (LukeConfig model) - **megatron-bert** -- `MegatronBertForMultipleChoice` (MegatronBertConfig model) - **mobilebert** -- `MobileBertForMultipleChoice` (MobileBertConfig model) - **modernbert** -- `ModernBertForMultipleChoice` (ModernBertConfig model) - **mpnet** -- `MPNetForMultipleChoice` (MPNetConfig model) - **mra** -- `MraForMultipleChoice` (MraConfig model) - **nystromformer** -- `NystromformerForMultipleChoice` (NystromformerConfig model) - **rembert** -- `RemBertForMultipleChoice` (RemBertConfig model) - **roberta** -- `RobertaForMultipleChoice` (RobertaConfig model) - **roberta-prelayernorm** -- `RobertaPreLayerNormForMultipleChoice` (RobertaPreLayerNormConfig model) - **roc_bert** -- `RoCBertForMultipleChoice` (RoCBertConfig model) - **roformer** -- `RoFormerForMultipleChoice` (RoFormerConfig model) - **squeezebert** -- `SqueezeBertForMultipleChoice` (SqueezeBertConfig model) - **xlm** -- `XLMForMultipleChoice` (XLMConfig model) - **xlm-roberta** -- `XLMRobertaForMultipleChoice` (XLMRobertaConfig model) - **xlm-roberta-xl** -- `XLMRobertaXLForMultipleChoice` (XLMRobertaXLConfig model) - **xlnet** -- `XLNetForMultipleChoice` (XLNetConfig model) - **xmod** -- `XmodForMultipleChoice` (XmodConfig model) - **yoso** -- `YosoForMultipleChoice` (YosoConfig model) The model is set in evaluation mode by default using `model.eval()` (so for instance, dropout modules are deactivated). To train the model, you should first set it back in training mode with `model.train()` Examples: ```python >>> from transformers import AutoConfig, AutoModelForMultipleChoice >>> # Download model and configuration from huggingface.co and cache. >>> model = AutoModelForMultipleChoice.from_pretrained("google-bert/bert-base-cased") >>> # Update configuration during loading >>> model = AutoModelForMultipleChoice.from_pretrained("google-bert/bert-base-cased", output_attentions=True) >>> model.config.output_attentions True ``` **Parameters:** pretrained_model_name_or_path (`str` or `os.PathLike`) : Can be either: - A string, the *model id* of a pretrained model hosted inside a model repo on huggingface.co. - A path to a *directory* containing model weights saved using [save_pretrained()](/docs/transformers/v5.6.0/ja/main_classes/model#transformers.PreTrainedModel.save_pretrained), e.g., `./my_model_directory/`. model_args (additional positional arguments, *optional*) : Will be passed along to the underlying model `__init__()` method. config ([PreTrainedConfig](/docs/transformers/v5.6.0/ja/main_classes/configuration#transformers.PreTrainedConfig), *optional*) : Configuration for the model to use instead of an automatically loaded configuration. Configuration can be automatically loaded when: - The model is a model provided by the library (loaded with the *model id* string of a pretrained model). - The model was saved using [save_pretrained()](/docs/transformers/v5.6.0/ja/main_classes/model#transformers.PreTrainedModel.save_pretrained) and is reloaded by supplying the save directory. - The model is loaded by supplying a local directory as `pretrained_model_name_or_path` and a configuration JSON file named *config.json* is found in the directory. state_dict (*dict[str, torch.Tensor]*, *optional*) : A state dictionary to use instead of a state dictionary loaded from saved weights file. This option can be used if you want to create a model from a pretrained configuration but load your own weights. In this case though, you should check if using [save_pretrained()](/docs/transformers/v5.6.0/ja/main_classes/model#transformers.PreTrainedModel.save_pretrained) and [from_pretrained()](/docs/transformers/v5.6.0/ja/main_classes/model#transformers.PreTrainedModel.from_pretrained) is not a simpler option. cache_dir (`str` or `os.PathLike`, *optional*) : Path to a directory in which a downloaded pretrained model configuration should be cached if the standard cache should not be used. force_download (`bool`, *optional*, defaults to `False`) : Whether or not to force the (re-)download of the model weights and configuration files, overriding the cached versions if they exist. proxies (`dict[str, str]`, *optional*) : A dictionary of proxy servers to use by protocol or endpoint, e.g., `{'http': 'foo.bar:3128', 'http://hostname': 'foo.bar:4012'}`. The proxies are used on each request. output_loading_info(`bool`, *optional*, defaults to `False`) : Whether or not to also return a dictionary containing missing keys, unexpected keys and error messages. local_files_only(`bool`, *optional*, defaults to `False`) : Whether or not to only look at local files (e.g., not try downloading the model). revision (`str`, *optional*, defaults to `"main"`) : The specific model version to use. It can be a branch name, a tag name, or a commit id, since we use a git-based system for storing models and other artifacts on huggingface.co, so `revision` can be any identifier allowed by git. trust_remote_code (`bool`, *optional*, defaults to `False`) : Whether or not to allow for custom models defined on the Hub in their own modeling files. This option should only be set to `True` for repositories you trust and in which you have read the code, as it will execute code present on the Hub on your local machine. code_revision (`str`, *optional*, defaults to `"main"`) : The specific revision to use for the code on the Hub, if the code leaves in a different repository than the rest of the model. It can be a branch name, a tag name, or a commit id, since we use a git-based system for storing models and other artifacts on huggingface.co, so `revision` can be any identifier allowed by git. kwargs (additional keyword arguments, *optional*) : Can be used to update the configuration object (after it being loaded) and initiate the model (e.g., `output_attentions=True`). Behaves differently depending on whether a `config` is provided or automatically loaded: - If a configuration is provided with `config`, `**kwargs` will be directly passed to the underlying model's `__init__` method (we assume all relevant updates to the configuration have already been done) - If a configuration is not provided, `kwargs` will be first passed to the configuration class initialization function ([from_pretrained()](/docs/transformers/v5.6.0/ja/main_classes/configuration#transformers.PreTrainedConfig.from_pretrained)). Each key of `kwargs` that corresponds to a configuration attribute will be used to override said attribute with the supplied `kwargs` value. Remaining keys that do not correspond to any configuration attribute will be passed to the underlying model's `__init__` function. ### AutoModelForNextSentencePrediction[[transformers.AutoModelForNextSentencePrediction]] #### transformers.AutoModelForNextSentencePrediction[[transformers.AutoModelForNextSentencePrediction]] [Source](https://github.com/huggingface/transformers/blob/v5.6.0/src/transformers/models/auto/modeling_auto.py#L2091) This is a generic model class that will be instantiated as one of the model classes of the library (with a next sentence prediction head) when created with the [from_pretrained()](/docs/transformers/v5.6.0/ja/model_doc/auto#transformers.AutoModel.from_pretrained) class method or the [from_config()](/docs/transformers/v5.6.0/ja/model_doc/auto#transformers.AutoModel.from_config) class method. This class cannot be instantiated directly using `__init__()` (throws an error). from_configtransformers.AutoModelForNextSentencePrediction.from_confighttps://github.com/huggingface/transformers/blob/v5.6.0/src/transformers/models/auto/auto_factory.py#L206[{"name": "**kwargs", "val": ""}]- **config** ([PreTrainedConfig](/docs/transformers/v5.6.0/ja/main_classes/configuration#transformers.PreTrainedConfig)) -- The model class to instantiate is selected based on the configuration class: - [BertConfig](/docs/transformers/v5.6.0/ja/model_doc/bert#transformers.BertConfig) configuration class: [BertForNextSentencePrediction](/docs/transformers/v5.6.0/ja/model_doc/bert#transformers.BertForNextSentencePrediction) (BertConfig model) - `ErnieConfig` configuration class: `ErnieForNextSentencePrediction` (ErnieConfig model) - `FNetConfig` configuration class: `FNetForNextSentencePrediction` (FNetConfig model) - `MegatronBertConfig` configuration class: `MegatronBertForNextSentencePrediction` (MegatronBertConfig model) - `MobileBertConfig` configuration class: `MobileBertForNextSentencePrediction` (MobileBertConfig model) - **attn_implementation** (`str`, *optional*) -- The attention implementation to use in the model (if relevant). Can be any of `"eager"` (manual implementation of the attention), `"sdpa"` (using [`F.scaled_dot_product_attention`](https://pytorch.org/docs/master/generated/torch.nn.functional.scaled_dot_product_attention.html)), `"flash_attention_2"` (using [Dao-AILab/flash-attention](https://github.com/Dao-AILab/flash-attention)), or `"flash_attention_3"` (using [Dao-AILab/flash-attention/hopper](https://github.com/Dao-AILab/flash-attention/tree/main/hopper)). By default, if available, SDPA will be used for torch>=2.1.1. The default is otherwise the manual `"eager"` implementation.0 Instantiates one of the model classes of the library (with a next sentence prediction head) from a configuration. Note: Loading a model from its configuration file does **not** load the model weights. It only affects the model's configuration. Use [from_pretrained()](/docs/transformers/v5.6.0/ja/model_doc/auto#transformers.AutoModel.from_pretrained) to load the model weights. Examples: ```python >>> from transformers import AutoConfig, AutoModelForNextSentencePrediction >>> # Download configuration from huggingface.co and cache. >>> config = AutoConfig.from_pretrained("google-bert/bert-base-cased") >>> model = AutoModelForNextSentencePrediction.from_config(config) ``` **Parameters:** config ([PreTrainedConfig](/docs/transformers/v5.6.0/ja/main_classes/configuration#transformers.PreTrainedConfig)) : The model class to instantiate is selected based on the configuration class: - [BertConfig](/docs/transformers/v5.6.0/ja/model_doc/bert#transformers.BertConfig) configuration class: [BertForNextSentencePrediction](/docs/transformers/v5.6.0/ja/model_doc/bert#transformers.BertForNextSentencePrediction) (BertConfig model) - `ErnieConfig` configuration class: `ErnieForNextSentencePrediction` (ErnieConfig model) - `FNetConfig` configuration class: `FNetForNextSentencePrediction` (FNetConfig model) - `MegatronBertConfig` configuration class: `MegatronBertForNextSentencePrediction` (MegatronBertConfig model) - `MobileBertConfig` configuration class: `MobileBertForNextSentencePrediction` (MobileBertConfig model) attn_implementation (`str`, *optional*) : The attention implementation to use in the model (if relevant). Can be any of `"eager"` (manual implementation of the attention), `"sdpa"` (using [`F.scaled_dot_product_attention`](https://pytorch.org/docs/master/generated/torch.nn.functional.scaled_dot_product_attention.html)), `"flash_attention_2"` (using [Dao-AILab/flash-attention](https://github.com/Dao-AILab/flash-attention)), or `"flash_attention_3"` (using [Dao-AILab/flash-attention/hopper](https://github.com/Dao-AILab/flash-attention/tree/main/hopper)). By default, if available, SDPA will be used for torch>=2.1.1. The default is otherwise the manual `"eager"` implementation. #### from_pretrained[[transformers.AutoModelForNextSentencePrediction.from_pretrained]] [Source](https://github.com/huggingface/transformers/blob/v5.6.0/src/transformers/models/auto/auto_factory.py#L253) Instantiate one of the model classes of the library (with a next sentence prediction head) from a pretrained model. The model class to instantiate is selected based on the `model_type` property of the config object (either passed as an argument or loaded from `pretrained_model_name_or_path` if possible), or when it's missing, by falling back to using pattern matching on `pretrained_model_name_or_path`: - **bert** -- [BertForNextSentencePrediction](/docs/transformers/v5.6.0/ja/model_doc/bert#transformers.BertForNextSentencePrediction) (BertConfig model) - **ernie** -- `ErnieForNextSentencePrediction` (ErnieConfig model) - **fnet** -- `FNetForNextSentencePrediction` (FNetConfig model) - **megatron-bert** -- `MegatronBertForNextSentencePrediction` (MegatronBertConfig model) - **mobilebert** -- `MobileBertForNextSentencePrediction` (MobileBertConfig model) The model is set in evaluation mode by default using `model.eval()` (so for instance, dropout modules are deactivated). To train the model, you should first set it back in training mode with `model.train()` Examples: ```python >>> from transformers import AutoConfig, AutoModelForNextSentencePrediction >>> # Download model and configuration from huggingface.co and cache. >>> model = AutoModelForNextSentencePrediction.from_pretrained("google-bert/bert-base-cased") >>> # Update configuration during loading >>> model = AutoModelForNextSentencePrediction.from_pretrained("google-bert/bert-base-cased", output_attentions=True) >>> model.config.output_attentions True ``` **Parameters:** pretrained_model_name_or_path (`str` or `os.PathLike`) : Can be either: - A string, the *model id* of a pretrained model hosted inside a model repo on huggingface.co. - A path to a *directory* containing model weights saved using [save_pretrained()](/docs/transformers/v5.6.0/ja/main_classes/model#transformers.PreTrainedModel.save_pretrained), e.g., `./my_model_directory/`. model_args (additional positional arguments, *optional*) : Will be passed along to the underlying model `__init__()` method. config ([PreTrainedConfig](/docs/transformers/v5.6.0/ja/main_classes/configuration#transformers.PreTrainedConfig), *optional*) : Configuration for the model to use instead of an automatically loaded configuration. Configuration can be automatically loaded when: - The model is a model provided by the library (loaded with the *model id* string of a pretrained model). - The model was saved using [save_pretrained()](/docs/transformers/v5.6.0/ja/main_classes/model#transformers.PreTrainedModel.save_pretrained) and is reloaded by supplying the save directory. - The model is loaded by supplying a local directory as `pretrained_model_name_or_path` and a configuration JSON file named *config.json* is found in the directory. state_dict (*dict[str, torch.Tensor]*, *optional*) : A state dictionary to use instead of a state dictionary loaded from saved weights file. This option can be used if you want to create a model from a pretrained configuration but load your own weights. In this case though, you should check if using [save_pretrained()](/docs/transformers/v5.6.0/ja/main_classes/model#transformers.PreTrainedModel.save_pretrained) and [from_pretrained()](/docs/transformers/v5.6.0/ja/main_classes/model#transformers.PreTrainedModel.from_pretrained) is not a simpler option. cache_dir (`str` or `os.PathLike`, *optional*) : Path to a directory in which a downloaded pretrained model configuration should be cached if the standard cache should not be used. force_download (`bool`, *optional*, defaults to `False`) : Whether or not to force the (re-)download of the model weights and configuration files, overriding the cached versions if they exist. proxies (`dict[str, str]`, *optional*) : A dictionary of proxy servers to use by protocol or endpoint, e.g., `{'http': 'foo.bar:3128', 'http://hostname': 'foo.bar:4012'}`. The proxies are used on each request. output_loading_info(`bool`, *optional*, defaults to `False`) : Whether or not to also return a dictionary containing missing keys, unexpected keys and error messages. local_files_only(`bool`, *optional*, defaults to `False`) : Whether or not to only look at local files (e.g., not try downloading the model). revision (`str`, *optional*, defaults to `"main"`) : The specific model version to use. It can be a branch name, a tag name, or a commit id, since we use a git-based system for storing models and other artifacts on huggingface.co, so `revision` can be any identifier allowed by git. trust_remote_code (`bool`, *optional*, defaults to `False`) : Whether or not to allow for custom models defined on the Hub in their own modeling files. This option should only be set to `True` for repositories you trust and in which you have read the code, as it will execute code present on the Hub on your local machine. code_revision (`str`, *optional*, defaults to `"main"`) : The specific revision to use for the code on the Hub, if the code leaves in a different repository than the rest of the model. It can be a branch name, a tag name, or a commit id, since we use a git-based system for storing models and other artifacts on huggingface.co, so `revision` can be any identifier allowed by git. kwargs (additional keyword arguments, *optional*) : Can be used to update the configuration object (after it being loaded) and initiate the model (e.g., `output_attentions=True`). Behaves differently depending on whether a `config` is provided or automatically loaded: - If a configuration is provided with `config`, `**kwargs` will be directly passed to the underlying model's `__init__` method (we assume all relevant updates to the configuration have already been done) - If a configuration is not provided, `kwargs` will be first passed to the configuration class initialization function ([from_pretrained()](/docs/transformers/v5.6.0/ja/main_classes/configuration#transformers.PreTrainedConfig.from_pretrained)). Each key of `kwargs` that corresponds to a configuration attribute will be used to override said attribute with the supplied `kwargs` value. Remaining keys that do not correspond to any configuration attribute will be passed to the underlying model's `__init__` function. ### AutoModelForTokenClassification[[transformers.AutoModelForTokenClassification]] #### transformers.AutoModelForTokenClassification[[transformers.AutoModelForTokenClassification]] [Source](https://github.com/huggingface/transformers/blob/v5.6.0/src/transformers/models/auto/modeling_auto.py#L2077) This is a generic model class that will be instantiated as one of the model classes of the library (with a token classification head) when created with the [from_pretrained()](/docs/transformers/v5.6.0/ja/model_doc/auto#transformers.AutoModel.from_pretrained) class method or the [from_config()](/docs/transformers/v5.6.0/ja/model_doc/auto#transformers.AutoModel.from_config) class method. This class cannot be instantiated directly using `__init__()` (throws an error). from_configtransformers.AutoModelForTokenClassification.from_confighttps://github.com/huggingface/transformers/blob/v5.6.0/src/transformers/models/auto/auto_factory.py#L206[{"name": "**kwargs", "val": ""}]- **config** ([PreTrainedConfig](/docs/transformers/v5.6.0/ja/main_classes/configuration#transformers.PreTrainedConfig)) -- The model class to instantiate is selected based on the configuration class: - [AlbertConfig](/docs/transformers/v5.6.0/ja/model_doc/albert#transformers.AlbertConfig) configuration class: [AlbertForTokenClassification](/docs/transformers/v5.6.0/ja/model_doc/albert#transformers.AlbertForTokenClassification) (AlbertConfig model) - `ApertusConfig` configuration class: `ApertusForTokenClassification` (ApertusConfig model) - `ArceeConfig` configuration class: `ArceeForTokenClassification` (ArceeConfig model) - [BertConfig](/docs/transformers/v5.6.0/ja/model_doc/bert#transformers.BertConfig) configuration class: [BertForTokenClassification](/docs/transformers/v5.6.0/ja/model_doc/bert#transformers.BertForTokenClassification) (BertConfig model) - [BigBirdConfig](/docs/transformers/v5.6.0/ja/model_doc/big_bird#transformers.BigBirdConfig) configuration class: [BigBirdForTokenClassification](/docs/transformers/v5.6.0/ja/model_doc/big_bird#transformers.BigBirdForTokenClassification) (BigBirdConfig model) - [BioGptConfig](/docs/transformers/v5.6.0/ja/model_doc/biogpt#transformers.BioGptConfig) configuration class: [BioGptForTokenClassification](/docs/transformers/v5.6.0/ja/model_doc/biogpt#transformers.BioGptForTokenClassification) (BioGptConfig model) - [BloomConfig](/docs/transformers/v5.6.0/ja/model_doc/bloom#transformers.BloomConfig) configuration class: [BloomForTokenClassification](/docs/transformers/v5.6.0/ja/model_doc/bloom#transformers.BloomForTokenClassification) (BloomConfig model) - [BrosConfig](/docs/transformers/v5.6.0/ja/model_doc/bros#transformers.BrosConfig) configuration class: [BrosForTokenClassification](/docs/transformers/v5.6.0/ja/model_doc/bros#transformers.BrosForTokenClassification) (BrosConfig model) - [CamembertConfig](/docs/transformers/v5.6.0/ja/model_doc/camembert#transformers.CamembertConfig) configuration class: [CamembertForTokenClassification](/docs/transformers/v5.6.0/ja/model_doc/camembert#transformers.CamembertForTokenClassification) (CamembertConfig model) - [CanineConfig](/docs/transformers/v5.6.0/ja/model_doc/canine#transformers.CanineConfig) configuration class: [CanineForTokenClassification](/docs/transformers/v5.6.0/ja/model_doc/canine#transformers.CanineForTokenClassification) (CanineConfig model) - [ConvBertConfig](/docs/transformers/v5.6.0/ja/model_doc/convbert#transformers.ConvBertConfig) configuration class: [ConvBertForTokenClassification](/docs/transformers/v5.6.0/ja/model_doc/convbert#transformers.ConvBertForTokenClassification) (ConvBertConfig model) - [Data2VecTextConfig](/docs/transformers/v5.6.0/ja/model_doc/data2vec#transformers.Data2VecTextConfig) configuration class: [Data2VecTextForTokenClassification](/docs/transformers/v5.6.0/ja/model_doc/data2vec#transformers.Data2VecTextForTokenClassification) (Data2VecTextConfig model) - [DebertaConfig](/docs/transformers/v5.6.0/ja/model_doc/deberta#transformers.DebertaConfig) configuration class: [DebertaForTokenClassification](/docs/transformers/v5.6.0/ja/model_doc/deberta#transformers.DebertaForTokenClassification) (DebertaConfig model) - [DebertaV2Config](/docs/transformers/v5.6.0/ja/model_doc/deberta-v2#transformers.DebertaV2Config) configuration class: [DebertaV2ForTokenClassification](/docs/transformers/v5.6.0/ja/model_doc/deberta-v2#transformers.DebertaV2ForTokenClassification) (DebertaV2Config model) - `DeepseekV3Config` configuration class: `DeepseekV3ForTokenClassification` (DeepseekV3Config model) - `DiffLlamaConfig` configuration class: `DiffLlamaForTokenClassification` (DiffLlamaConfig model) - `DistilBertConfig` configuration class: `DistilBertForTokenClassification` (DistilBertConfig model) - `ElectraConfig` configuration class: `ElectraForTokenClassification` (ElectraConfig model) - `ErnieConfig` configuration class: `ErnieForTokenClassification` (ErnieConfig model) - `EsmConfig` configuration class: `EsmForTokenClassification` (EsmConfig model) - `EuroBertConfig` configuration class: `EuroBertForTokenClassification` (EuroBertConfig model) - `Exaone4Config` configuration class: `Exaone4ForTokenClassification` (Exaone4Config model) - `FNetConfig` configuration class: `FNetForTokenClassification` (FNetConfig model) - `FalconConfig` configuration class: `FalconForTokenClassification` (FalconConfig model) - `FlaubertConfig` configuration class: `FlaubertForTokenClassification` (FlaubertConfig model) - `FunnelConfig` configuration class: `FunnelForTokenClassification` (FunnelConfig model) - `GPT2Config` configuration class: `GPT2ForTokenClassification` (GPT2Config model) - `GPTBigCodeConfig` configuration class: `GPTBigCodeForTokenClassification` (GPTBigCodeConfig model) - `GPTNeoConfig` configuration class: `GPTNeoForTokenClassification` (GPTNeoConfig model) - `GPTNeoXConfig` configuration class: `GPTNeoXForTokenClassification` (GPTNeoXConfig model) - `Gemma2Config` configuration class: `Gemma2ForTokenClassification` (Gemma2Config model) - `GemmaConfig` configuration class: `GemmaForTokenClassification` (GemmaConfig model) - `Glm4Config` configuration class: `Glm4ForTokenClassification` (Glm4Config model) - `GlmConfig` configuration class: `GlmForTokenClassification` (GlmConfig model) - `GptOssConfig` configuration class: `GptOssForTokenClassification` (GptOssConfig model) - `HeliumConfig` configuration class: `HeliumForTokenClassification` (HeliumConfig model) - `IBertConfig` configuration class: `IBertForTokenClassification` (IBertConfig model) - `JinaEmbeddingsV3Config` configuration class: `JinaEmbeddingsV3ForTokenClassification` (JinaEmbeddingsV3Config model) - `LayoutLMConfig` configuration class: `LayoutLMForTokenClassification` (LayoutLMConfig model) - `LayoutLMv2Config` configuration class: `LayoutLMv2ForTokenClassification` (LayoutLMv2Config model) - `LayoutLMv3Config` configuration class: `LayoutLMv3ForTokenClassification` (LayoutLMv3Config model) - `LiltConfig` configuration class: `LiltForTokenClassification` (LiltConfig model) - `LlamaConfig` configuration class: `LlamaForTokenClassification` (LlamaConfig model) - `LongformerConfig` configuration class: `LongformerForTokenClassification` (LongformerConfig model) - `LukeConfig` configuration class: `LukeForTokenClassification` (LukeConfig model) - `MPNetConfig` configuration class: `MPNetForTokenClassification` (MPNetConfig model) - `MT5Config` configuration class: `MT5ForTokenClassification` (MT5Config model) - `MarkupLMConfig` configuration class: `MarkupLMForTokenClassification` (MarkupLMConfig model) - `MegatronBertConfig` configuration class: `MegatronBertForTokenClassification` (MegatronBertConfig model) - `MiniMaxConfig` configuration class: `MiniMaxForTokenClassification` (MiniMaxConfig model) - `Ministral3Config` configuration class: `Ministral3ForTokenClassification` (Ministral3Config model) - `MinistralConfig` configuration class: `MinistralForTokenClassification` (MinistralConfig model) - `Mistral4Config` configuration class: `Mistral4ForTokenClassification` (Mistral4Config model) - `MistralConfig` configuration class: `MistralForTokenClassification` (MistralConfig model) - `MixtralConfig` configuration class: `MixtralForTokenClassification` (MixtralConfig model) - `MobileBertConfig` configuration class: `MobileBertForTokenClassification` (MobileBertConfig model) - `ModernBertConfig` configuration class: `ModernBertForTokenClassification` (ModernBertConfig model) - `ModernVBertConfig` configuration class: `ModernVBertForTokenClassification` (ModernVBertConfig model) - `MptConfig` configuration class: `MptForTokenClassification` (MptConfig model) - `MraConfig` configuration class: `MraForTokenClassification` (MraConfig model) - `NemotronConfig` configuration class: `NemotronForTokenClassification` (NemotronConfig model) - `NomicBertConfig` configuration class: `NomicBertForTokenClassification` (NomicBertConfig model) - `NystromformerConfig` configuration class: `NystromformerForTokenClassification` (NystromformerConfig model) - `OpenAIPrivacyFilterConfig` configuration class: `OpenAIPrivacyFilterForTokenClassification` (OpenAIPrivacyFilterConfig model) - `PersimmonConfig` configuration class: `PersimmonForTokenClassification` (PersimmonConfig model) - `Phi3Config` configuration class: `Phi3ForTokenClassification` (Phi3Config model) - `PhiConfig` configuration class: `PhiForTokenClassification` (PhiConfig model) - `Qwen2Config` configuration class: `Qwen2ForTokenClassification` (Qwen2Config model) - `Qwen2MoeConfig` configuration class: `Qwen2MoeForTokenClassification` (Qwen2MoeConfig model) - `Qwen3Config` configuration class: `Qwen3ForTokenClassification` (Qwen3Config model) - `Qwen3MoeConfig` configuration class: `Qwen3MoeForTokenClassification` (Qwen3MoeConfig model) - `Qwen3NextConfig` configuration class: `Qwen3NextForTokenClassification` (Qwen3NextConfig model) - `RemBertConfig` configuration class: `RemBertForTokenClassification` (RemBertConfig model) - `RoCBertConfig` configuration class: `RoCBertForTokenClassification` (RoCBertConfig model) - `RoFormerConfig` configuration class: `RoFormerForTokenClassification` (RoFormerConfig model) - `RobertaConfig` configuration class: `RobertaForTokenClassification` (RobertaConfig model) - `RobertaPreLayerNormConfig` configuration class: `RobertaPreLayerNormForTokenClassification` (RobertaPreLayerNormConfig model) - `SeedOssConfig` configuration class: `SeedOssForTokenClassification` (SeedOssConfig model) - `SmolLM3Config` configuration class: `SmolLM3ForTokenClassification` (SmolLM3Config model) - `SqueezeBertConfig` configuration class: `SqueezeBertForTokenClassification` (SqueezeBertConfig model) - `StableLmConfig` configuration class: `StableLmForTokenClassification` (StableLmConfig model) - `Starcoder2Config` configuration class: `Starcoder2ForTokenClassification` (Starcoder2Config model) - `T5Config` configuration class: `T5ForTokenClassification` (T5Config model) - `T5Gemma2Config` configuration class: `T5Gemma2ForTokenClassification` (T5Gemma2Config model) - `T5GemmaConfig` configuration class: `T5GemmaForTokenClassification` (T5GemmaConfig model) - `UMT5Config` configuration class: `UMT5ForTokenClassification` (UMT5Config model) - `XLMConfig` configuration class: `XLMForTokenClassification` (XLMConfig model) - `XLMRobertaConfig` configuration class: `XLMRobertaForTokenClassification` (XLMRobertaConfig model) - `XLMRobertaXLConfig` configuration class: `XLMRobertaXLForTokenClassification` (XLMRobertaXLConfig model) - `XLNetConfig` configuration class: `XLNetForTokenClassification` (XLNetConfig model) - `XmodConfig` configuration class: `XmodForTokenClassification` (XmodConfig model) - `YosoConfig` configuration class: `YosoForTokenClassification` (YosoConfig model) - **attn_implementation** (`str`, *optional*) -- The attention implementation to use in the model (if relevant). Can be any of `"eager"` (manual implementation of the attention), `"sdpa"` (using [`F.scaled_dot_product_attention`](https://pytorch.org/docs/master/generated/torch.nn.functional.scaled_dot_product_attention.html)), `"flash_attention_2"` (using [Dao-AILab/flash-attention](https://github.com/Dao-AILab/flash-attention)), or `"flash_attention_3"` (using [Dao-AILab/flash-attention/hopper](https://github.com/Dao-AILab/flash-attention/tree/main/hopper)). By default, if available, SDPA will be used for torch>=2.1.1. The default is otherwise the manual `"eager"` implementation.0 Instantiates one of the model classes of the library (with a token classification head) from a configuration. Note: Loading a model from its configuration file does **not** load the model weights. It only affects the model's configuration. Use [from_pretrained()](/docs/transformers/v5.6.0/ja/model_doc/auto#transformers.AutoModel.from_pretrained) to load the model weights. Examples: ```python >>> from transformers import AutoConfig, AutoModelForTokenClassification >>> # Download configuration from huggingface.co and cache. >>> config = AutoConfig.from_pretrained("google-bert/bert-base-cased") >>> model = AutoModelForTokenClassification.from_config(config) ``` **Parameters:** config ([PreTrainedConfig](/docs/transformers/v5.6.0/ja/main_classes/configuration#transformers.PreTrainedConfig)) : The model class to instantiate is selected based on the configuration class: - [AlbertConfig](/docs/transformers/v5.6.0/ja/model_doc/albert#transformers.AlbertConfig) configuration class: [AlbertForTokenClassification](/docs/transformers/v5.6.0/ja/model_doc/albert#transformers.AlbertForTokenClassification) (AlbertConfig model) - `ApertusConfig` configuration class: `ApertusForTokenClassification` (ApertusConfig model) - `ArceeConfig` configuration class: `ArceeForTokenClassification` (ArceeConfig model) - [BertConfig](/docs/transformers/v5.6.0/ja/model_doc/bert#transformers.BertConfig) configuration class: [BertForTokenClassification](/docs/transformers/v5.6.0/ja/model_doc/bert#transformers.BertForTokenClassification) (BertConfig model) - [BigBirdConfig](/docs/transformers/v5.6.0/ja/model_doc/big_bird#transformers.BigBirdConfig) configuration class: [BigBirdForTokenClassification](/docs/transformers/v5.6.0/ja/model_doc/big_bird#transformers.BigBirdForTokenClassification) (BigBirdConfig model) - [BioGptConfig](/docs/transformers/v5.6.0/ja/model_doc/biogpt#transformers.BioGptConfig) configuration class: [BioGptForTokenClassification](/docs/transformers/v5.6.0/ja/model_doc/biogpt#transformers.BioGptForTokenClassification) (BioGptConfig model) - [BloomConfig](/docs/transformers/v5.6.0/ja/model_doc/bloom#transformers.BloomConfig) configuration class: [BloomForTokenClassification](/docs/transformers/v5.6.0/ja/model_doc/bloom#transformers.BloomForTokenClassification) (BloomConfig model) - [BrosConfig](/docs/transformers/v5.6.0/ja/model_doc/bros#transformers.BrosConfig) configuration class: [BrosForTokenClassification](/docs/transformers/v5.6.0/ja/model_doc/bros#transformers.BrosForTokenClassification) (BrosConfig model) - [CamembertConfig](/docs/transformers/v5.6.0/ja/model_doc/camembert#transformers.CamembertConfig) configuration class: [CamembertForTokenClassification](/docs/transformers/v5.6.0/ja/model_doc/camembert#transformers.CamembertForTokenClassification) (CamembertConfig model) - [CanineConfig](/docs/transformers/v5.6.0/ja/model_doc/canine#transformers.CanineConfig) configuration class: [CanineForTokenClassification](/docs/transformers/v5.6.0/ja/model_doc/canine#transformers.CanineForTokenClassification) (CanineConfig model) - [ConvBertConfig](/docs/transformers/v5.6.0/ja/model_doc/convbert#transformers.ConvBertConfig) configuration class: [ConvBertForTokenClassification](/docs/transformers/v5.6.0/ja/model_doc/convbert#transformers.ConvBertForTokenClassification) (ConvBertConfig model) - [Data2VecTextConfig](/docs/transformers/v5.6.0/ja/model_doc/data2vec#transformers.Data2VecTextConfig) configuration class: [Data2VecTextForTokenClassification](/docs/transformers/v5.6.0/ja/model_doc/data2vec#transformers.Data2VecTextForTokenClassification) (Data2VecTextConfig model) - [DebertaConfig](/docs/transformers/v5.6.0/ja/model_doc/deberta#transformers.DebertaConfig) configuration class: [DebertaForTokenClassification](/docs/transformers/v5.6.0/ja/model_doc/deberta#transformers.DebertaForTokenClassification) (DebertaConfig model) - [DebertaV2Config](/docs/transformers/v5.6.0/ja/model_doc/deberta-v2#transformers.DebertaV2Config) configuration class: [DebertaV2ForTokenClassification](/docs/transformers/v5.6.0/ja/model_doc/deberta-v2#transformers.DebertaV2ForTokenClassification) (DebertaV2Config model) - `DeepseekV3Config` configuration class: `DeepseekV3ForTokenClassification` (DeepseekV3Config model) - `DiffLlamaConfig` configuration class: `DiffLlamaForTokenClassification` (DiffLlamaConfig model) - `DistilBertConfig` configuration class: `DistilBertForTokenClassification` (DistilBertConfig model) - `ElectraConfig` configuration class: `ElectraForTokenClassification` (ElectraConfig model) - `ErnieConfig` configuration class: `ErnieForTokenClassification` (ErnieConfig model) - `EsmConfig` configuration class: `EsmForTokenClassification` (EsmConfig model) - `EuroBertConfig` configuration class: `EuroBertForTokenClassification` (EuroBertConfig model) - `Exaone4Config` configuration class: `Exaone4ForTokenClassification` (Exaone4Config model) - `FNetConfig` configuration class: `FNetForTokenClassification` (FNetConfig model) - `FalconConfig` configuration class: `FalconForTokenClassification` (FalconConfig model) - `FlaubertConfig` configuration class: `FlaubertForTokenClassification` (FlaubertConfig model) - `FunnelConfig` configuration class: `FunnelForTokenClassification` (FunnelConfig model) - `GPT2Config` configuration class: `GPT2ForTokenClassification` (GPT2Config model) - `GPTBigCodeConfig` configuration class: `GPTBigCodeForTokenClassification` (GPTBigCodeConfig model) - `GPTNeoConfig` configuration class: `GPTNeoForTokenClassification` (GPTNeoConfig model) - `GPTNeoXConfig` configuration class: `GPTNeoXForTokenClassification` (GPTNeoXConfig model) - `Gemma2Config` configuration class: `Gemma2ForTokenClassification` (Gemma2Config model) - `GemmaConfig` configuration class: `GemmaForTokenClassification` (GemmaConfig model) - `Glm4Config` configuration class: `Glm4ForTokenClassification` (Glm4Config model) - `GlmConfig` configuration class: `GlmForTokenClassification` (GlmConfig model) - `GptOssConfig` configuration class: `GptOssForTokenClassification` (GptOssConfig model) - `HeliumConfig` configuration class: `HeliumForTokenClassification` (HeliumConfig model) - `IBertConfig` configuration class: `IBertForTokenClassification` (IBertConfig model) - `JinaEmbeddingsV3Config` configuration class: `JinaEmbeddingsV3ForTokenClassification` (JinaEmbeddingsV3Config model) - `LayoutLMConfig` configuration class: `LayoutLMForTokenClassification` (LayoutLMConfig model) - `LayoutLMv2Config` configuration class: `LayoutLMv2ForTokenClassification` (LayoutLMv2Config model) - `LayoutLMv3Config` configuration class: `LayoutLMv3ForTokenClassification` (LayoutLMv3Config model) - `LiltConfig` configuration class: `LiltForTokenClassification` (LiltConfig model) - `LlamaConfig` configuration class: `LlamaForTokenClassification` (LlamaConfig model) - `LongformerConfig` configuration class: `LongformerForTokenClassification` (LongformerConfig model) - `LukeConfig` configuration class: `LukeForTokenClassification` (LukeConfig model) - `MPNetConfig` configuration class: `MPNetForTokenClassification` (MPNetConfig model) - `MT5Config` configuration class: `MT5ForTokenClassification` (MT5Config model) - `MarkupLMConfig` configuration class: `MarkupLMForTokenClassification` (MarkupLMConfig model) - `MegatronBertConfig` configuration class: `MegatronBertForTokenClassification` (MegatronBertConfig model) - `MiniMaxConfig` configuration class: `MiniMaxForTokenClassification` (MiniMaxConfig model) - `Ministral3Config` configuration class: `Ministral3ForTokenClassification` (Ministral3Config model) - `MinistralConfig` configuration class: `MinistralForTokenClassification` (MinistralConfig model) - `Mistral4Config` configuration class: `Mistral4ForTokenClassification` (Mistral4Config model) - `MistralConfig` configuration class: `MistralForTokenClassification` (MistralConfig model) - `MixtralConfig` configuration class: `MixtralForTokenClassification` (MixtralConfig model) - `MobileBertConfig` configuration class: `MobileBertForTokenClassification` (MobileBertConfig model) - `ModernBertConfig` configuration class: `ModernBertForTokenClassification` (ModernBertConfig model) - `ModernVBertConfig` configuration class: `ModernVBertForTokenClassification` (ModernVBertConfig model) - `MptConfig` configuration class: `MptForTokenClassification` (MptConfig model) - `MraConfig` configuration class: `MraForTokenClassification` (MraConfig model) - `NemotronConfig` configuration class: `NemotronForTokenClassification` (NemotronConfig model) - `NomicBertConfig` configuration class: `NomicBertForTokenClassification` (NomicBertConfig model) - `NystromformerConfig` configuration class: `NystromformerForTokenClassification` (NystromformerConfig model) - `OpenAIPrivacyFilterConfig` configuration class: `OpenAIPrivacyFilterForTokenClassification` (OpenAIPrivacyFilterConfig model) - `PersimmonConfig` configuration class: `PersimmonForTokenClassification` (PersimmonConfig model) - `Phi3Config` configuration class: `Phi3ForTokenClassification` (Phi3Config model) - `PhiConfig` configuration class: `PhiForTokenClassification` (PhiConfig model) - `Qwen2Config` configuration class: `Qwen2ForTokenClassification` (Qwen2Config model) - `Qwen2MoeConfig` configuration class: `Qwen2MoeForTokenClassification` (Qwen2MoeConfig model) - `Qwen3Config` configuration class: `Qwen3ForTokenClassification` (Qwen3Config model) - `Qwen3MoeConfig` configuration class: `Qwen3MoeForTokenClassification` (Qwen3MoeConfig model) - `Qwen3NextConfig` configuration class: `Qwen3NextForTokenClassification` (Qwen3NextConfig model) - `RemBertConfig` configuration class: `RemBertForTokenClassification` (RemBertConfig model) - `RoCBertConfig` configuration class: `RoCBertForTokenClassification` (RoCBertConfig model) - `RoFormerConfig` configuration class: `RoFormerForTokenClassification` (RoFormerConfig model) - `RobertaConfig` configuration class: `RobertaForTokenClassification` (RobertaConfig model) - `RobertaPreLayerNormConfig` configuration class: `RobertaPreLayerNormForTokenClassification` (RobertaPreLayerNormConfig model) - `SeedOssConfig` configuration class: `SeedOssForTokenClassification` (SeedOssConfig model) - `SmolLM3Config` configuration class: `SmolLM3ForTokenClassification` (SmolLM3Config model) - `SqueezeBertConfig` configuration class: `SqueezeBertForTokenClassification` (SqueezeBertConfig model) - `StableLmConfig` configuration class: `StableLmForTokenClassification` (StableLmConfig model) - `Starcoder2Config` configuration class: `Starcoder2ForTokenClassification` (Starcoder2Config model) - `T5Config` configuration class: `T5ForTokenClassification` (T5Config model) - `T5Gemma2Config` configuration class: `T5Gemma2ForTokenClassification` (T5Gemma2Config model) - `T5GemmaConfig` configuration class: `T5GemmaForTokenClassification` (T5GemmaConfig model) - `UMT5Config` configuration class: `UMT5ForTokenClassification` (UMT5Config model) - `XLMConfig` configuration class: `XLMForTokenClassification` (XLMConfig model) - `XLMRobertaConfig` configuration class: `XLMRobertaForTokenClassification` (XLMRobertaConfig model) - `XLMRobertaXLConfig` configuration class: `XLMRobertaXLForTokenClassification` (XLMRobertaXLConfig model) - `XLNetConfig` configuration class: `XLNetForTokenClassification` (XLNetConfig model) - `XmodConfig` configuration class: `XmodForTokenClassification` (XmodConfig model) - `YosoConfig` configuration class: `YosoForTokenClassification` (YosoConfig model) attn_implementation (`str`, *optional*) : The attention implementation to use in the model (if relevant). Can be any of `"eager"` (manual implementation of the attention), `"sdpa"` (using [`F.scaled_dot_product_attention`](https://pytorch.org/docs/master/generated/torch.nn.functional.scaled_dot_product_attention.html)), `"flash_attention_2"` (using [Dao-AILab/flash-attention](https://github.com/Dao-AILab/flash-attention)), or `"flash_attention_3"` (using [Dao-AILab/flash-attention/hopper](https://github.com/Dao-AILab/flash-attention/tree/main/hopper)). By default, if available, SDPA will be used for torch>=2.1.1. The default is otherwise the manual `"eager"` implementation. #### from_pretrained[[transformers.AutoModelForTokenClassification.from_pretrained]] [Source](https://github.com/huggingface/transformers/blob/v5.6.0/src/transformers/models/auto/auto_factory.py#L253) Instantiate one of the model classes of the library (with a token classification head) from a pretrained model. The model class to instantiate is selected based on the `model_type` property of the config object (either passed as an argument or loaded from `pretrained_model_name_or_path` if possible), or when it's missing, by falling back to using pattern matching on `pretrained_model_name_or_path`: - **albert** -- [AlbertForTokenClassification](/docs/transformers/v5.6.0/ja/model_doc/albert#transformers.AlbertForTokenClassification) (AlbertConfig model) - **apertus** -- `ApertusForTokenClassification` (ApertusConfig model) - **arcee** -- `ArceeForTokenClassification` (ArceeConfig model) - **bert** -- [BertForTokenClassification](/docs/transformers/v5.6.0/ja/model_doc/bert#transformers.BertForTokenClassification) (BertConfig model) - **big_bird** -- [BigBirdForTokenClassification](/docs/transformers/v5.6.0/ja/model_doc/big_bird#transformers.BigBirdForTokenClassification) (BigBirdConfig model) - **biogpt** -- [BioGptForTokenClassification](/docs/transformers/v5.6.0/ja/model_doc/biogpt#transformers.BioGptForTokenClassification) (BioGptConfig model) - **bloom** -- [BloomForTokenClassification](/docs/transformers/v5.6.0/ja/model_doc/bloom#transformers.BloomForTokenClassification) (BloomConfig model) - **bros** -- [BrosForTokenClassification](/docs/transformers/v5.6.0/ja/model_doc/bros#transformers.BrosForTokenClassification) (BrosConfig model) - **camembert** -- [CamembertForTokenClassification](/docs/transformers/v5.6.0/ja/model_doc/camembert#transformers.CamembertForTokenClassification) (CamembertConfig model) - **canine** -- [CanineForTokenClassification](/docs/transformers/v5.6.0/ja/model_doc/canine#transformers.CanineForTokenClassification) (CanineConfig model) - **convbert** -- [ConvBertForTokenClassification](/docs/transformers/v5.6.0/ja/model_doc/convbert#transformers.ConvBertForTokenClassification) (ConvBertConfig model) - **data2vec-text** -- [Data2VecTextForTokenClassification](/docs/transformers/v5.6.0/ja/model_doc/data2vec#transformers.Data2VecTextForTokenClassification) (Data2VecTextConfig model) - **deberta** -- [DebertaForTokenClassification](/docs/transformers/v5.6.0/ja/model_doc/deberta#transformers.DebertaForTokenClassification) (DebertaConfig model) - **deberta-v2** -- [DebertaV2ForTokenClassification](/docs/transformers/v5.6.0/ja/model_doc/deberta-v2#transformers.DebertaV2ForTokenClassification) (DebertaV2Config model) - **deepseek_v3** -- `DeepseekV3ForTokenClassification` (DeepseekV3Config model) - **diffllama** -- `DiffLlamaForTokenClassification` (DiffLlamaConfig model) - **distilbert** -- `DistilBertForTokenClassification` (DistilBertConfig model) - **electra** -- `ElectraForTokenClassification` (ElectraConfig model) - **ernie** -- `ErnieForTokenClassification` (ErnieConfig model) - **esm** -- `EsmForTokenClassification` (EsmConfig model) - **eurobert** -- `EuroBertForTokenClassification` (EuroBertConfig model) - **exaone4** -- `Exaone4ForTokenClassification` (Exaone4Config model) - **falcon** -- `FalconForTokenClassification` (FalconConfig model) - **flaubert** -- `FlaubertForTokenClassification` (FlaubertConfig model) - **fnet** -- `FNetForTokenClassification` (FNetConfig model) - **funnel** -- `FunnelForTokenClassification` (FunnelConfig model) - **gemma** -- `GemmaForTokenClassification` (GemmaConfig model) - **gemma2** -- `Gemma2ForTokenClassification` (Gemma2Config model) - **glm** -- `GlmForTokenClassification` (GlmConfig model) - **glm4** -- `Glm4ForTokenClassification` (Glm4Config model) - **gpt-sw3** -- `GPT2ForTokenClassification` (GPT2Config model) - **gpt2** -- `GPT2ForTokenClassification` (GPT2Config model) - **gpt_bigcode** -- `GPTBigCodeForTokenClassification` (GPTBigCodeConfig model) - **gpt_neo** -- `GPTNeoForTokenClassification` (GPTNeoConfig model) - **gpt_neox** -- `GPTNeoXForTokenClassification` (GPTNeoXConfig model) - **gpt_oss** -- `GptOssForTokenClassification` (GptOssConfig model) - **helium** -- `HeliumForTokenClassification` (HeliumConfig model) - **ibert** -- `IBertForTokenClassification` (IBertConfig model) - **jina_embeddings_v3** -- `JinaEmbeddingsV3ForTokenClassification` (JinaEmbeddingsV3Config model) - **layoutlm** -- `LayoutLMForTokenClassification` (LayoutLMConfig model) - **layoutlmv2** -- `LayoutLMv2ForTokenClassification` (LayoutLMv2Config model) - **layoutlmv3** -- `LayoutLMv3ForTokenClassification` (LayoutLMv3Config model) - **lilt** -- `LiltForTokenClassification` (LiltConfig model) - **llama** -- `LlamaForTokenClassification` (LlamaConfig model) - **longformer** -- `LongformerForTokenClassification` (LongformerConfig model) - **luke** -- `LukeForTokenClassification` (LukeConfig model) - **markuplm** -- `MarkupLMForTokenClassification` (MarkupLMConfig model) - **megatron-bert** -- `MegatronBertForTokenClassification` (MegatronBertConfig model) - **minimax** -- `MiniMaxForTokenClassification` (MiniMaxConfig model) - **ministral** -- `MinistralForTokenClassification` (MinistralConfig model) - **ministral3** -- `Ministral3ForTokenClassification` (Ministral3Config model) - **mistral** -- `MistralForTokenClassification` (MistralConfig model) - **mistral4** -- `Mistral4ForTokenClassification` (Mistral4Config model) - **mixtral** -- `MixtralForTokenClassification` (MixtralConfig model) - **mobilebert** -- `MobileBertForTokenClassification` (MobileBertConfig model) - **modernbert** -- `ModernBertForTokenClassification` (ModernBertConfig model) - **modernvbert** -- `ModernVBertForTokenClassification` (ModernVBertConfig model) - **mpnet** -- `MPNetForTokenClassification` (MPNetConfig model) - **mpt** -- `MptForTokenClassification` (MptConfig model) - **mra** -- `MraForTokenClassification` (MraConfig model) - **mt5** -- `MT5ForTokenClassification` (MT5Config model) - **nemotron** -- `NemotronForTokenClassification` (NemotronConfig model) - **nomic_bert** -- `NomicBertForTokenClassification` (NomicBertConfig model) - **nystromformer** -- `NystromformerForTokenClassification` (NystromformerConfig model) - **openai_privacy_filter** -- `OpenAIPrivacyFilterForTokenClassification` (OpenAIPrivacyFilterConfig model) - **persimmon** -- `PersimmonForTokenClassification` (PersimmonConfig model) - **phi** -- `PhiForTokenClassification` (PhiConfig model) - **phi3** -- `Phi3ForTokenClassification` (Phi3Config model) - **qwen2** -- `Qwen2ForTokenClassification` (Qwen2Config model) - **qwen2_moe** -- `Qwen2MoeForTokenClassification` (Qwen2MoeConfig model) - **qwen3** -- `Qwen3ForTokenClassification` (Qwen3Config model) - **qwen3_moe** -- `Qwen3MoeForTokenClassification` (Qwen3MoeConfig model) - **qwen3_next** -- `Qwen3NextForTokenClassification` (Qwen3NextConfig model) - **rembert** -- `RemBertForTokenClassification` (RemBertConfig model) - **roberta** -- `RobertaForTokenClassification` (RobertaConfig model) - **roberta-prelayernorm** -- `RobertaPreLayerNormForTokenClassification` (RobertaPreLayerNormConfig model) - **roc_bert** -- `RoCBertForTokenClassification` (RoCBertConfig model) - **roformer** -- `RoFormerForTokenClassification` (RoFormerConfig model) - **seed_oss** -- `SeedOssForTokenClassification` (SeedOssConfig model) - **smollm3** -- `SmolLM3ForTokenClassification` (SmolLM3Config model) - **squeezebert** -- `SqueezeBertForTokenClassification` (SqueezeBertConfig model) - **stablelm** -- `StableLmForTokenClassification` (StableLmConfig model) - **starcoder2** -- `Starcoder2ForTokenClassification` (Starcoder2Config model) - **t5** -- `T5ForTokenClassification` (T5Config model) - **t5gemma** -- `T5GemmaForTokenClassification` (T5GemmaConfig model) - **t5gemma2** -- `T5Gemma2ForTokenClassification` (T5Gemma2Config model) - **umt5** -- `UMT5ForTokenClassification` (UMT5Config model) - **xlm** -- `XLMForTokenClassification` (XLMConfig model) - **xlm-roberta** -- `XLMRobertaForTokenClassification` (XLMRobertaConfig model) - **xlm-roberta-xl** -- `XLMRobertaXLForTokenClassification` (XLMRobertaXLConfig model) - **xlnet** -- `XLNetForTokenClassification` (XLNetConfig model) - **xmod** -- `XmodForTokenClassification` (XmodConfig model) - **yoso** -- `YosoForTokenClassification` (YosoConfig model) The model is set in evaluation mode by default using `model.eval()` (so for instance, dropout modules are deactivated). To train the model, you should first set it back in training mode with `model.train()` Examples: ```python >>> from transformers import AutoConfig, AutoModelForTokenClassification >>> # Download model and configuration from huggingface.co and cache. >>> model = AutoModelForTokenClassification.from_pretrained("google-bert/bert-base-cased") >>> # Update configuration during loading >>> model = AutoModelForTokenClassification.from_pretrained("google-bert/bert-base-cased", output_attentions=True) >>> model.config.output_attentions True ``` **Parameters:** pretrained_model_name_or_path (`str` or `os.PathLike`) : Can be either: - A string, the *model id* of a pretrained model hosted inside a model repo on huggingface.co. - A path to a *directory* containing model weights saved using [save_pretrained()](/docs/transformers/v5.6.0/ja/main_classes/model#transformers.PreTrainedModel.save_pretrained), e.g., `./my_model_directory/`. model_args (additional positional arguments, *optional*) : Will be passed along to the underlying model `__init__()` method. config ([PreTrainedConfig](/docs/transformers/v5.6.0/ja/main_classes/configuration#transformers.PreTrainedConfig), *optional*) : Configuration for the model to use instead of an automatically loaded configuration. Configuration can be automatically loaded when: - The model is a model provided by the library (loaded with the *model id* string of a pretrained model). - The model was saved using [save_pretrained()](/docs/transformers/v5.6.0/ja/main_classes/model#transformers.PreTrainedModel.save_pretrained) and is reloaded by supplying the save directory. - The model is loaded by supplying a local directory as `pretrained_model_name_or_path` and a configuration JSON file named *config.json* is found in the directory. state_dict (*dict[str, torch.Tensor]*, *optional*) : A state dictionary to use instead of a state dictionary loaded from saved weights file. This option can be used if you want to create a model from a pretrained configuration but load your own weights. In this case though, you should check if using [save_pretrained()](/docs/transformers/v5.6.0/ja/main_classes/model#transformers.PreTrainedModel.save_pretrained) and [from_pretrained()](/docs/transformers/v5.6.0/ja/main_classes/model#transformers.PreTrainedModel.from_pretrained) is not a simpler option. cache_dir (`str` or `os.PathLike`, *optional*) : Path to a directory in which a downloaded pretrained model configuration should be cached if the standard cache should not be used. force_download (`bool`, *optional*, defaults to `False`) : Whether or not to force the (re-)download of the model weights and configuration files, overriding the cached versions if they exist. proxies (`dict[str, str]`, *optional*) : A dictionary of proxy servers to use by protocol or endpoint, e.g., `{'http': 'foo.bar:3128', 'http://hostname': 'foo.bar:4012'}`. The proxies are used on each request. output_loading_info(`bool`, *optional*, defaults to `False`) : Whether or not to also return a dictionary containing missing keys, unexpected keys and error messages. local_files_only(`bool`, *optional*, defaults to `False`) : Whether or not to only look at local files (e.g., not try downloading the model). revision (`str`, *optional*, defaults to `"main"`) : The specific model version to use. It can be a branch name, a tag name, or a commit id, since we use a git-based system for storing models and other artifacts on huggingface.co, so `revision` can be any identifier allowed by git. trust_remote_code (`bool`, *optional*, defaults to `False`) : Whether or not to allow for custom models defined on the Hub in their own modeling files. This option should only be set to `True` for repositories you trust and in which you have read the code, as it will execute code present on the Hub on your local machine. code_revision (`str`, *optional*, defaults to `"main"`) : The specific revision to use for the code on the Hub, if the code leaves in a different repository than the rest of the model. It can be a branch name, a tag name, or a commit id, since we use a git-based system for storing models and other artifacts on huggingface.co, so `revision` can be any identifier allowed by git. kwargs (additional keyword arguments, *optional*) : Can be used to update the configuration object (after it being loaded) and initiate the model (e.g., `output_attentions=True`). Behaves differently depending on whether a `config` is provided or automatically loaded: - If a configuration is provided with `config`, `**kwargs` will be directly passed to the underlying model's `__init__` method (we assume all relevant updates to the configuration have already been done) - If a configuration is not provided, `kwargs` will be first passed to the configuration class initialization function ([from_pretrained()](/docs/transformers/v5.6.0/ja/main_classes/configuration#transformers.PreTrainedConfig.from_pretrained)). Each key of `kwargs` that corresponds to a configuration attribute will be used to override said attribute with the supplied `kwargs` value. Remaining keys that do not correspond to any configuration attribute will be passed to the underlying model's `__init__` function. ### AutoModelForQuestionAnswering[[transformers.AutoModelForQuestionAnswering]] #### transformers.AutoModelForQuestionAnswering[[transformers.AutoModelForQuestionAnswering]] [Source](https://github.com/huggingface/transformers/blob/v5.6.0/src/transformers/models/auto/modeling_auto.py#L2037) This is a generic model class that will be instantiated as one of the model classes of the library (with a question answering head) when created with the [from_pretrained()](/docs/transformers/v5.6.0/ja/model_doc/auto#transformers.AutoModel.from_pretrained) class method or the [from_config()](/docs/transformers/v5.6.0/ja/model_doc/auto#transformers.AutoModel.from_config) class method. This class cannot be instantiated directly using `__init__()` (throws an error). from_configtransformers.AutoModelForQuestionAnswering.from_confighttps://github.com/huggingface/transformers/blob/v5.6.0/src/transformers/models/auto/auto_factory.py#L206[{"name": "**kwargs", "val": ""}]- **config** ([PreTrainedConfig](/docs/transformers/v5.6.0/ja/main_classes/configuration#transformers.PreTrainedConfig)) -- The model class to instantiate is selected based on the configuration class: - [AlbertConfig](/docs/transformers/v5.6.0/ja/model_doc/albert#transformers.AlbertConfig) configuration class: [AlbertForQuestionAnswering](/docs/transformers/v5.6.0/ja/model_doc/albert#transformers.AlbertForQuestionAnswering) (AlbertConfig model) - `ArceeConfig` configuration class: `ArceeForQuestionAnswering` (ArceeConfig model) - [BartConfig](/docs/transformers/v5.6.0/ja/model_doc/bart#transformers.BartConfig) configuration class: [BartForQuestionAnswering](/docs/transformers/v5.6.0/ja/model_doc/bart#transformers.BartForQuestionAnswering) (BartConfig model) - [BertConfig](/docs/transformers/v5.6.0/ja/model_doc/bert#transformers.BertConfig) configuration class: [BertForQuestionAnswering](/docs/transformers/v5.6.0/ja/model_doc/bert#transformers.BertForQuestionAnswering) (BertConfig model) - [BigBirdConfig](/docs/transformers/v5.6.0/ja/model_doc/big_bird#transformers.BigBirdConfig) configuration class: [BigBirdForQuestionAnswering](/docs/transformers/v5.6.0/ja/model_doc/big_bird#transformers.BigBirdForQuestionAnswering) (BigBirdConfig model) - [BigBirdPegasusConfig](/docs/transformers/v5.6.0/ja/model_doc/bigbird_pegasus#transformers.BigBirdPegasusConfig) configuration class: [BigBirdPegasusForQuestionAnswering](/docs/transformers/v5.6.0/ja/model_doc/bigbird_pegasus#transformers.BigBirdPegasusForQuestionAnswering) (BigBirdPegasusConfig model) - [BloomConfig](/docs/transformers/v5.6.0/ja/model_doc/bloom#transformers.BloomConfig) configuration class: [BloomForQuestionAnswering](/docs/transformers/v5.6.0/ja/model_doc/bloom#transformers.BloomForQuestionAnswering) (BloomConfig model) - [CamembertConfig](/docs/transformers/v5.6.0/ja/model_doc/camembert#transformers.CamembertConfig) configuration class: [CamembertForQuestionAnswering](/docs/transformers/v5.6.0/ja/model_doc/camembert#transformers.CamembertForQuestionAnswering) (CamembertConfig model) - [CanineConfig](/docs/transformers/v5.6.0/ja/model_doc/canine#transformers.CanineConfig) configuration class: [CanineForQuestionAnswering](/docs/transformers/v5.6.0/ja/model_doc/canine#transformers.CanineForQuestionAnswering) (CanineConfig model) - [ConvBertConfig](/docs/transformers/v5.6.0/ja/model_doc/convbert#transformers.ConvBertConfig) configuration class: [ConvBertForQuestionAnswering](/docs/transformers/v5.6.0/ja/model_doc/convbert#transformers.ConvBertForQuestionAnswering) (ConvBertConfig model) - [Data2VecTextConfig](/docs/transformers/v5.6.0/ja/model_doc/data2vec#transformers.Data2VecTextConfig) configuration class: [Data2VecTextForQuestionAnswering](/docs/transformers/v5.6.0/ja/model_doc/data2vec#transformers.Data2VecTextForQuestionAnswering) (Data2VecTextConfig model) - [DebertaConfig](/docs/transformers/v5.6.0/ja/model_doc/deberta#transformers.DebertaConfig) configuration class: [DebertaForQuestionAnswering](/docs/transformers/v5.6.0/ja/model_doc/deberta#transformers.DebertaForQuestionAnswering) (DebertaConfig model) - [DebertaV2Config](/docs/transformers/v5.6.0/ja/model_doc/deberta-v2#transformers.DebertaV2Config) configuration class: [DebertaV2ForQuestionAnswering](/docs/transformers/v5.6.0/ja/model_doc/deberta-v2#transformers.DebertaV2ForQuestionAnswering) (DebertaV2Config model) - `DiffLlamaConfig` configuration class: `DiffLlamaForQuestionAnswering` (DiffLlamaConfig model) - `DistilBertConfig` configuration class: `DistilBertForQuestionAnswering` (DistilBertConfig model) - `ElectraConfig` configuration class: `ElectraForQuestionAnswering` (ElectraConfig model) - `ErnieConfig` configuration class: `ErnieForQuestionAnswering` (ErnieConfig model) - `Exaone4Config` configuration class: `Exaone4ForQuestionAnswering` (Exaone4Config model) - `FNetConfig` configuration class: `FNetForQuestionAnswering` (FNetConfig model) - `FalconConfig` configuration class: `FalconForQuestionAnswering` (FalconConfig model) - `FlaubertConfig` configuration class: `FlaubertForQuestionAnsweringSimple` (FlaubertConfig model) - `FunnelConfig` configuration class: `FunnelForQuestionAnswering` (FunnelConfig model) - `GPT2Config` configuration class: `GPT2ForQuestionAnswering` (GPT2Config model) - `GPTJConfig` configuration class: `GPTJForQuestionAnswering` (GPTJConfig model) - `GPTNeoConfig` configuration class: `GPTNeoForQuestionAnswering` (GPTNeoConfig model) - `GPTNeoXConfig` configuration class: `GPTNeoXForQuestionAnswering` (GPTNeoXConfig model) - `IBertConfig` configuration class: `IBertForQuestionAnswering` (IBertConfig model) - `JinaEmbeddingsV3Config` configuration class: `JinaEmbeddingsV3ForQuestionAnswering` (JinaEmbeddingsV3Config model) - `LEDConfig` configuration class: `LEDForQuestionAnswering` (LEDConfig model) - `LayoutLMv2Config` configuration class: `LayoutLMv2ForQuestionAnswering` (LayoutLMv2Config model) - `LayoutLMv3Config` configuration class: `LayoutLMv3ForQuestionAnswering` (LayoutLMv3Config model) - `LiltConfig` configuration class: `LiltForQuestionAnswering` (LiltConfig model) - `LlamaConfig` configuration class: `LlamaForQuestionAnswering` (LlamaConfig model) - `LongformerConfig` configuration class: `LongformerForQuestionAnswering` (LongformerConfig model) - `LukeConfig` configuration class: `LukeForQuestionAnswering` (LukeConfig model) - `LxmertConfig` configuration class: `LxmertForQuestionAnswering` (LxmertConfig model) - `MBartConfig` configuration class: `MBartForQuestionAnswering` (MBartConfig model) - `MPNetConfig` configuration class: `MPNetForQuestionAnswering` (MPNetConfig model) - `MT5Config` configuration class: `MT5ForQuestionAnswering` (MT5Config model) - `MarkupLMConfig` configuration class: `MarkupLMForQuestionAnswering` (MarkupLMConfig model) - `MegatronBertConfig` configuration class: `MegatronBertForQuestionAnswering` (MegatronBertConfig model) - `MiniMaxConfig` configuration class: `MiniMaxForQuestionAnswering` (MiniMaxConfig model) - `Ministral3Config` configuration class: `Ministral3ForQuestionAnswering` (Ministral3Config model) - `MinistralConfig` configuration class: `MinistralForQuestionAnswering` (MinistralConfig model) - `MistralConfig` configuration class: `MistralForQuestionAnswering` (MistralConfig model) - `MixtralConfig` configuration class: `MixtralForQuestionAnswering` (MixtralConfig model) - `MobileBertConfig` configuration class: `MobileBertForQuestionAnswering` (MobileBertConfig model) - `ModernBertConfig` configuration class: `ModernBertForQuestionAnswering` (ModernBertConfig model) - `MptConfig` configuration class: `MptForQuestionAnswering` (MptConfig model) - `MraConfig` configuration class: `MraForQuestionAnswering` (MraConfig model) - `MvpConfig` configuration class: `MvpForQuestionAnswering` (MvpConfig model) - `NemotronConfig` configuration class: `NemotronForQuestionAnswering` (NemotronConfig model) - `NystromformerConfig` configuration class: `NystromformerForQuestionAnswering` (NystromformerConfig model) - `OPTConfig` configuration class: `OPTForQuestionAnswering` (OPTConfig model) - `Qwen2Config` configuration class: `Qwen2ForQuestionAnswering` (Qwen2Config model) - `Qwen2MoeConfig` configuration class: `Qwen2MoeForQuestionAnswering` (Qwen2MoeConfig model) - `Qwen3Config` configuration class: `Qwen3ForQuestionAnswering` (Qwen3Config model) - `Qwen3MoeConfig` configuration class: `Qwen3MoeForQuestionAnswering` (Qwen3MoeConfig model) - `Qwen3NextConfig` configuration class: `Qwen3NextForQuestionAnswering` (Qwen3NextConfig model) - `ReformerConfig` configuration class: `ReformerForQuestionAnswering` (ReformerConfig model) - `RemBertConfig` configuration class: `RemBertForQuestionAnswering` (RemBertConfig model) - `RoCBertConfig` configuration class: `RoCBertForQuestionAnswering` (RoCBertConfig model) - `RoFormerConfig` configuration class: `RoFormerForQuestionAnswering` (RoFormerConfig model) - `RobertaConfig` configuration class: `RobertaForQuestionAnswering` (RobertaConfig model) - `RobertaPreLayerNormConfig` configuration class: `RobertaPreLayerNormForQuestionAnswering` (RobertaPreLayerNormConfig model) - `SeedOssConfig` configuration class: `SeedOssForQuestionAnswering` (SeedOssConfig model) - `SmolLM3Config` configuration class: `SmolLM3ForQuestionAnswering` (SmolLM3Config model) - `SplinterConfig` configuration class: `SplinterForQuestionAnswering` (SplinterConfig model) - `SqueezeBertConfig` configuration class: `SqueezeBertForQuestionAnswering` (SqueezeBertConfig model) - `T5Config` configuration class: `T5ForQuestionAnswering` (T5Config model) - `UMT5Config` configuration class: `UMT5ForQuestionAnswering` (UMT5Config model) - `XLMConfig` configuration class: `XLMForQuestionAnsweringSimple` (XLMConfig model) - `XLMRobertaConfig` configuration class: `XLMRobertaForQuestionAnswering` (XLMRobertaConfig model) - `XLMRobertaXLConfig` configuration class: `XLMRobertaXLForQuestionAnswering` (XLMRobertaXLConfig model) - `XLNetConfig` configuration class: `XLNetForQuestionAnsweringSimple` (XLNetConfig model) - `XmodConfig` configuration class: `XmodForQuestionAnswering` (XmodConfig model) - `YosoConfig` configuration class: `YosoForQuestionAnswering` (YosoConfig model) - **attn_implementation** (`str`, *optional*) -- The attention implementation to use in the model (if relevant). Can be any of `"eager"` (manual implementation of the attention), `"sdpa"` (using [`F.scaled_dot_product_attention`](https://pytorch.org/docs/master/generated/torch.nn.functional.scaled_dot_product_attention.html)), `"flash_attention_2"` (using [Dao-AILab/flash-attention](https://github.com/Dao-AILab/flash-attention)), or `"flash_attention_3"` (using [Dao-AILab/flash-attention/hopper](https://github.com/Dao-AILab/flash-attention/tree/main/hopper)). By default, if available, SDPA will be used for torch>=2.1.1. The default is otherwise the manual `"eager"` implementation.0 Instantiates one of the model classes of the library (with a question answering head) from a configuration. Note: Loading a model from its configuration file does **not** load the model weights. It only affects the model's configuration. Use [from_pretrained()](/docs/transformers/v5.6.0/ja/model_doc/auto#transformers.AutoModel.from_pretrained) to load the model weights. Examples: ```python >>> from transformers import AutoConfig, AutoModelForQuestionAnswering >>> # Download configuration from huggingface.co and cache. >>> config = AutoConfig.from_pretrained("google-bert/bert-base-cased") >>> model = AutoModelForQuestionAnswering.from_config(config) ``` **Parameters:** config ([PreTrainedConfig](/docs/transformers/v5.6.0/ja/main_classes/configuration#transformers.PreTrainedConfig)) : The model class to instantiate is selected based on the configuration class: - [AlbertConfig](/docs/transformers/v5.6.0/ja/model_doc/albert#transformers.AlbertConfig) configuration class: [AlbertForQuestionAnswering](/docs/transformers/v5.6.0/ja/model_doc/albert#transformers.AlbertForQuestionAnswering) (AlbertConfig model) - `ArceeConfig` configuration class: `ArceeForQuestionAnswering` (ArceeConfig model) - [BartConfig](/docs/transformers/v5.6.0/ja/model_doc/bart#transformers.BartConfig) configuration class: [BartForQuestionAnswering](/docs/transformers/v5.6.0/ja/model_doc/bart#transformers.BartForQuestionAnswering) (BartConfig model) - [BertConfig](/docs/transformers/v5.6.0/ja/model_doc/bert#transformers.BertConfig) configuration class: [BertForQuestionAnswering](/docs/transformers/v5.6.0/ja/model_doc/bert#transformers.BertForQuestionAnswering) (BertConfig model) - [BigBirdConfig](/docs/transformers/v5.6.0/ja/model_doc/big_bird#transformers.BigBirdConfig) configuration class: [BigBirdForQuestionAnswering](/docs/transformers/v5.6.0/ja/model_doc/big_bird#transformers.BigBirdForQuestionAnswering) (BigBirdConfig model) - [BigBirdPegasusConfig](/docs/transformers/v5.6.0/ja/model_doc/bigbird_pegasus#transformers.BigBirdPegasusConfig) configuration class: [BigBirdPegasusForQuestionAnswering](/docs/transformers/v5.6.0/ja/model_doc/bigbird_pegasus#transformers.BigBirdPegasusForQuestionAnswering) (BigBirdPegasusConfig model) - [BloomConfig](/docs/transformers/v5.6.0/ja/model_doc/bloom#transformers.BloomConfig) configuration class: [BloomForQuestionAnswering](/docs/transformers/v5.6.0/ja/model_doc/bloom#transformers.BloomForQuestionAnswering) (BloomConfig model) - [CamembertConfig](/docs/transformers/v5.6.0/ja/model_doc/camembert#transformers.CamembertConfig) configuration class: [CamembertForQuestionAnswering](/docs/transformers/v5.6.0/ja/model_doc/camembert#transformers.CamembertForQuestionAnswering) (CamembertConfig model) - [CanineConfig](/docs/transformers/v5.6.0/ja/model_doc/canine#transformers.CanineConfig) configuration class: [CanineForQuestionAnswering](/docs/transformers/v5.6.0/ja/model_doc/canine#transformers.CanineForQuestionAnswering) (CanineConfig model) - [ConvBertConfig](/docs/transformers/v5.6.0/ja/model_doc/convbert#transformers.ConvBertConfig) configuration class: [ConvBertForQuestionAnswering](/docs/transformers/v5.6.0/ja/model_doc/convbert#transformers.ConvBertForQuestionAnswering) (ConvBertConfig model) - [Data2VecTextConfig](/docs/transformers/v5.6.0/ja/model_doc/data2vec#transformers.Data2VecTextConfig) configuration class: [Data2VecTextForQuestionAnswering](/docs/transformers/v5.6.0/ja/model_doc/data2vec#transformers.Data2VecTextForQuestionAnswering) (Data2VecTextConfig model) - [DebertaConfig](/docs/transformers/v5.6.0/ja/model_doc/deberta#transformers.DebertaConfig) configuration class: [DebertaForQuestionAnswering](/docs/transformers/v5.6.0/ja/model_doc/deberta#transformers.DebertaForQuestionAnswering) (DebertaConfig model) - [DebertaV2Config](/docs/transformers/v5.6.0/ja/model_doc/deberta-v2#transformers.DebertaV2Config) configuration class: [DebertaV2ForQuestionAnswering](/docs/transformers/v5.6.0/ja/model_doc/deberta-v2#transformers.DebertaV2ForQuestionAnswering) (DebertaV2Config model) - `DiffLlamaConfig` configuration class: `DiffLlamaForQuestionAnswering` (DiffLlamaConfig model) - `DistilBertConfig` configuration class: `DistilBertForQuestionAnswering` (DistilBertConfig model) - `ElectraConfig` configuration class: `ElectraForQuestionAnswering` (ElectraConfig model) - `ErnieConfig` configuration class: `ErnieForQuestionAnswering` (ErnieConfig model) - `Exaone4Config` configuration class: `Exaone4ForQuestionAnswering` (Exaone4Config model) - `FNetConfig` configuration class: `FNetForQuestionAnswering` (FNetConfig model) - `FalconConfig` configuration class: `FalconForQuestionAnswering` (FalconConfig model) - `FlaubertConfig` configuration class: `FlaubertForQuestionAnsweringSimple` (FlaubertConfig model) - `FunnelConfig` configuration class: `FunnelForQuestionAnswering` (FunnelConfig model) - `GPT2Config` configuration class: `GPT2ForQuestionAnswering` (GPT2Config model) - `GPTJConfig` configuration class: `GPTJForQuestionAnswering` (GPTJConfig model) - `GPTNeoConfig` configuration class: `GPTNeoForQuestionAnswering` (GPTNeoConfig model) - `GPTNeoXConfig` configuration class: `GPTNeoXForQuestionAnswering` (GPTNeoXConfig model) - `IBertConfig` configuration class: `IBertForQuestionAnswering` (IBertConfig model) - `JinaEmbeddingsV3Config` configuration class: `JinaEmbeddingsV3ForQuestionAnswering` (JinaEmbeddingsV3Config model) - `LEDConfig` configuration class: `LEDForQuestionAnswering` (LEDConfig model) - `LayoutLMv2Config` configuration class: `LayoutLMv2ForQuestionAnswering` (LayoutLMv2Config model) - `LayoutLMv3Config` configuration class: `LayoutLMv3ForQuestionAnswering` (LayoutLMv3Config model) - `LiltConfig` configuration class: `LiltForQuestionAnswering` (LiltConfig model) - `LlamaConfig` configuration class: `LlamaForQuestionAnswering` (LlamaConfig model) - `LongformerConfig` configuration class: `LongformerForQuestionAnswering` (LongformerConfig model) - `LukeConfig` configuration class: `LukeForQuestionAnswering` (LukeConfig model) - `LxmertConfig` configuration class: `LxmertForQuestionAnswering` (LxmertConfig model) - `MBartConfig` configuration class: `MBartForQuestionAnswering` (MBartConfig model) - `MPNetConfig` configuration class: `MPNetForQuestionAnswering` (MPNetConfig model) - `MT5Config` configuration class: `MT5ForQuestionAnswering` (MT5Config model) - `MarkupLMConfig` configuration class: `MarkupLMForQuestionAnswering` (MarkupLMConfig model) - `MegatronBertConfig` configuration class: `MegatronBertForQuestionAnswering` (MegatronBertConfig model) - `MiniMaxConfig` configuration class: `MiniMaxForQuestionAnswering` (MiniMaxConfig model) - `Ministral3Config` configuration class: `Ministral3ForQuestionAnswering` (Ministral3Config model) - `MinistralConfig` configuration class: `MinistralForQuestionAnswering` (MinistralConfig model) - `MistralConfig` configuration class: `MistralForQuestionAnswering` (MistralConfig model) - `MixtralConfig` configuration class: `MixtralForQuestionAnswering` (MixtralConfig model) - `MobileBertConfig` configuration class: `MobileBertForQuestionAnswering` (MobileBertConfig model) - `ModernBertConfig` configuration class: `ModernBertForQuestionAnswering` (ModernBertConfig model) - `MptConfig` configuration class: `MptForQuestionAnswering` (MptConfig model) - `MraConfig` configuration class: `MraForQuestionAnswering` (MraConfig model) - `MvpConfig` configuration class: `MvpForQuestionAnswering` (MvpConfig model) - `NemotronConfig` configuration class: `NemotronForQuestionAnswering` (NemotronConfig model) - `NystromformerConfig` configuration class: `NystromformerForQuestionAnswering` (NystromformerConfig model) - `OPTConfig` configuration class: `OPTForQuestionAnswering` (OPTConfig model) - `Qwen2Config` configuration class: `Qwen2ForQuestionAnswering` (Qwen2Config model) - `Qwen2MoeConfig` configuration class: `Qwen2MoeForQuestionAnswering` (Qwen2MoeConfig model) - `Qwen3Config` configuration class: `Qwen3ForQuestionAnswering` (Qwen3Config model) - `Qwen3MoeConfig` configuration class: `Qwen3MoeForQuestionAnswering` (Qwen3MoeConfig model) - `Qwen3NextConfig` configuration class: `Qwen3NextForQuestionAnswering` (Qwen3NextConfig model) - `ReformerConfig` configuration class: `ReformerForQuestionAnswering` (ReformerConfig model) - `RemBertConfig` configuration class: `RemBertForQuestionAnswering` (RemBertConfig model) - `RoCBertConfig` configuration class: `RoCBertForQuestionAnswering` (RoCBertConfig model) - `RoFormerConfig` configuration class: `RoFormerForQuestionAnswering` (RoFormerConfig model) - `RobertaConfig` configuration class: `RobertaForQuestionAnswering` (RobertaConfig model) - `RobertaPreLayerNormConfig` configuration class: `RobertaPreLayerNormForQuestionAnswering` (RobertaPreLayerNormConfig model) - `SeedOssConfig` configuration class: `SeedOssForQuestionAnswering` (SeedOssConfig model) - `SmolLM3Config` configuration class: `SmolLM3ForQuestionAnswering` (SmolLM3Config model) - `SplinterConfig` configuration class: `SplinterForQuestionAnswering` (SplinterConfig model) - `SqueezeBertConfig` configuration class: `SqueezeBertForQuestionAnswering` (SqueezeBertConfig model) - `T5Config` configuration class: `T5ForQuestionAnswering` (T5Config model) - `UMT5Config` configuration class: `UMT5ForQuestionAnswering` (UMT5Config model) - `XLMConfig` configuration class: `XLMForQuestionAnsweringSimple` (XLMConfig model) - `XLMRobertaConfig` configuration class: `XLMRobertaForQuestionAnswering` (XLMRobertaConfig model) - `XLMRobertaXLConfig` configuration class: `XLMRobertaXLForQuestionAnswering` (XLMRobertaXLConfig model) - `XLNetConfig` configuration class: `XLNetForQuestionAnsweringSimple` (XLNetConfig model) - `XmodConfig` configuration class: `XmodForQuestionAnswering` (XmodConfig model) - `YosoConfig` configuration class: `YosoForQuestionAnswering` (YosoConfig model) attn_implementation (`str`, *optional*) : The attention implementation to use in the model (if relevant). Can be any of `"eager"` (manual implementation of the attention), `"sdpa"` (using [`F.scaled_dot_product_attention`](https://pytorch.org/docs/master/generated/torch.nn.functional.scaled_dot_product_attention.html)), `"flash_attention_2"` (using [Dao-AILab/flash-attention](https://github.com/Dao-AILab/flash-attention)), or `"flash_attention_3"` (using [Dao-AILab/flash-attention/hopper](https://github.com/Dao-AILab/flash-attention/tree/main/hopper)). By default, if available, SDPA will be used for torch>=2.1.1. The default is otherwise the manual `"eager"` implementation. #### from_pretrained[[transformers.AutoModelForQuestionAnswering.from_pretrained]] [Source](https://github.com/huggingface/transformers/blob/v5.6.0/src/transformers/models/auto/auto_factory.py#L253) Instantiate one of the model classes of the library (with a question answering head) from a pretrained model. The model class to instantiate is selected based on the `model_type` property of the config object (either passed as an argument or loaded from `pretrained_model_name_or_path` if possible), or when it's missing, by falling back to using pattern matching on `pretrained_model_name_or_path`: - **albert** -- [AlbertForQuestionAnswering](/docs/transformers/v5.6.0/ja/model_doc/albert#transformers.AlbertForQuestionAnswering) (AlbertConfig model) - **arcee** -- `ArceeForQuestionAnswering` (ArceeConfig model) - **bart** -- [BartForQuestionAnswering](/docs/transformers/v5.6.0/ja/model_doc/bart#transformers.BartForQuestionAnswering) (BartConfig model) - **bert** -- [BertForQuestionAnswering](/docs/transformers/v5.6.0/ja/model_doc/bert#transformers.BertForQuestionAnswering) (BertConfig model) - **big_bird** -- [BigBirdForQuestionAnswering](/docs/transformers/v5.6.0/ja/model_doc/big_bird#transformers.BigBirdForQuestionAnswering) (BigBirdConfig model) - **bigbird_pegasus** -- [BigBirdPegasusForQuestionAnswering](/docs/transformers/v5.6.0/ja/model_doc/bigbird_pegasus#transformers.BigBirdPegasusForQuestionAnswering) (BigBirdPegasusConfig model) - **bloom** -- [BloomForQuestionAnswering](/docs/transformers/v5.6.0/ja/model_doc/bloom#transformers.BloomForQuestionAnswering) (BloomConfig model) - **camembert** -- [CamembertForQuestionAnswering](/docs/transformers/v5.6.0/ja/model_doc/camembert#transformers.CamembertForQuestionAnswering) (CamembertConfig model) - **canine** -- [CanineForQuestionAnswering](/docs/transformers/v5.6.0/ja/model_doc/canine#transformers.CanineForQuestionAnswering) (CanineConfig model) - **convbert** -- [ConvBertForQuestionAnswering](/docs/transformers/v5.6.0/ja/model_doc/convbert#transformers.ConvBertForQuestionAnswering) (ConvBertConfig model) - **data2vec-text** -- [Data2VecTextForQuestionAnswering](/docs/transformers/v5.6.0/ja/model_doc/data2vec#transformers.Data2VecTextForQuestionAnswering) (Data2VecTextConfig model) - **deberta** -- [DebertaForQuestionAnswering](/docs/transformers/v5.6.0/ja/model_doc/deberta#transformers.DebertaForQuestionAnswering) (DebertaConfig model) - **deberta-v2** -- [DebertaV2ForQuestionAnswering](/docs/transformers/v5.6.0/ja/model_doc/deberta-v2#transformers.DebertaV2ForQuestionAnswering) (DebertaV2Config model) - **diffllama** -- `DiffLlamaForQuestionAnswering` (DiffLlamaConfig model) - **distilbert** -- `DistilBertForQuestionAnswering` (DistilBertConfig model) - **electra** -- `ElectraForQuestionAnswering` (ElectraConfig model) - **ernie** -- `ErnieForQuestionAnswering` (ErnieConfig model) - **exaone4** -- `Exaone4ForQuestionAnswering` (Exaone4Config model) - **falcon** -- `FalconForQuestionAnswering` (FalconConfig model) - **flaubert** -- `FlaubertForQuestionAnsweringSimple` (FlaubertConfig model) - **fnet** -- `FNetForQuestionAnswering` (FNetConfig model) - **funnel** -- `FunnelForQuestionAnswering` (FunnelConfig model) - **gpt2** -- `GPT2ForQuestionAnswering` (GPT2Config model) - **gpt_neo** -- `GPTNeoForQuestionAnswering` (GPTNeoConfig model) - **gpt_neox** -- `GPTNeoXForQuestionAnswering` (GPTNeoXConfig model) - **gptj** -- `GPTJForQuestionAnswering` (GPTJConfig model) - **ibert** -- `IBertForQuestionAnswering` (IBertConfig model) - **jina_embeddings_v3** -- `JinaEmbeddingsV3ForQuestionAnswering` (JinaEmbeddingsV3Config model) - **layoutlmv2** -- `LayoutLMv2ForQuestionAnswering` (LayoutLMv2Config model) - **layoutlmv3** -- `LayoutLMv3ForQuestionAnswering` (LayoutLMv3Config model) - **led** -- `LEDForQuestionAnswering` (LEDConfig model) - **lilt** -- `LiltForQuestionAnswering` (LiltConfig model) - **llama** -- `LlamaForQuestionAnswering` (LlamaConfig model) - **longformer** -- `LongformerForQuestionAnswering` (LongformerConfig model) - **luke** -- `LukeForQuestionAnswering` (LukeConfig model) - **lxmert** -- `LxmertForQuestionAnswering` (LxmertConfig model) - **markuplm** -- `MarkupLMForQuestionAnswering` (MarkupLMConfig model) - **mbart** -- `MBartForQuestionAnswering` (MBartConfig model) - **megatron-bert** -- `MegatronBertForQuestionAnswering` (MegatronBertConfig model) - **minimax** -- `MiniMaxForQuestionAnswering` (MiniMaxConfig model) - **ministral** -- `MinistralForQuestionAnswering` (MinistralConfig model) - **ministral3** -- `Ministral3ForQuestionAnswering` (Ministral3Config model) - **mistral** -- `MistralForQuestionAnswering` (MistralConfig model) - **mixtral** -- `MixtralForQuestionAnswering` (MixtralConfig model) - **mobilebert** -- `MobileBertForQuestionAnswering` (MobileBertConfig model) - **modernbert** -- `ModernBertForQuestionAnswering` (ModernBertConfig model) - **mpnet** -- `MPNetForQuestionAnswering` (MPNetConfig model) - **mpt** -- `MptForQuestionAnswering` (MptConfig model) - **mra** -- `MraForQuestionAnswering` (MraConfig model) - **mt5** -- `MT5ForQuestionAnswering` (MT5Config model) - **mvp** -- `MvpForQuestionAnswering` (MvpConfig model) - **nemotron** -- `NemotronForQuestionAnswering` (NemotronConfig model) - **nystromformer** -- `NystromformerForQuestionAnswering` (NystromformerConfig model) - **opt** -- `OPTForQuestionAnswering` (OPTConfig model) - **qwen2** -- `Qwen2ForQuestionAnswering` (Qwen2Config model) - **qwen2_moe** -- `Qwen2MoeForQuestionAnswering` (Qwen2MoeConfig model) - **qwen3** -- `Qwen3ForQuestionAnswering` (Qwen3Config model) - **qwen3_moe** -- `Qwen3MoeForQuestionAnswering` (Qwen3MoeConfig model) - **qwen3_next** -- `Qwen3NextForQuestionAnswering` (Qwen3NextConfig model) - **reformer** -- `ReformerForQuestionAnswering` (ReformerConfig model) - **rembert** -- `RemBertForQuestionAnswering` (RemBertConfig model) - **roberta** -- `RobertaForQuestionAnswering` (RobertaConfig model) - **roberta-prelayernorm** -- `RobertaPreLayerNormForQuestionAnswering` (RobertaPreLayerNormConfig model) - **roc_bert** -- `RoCBertForQuestionAnswering` (RoCBertConfig model) - **roformer** -- `RoFormerForQuestionAnswering` (RoFormerConfig model) - **seed_oss** -- `SeedOssForQuestionAnswering` (SeedOssConfig model) - **smollm3** -- `SmolLM3ForQuestionAnswering` (SmolLM3Config model) - **splinter** -- `SplinterForQuestionAnswering` (SplinterConfig model) - **squeezebert** -- `SqueezeBertForQuestionAnswering` (SqueezeBertConfig model) - **t5** -- `T5ForQuestionAnswering` (T5Config model) - **umt5** -- `UMT5ForQuestionAnswering` (UMT5Config model) - **xlm** -- `XLMForQuestionAnsweringSimple` (XLMConfig model) - **xlm-roberta** -- `XLMRobertaForQuestionAnswering` (XLMRobertaConfig model) - **xlm-roberta-xl** -- `XLMRobertaXLForQuestionAnswering` (XLMRobertaXLConfig model) - **xlnet** -- `XLNetForQuestionAnsweringSimple` (XLNetConfig model) - **xmod** -- `XmodForQuestionAnswering` (XmodConfig model) - **yoso** -- `YosoForQuestionAnswering` (YosoConfig model) The model is set in evaluation mode by default using `model.eval()` (so for instance, dropout modules are deactivated). To train the model, you should first set it back in training mode with `model.train()` Examples: ```python >>> from transformers import AutoConfig, AutoModelForQuestionAnswering >>> # Download model and configuration from huggingface.co and cache. >>> model = AutoModelForQuestionAnswering.from_pretrained("google-bert/bert-base-cased") >>> # Update configuration during loading >>> model = AutoModelForQuestionAnswering.from_pretrained("google-bert/bert-base-cased", output_attentions=True) >>> model.config.output_attentions True ``` **Parameters:** pretrained_model_name_or_path (`str` or `os.PathLike`) : Can be either: - A string, the *model id* of a pretrained model hosted inside a model repo on huggingface.co. - A path to a *directory* containing model weights saved using [save_pretrained()](/docs/transformers/v5.6.0/ja/main_classes/model#transformers.PreTrainedModel.save_pretrained), e.g., `./my_model_directory/`. model_args (additional positional arguments, *optional*) : Will be passed along to the underlying model `__init__()` method. config ([PreTrainedConfig](/docs/transformers/v5.6.0/ja/main_classes/configuration#transformers.PreTrainedConfig), *optional*) : Configuration for the model to use instead of an automatically loaded configuration. Configuration can be automatically loaded when: - The model is a model provided by the library (loaded with the *model id* string of a pretrained model). - The model was saved using [save_pretrained()](/docs/transformers/v5.6.0/ja/main_classes/model#transformers.PreTrainedModel.save_pretrained) and is reloaded by supplying the save directory. - The model is loaded by supplying a local directory as `pretrained_model_name_or_path` and a configuration JSON file named *config.json* is found in the directory. state_dict (*dict[str, torch.Tensor]*, *optional*) : A state dictionary to use instead of a state dictionary loaded from saved weights file. This option can be used if you want to create a model from a pretrained configuration but load your own weights. In this case though, you should check if using [save_pretrained()](/docs/transformers/v5.6.0/ja/main_classes/model#transformers.PreTrainedModel.save_pretrained) and [from_pretrained()](/docs/transformers/v5.6.0/ja/main_classes/model#transformers.PreTrainedModel.from_pretrained) is not a simpler option. cache_dir (`str` or `os.PathLike`, *optional*) : Path to a directory in which a downloaded pretrained model configuration should be cached if the standard cache should not be used. force_download (`bool`, *optional*, defaults to `False`) : Whether or not to force the (re-)download of the model weights and configuration files, overriding the cached versions if they exist. proxies (`dict[str, str]`, *optional*) : A dictionary of proxy servers to use by protocol or endpoint, e.g., `{'http': 'foo.bar:3128', 'http://hostname': 'foo.bar:4012'}`. The proxies are used on each request. output_loading_info(`bool`, *optional*, defaults to `False`) : Whether or not to also return a dictionary containing missing keys, unexpected keys and error messages. local_files_only(`bool`, *optional*, defaults to `False`) : Whether or not to only look at local files (e.g., not try downloading the model). revision (`str`, *optional*, defaults to `"main"`) : The specific model version to use. It can be a branch name, a tag name, or a commit id, since we use a git-based system for storing models and other artifacts on huggingface.co, so `revision` can be any identifier allowed by git. trust_remote_code (`bool`, *optional*, defaults to `False`) : Whether or not to allow for custom models defined on the Hub in their own modeling files. This option should only be set to `True` for repositories you trust and in which you have read the code, as it will execute code present on the Hub on your local machine. code_revision (`str`, *optional*, defaults to `"main"`) : The specific revision to use for the code on the Hub, if the code leaves in a different repository than the rest of the model. It can be a branch name, a tag name, or a commit id, since we use a git-based system for storing models and other artifacts on huggingface.co, so `revision` can be any identifier allowed by git. kwargs (additional keyword arguments, *optional*) : Can be used to update the configuration object (after it being loaded) and initiate the model (e.g., `output_attentions=True`). Behaves differently depending on whether a `config` is provided or automatically loaded: - If a configuration is provided with `config`, `**kwargs` will be directly passed to the underlying model's `__init__` method (we assume all relevant updates to the configuration have already been done) - If a configuration is not provided, `kwargs` will be first passed to the configuration class initialization function ([from_pretrained()](/docs/transformers/v5.6.0/ja/main_classes/configuration#transformers.PreTrainedConfig.from_pretrained)). Each key of `kwargs` that corresponds to a configuration attribute will be used to override said attribute with the supplied `kwargs` value. Remaining keys that do not correspond to any configuration attribute will be passed to the underlying model's `__init__` function. ### AutoModelForTextEncoding[[transformers.AutoModelForTextEncoding]] #### transformers.AutoModelForTextEncoding[[transformers.AutoModelForTextEncoding]] [Source](https://github.com/huggingface/transformers/blob/v5.6.0/src/transformers/models/auto/modeling_auto.py#L1971) ## Computer vision 以下の自動クラスは、次のコンピュータービジョンタスクに利用可能です。 ### AutoModelForDepthEstimation[[transformers.AutoModelForDepthEstimation]] #### transformers.AutoModelForDepthEstimation[[transformers.AutoModelForDepthEstimation]] [Source](https://github.com/huggingface/transformers/blob/v5.6.0/src/transformers/models/auto/modeling_auto.py#L2175) This is a generic model class that will be instantiated as one of the model classes of the library (with a depth estimation head) when created with the [from_pretrained()](/docs/transformers/v5.6.0/ja/model_doc/auto#transformers.AutoModel.from_pretrained) class method or the [from_config()](/docs/transformers/v5.6.0/ja/model_doc/auto#transformers.AutoModel.from_config) class method. This class cannot be instantiated directly using `__init__()` (throws an error). from_configtransformers.AutoModelForDepthEstimation.from_confighttps://github.com/huggingface/transformers/blob/v5.6.0/src/transformers/models/auto/auto_factory.py#L206[{"name": "**kwargs", "val": ""}]- **config** ([PreTrainedConfig](/docs/transformers/v5.6.0/ja/main_classes/configuration#transformers.PreTrainedConfig)) -- The model class to instantiate is selected based on the configuration class: - `CHMv2Config` configuration class: `CHMv2ForDepthEstimation` (CHMv2Config model) - `DPTConfig` configuration class: `DPTForDepthEstimation` (DPTConfig model) - `DepthAnythingConfig` configuration class: `DepthAnythingForDepthEstimation` (DepthAnythingConfig model) - `DepthProConfig` configuration class: `DepthProForDepthEstimation` (DepthProConfig model) - `GLPNConfig` configuration class: `GLPNForDepthEstimation` (GLPNConfig model) - `PromptDepthAnythingConfig` configuration class: `PromptDepthAnythingForDepthEstimation` (PromptDepthAnythingConfig model) - `ZoeDepthConfig` configuration class: `ZoeDepthForDepthEstimation` (ZoeDepthConfig model) - **attn_implementation** (`str`, *optional*) -- The attention implementation to use in the model (if relevant). Can be any of `"eager"` (manual implementation of the attention), `"sdpa"` (using [`F.scaled_dot_product_attention`](https://pytorch.org/docs/master/generated/torch.nn.functional.scaled_dot_product_attention.html)), `"flash_attention_2"` (using [Dao-AILab/flash-attention](https://github.com/Dao-AILab/flash-attention)), or `"flash_attention_3"` (using [Dao-AILab/flash-attention/hopper](https://github.com/Dao-AILab/flash-attention/tree/main/hopper)). By default, if available, SDPA will be used for torch>=2.1.1. The default is otherwise the manual `"eager"` implementation.0 Instantiates one of the model classes of the library (with a depth estimation head) from a configuration. Note: Loading a model from its configuration file does **not** load the model weights. It only affects the model's configuration. Use [from_pretrained()](/docs/transformers/v5.6.0/ja/model_doc/auto#transformers.AutoModel.from_pretrained) to load the model weights. Examples: ```python >>> from transformers import AutoConfig, AutoModelForDepthEstimation >>> # Download configuration from huggingface.co and cache. >>> config = AutoConfig.from_pretrained("google-bert/bert-base-cased") >>> model = AutoModelForDepthEstimation.from_config(config) ``` **Parameters:** config ([PreTrainedConfig](/docs/transformers/v5.6.0/ja/main_classes/configuration#transformers.PreTrainedConfig)) : The model class to instantiate is selected based on the configuration class: - `CHMv2Config` configuration class: `CHMv2ForDepthEstimation` (CHMv2Config model) - `DPTConfig` configuration class: `DPTForDepthEstimation` (DPTConfig model) - `DepthAnythingConfig` configuration class: `DepthAnythingForDepthEstimation` (DepthAnythingConfig model) - `DepthProConfig` configuration class: `DepthProForDepthEstimation` (DepthProConfig model) - `GLPNConfig` configuration class: `GLPNForDepthEstimation` (GLPNConfig model) - `PromptDepthAnythingConfig` configuration class: `PromptDepthAnythingForDepthEstimation` (PromptDepthAnythingConfig model) - `ZoeDepthConfig` configuration class: `ZoeDepthForDepthEstimation` (ZoeDepthConfig model) attn_implementation (`str`, *optional*) : The attention implementation to use in the model (if relevant). Can be any of `"eager"` (manual implementation of the attention), `"sdpa"` (using [`F.scaled_dot_product_attention`](https://pytorch.org/docs/master/generated/torch.nn.functional.scaled_dot_product_attention.html)), `"flash_attention_2"` (using [Dao-AILab/flash-attention](https://github.com/Dao-AILab/flash-attention)), or `"flash_attention_3"` (using [Dao-AILab/flash-attention/hopper](https://github.com/Dao-AILab/flash-attention/tree/main/hopper)). By default, if available, SDPA will be used for torch>=2.1.1. The default is otherwise the manual `"eager"` implementation. #### from_pretrained[[transformers.AutoModelForDepthEstimation.from_pretrained]] [Source](https://github.com/huggingface/transformers/blob/v5.6.0/src/transformers/models/auto/auto_factory.py#L253) Instantiate one of the model classes of the library (with a depth estimation head) from a pretrained model. The model class to instantiate is selected based on the `model_type` property of the config object (either passed as an argument or loaded from `pretrained_model_name_or_path` if possible), or when it's missing, by falling back to using pattern matching on `pretrained_model_name_or_path`: - **chmv2** -- `CHMv2ForDepthEstimation` (CHMv2Config model) - **depth_anything** -- `DepthAnythingForDepthEstimation` (DepthAnythingConfig model) - **depth_pro** -- `DepthProForDepthEstimation` (DepthProConfig model) - **dpt** -- `DPTForDepthEstimation` (DPTConfig model) - **glpn** -- `GLPNForDepthEstimation` (GLPNConfig model) - **prompt_depth_anything** -- `PromptDepthAnythingForDepthEstimation` (PromptDepthAnythingConfig model) - **zoedepth** -- `ZoeDepthForDepthEstimation` (ZoeDepthConfig model) The model is set in evaluation mode by default using `model.eval()` (so for instance, dropout modules are deactivated). To train the model, you should first set it back in training mode with `model.train()` Examples: ```python >>> from transformers import AutoConfig, AutoModelForDepthEstimation >>> # Download model and configuration from huggingface.co and cache. >>> model = AutoModelForDepthEstimation.from_pretrained("google-bert/bert-base-cased") >>> # Update configuration during loading >>> model = AutoModelForDepthEstimation.from_pretrained("google-bert/bert-base-cased", output_attentions=True) >>> model.config.output_attentions True ``` **Parameters:** pretrained_model_name_or_path (`str` or `os.PathLike`) : Can be either: - A string, the *model id* of a pretrained model hosted inside a model repo on huggingface.co. - A path to a *directory* containing model weights saved using [save_pretrained()](/docs/transformers/v5.6.0/ja/main_classes/model#transformers.PreTrainedModel.save_pretrained), e.g., `./my_model_directory/`. model_args (additional positional arguments, *optional*) : Will be passed along to the underlying model `__init__()` method. config ([PreTrainedConfig](/docs/transformers/v5.6.0/ja/main_classes/configuration#transformers.PreTrainedConfig), *optional*) : Configuration for the model to use instead of an automatically loaded configuration. Configuration can be automatically loaded when: - The model is a model provided by the library (loaded with the *model id* string of a pretrained model). - The model was saved using [save_pretrained()](/docs/transformers/v5.6.0/ja/main_classes/model#transformers.PreTrainedModel.save_pretrained) and is reloaded by supplying the save directory. - The model is loaded by supplying a local directory as `pretrained_model_name_or_path` and a configuration JSON file named *config.json* is found in the directory. state_dict (*dict[str, torch.Tensor]*, *optional*) : A state dictionary to use instead of a state dictionary loaded from saved weights file. This option can be used if you want to create a model from a pretrained configuration but load your own weights. In this case though, you should check if using [save_pretrained()](/docs/transformers/v5.6.0/ja/main_classes/model#transformers.PreTrainedModel.save_pretrained) and [from_pretrained()](/docs/transformers/v5.6.0/ja/main_classes/model#transformers.PreTrainedModel.from_pretrained) is not a simpler option. cache_dir (`str` or `os.PathLike`, *optional*) : Path to a directory in which a downloaded pretrained model configuration should be cached if the standard cache should not be used. force_download (`bool`, *optional*, defaults to `False`) : Whether or not to force the (re-)download of the model weights and configuration files, overriding the cached versions if they exist. proxies (`dict[str, str]`, *optional*) : A dictionary of proxy servers to use by protocol or endpoint, e.g., `{'http': 'foo.bar:3128', 'http://hostname': 'foo.bar:4012'}`. The proxies are used on each request. output_loading_info(`bool`, *optional*, defaults to `False`) : Whether or not to also return a dictionary containing missing keys, unexpected keys and error messages. local_files_only(`bool`, *optional*, defaults to `False`) : Whether or not to only look at local files (e.g., not try downloading the model). revision (`str`, *optional*, defaults to `"main"`) : The specific model version to use. It can be a branch name, a tag name, or a commit id, since we use a git-based system for storing models and other artifacts on huggingface.co, so `revision` can be any identifier allowed by git. trust_remote_code (`bool`, *optional*, defaults to `False`) : Whether or not to allow for custom models defined on the Hub in their own modeling files. This option should only be set to `True` for repositories you trust and in which you have read the code, as it will execute code present on the Hub on your local machine. code_revision (`str`, *optional*, defaults to `"main"`) : The specific revision to use for the code on the Hub, if the code leaves in a different repository than the rest of the model. It can be a branch name, a tag name, or a commit id, since we use a git-based system for storing models and other artifacts on huggingface.co, so `revision` can be any identifier allowed by git. kwargs (additional keyword arguments, *optional*) : Can be used to update the configuration object (after it being loaded) and initiate the model (e.g., `output_attentions=True`). Behaves differently depending on whether a `config` is provided or automatically loaded: - If a configuration is provided with `config`, `**kwargs` will be directly passed to the underlying model's `__init__` method (we assume all relevant updates to the configuration have already been done) - If a configuration is not provided, `kwargs` will be first passed to the configuration class initialization function ([from_pretrained()](/docs/transformers/v5.6.0/ja/main_classes/configuration#transformers.PreTrainedConfig.from_pretrained)). Each key of `kwargs` that corresponds to a configuration attribute will be used to override said attribute with the supplied `kwargs` value. Remaining keys that do not correspond to any configuration attribute will be passed to the underlying model's `__init__` function. ### AutoModelForImageClassification[[transformers.AutoModelForImageClassification]] #### transformers.AutoModelForImageClassification[[transformers.AutoModelForImageClassification]] [Source](https://github.com/huggingface/transformers/blob/v5.6.0/src/transformers/models/auto/modeling_auto.py#L2100) This is a generic model class that will be instantiated as one of the model classes of the library (with a image classification head) when created with the [from_pretrained()](/docs/transformers/v5.6.0/ja/model_doc/auto#transformers.AutoModel.from_pretrained) class method or the [from_config()](/docs/transformers/v5.6.0/ja/model_doc/auto#transformers.AutoModel.from_config) class method. This class cannot be instantiated directly using `__init__()` (throws an error). from_configtransformers.AutoModelForImageClassification.from_confighttps://github.com/huggingface/transformers/blob/v5.6.0/src/transformers/models/auto/auto_factory.py#L206[{"name": "**kwargs", "val": ""}]- **config** ([PreTrainedConfig](/docs/transformers/v5.6.0/ja/main_classes/configuration#transformers.PreTrainedConfig)) -- The model class to instantiate is selected based on the configuration class: - [BeitConfig](/docs/transformers/v5.6.0/ja/model_doc/beit#transformers.BeitConfig) configuration class: [BeitForImageClassification](/docs/transformers/v5.6.0/ja/model_doc/beit#transformers.BeitForImageClassification) (BeitConfig model) - [BitConfig](/docs/transformers/v5.6.0/ja/model_doc/bit#transformers.BitConfig) configuration class: [BitForImageClassification](/docs/transformers/v5.6.0/ja/model_doc/bit#transformers.BitForImageClassification) (BitConfig model) - [CLIPConfig](/docs/transformers/v5.6.0/ja/model_doc/clip#transformers.CLIPConfig) configuration class: `CLIPForImageClassification` (CLIPConfig model) - [ConvNextConfig](/docs/transformers/v5.6.0/ja/model_doc/convnext#transformers.ConvNextConfig) configuration class: [ConvNextForImageClassification](/docs/transformers/v5.6.0/ja/model_doc/convnext#transformers.ConvNextForImageClassification) (ConvNextConfig model) - [ConvNextV2Config](/docs/transformers/v5.6.0/ja/model_doc/convnextv2#transformers.ConvNextV2Config) configuration class: [ConvNextV2ForImageClassification](/docs/transformers/v5.6.0/ja/model_doc/convnextv2#transformers.ConvNextV2ForImageClassification) (ConvNextV2Config model) - [CvtConfig](/docs/transformers/v5.6.0/ja/model_doc/cvt#transformers.CvtConfig) configuration class: [CvtForImageClassification](/docs/transformers/v5.6.0/ja/model_doc/cvt#transformers.CvtForImageClassification) (CvtConfig model) - [Data2VecVisionConfig](/docs/transformers/v5.6.0/ja/model_doc/data2vec#transformers.Data2VecVisionConfig) configuration class: [Data2VecVisionForImageClassification](/docs/transformers/v5.6.0/ja/model_doc/data2vec#transformers.Data2VecVisionForImageClassification) (Data2VecVisionConfig model) - [DeiTConfig](/docs/transformers/v5.6.0/ja/model_doc/deit#transformers.DeiTConfig) configuration class: [DeiTForImageClassification](/docs/transformers/v5.6.0/ja/model_doc/deit#transformers.DeiTForImageClassification) or [DeiTForImageClassificationWithTeacher](/docs/transformers/v5.6.0/ja/model_doc/deit#transformers.DeiTForImageClassificationWithTeacher) (DeiTConfig model) - [DinatConfig](/docs/transformers/v5.6.0/ja/model_doc/dinat#transformers.DinatConfig) configuration class: [DinatForImageClassification](/docs/transformers/v5.6.0/ja/model_doc/dinat#transformers.DinatForImageClassification) (DinatConfig model) - `Dinov2Config` configuration class: `Dinov2ForImageClassification` (Dinov2Config model) - `Dinov2WithRegistersConfig` configuration class: `Dinov2WithRegistersForImageClassification` (Dinov2WithRegistersConfig model) - `DonutSwinConfig` configuration class: `DonutSwinForImageClassification` (DonutSwinConfig model) - `EfficientNetConfig` configuration class: `EfficientNetForImageClassification` (EfficientNetConfig model) - `FocalNetConfig` configuration class: `FocalNetForImageClassification` (FocalNetConfig model) - `HGNetV2Config` configuration class: `HGNetV2ForImageClassification` (HGNetV2Config model) - `HieraConfig` configuration class: `HieraForImageClassification` (HieraConfig model) - `IJepaConfig` configuration class: `IJepaForImageClassification` (IJepaConfig model) - `ImageGPTConfig` configuration class: `ImageGPTForImageClassification` (ImageGPTConfig model) - `LevitConfig` configuration class: `LevitForImageClassification` or `LevitForImageClassificationWithTeacher` (LevitConfig model) - `MetaClip2Config` configuration class: `MetaClip2ForImageClassification` (MetaClip2Config model) - `MobileNetV1Config` configuration class: `MobileNetV1ForImageClassification` (MobileNetV1Config model) - `MobileNetV2Config` configuration class: `MobileNetV2ForImageClassification` (MobileNetV2Config model) - `MobileViTConfig` configuration class: `MobileViTForImageClassification` (MobileViTConfig model) - `MobileViTV2Config` configuration class: `MobileViTV2ForImageClassification` (MobileViTV2Config model) - `PPLCNetConfig` configuration class: `PPLCNetForImageClassification` (PPLCNetConfig model) - `PerceiverConfig` configuration class: `PerceiverForImageClassificationLearned` or `PerceiverForImageClassificationFourier` or `PerceiverForImageClassificationConvProcessing` (PerceiverConfig model) - `PoolFormerConfig` configuration class: `PoolFormerForImageClassification` (PoolFormerConfig model) - `PvtConfig` configuration class: `PvtForImageClassification` (PvtConfig model) - `PvtV2Config` configuration class: `PvtV2ForImageClassification` (PvtV2Config model) - `RegNetConfig` configuration class: `RegNetForImageClassification` (RegNetConfig model) - `ResNetConfig` configuration class: `ResNetForImageClassification` (ResNetConfig model) - `SegformerConfig` configuration class: `SegformerForImageClassification` (SegformerConfig model) - `ShieldGemma2Config` configuration class: `ShieldGemma2ForImageClassification` (ShieldGemma2Config model) - `Siglip2Config` configuration class: `Siglip2ForImageClassification` (Siglip2Config model) - `SiglipConfig` configuration class: `SiglipForImageClassification` (SiglipConfig model) - `SwiftFormerConfig` configuration class: `SwiftFormerForImageClassification` (SwiftFormerConfig model) - `SwinConfig` configuration class: `SwinForImageClassification` (SwinConfig model) - `Swinv2Config` configuration class: `Swinv2ForImageClassification` (Swinv2Config model) - `TextNetConfig` configuration class: `TextNetForImageClassification` (TextNetConfig model) - `TimmWrapperConfig` configuration class: `TimmWrapperForImageClassification` (TimmWrapperConfig model) - `ViTConfig` configuration class: `ViTForImageClassification` (ViTConfig model) - `ViTMSNConfig` configuration class: `ViTMSNForImageClassification` (ViTMSNConfig model) - **attn_implementation** (`str`, *optional*) -- The attention implementation to use in the model (if relevant). Can be any of `"eager"` (manual implementation of the attention), `"sdpa"` (using [`F.scaled_dot_product_attention`](https://pytorch.org/docs/master/generated/torch.nn.functional.scaled_dot_product_attention.html)), `"flash_attention_2"` (using [Dao-AILab/flash-attention](https://github.com/Dao-AILab/flash-attention)), or `"flash_attention_3"` (using [Dao-AILab/flash-attention/hopper](https://github.com/Dao-AILab/flash-attention/tree/main/hopper)). By default, if available, SDPA will be used for torch>=2.1.1. The default is otherwise the manual `"eager"` implementation.0 Instantiates one of the model classes of the library (with a image classification head) from a configuration. Note: Loading a model from its configuration file does **not** load the model weights. It only affects the model's configuration. Use [from_pretrained()](/docs/transformers/v5.6.0/ja/model_doc/auto#transformers.AutoModel.from_pretrained) to load the model weights. Examples: ```python >>> from transformers import AutoConfig, AutoModelForImageClassification >>> # Download configuration from huggingface.co and cache. >>> config = AutoConfig.from_pretrained("google-bert/bert-base-cased") >>> model = AutoModelForImageClassification.from_config(config) ``` **Parameters:** config ([PreTrainedConfig](/docs/transformers/v5.6.0/ja/main_classes/configuration#transformers.PreTrainedConfig)) : The model class to instantiate is selected based on the configuration class: - [BeitConfig](/docs/transformers/v5.6.0/ja/model_doc/beit#transformers.BeitConfig) configuration class: [BeitForImageClassification](/docs/transformers/v5.6.0/ja/model_doc/beit#transformers.BeitForImageClassification) (BeitConfig model) - [BitConfig](/docs/transformers/v5.6.0/ja/model_doc/bit#transformers.BitConfig) configuration class: [BitForImageClassification](/docs/transformers/v5.6.0/ja/model_doc/bit#transformers.BitForImageClassification) (BitConfig model) - [CLIPConfig](/docs/transformers/v5.6.0/ja/model_doc/clip#transformers.CLIPConfig) configuration class: `CLIPForImageClassification` (CLIPConfig model) - [ConvNextConfig](/docs/transformers/v5.6.0/ja/model_doc/convnext#transformers.ConvNextConfig) configuration class: [ConvNextForImageClassification](/docs/transformers/v5.6.0/ja/model_doc/convnext#transformers.ConvNextForImageClassification) (ConvNextConfig model) - [ConvNextV2Config](/docs/transformers/v5.6.0/ja/model_doc/convnextv2#transformers.ConvNextV2Config) configuration class: [ConvNextV2ForImageClassification](/docs/transformers/v5.6.0/ja/model_doc/convnextv2#transformers.ConvNextV2ForImageClassification) (ConvNextV2Config model) - [CvtConfig](/docs/transformers/v5.6.0/ja/model_doc/cvt#transformers.CvtConfig) configuration class: [CvtForImageClassification](/docs/transformers/v5.6.0/ja/model_doc/cvt#transformers.CvtForImageClassification) (CvtConfig model) - [Data2VecVisionConfig](/docs/transformers/v5.6.0/ja/model_doc/data2vec#transformers.Data2VecVisionConfig) configuration class: [Data2VecVisionForImageClassification](/docs/transformers/v5.6.0/ja/model_doc/data2vec#transformers.Data2VecVisionForImageClassification) (Data2VecVisionConfig model) - [DeiTConfig](/docs/transformers/v5.6.0/ja/model_doc/deit#transformers.DeiTConfig) configuration class: [DeiTForImageClassification](/docs/transformers/v5.6.0/ja/model_doc/deit#transformers.DeiTForImageClassification) or [DeiTForImageClassificationWithTeacher](/docs/transformers/v5.6.0/ja/model_doc/deit#transformers.DeiTForImageClassificationWithTeacher) (DeiTConfig model) - [DinatConfig](/docs/transformers/v5.6.0/ja/model_doc/dinat#transformers.DinatConfig) configuration class: [DinatForImageClassification](/docs/transformers/v5.6.0/ja/model_doc/dinat#transformers.DinatForImageClassification) (DinatConfig model) - `Dinov2Config` configuration class: `Dinov2ForImageClassification` (Dinov2Config model) - `Dinov2WithRegistersConfig` configuration class: `Dinov2WithRegistersForImageClassification` (Dinov2WithRegistersConfig model) - `DonutSwinConfig` configuration class: `DonutSwinForImageClassification` (DonutSwinConfig model) - `EfficientNetConfig` configuration class: `EfficientNetForImageClassification` (EfficientNetConfig model) - `FocalNetConfig` configuration class: `FocalNetForImageClassification` (FocalNetConfig model) - `HGNetV2Config` configuration class: `HGNetV2ForImageClassification` (HGNetV2Config model) - `HieraConfig` configuration class: `HieraForImageClassification` (HieraConfig model) - `IJepaConfig` configuration class: `IJepaForImageClassification` (IJepaConfig model) - `ImageGPTConfig` configuration class: `ImageGPTForImageClassification` (ImageGPTConfig model) - `LevitConfig` configuration class: `LevitForImageClassification` or `LevitForImageClassificationWithTeacher` (LevitConfig model) - `MetaClip2Config` configuration class: `MetaClip2ForImageClassification` (MetaClip2Config model) - `MobileNetV1Config` configuration class: `MobileNetV1ForImageClassification` (MobileNetV1Config model) - `MobileNetV2Config` configuration class: `MobileNetV2ForImageClassification` (MobileNetV2Config model) - `MobileViTConfig` configuration class: `MobileViTForImageClassification` (MobileViTConfig model) - `MobileViTV2Config` configuration class: `MobileViTV2ForImageClassification` (MobileViTV2Config model) - `PPLCNetConfig` configuration class: `PPLCNetForImageClassification` (PPLCNetConfig model) - `PerceiverConfig` configuration class: `PerceiverForImageClassificationLearned` or `PerceiverForImageClassificationFourier` or `PerceiverForImageClassificationConvProcessing` (PerceiverConfig model) - `PoolFormerConfig` configuration class: `PoolFormerForImageClassification` (PoolFormerConfig model) - `PvtConfig` configuration class: `PvtForImageClassification` (PvtConfig model) - `PvtV2Config` configuration class: `PvtV2ForImageClassification` (PvtV2Config model) - `RegNetConfig` configuration class: `RegNetForImageClassification` (RegNetConfig model) - `ResNetConfig` configuration class: `ResNetForImageClassification` (ResNetConfig model) - `SegformerConfig` configuration class: `SegformerForImageClassification` (SegformerConfig model) - `ShieldGemma2Config` configuration class: `ShieldGemma2ForImageClassification` (ShieldGemma2Config model) - `Siglip2Config` configuration class: `Siglip2ForImageClassification` (Siglip2Config model) - `SiglipConfig` configuration class: `SiglipForImageClassification` (SiglipConfig model) - `SwiftFormerConfig` configuration class: `SwiftFormerForImageClassification` (SwiftFormerConfig model) - `SwinConfig` configuration class: `SwinForImageClassification` (SwinConfig model) - `Swinv2Config` configuration class: `Swinv2ForImageClassification` (Swinv2Config model) - `TextNetConfig` configuration class: `TextNetForImageClassification` (TextNetConfig model) - `TimmWrapperConfig` configuration class: `TimmWrapperForImageClassification` (TimmWrapperConfig model) - `ViTConfig` configuration class: `ViTForImageClassification` (ViTConfig model) - `ViTMSNConfig` configuration class: `ViTMSNForImageClassification` (ViTMSNConfig model) attn_implementation (`str`, *optional*) : The attention implementation to use in the model (if relevant). Can be any of `"eager"` (manual implementation of the attention), `"sdpa"` (using [`F.scaled_dot_product_attention`](https://pytorch.org/docs/master/generated/torch.nn.functional.scaled_dot_product_attention.html)), `"flash_attention_2"` (using [Dao-AILab/flash-attention](https://github.com/Dao-AILab/flash-attention)), or `"flash_attention_3"` (using [Dao-AILab/flash-attention/hopper](https://github.com/Dao-AILab/flash-attention/tree/main/hopper)). By default, if available, SDPA will be used for torch>=2.1.1. The default is otherwise the manual `"eager"` implementation. #### from_pretrained[[transformers.AutoModelForImageClassification.from_pretrained]] [Source](https://github.com/huggingface/transformers/blob/v5.6.0/src/transformers/models/auto/auto_factory.py#L253) Instantiate one of the model classes of the library (with a image classification head) from a pretrained model. The model class to instantiate is selected based on the `model_type` property of the config object (either passed as an argument or loaded from `pretrained_model_name_or_path` if possible), or when it's missing, by falling back to using pattern matching on `pretrained_model_name_or_path`: - **beit** -- [BeitForImageClassification](/docs/transformers/v5.6.0/ja/model_doc/beit#transformers.BeitForImageClassification) (BeitConfig model) - **bit** -- [BitForImageClassification](/docs/transformers/v5.6.0/ja/model_doc/bit#transformers.BitForImageClassification) (BitConfig model) - **clip** -- `CLIPForImageClassification` (CLIPConfig model) - **convnext** -- [ConvNextForImageClassification](/docs/transformers/v5.6.0/ja/model_doc/convnext#transformers.ConvNextForImageClassification) (ConvNextConfig model) - **convnextv2** -- [ConvNextV2ForImageClassification](/docs/transformers/v5.6.0/ja/model_doc/convnextv2#transformers.ConvNextV2ForImageClassification) (ConvNextV2Config model) - **cvt** -- [CvtForImageClassification](/docs/transformers/v5.6.0/ja/model_doc/cvt#transformers.CvtForImageClassification) (CvtConfig model) - **data2vec-vision** -- [Data2VecVisionForImageClassification](/docs/transformers/v5.6.0/ja/model_doc/data2vec#transformers.Data2VecVisionForImageClassification) (Data2VecVisionConfig model) - **deit** -- [DeiTForImageClassification](/docs/transformers/v5.6.0/ja/model_doc/deit#transformers.DeiTForImageClassification) or [DeiTForImageClassificationWithTeacher](/docs/transformers/v5.6.0/ja/model_doc/deit#transformers.DeiTForImageClassificationWithTeacher) (DeiTConfig model) - **dinat** -- [DinatForImageClassification](/docs/transformers/v5.6.0/ja/model_doc/dinat#transformers.DinatForImageClassification) (DinatConfig model) - **dinov2** -- `Dinov2ForImageClassification` (Dinov2Config model) - **dinov2_with_registers** -- `Dinov2WithRegistersForImageClassification` (Dinov2WithRegistersConfig model) - **donut-swin** -- `DonutSwinForImageClassification` (DonutSwinConfig model) - **efficientnet** -- `EfficientNetForImageClassification` (EfficientNetConfig model) - **focalnet** -- `FocalNetForImageClassification` (FocalNetConfig model) - **hgnet_v2** -- `HGNetV2ForImageClassification` (HGNetV2Config model) - **hiera** -- `HieraForImageClassification` (HieraConfig model) - **ijepa** -- `IJepaForImageClassification` (IJepaConfig model) - **imagegpt** -- `ImageGPTForImageClassification` (ImageGPTConfig model) - **levit** -- `LevitForImageClassification` or `LevitForImageClassificationWithTeacher` (LevitConfig model) - **metaclip_2** -- `MetaClip2ForImageClassification` (MetaClip2Config model) - **mobilenet_v1** -- `MobileNetV1ForImageClassification` (MobileNetV1Config model) - **mobilenet_v2** -- `MobileNetV2ForImageClassification` (MobileNetV2Config model) - **mobilevit** -- `MobileViTForImageClassification` (MobileViTConfig model) - **mobilevitv2** -- `MobileViTV2ForImageClassification` (MobileViTV2Config model) - **perceiver** -- `PerceiverForImageClassificationLearned` or `PerceiverForImageClassificationFourier` or `PerceiverForImageClassificationConvProcessing` (PerceiverConfig model) - **poolformer** -- `PoolFormerForImageClassification` (PoolFormerConfig model) - **pp_lcnet** -- `PPLCNetForImageClassification` (PPLCNetConfig model) - **pvt** -- `PvtForImageClassification` (PvtConfig model) - **pvt_v2** -- `PvtV2ForImageClassification` (PvtV2Config model) - **regnet** -- `RegNetForImageClassification` (RegNetConfig model) - **resnet** -- `ResNetForImageClassification` (ResNetConfig model) - **segformer** -- `SegformerForImageClassification` (SegformerConfig model) - **shieldgemma2** -- `ShieldGemma2ForImageClassification` (ShieldGemma2Config model) - **siglip** -- `SiglipForImageClassification` (SiglipConfig model) - **siglip2** -- `Siglip2ForImageClassification` (Siglip2Config model) - **swiftformer** -- `SwiftFormerForImageClassification` (SwiftFormerConfig model) - **swin** -- `SwinForImageClassification` (SwinConfig model) - **swinv2** -- `Swinv2ForImageClassification` (Swinv2Config model) - **textnet** -- `TextNetForImageClassification` (TextNetConfig model) - **timm_wrapper** -- `TimmWrapperForImageClassification` (TimmWrapperConfig model) - **vit** -- `ViTForImageClassification` (ViTConfig model) - **vit_msn** -- `ViTMSNForImageClassification` (ViTMSNConfig model) The model is set in evaluation mode by default using `model.eval()` (so for instance, dropout modules are deactivated). To train the model, you should first set it back in training mode with `model.train()` Examples: ```python >>> from transformers import AutoConfig, AutoModelForImageClassification >>> # Download model and configuration from huggingface.co and cache. >>> model = AutoModelForImageClassification.from_pretrained("google-bert/bert-base-cased") >>> # Update configuration during loading >>> model = AutoModelForImageClassification.from_pretrained("google-bert/bert-base-cased", output_attentions=True) >>> model.config.output_attentions True ``` **Parameters:** pretrained_model_name_or_path (`str` or `os.PathLike`) : Can be either: - A string, the *model id* of a pretrained model hosted inside a model repo on huggingface.co. - A path to a *directory* containing model weights saved using [save_pretrained()](/docs/transformers/v5.6.0/ja/main_classes/model#transformers.PreTrainedModel.save_pretrained), e.g., `./my_model_directory/`. model_args (additional positional arguments, *optional*) : Will be passed along to the underlying model `__init__()` method. config ([PreTrainedConfig](/docs/transformers/v5.6.0/ja/main_classes/configuration#transformers.PreTrainedConfig), *optional*) : Configuration for the model to use instead of an automatically loaded configuration. Configuration can be automatically loaded when: - The model is a model provided by the library (loaded with the *model id* string of a pretrained model). - The model was saved using [save_pretrained()](/docs/transformers/v5.6.0/ja/main_classes/model#transformers.PreTrainedModel.save_pretrained) and is reloaded by supplying the save directory. - The model is loaded by supplying a local directory as `pretrained_model_name_or_path` and a configuration JSON file named *config.json* is found in the directory. state_dict (*dict[str, torch.Tensor]*, *optional*) : A state dictionary to use instead of a state dictionary loaded from saved weights file. This option can be used if you want to create a model from a pretrained configuration but load your own weights. In this case though, you should check if using [save_pretrained()](/docs/transformers/v5.6.0/ja/main_classes/model#transformers.PreTrainedModel.save_pretrained) and [from_pretrained()](/docs/transformers/v5.6.0/ja/main_classes/model#transformers.PreTrainedModel.from_pretrained) is not a simpler option. cache_dir (`str` or `os.PathLike`, *optional*) : Path to a directory in which a downloaded pretrained model configuration should be cached if the standard cache should not be used. force_download (`bool`, *optional*, defaults to `False`) : Whether or not to force the (re-)download of the model weights and configuration files, overriding the cached versions if they exist. proxies (`dict[str, str]`, *optional*) : A dictionary of proxy servers to use by protocol or endpoint, e.g., `{'http': 'foo.bar:3128', 'http://hostname': 'foo.bar:4012'}`. The proxies are used on each request. output_loading_info(`bool`, *optional*, defaults to `False`) : Whether or not to also return a dictionary containing missing keys, unexpected keys and error messages. local_files_only(`bool`, *optional*, defaults to `False`) : Whether or not to only look at local files (e.g., not try downloading the model). revision (`str`, *optional*, defaults to `"main"`) : The specific model version to use. It can be a branch name, a tag name, or a commit id, since we use a git-based system for storing models and other artifacts on huggingface.co, so `revision` can be any identifier allowed by git. trust_remote_code (`bool`, *optional*, defaults to `False`) : Whether or not to allow for custom models defined on the Hub in their own modeling files. This option should only be set to `True` for repositories you trust and in which you have read the code, as it will execute code present on the Hub on your local machine. code_revision (`str`, *optional*, defaults to `"main"`) : The specific revision to use for the code on the Hub, if the code leaves in a different repository than the rest of the model. It can be a branch name, a tag name, or a commit id, since we use a git-based system for storing models and other artifacts on huggingface.co, so `revision` can be any identifier allowed by git. kwargs (additional keyword arguments, *optional*) : Can be used to update the configuration object (after it being loaded) and initiate the model (e.g., `output_attentions=True`). Behaves differently depending on whether a `config` is provided or automatically loaded: - If a configuration is provided with `config`, `**kwargs` will be directly passed to the underlying model's `__init__` method (we assume all relevant updates to the configuration have already been done) - If a configuration is not provided, `kwargs` will be first passed to the configuration class initialization function ([from_pretrained()](/docs/transformers/v5.6.0/ja/main_classes/configuration#transformers.PreTrainedConfig.from_pretrained)). Each key of `kwargs` that corresponds to a configuration attribute will be used to override said attribute with the supplied `kwargs` value. Remaining keys that do not correspond to any configuration attribute will be passed to the underlying model's `__init__` function. ### AutoModelForVideoClassification[[transformers.AutoModelForVideoClassification]] #### transformers.AutoModelForVideoClassification[[transformers.AutoModelForVideoClassification]] [Source](https://github.com/huggingface/transformers/blob/v5.6.0/src/transformers/models/auto/modeling_auto.py#L2196) This is a generic model class that will be instantiated as one of the model classes of the library (with a video classification head) when created with the [from_pretrained()](/docs/transformers/v5.6.0/ja/model_doc/auto#transformers.AutoModel.from_pretrained) class method or the [from_config()](/docs/transformers/v5.6.0/ja/model_doc/auto#transformers.AutoModel.from_config) class method. This class cannot be instantiated directly using `__init__()` (throws an error). from_configtransformers.AutoModelForVideoClassification.from_confighttps://github.com/huggingface/transformers/blob/v5.6.0/src/transformers/models/auto/auto_factory.py#L206[{"name": "**kwargs", "val": ""}]- **config** ([PreTrainedConfig](/docs/transformers/v5.6.0/ja/main_classes/configuration#transformers.PreTrainedConfig)) -- The model class to instantiate is selected based on the configuration class: - `TimesformerConfig` configuration class: `TimesformerForVideoClassification` (TimesformerConfig model) - `VJEPA2Config` configuration class: `VJEPA2ForVideoClassification` (VJEPA2Config model) - `VideoMAEConfig` configuration class: `VideoMAEForVideoClassification` (VideoMAEConfig model) - `VivitConfig` configuration class: `VivitForVideoClassification` (VivitConfig model) - **attn_implementation** (`str`, *optional*) -- The attention implementation to use in the model (if relevant). Can be any of `"eager"` (manual implementation of the attention), `"sdpa"` (using [`F.scaled_dot_product_attention`](https://pytorch.org/docs/master/generated/torch.nn.functional.scaled_dot_product_attention.html)), `"flash_attention_2"` (using [Dao-AILab/flash-attention](https://github.com/Dao-AILab/flash-attention)), or `"flash_attention_3"` (using [Dao-AILab/flash-attention/hopper](https://github.com/Dao-AILab/flash-attention/tree/main/hopper)). By default, if available, SDPA will be used for torch>=2.1.1. The default is otherwise the manual `"eager"` implementation.0 Instantiates one of the model classes of the library (with a video classification head) from a configuration. Note: Loading a model from its configuration file does **not** load the model weights. It only affects the model's configuration. Use [from_pretrained()](/docs/transformers/v5.6.0/ja/model_doc/auto#transformers.AutoModel.from_pretrained) to load the model weights. Examples: ```python >>> from transformers import AutoConfig, AutoModelForVideoClassification >>> # Download configuration from huggingface.co and cache. >>> config = AutoConfig.from_pretrained("google-bert/bert-base-cased") >>> model = AutoModelForVideoClassification.from_config(config) ``` **Parameters:** config ([PreTrainedConfig](/docs/transformers/v5.6.0/ja/main_classes/configuration#transformers.PreTrainedConfig)) : The model class to instantiate is selected based on the configuration class: - `TimesformerConfig` configuration class: `TimesformerForVideoClassification` (TimesformerConfig model) - `VJEPA2Config` configuration class: `VJEPA2ForVideoClassification` (VJEPA2Config model) - `VideoMAEConfig` configuration class: `VideoMAEForVideoClassification` (VideoMAEConfig model) - `VivitConfig` configuration class: `VivitForVideoClassification` (VivitConfig model) attn_implementation (`str`, *optional*) : The attention implementation to use in the model (if relevant). Can be any of `"eager"` (manual implementation of the attention), `"sdpa"` (using [`F.scaled_dot_product_attention`](https://pytorch.org/docs/master/generated/torch.nn.functional.scaled_dot_product_attention.html)), `"flash_attention_2"` (using [Dao-AILab/flash-attention](https://github.com/Dao-AILab/flash-attention)), or `"flash_attention_3"` (using [Dao-AILab/flash-attention/hopper](https://github.com/Dao-AILab/flash-attention/tree/main/hopper)). By default, if available, SDPA will be used for torch>=2.1.1. The default is otherwise the manual `"eager"` implementation. #### from_pretrained[[transformers.AutoModelForVideoClassification.from_pretrained]] [Source](https://github.com/huggingface/transformers/blob/v5.6.0/src/transformers/models/auto/auto_factory.py#L253) Instantiate one of the model classes of the library (with a video classification head) from a pretrained model. The model class to instantiate is selected based on the `model_type` property of the config object (either passed as an argument or loaded from `pretrained_model_name_or_path` if possible), or when it's missing, by falling back to using pattern matching on `pretrained_model_name_or_path`: - **timesformer** -- `TimesformerForVideoClassification` (TimesformerConfig model) - **videomae** -- `VideoMAEForVideoClassification` (VideoMAEConfig model) - **vivit** -- `VivitForVideoClassification` (VivitConfig model) - **vjepa2** -- `VJEPA2ForVideoClassification` (VJEPA2Config model) The model is set in evaluation mode by default using `model.eval()` (so for instance, dropout modules are deactivated). To train the model, you should first set it back in training mode with `model.train()` Examples: ```python >>> from transformers import AutoConfig, AutoModelForVideoClassification >>> # Download model and configuration from huggingface.co and cache. >>> model = AutoModelForVideoClassification.from_pretrained("google-bert/bert-base-cased") >>> # Update configuration during loading >>> model = AutoModelForVideoClassification.from_pretrained("google-bert/bert-base-cased", output_attentions=True) >>> model.config.output_attentions True ``` **Parameters:** pretrained_model_name_or_path (`str` or `os.PathLike`) : Can be either: - A string, the *model id* of a pretrained model hosted inside a model repo on huggingface.co. - A path to a *directory* containing model weights saved using [save_pretrained()](/docs/transformers/v5.6.0/ja/main_classes/model#transformers.PreTrainedModel.save_pretrained), e.g., `./my_model_directory/`. model_args (additional positional arguments, *optional*) : Will be passed along to the underlying model `__init__()` method. config ([PreTrainedConfig](/docs/transformers/v5.6.0/ja/main_classes/configuration#transformers.PreTrainedConfig), *optional*) : Configuration for the model to use instead of an automatically loaded configuration. Configuration can be automatically loaded when: - The model is a model provided by the library (loaded with the *model id* string of a pretrained model). - The model was saved using [save_pretrained()](/docs/transformers/v5.6.0/ja/main_classes/model#transformers.PreTrainedModel.save_pretrained) and is reloaded by supplying the save directory. - The model is loaded by supplying a local directory as `pretrained_model_name_or_path` and a configuration JSON file named *config.json* is found in the directory. state_dict (*dict[str, torch.Tensor]*, *optional*) : A state dictionary to use instead of a state dictionary loaded from saved weights file. This option can be used if you want to create a model from a pretrained configuration but load your own weights. In this case though, you should check if using [save_pretrained()](/docs/transformers/v5.6.0/ja/main_classes/model#transformers.PreTrainedModel.save_pretrained) and [from_pretrained()](/docs/transformers/v5.6.0/ja/main_classes/model#transformers.PreTrainedModel.from_pretrained) is not a simpler option. cache_dir (`str` or `os.PathLike`, *optional*) : Path to a directory in which a downloaded pretrained model configuration should be cached if the standard cache should not be used. force_download (`bool`, *optional*, defaults to `False`) : Whether or not to force the (re-)download of the model weights and configuration files, overriding the cached versions if they exist. proxies (`dict[str, str]`, *optional*) : A dictionary of proxy servers to use by protocol or endpoint, e.g., `{'http': 'foo.bar:3128', 'http://hostname': 'foo.bar:4012'}`. The proxies are used on each request. output_loading_info(`bool`, *optional*, defaults to `False`) : Whether or not to also return a dictionary containing missing keys, unexpected keys and error messages. local_files_only(`bool`, *optional*, defaults to `False`) : Whether or not to only look at local files (e.g., not try downloading the model). revision (`str`, *optional*, defaults to `"main"`) : The specific model version to use. It can be a branch name, a tag name, or a commit id, since we use a git-based system for storing models and other artifacts on huggingface.co, so `revision` can be any identifier allowed by git. trust_remote_code (`bool`, *optional*, defaults to `False`) : Whether or not to allow for custom models defined on the Hub in their own modeling files. This option should only be set to `True` for repositories you trust and in which you have read the code, as it will execute code present on the Hub on your local machine. code_revision (`str`, *optional*, defaults to `"main"`) : The specific revision to use for the code on the Hub, if the code leaves in a different repository than the rest of the model. It can be a branch name, a tag name, or a commit id, since we use a git-based system for storing models and other artifacts on huggingface.co, so `revision` can be any identifier allowed by git. kwargs (additional keyword arguments, *optional*) : Can be used to update the configuration object (after it being loaded) and initiate the model (e.g., `output_attentions=True`). Behaves differently depending on whether a `config` is provided or automatically loaded: - If a configuration is provided with `config`, `**kwargs` will be directly passed to the underlying model's `__init__` method (we assume all relevant updates to the configuration have already been done) - If a configuration is not provided, `kwargs` will be first passed to the configuration class initialization function ([from_pretrained()](/docs/transformers/v5.6.0/ja/main_classes/configuration#transformers.PreTrainedConfig.from_pretrained)). Each key of `kwargs` that corresponds to a configuration attribute will be used to override said attribute with the supplied `kwargs` value. Remaining keys that do not correspond to any configuration attribute will be passed to the underlying model's `__init__` function. ### AutoModelForMaskedImageModeling[[transformers.AutoModelForMaskedImageModeling]] #### transformers.AutoModelForMaskedImageModeling[[transformers.AutoModelForMaskedImageModeling]] [Source](https://github.com/huggingface/transformers/blob/v5.6.0/src/transformers/models/auto/modeling_auto.py#L2278) This is a generic model class that will be instantiated as one of the model classes of the library (with a masked image modeling head) when created with the [from_pretrained()](/docs/transformers/v5.6.0/ja/model_doc/auto#transformers.AutoModel.from_pretrained) class method or the [from_config()](/docs/transformers/v5.6.0/ja/model_doc/auto#transformers.AutoModel.from_config) class method. This class cannot be instantiated directly using `__init__()` (throws an error). from_configtransformers.AutoModelForMaskedImageModeling.from_confighttps://github.com/huggingface/transformers/blob/v5.6.0/src/transformers/models/auto/auto_factory.py#L206[{"name": "**kwargs", "val": ""}]- **config** ([PreTrainedConfig](/docs/transformers/v5.6.0/ja/main_classes/configuration#transformers.PreTrainedConfig)) -- The model class to instantiate is selected based on the configuration class: - [DeiTConfig](/docs/transformers/v5.6.0/ja/model_doc/deit#transformers.DeiTConfig) configuration class: [DeiTForMaskedImageModeling](/docs/transformers/v5.6.0/ja/model_doc/deit#transformers.DeiTForMaskedImageModeling) (DeiTConfig model) - `FocalNetConfig` configuration class: `FocalNetForMaskedImageModeling` (FocalNetConfig model) - `SwinConfig` configuration class: `SwinForMaskedImageModeling` (SwinConfig model) - `Swinv2Config` configuration class: `Swinv2ForMaskedImageModeling` (Swinv2Config model) - `ViTConfig` configuration class: `ViTForMaskedImageModeling` (ViTConfig model) - **attn_implementation** (`str`, *optional*) -- The attention implementation to use in the model (if relevant). Can be any of `"eager"` (manual implementation of the attention), `"sdpa"` (using [`F.scaled_dot_product_attention`](https://pytorch.org/docs/master/generated/torch.nn.functional.scaled_dot_product_attention.html)), `"flash_attention_2"` (using [Dao-AILab/flash-attention](https://github.com/Dao-AILab/flash-attention)), or `"flash_attention_3"` (using [Dao-AILab/flash-attention/hopper](https://github.com/Dao-AILab/flash-attention/tree/main/hopper)). By default, if available, SDPA will be used for torch>=2.1.1. The default is otherwise the manual `"eager"` implementation.0 Instantiates one of the model classes of the library (with a masked image modeling head) from a configuration. Note: Loading a model from its configuration file does **not** load the model weights. It only affects the model's configuration. Use [from_pretrained()](/docs/transformers/v5.6.0/ja/model_doc/auto#transformers.AutoModel.from_pretrained) to load the model weights. Examples: ```python >>> from transformers import AutoConfig, AutoModelForMaskedImageModeling >>> # Download configuration from huggingface.co and cache. >>> config = AutoConfig.from_pretrained("google-bert/bert-base-cased") >>> model = AutoModelForMaskedImageModeling.from_config(config) ``` **Parameters:** config ([PreTrainedConfig](/docs/transformers/v5.6.0/ja/main_classes/configuration#transformers.PreTrainedConfig)) : The model class to instantiate is selected based on the configuration class: - [DeiTConfig](/docs/transformers/v5.6.0/ja/model_doc/deit#transformers.DeiTConfig) configuration class: [DeiTForMaskedImageModeling](/docs/transformers/v5.6.0/ja/model_doc/deit#transformers.DeiTForMaskedImageModeling) (DeiTConfig model) - `FocalNetConfig` configuration class: `FocalNetForMaskedImageModeling` (FocalNetConfig model) - `SwinConfig` configuration class: `SwinForMaskedImageModeling` (SwinConfig model) - `Swinv2Config` configuration class: `Swinv2ForMaskedImageModeling` (Swinv2Config model) - `ViTConfig` configuration class: `ViTForMaskedImageModeling` (ViTConfig model) attn_implementation (`str`, *optional*) : The attention implementation to use in the model (if relevant). Can be any of `"eager"` (manual implementation of the attention), `"sdpa"` (using [`F.scaled_dot_product_attention`](https://pytorch.org/docs/master/generated/torch.nn.functional.scaled_dot_product_attention.html)), `"flash_attention_2"` (using [Dao-AILab/flash-attention](https://github.com/Dao-AILab/flash-attention)), or `"flash_attention_3"` (using [Dao-AILab/flash-attention/hopper](https://github.com/Dao-AILab/flash-attention/tree/main/hopper)). By default, if available, SDPA will be used for torch>=2.1.1. The default is otherwise the manual `"eager"` implementation. #### from_pretrained[[transformers.AutoModelForMaskedImageModeling.from_pretrained]] [Source](https://github.com/huggingface/transformers/blob/v5.6.0/src/transformers/models/auto/auto_factory.py#L253) Instantiate one of the model classes of the library (with a masked image modeling head) from a pretrained model. The model class to instantiate is selected based on the `model_type` property of the config object (either passed as an argument or loaded from `pretrained_model_name_or_path` if possible), or when it's missing, by falling back to using pattern matching on `pretrained_model_name_or_path`: - **deit** -- [DeiTForMaskedImageModeling](/docs/transformers/v5.6.0/ja/model_doc/deit#transformers.DeiTForMaskedImageModeling) (DeiTConfig model) - **focalnet** -- `FocalNetForMaskedImageModeling` (FocalNetConfig model) - **swin** -- `SwinForMaskedImageModeling` (SwinConfig model) - **swinv2** -- `Swinv2ForMaskedImageModeling` (Swinv2Config model) - **vit** -- `ViTForMaskedImageModeling` (ViTConfig model) The model is set in evaluation mode by default using `model.eval()` (so for instance, dropout modules are deactivated). To train the model, you should first set it back in training mode with `model.train()` Examples: ```python >>> from transformers import AutoConfig, AutoModelForMaskedImageModeling >>> # Download model and configuration from huggingface.co and cache. >>> model = AutoModelForMaskedImageModeling.from_pretrained("google-bert/bert-base-cased") >>> # Update configuration during loading >>> model = AutoModelForMaskedImageModeling.from_pretrained("google-bert/bert-base-cased", output_attentions=True) >>> model.config.output_attentions True ``` **Parameters:** pretrained_model_name_or_path (`str` or `os.PathLike`) : Can be either: - A string, the *model id* of a pretrained model hosted inside a model repo on huggingface.co. - A path to a *directory* containing model weights saved using [save_pretrained()](/docs/transformers/v5.6.0/ja/main_classes/model#transformers.PreTrainedModel.save_pretrained), e.g., `./my_model_directory/`. model_args (additional positional arguments, *optional*) : Will be passed along to the underlying model `__init__()` method. config ([PreTrainedConfig](/docs/transformers/v5.6.0/ja/main_classes/configuration#transformers.PreTrainedConfig), *optional*) : Configuration for the model to use instead of an automatically loaded configuration. Configuration can be automatically loaded when: - The model is a model provided by the library (loaded with the *model id* string of a pretrained model). - The model was saved using [save_pretrained()](/docs/transformers/v5.6.0/ja/main_classes/model#transformers.PreTrainedModel.save_pretrained) and is reloaded by supplying the save directory. - The model is loaded by supplying a local directory as `pretrained_model_name_or_path` and a configuration JSON file named *config.json* is found in the directory. state_dict (*dict[str, torch.Tensor]*, *optional*) : A state dictionary to use instead of a state dictionary loaded from saved weights file. This option can be used if you want to create a model from a pretrained configuration but load your own weights. In this case though, you should check if using [save_pretrained()](/docs/transformers/v5.6.0/ja/main_classes/model#transformers.PreTrainedModel.save_pretrained) and [from_pretrained()](/docs/transformers/v5.6.0/ja/main_classes/model#transformers.PreTrainedModel.from_pretrained) is not a simpler option. cache_dir (`str` or `os.PathLike`, *optional*) : Path to a directory in which a downloaded pretrained model configuration should be cached if the standard cache should not be used. force_download (`bool`, *optional*, defaults to `False`) : Whether or not to force the (re-)download of the model weights and configuration files, overriding the cached versions if they exist. proxies (`dict[str, str]`, *optional*) : A dictionary of proxy servers to use by protocol or endpoint, e.g., `{'http': 'foo.bar:3128', 'http://hostname': 'foo.bar:4012'}`. The proxies are used on each request. output_loading_info(`bool`, *optional*, defaults to `False`) : Whether or not to also return a dictionary containing missing keys, unexpected keys and error messages. local_files_only(`bool`, *optional*, defaults to `False`) : Whether or not to only look at local files (e.g., not try downloading the model). revision (`str`, *optional*, defaults to `"main"`) : The specific model version to use. It can be a branch name, a tag name, or a commit id, since we use a git-based system for storing models and other artifacts on huggingface.co, so `revision` can be any identifier allowed by git. trust_remote_code (`bool`, *optional*, defaults to `False`) : Whether or not to allow for custom models defined on the Hub in their own modeling files. This option should only be set to `True` for repositories you trust and in which you have read the code, as it will execute code present on the Hub on your local machine. code_revision (`str`, *optional*, defaults to `"main"`) : The specific revision to use for the code on the Hub, if the code leaves in a different repository than the rest of the model. It can be a branch name, a tag name, or a commit id, since we use a git-based system for storing models and other artifacts on huggingface.co, so `revision` can be any identifier allowed by git. kwargs (additional keyword arguments, *optional*) : Can be used to update the configuration object (after it being loaded) and initiate the model (e.g., `output_attentions=True`). Behaves differently depending on whether a `config` is provided or automatically loaded: - If a configuration is provided with `config`, `**kwargs` will be directly passed to the underlying model's `__init__` method (we assume all relevant updates to the configuration have already been done) - If a configuration is not provided, `kwargs` will be first passed to the configuration class initialization function ([from_pretrained()](/docs/transformers/v5.6.0/ja/main_classes/configuration#transformers.PreTrainedConfig.from_pretrained)). Each key of `kwargs` that corresponds to a configuration attribute will be used to override said attribute with the supplied `kwargs` value. Remaining keys that do not correspond to any configuration attribute will be passed to the underlying model's `__init__` function. ### AutoModelForObjectDetection[[transformers.AutoModelForObjectDetection]] #### transformers.AutoModelForObjectDetection[[transformers.AutoModelForObjectDetection]] [Source](https://github.com/huggingface/transformers/blob/v5.6.0/src/transformers/models/auto/modeling_auto.py#L2159) This is a generic model class that will be instantiated as one of the model classes of the library (with a object detection head) when created with the [from_pretrained()](/docs/transformers/v5.6.0/ja/model_doc/auto#transformers.AutoModel.from_pretrained) class method or the [from_config()](/docs/transformers/v5.6.0/ja/model_doc/auto#transformers.AutoModel.from_config) class method. This class cannot be instantiated directly using `__init__()` (throws an error). from_configtransformers.AutoModelForObjectDetection.from_confighttps://github.com/huggingface/transformers/blob/v5.6.0/src/transformers/models/auto/auto_factory.py#L206[{"name": "**kwargs", "val": ""}]- **config** ([PreTrainedConfig](/docs/transformers/v5.6.0/ja/main_classes/configuration#transformers.PreTrainedConfig)) -- The model class to instantiate is selected based on the configuration class: - [ConditionalDetrConfig](/docs/transformers/v5.6.0/ja/model_doc/conditional_detr#transformers.ConditionalDetrConfig) configuration class: [ConditionalDetrForObjectDetection](/docs/transformers/v5.6.0/ja/model_doc/conditional_detr#transformers.ConditionalDetrForObjectDetection) (ConditionalDetrConfig model) - `DFineConfig` configuration class: `DFineForObjectDetection` (DFineConfig model) - `DabDetrConfig` configuration class: `DabDetrForObjectDetection` (DabDetrConfig model) - [DeformableDetrConfig](/docs/transformers/v5.6.0/ja/model_doc/deformable_detr#transformers.DeformableDetrConfig) configuration class: [DeformableDetrForObjectDetection](/docs/transformers/v5.6.0/ja/model_doc/deformable_detr#transformers.DeformableDetrForObjectDetection) (DeformableDetrConfig model) - [DetrConfig](/docs/transformers/v5.6.0/ja/model_doc/detr#transformers.DetrConfig) configuration class: [DetrForObjectDetection](/docs/transformers/v5.6.0/ja/model_doc/detr#transformers.DetrForObjectDetection) (DetrConfig model) - `LwDetrConfig` configuration class: `LwDetrForObjectDetection` (LwDetrConfig model) - `PPDocLayoutV2Config` configuration class: `PPDocLayoutV2ForObjectDetection` (PPDocLayoutV2Config model) - `PPDocLayoutV3Config` configuration class: `PPDocLayoutV3ForObjectDetection` (PPDocLayoutV3Config model) - `PPOCRV5MobileDetConfig` configuration class: `PPOCRV5MobileDetForObjectDetection` (PPOCRV5MobileDetConfig model) - `PPOCRV5ServerDetConfig` configuration class: `PPOCRV5ServerDetForObjectDetection` (PPOCRV5ServerDetConfig model) - `RTDetrConfig` configuration class: `RTDetrForObjectDetection` (RTDetrConfig model) - `RTDetrV2Config` configuration class: `RTDetrV2ForObjectDetection` (RTDetrV2Config model) - `TableTransformerConfig` configuration class: `TableTransformerForObjectDetection` (TableTransformerConfig model) - `YolosConfig` configuration class: `YolosForObjectDetection` (YolosConfig model) - **attn_implementation** (`str`, *optional*) -- The attention implementation to use in the model (if relevant). Can be any of `"eager"` (manual implementation of the attention), `"sdpa"` (using [`F.scaled_dot_product_attention`](https://pytorch.org/docs/master/generated/torch.nn.functional.scaled_dot_product_attention.html)), `"flash_attention_2"` (using [Dao-AILab/flash-attention](https://github.com/Dao-AILab/flash-attention)), or `"flash_attention_3"` (using [Dao-AILab/flash-attention/hopper](https://github.com/Dao-AILab/flash-attention/tree/main/hopper)). By default, if available, SDPA will be used for torch>=2.1.1. The default is otherwise the manual `"eager"` implementation.0 Instantiates one of the model classes of the library (with a object detection head) from a configuration. Note: Loading a model from its configuration file does **not** load the model weights. It only affects the model's configuration. Use [from_pretrained()](/docs/transformers/v5.6.0/ja/model_doc/auto#transformers.AutoModel.from_pretrained) to load the model weights. Examples: ```python >>> from transformers import AutoConfig, AutoModelForObjectDetection >>> # Download configuration from huggingface.co and cache. >>> config = AutoConfig.from_pretrained("google-bert/bert-base-cased") >>> model = AutoModelForObjectDetection.from_config(config) ``` **Parameters:** config ([PreTrainedConfig](/docs/transformers/v5.6.0/ja/main_classes/configuration#transformers.PreTrainedConfig)) : The model class to instantiate is selected based on the configuration class: - [ConditionalDetrConfig](/docs/transformers/v5.6.0/ja/model_doc/conditional_detr#transformers.ConditionalDetrConfig) configuration class: [ConditionalDetrForObjectDetection](/docs/transformers/v5.6.0/ja/model_doc/conditional_detr#transformers.ConditionalDetrForObjectDetection) (ConditionalDetrConfig model) - `DFineConfig` configuration class: `DFineForObjectDetection` (DFineConfig model) - `DabDetrConfig` configuration class: `DabDetrForObjectDetection` (DabDetrConfig model) - [DeformableDetrConfig](/docs/transformers/v5.6.0/ja/model_doc/deformable_detr#transformers.DeformableDetrConfig) configuration class: [DeformableDetrForObjectDetection](/docs/transformers/v5.6.0/ja/model_doc/deformable_detr#transformers.DeformableDetrForObjectDetection) (DeformableDetrConfig model) - [DetrConfig](/docs/transformers/v5.6.0/ja/model_doc/detr#transformers.DetrConfig) configuration class: [DetrForObjectDetection](/docs/transformers/v5.6.0/ja/model_doc/detr#transformers.DetrForObjectDetection) (DetrConfig model) - `LwDetrConfig` configuration class: `LwDetrForObjectDetection` (LwDetrConfig model) - `PPDocLayoutV2Config` configuration class: `PPDocLayoutV2ForObjectDetection` (PPDocLayoutV2Config model) - `PPDocLayoutV3Config` configuration class: `PPDocLayoutV3ForObjectDetection` (PPDocLayoutV3Config model) - `PPOCRV5MobileDetConfig` configuration class: `PPOCRV5MobileDetForObjectDetection` (PPOCRV5MobileDetConfig model) - `PPOCRV5ServerDetConfig` configuration class: `PPOCRV5ServerDetForObjectDetection` (PPOCRV5ServerDetConfig model) - `RTDetrConfig` configuration class: `RTDetrForObjectDetection` (RTDetrConfig model) - `RTDetrV2Config` configuration class: `RTDetrV2ForObjectDetection` (RTDetrV2Config model) - `TableTransformerConfig` configuration class: `TableTransformerForObjectDetection` (TableTransformerConfig model) - `YolosConfig` configuration class: `YolosForObjectDetection` (YolosConfig model) attn_implementation (`str`, *optional*) : The attention implementation to use in the model (if relevant). Can be any of `"eager"` (manual implementation of the attention), `"sdpa"` (using [`F.scaled_dot_product_attention`](https://pytorch.org/docs/master/generated/torch.nn.functional.scaled_dot_product_attention.html)), `"flash_attention_2"` (using [Dao-AILab/flash-attention](https://github.com/Dao-AILab/flash-attention)), or `"flash_attention_3"` (using [Dao-AILab/flash-attention/hopper](https://github.com/Dao-AILab/flash-attention/tree/main/hopper)). By default, if available, SDPA will be used for torch>=2.1.1. The default is otherwise the manual `"eager"` implementation. #### from_pretrained[[transformers.AutoModelForObjectDetection.from_pretrained]] [Source](https://github.com/huggingface/transformers/blob/v5.6.0/src/transformers/models/auto/auto_factory.py#L253) Instantiate one of the model classes of the library (with a object detection head) from a pretrained model. The model class to instantiate is selected based on the `model_type` property of the config object (either passed as an argument or loaded from `pretrained_model_name_or_path` if possible), or when it's missing, by falling back to using pattern matching on `pretrained_model_name_or_path`: - **conditional_detr** -- [ConditionalDetrForObjectDetection](/docs/transformers/v5.6.0/ja/model_doc/conditional_detr#transformers.ConditionalDetrForObjectDetection) (ConditionalDetrConfig model) - **d_fine** -- `DFineForObjectDetection` (DFineConfig model) - **dab-detr** -- `DabDetrForObjectDetection` (DabDetrConfig model) - **deformable_detr** -- [DeformableDetrForObjectDetection](/docs/transformers/v5.6.0/ja/model_doc/deformable_detr#transformers.DeformableDetrForObjectDetection) (DeformableDetrConfig model) - **detr** -- [DetrForObjectDetection](/docs/transformers/v5.6.0/ja/model_doc/detr#transformers.DetrForObjectDetection) (DetrConfig model) - **lw_detr** -- `LwDetrForObjectDetection` (LwDetrConfig model) - **pp_doclayout_v2** -- `PPDocLayoutV2ForObjectDetection` (PPDocLayoutV2Config model) - **pp_doclayout_v3** -- `PPDocLayoutV3ForObjectDetection` (PPDocLayoutV3Config model) - **pp_ocrv5_mobile_det** -- `PPOCRV5MobileDetForObjectDetection` (PPOCRV5MobileDetConfig model) - **pp_ocrv5_server_det** -- `PPOCRV5ServerDetForObjectDetection` (PPOCRV5ServerDetConfig model) - **rt_detr** -- `RTDetrForObjectDetection` (RTDetrConfig model) - **rt_detr_v2** -- `RTDetrV2ForObjectDetection` (RTDetrV2Config model) - **table-transformer** -- `TableTransformerForObjectDetection` (TableTransformerConfig model) - **yolos** -- `YolosForObjectDetection` (YolosConfig model) The model is set in evaluation mode by default using `model.eval()` (so for instance, dropout modules are deactivated). To train the model, you should first set it back in training mode with `model.train()` Examples: ```python >>> from transformers import AutoConfig, AutoModelForObjectDetection >>> # Download model and configuration from huggingface.co and cache. >>> model = AutoModelForObjectDetection.from_pretrained("google-bert/bert-base-cased") >>> # Update configuration during loading >>> model = AutoModelForObjectDetection.from_pretrained("google-bert/bert-base-cased", output_attentions=True) >>> model.config.output_attentions True ``` **Parameters:** pretrained_model_name_or_path (`str` or `os.PathLike`) : Can be either: - A string, the *model id* of a pretrained model hosted inside a model repo on huggingface.co. - A path to a *directory* containing model weights saved using [save_pretrained()](/docs/transformers/v5.6.0/ja/main_classes/model#transformers.PreTrainedModel.save_pretrained), e.g., `./my_model_directory/`. model_args (additional positional arguments, *optional*) : Will be passed along to the underlying model `__init__()` method. config ([PreTrainedConfig](/docs/transformers/v5.6.0/ja/main_classes/configuration#transformers.PreTrainedConfig), *optional*) : Configuration for the model to use instead of an automatically loaded configuration. Configuration can be automatically loaded when: - The model is a model provided by the library (loaded with the *model id* string of a pretrained model). - The model was saved using [save_pretrained()](/docs/transformers/v5.6.0/ja/main_classes/model#transformers.PreTrainedModel.save_pretrained) and is reloaded by supplying the save directory. - The model is loaded by supplying a local directory as `pretrained_model_name_or_path` and a configuration JSON file named *config.json* is found in the directory. state_dict (*dict[str, torch.Tensor]*, *optional*) : A state dictionary to use instead of a state dictionary loaded from saved weights file. This option can be used if you want to create a model from a pretrained configuration but load your own weights. In this case though, you should check if using [save_pretrained()](/docs/transformers/v5.6.0/ja/main_classes/model#transformers.PreTrainedModel.save_pretrained) and [from_pretrained()](/docs/transformers/v5.6.0/ja/main_classes/model#transformers.PreTrainedModel.from_pretrained) is not a simpler option. cache_dir (`str` or `os.PathLike`, *optional*) : Path to a directory in which a downloaded pretrained model configuration should be cached if the standard cache should not be used. force_download (`bool`, *optional*, defaults to `False`) : Whether or not to force the (re-)download of the model weights and configuration files, overriding the cached versions if they exist. proxies (`dict[str, str]`, *optional*) : A dictionary of proxy servers to use by protocol or endpoint, e.g., `{'http': 'foo.bar:3128', 'http://hostname': 'foo.bar:4012'}`. The proxies are used on each request. output_loading_info(`bool`, *optional*, defaults to `False`) : Whether or not to also return a dictionary containing missing keys, unexpected keys and error messages. local_files_only(`bool`, *optional*, defaults to `False`) : Whether or not to only look at local files (e.g., not try downloading the model). revision (`str`, *optional*, defaults to `"main"`) : The specific model version to use. It can be a branch name, a tag name, or a commit id, since we use a git-based system for storing models and other artifacts on huggingface.co, so `revision` can be any identifier allowed by git. trust_remote_code (`bool`, *optional*, defaults to `False`) : Whether or not to allow for custom models defined on the Hub in their own modeling files. This option should only be set to `True` for repositories you trust and in which you have read the code, as it will execute code present on the Hub on your local machine. code_revision (`str`, *optional*, defaults to `"main"`) : The specific revision to use for the code on the Hub, if the code leaves in a different repository than the rest of the model. It can be a branch name, a tag name, or a commit id, since we use a git-based system for storing models and other artifacts on huggingface.co, so `revision` can be any identifier allowed by git. kwargs (additional keyword arguments, *optional*) : Can be used to update the configuration object (after it being loaded) and initiate the model (e.g., `output_attentions=True`). Behaves differently depending on whether a `config` is provided or automatically loaded: - If a configuration is provided with `config`, `**kwargs` will be directly passed to the underlying model's `__init__` method (we assume all relevant updates to the configuration have already been done) - If a configuration is not provided, `kwargs` will be first passed to the configuration class initialization function ([from_pretrained()](/docs/transformers/v5.6.0/ja/main_classes/configuration#transformers.PreTrainedConfig.from_pretrained)). Each key of `kwargs` that corresponds to a configuration attribute will be used to override said attribute with the supplied `kwargs` value. Remaining keys that do not correspond to any configuration attribute will be passed to the underlying model's `__init__` function. ### AutoModelForImageSegmentation[[transformers.AutoModelForImageSegmentation]] #### transformers.AutoModelForImageSegmentation[[transformers.AutoModelForImageSegmentation]] [Source](https://github.com/huggingface/transformers/blob/v5.6.0/src/transformers/models/auto/modeling_auto.py#L2116) This is a generic model class that will be instantiated as one of the model classes of the library (with a image segmentation head) when created with the [from_pretrained()](/docs/transformers/v5.6.0/ja/model_doc/auto#transformers.AutoModel.from_pretrained) class method or the [from_config()](/docs/transformers/v5.6.0/ja/model_doc/auto#transformers.AutoModel.from_config) class method. This class cannot be instantiated directly using `__init__()` (throws an error). from_configtransformers.AutoModelForImageSegmentation.from_confighttps://github.com/huggingface/transformers/blob/v5.6.0/src/transformers/models/auto/auto_factory.py#L206[{"name": "**kwargs", "val": ""}]- **config** ([PreTrainedConfig](/docs/transformers/v5.6.0/ja/main_classes/configuration#transformers.PreTrainedConfig)) -- The model class to instantiate is selected based on the configuration class: - [DetrConfig](/docs/transformers/v5.6.0/ja/model_doc/detr#transformers.DetrConfig) configuration class: [DetrForSegmentation](/docs/transformers/v5.6.0/ja/model_doc/detr#transformers.DetrForSegmentation) (DetrConfig model) - **attn_implementation** (`str`, *optional*) -- The attention implementation to use in the model (if relevant). Can be any of `"eager"` (manual implementation of the attention), `"sdpa"` (using [`F.scaled_dot_product_attention`](https://pytorch.org/docs/master/generated/torch.nn.functional.scaled_dot_product_attention.html)), `"flash_attention_2"` (using [Dao-AILab/flash-attention](https://github.com/Dao-AILab/flash-attention)), or `"flash_attention_3"` (using [Dao-AILab/flash-attention/hopper](https://github.com/Dao-AILab/flash-attention/tree/main/hopper)). By default, if available, SDPA will be used for torch>=2.1.1. The default is otherwise the manual `"eager"` implementation.0 Instantiates one of the model classes of the library (with a image segmentation head) from a configuration. Note: Loading a model from its configuration file does **not** load the model weights. It only affects the model's configuration. Use [from_pretrained()](/docs/transformers/v5.6.0/ja/model_doc/auto#transformers.AutoModel.from_pretrained) to load the model weights. Examples: ```python >>> from transformers import AutoConfig, AutoModelForImageSegmentation >>> # Download configuration from huggingface.co and cache. >>> config = AutoConfig.from_pretrained("google-bert/bert-base-cased") >>> model = AutoModelForImageSegmentation.from_config(config) ``` **Parameters:** config ([PreTrainedConfig](/docs/transformers/v5.6.0/ja/main_classes/configuration#transformers.PreTrainedConfig)) : The model class to instantiate is selected based on the configuration class: - [DetrConfig](/docs/transformers/v5.6.0/ja/model_doc/detr#transformers.DetrConfig) configuration class: [DetrForSegmentation](/docs/transformers/v5.6.0/ja/model_doc/detr#transformers.DetrForSegmentation) (DetrConfig model) attn_implementation (`str`, *optional*) : The attention implementation to use in the model (if relevant). Can be any of `"eager"` (manual implementation of the attention), `"sdpa"` (using [`F.scaled_dot_product_attention`](https://pytorch.org/docs/master/generated/torch.nn.functional.scaled_dot_product_attention.html)), `"flash_attention_2"` (using [Dao-AILab/flash-attention](https://github.com/Dao-AILab/flash-attention)), or `"flash_attention_3"` (using [Dao-AILab/flash-attention/hopper](https://github.com/Dao-AILab/flash-attention/tree/main/hopper)). By default, if available, SDPA will be used for torch>=2.1.1. The default is otherwise the manual `"eager"` implementation. #### from_pretrained[[transformers.AutoModelForImageSegmentation.from_pretrained]] [Source](https://github.com/huggingface/transformers/blob/v5.6.0/src/transformers/models/auto/auto_factory.py#L253) Instantiate one of the model classes of the library (with a image segmentation head) from a pretrained model. The model class to instantiate is selected based on the `model_type` property of the config object (either passed as an argument or loaded from `pretrained_model_name_or_path` if possible), or when it's missing, by falling back to using pattern matching on `pretrained_model_name_or_path`: - **detr** -- [DetrForSegmentation](/docs/transformers/v5.6.0/ja/model_doc/detr#transformers.DetrForSegmentation) (DetrConfig model) The model is set in evaluation mode by default using `model.eval()` (so for instance, dropout modules are deactivated). To train the model, you should first set it back in training mode with `model.train()` Examples: ```python >>> from transformers import AutoConfig, AutoModelForImageSegmentation >>> # Download model and configuration from huggingface.co and cache. >>> model = AutoModelForImageSegmentation.from_pretrained("google-bert/bert-base-cased") >>> # Update configuration during loading >>> model = AutoModelForImageSegmentation.from_pretrained("google-bert/bert-base-cased", output_attentions=True) >>> model.config.output_attentions True ``` **Parameters:** pretrained_model_name_or_path (`str` or `os.PathLike`) : Can be either: - A string, the *model id* of a pretrained model hosted inside a model repo on huggingface.co. - A path to a *directory* containing model weights saved using [save_pretrained()](/docs/transformers/v5.6.0/ja/main_classes/model#transformers.PreTrainedModel.save_pretrained), e.g., `./my_model_directory/`. model_args (additional positional arguments, *optional*) : Will be passed along to the underlying model `__init__()` method. config ([PreTrainedConfig](/docs/transformers/v5.6.0/ja/main_classes/configuration#transformers.PreTrainedConfig), *optional*) : Configuration for the model to use instead of an automatically loaded configuration. Configuration can be automatically loaded when: - The model is a model provided by the library (loaded with the *model id* string of a pretrained model). - The model was saved using [save_pretrained()](/docs/transformers/v5.6.0/ja/main_classes/model#transformers.PreTrainedModel.save_pretrained) and is reloaded by supplying the save directory. - The model is loaded by supplying a local directory as `pretrained_model_name_or_path` and a configuration JSON file named *config.json* is found in the directory. state_dict (*dict[str, torch.Tensor]*, *optional*) : A state dictionary to use instead of a state dictionary loaded from saved weights file. This option can be used if you want to create a model from a pretrained configuration but load your own weights. In this case though, you should check if using [save_pretrained()](/docs/transformers/v5.6.0/ja/main_classes/model#transformers.PreTrainedModel.save_pretrained) and [from_pretrained()](/docs/transformers/v5.6.0/ja/main_classes/model#transformers.PreTrainedModel.from_pretrained) is not a simpler option. cache_dir (`str` or `os.PathLike`, *optional*) : Path to a directory in which a downloaded pretrained model configuration should be cached if the standard cache should not be used. force_download (`bool`, *optional*, defaults to `False`) : Whether or not to force the (re-)download of the model weights and configuration files, overriding the cached versions if they exist. proxies (`dict[str, str]`, *optional*) : A dictionary of proxy servers to use by protocol or endpoint, e.g., `{'http': 'foo.bar:3128', 'http://hostname': 'foo.bar:4012'}`. The proxies are used on each request. output_loading_info(`bool`, *optional*, defaults to `False`) : Whether or not to also return a dictionary containing missing keys, unexpected keys and error messages. local_files_only(`bool`, *optional*, defaults to `False`) : Whether or not to only look at local files (e.g., not try downloading the model). revision (`str`, *optional*, defaults to `"main"`) : The specific model version to use. It can be a branch name, a tag name, or a commit id, since we use a git-based system for storing models and other artifacts on huggingface.co, so `revision` can be any identifier allowed by git. trust_remote_code (`bool`, *optional*, defaults to `False`) : Whether or not to allow for custom models defined on the Hub in their own modeling files. This option should only be set to `True` for repositories you trust and in which you have read the code, as it will execute code present on the Hub on your local machine. code_revision (`str`, *optional*, defaults to `"main"`) : The specific revision to use for the code on the Hub, if the code leaves in a different repository than the rest of the model. It can be a branch name, a tag name, or a commit id, since we use a git-based system for storing models and other artifacts on huggingface.co, so `revision` can be any identifier allowed by git. kwargs (additional keyword arguments, *optional*) : Can be used to update the configuration object (after it being loaded) and initiate the model (e.g., `output_attentions=True`). Behaves differently depending on whether a `config` is provided or automatically loaded: - If a configuration is provided with `config`, `**kwargs` will be directly passed to the underlying model's `__init__` method (we assume all relevant updates to the configuration have already been done) - If a configuration is not provided, `kwargs` will be first passed to the configuration class initialization function ([from_pretrained()](/docs/transformers/v5.6.0/ja/main_classes/configuration#transformers.PreTrainedConfig.from_pretrained)). Each key of `kwargs` that corresponds to a configuration attribute will be used to override said attribute with the supplied `kwargs` value. Remaining keys that do not correspond to any configuration attribute will be passed to the underlying model's `__init__` function. ### AutoModelForImageToImage[[transformers.AutoModelForImageToImage]] #### transformers.AutoModelForImageToImage[[transformers.AutoModelForImageToImage]] [Source](https://github.com/huggingface/transformers/blob/v5.6.0/src/transformers/models/auto/modeling_auto.py#L1975) ### AutoModelForSemanticSegmentation[[transformers.AutoModelForSemanticSegmentation]] #### transformers.AutoModelForSemanticSegmentation[[transformers.AutoModelForSemanticSegmentation]] [Source](https://github.com/huggingface/transformers/blob/v5.6.0/src/transformers/models/auto/modeling_auto.py#L2123) This is a generic model class that will be instantiated as one of the model classes of the library (with a semantic segmentation head) when created with the [from_pretrained()](/docs/transformers/v5.6.0/ja/model_doc/auto#transformers.AutoModel.from_pretrained) class method or the [from_config()](/docs/transformers/v5.6.0/ja/model_doc/auto#transformers.AutoModel.from_config) class method. This class cannot be instantiated directly using `__init__()` (throws an error). from_configtransformers.AutoModelForSemanticSegmentation.from_confighttps://github.com/huggingface/transformers/blob/v5.6.0/src/transformers/models/auto/auto_factory.py#L206[{"name": "**kwargs", "val": ""}]- **config** ([PreTrainedConfig](/docs/transformers/v5.6.0/ja/main_classes/configuration#transformers.PreTrainedConfig)) -- The model class to instantiate is selected based on the configuration class: - [BeitConfig](/docs/transformers/v5.6.0/ja/model_doc/beit#transformers.BeitConfig) configuration class: [BeitForSemanticSegmentation](/docs/transformers/v5.6.0/ja/model_doc/beit#transformers.BeitForSemanticSegmentation) (BeitConfig model) - `DPTConfig` configuration class: `DPTForSemanticSegmentation` (DPTConfig model) - [Data2VecVisionConfig](/docs/transformers/v5.6.0/ja/model_doc/data2vec#transformers.Data2VecVisionConfig) configuration class: [Data2VecVisionForSemanticSegmentation](/docs/transformers/v5.6.0/ja/model_doc/data2vec#transformers.Data2VecVisionForSemanticSegmentation) (Data2VecVisionConfig model) - `MobileNetV2Config` configuration class: `MobileNetV2ForSemanticSegmentation` (MobileNetV2Config model) - `MobileViTConfig` configuration class: `MobileViTForSemanticSegmentation` (MobileViTConfig model) - `MobileViTV2Config` configuration class: `MobileViTV2ForSemanticSegmentation` (MobileViTV2Config model) - `SegformerConfig` configuration class: `SegformerForSemanticSegmentation` (SegformerConfig model) - `UperNetConfig` configuration class: `UperNetForSemanticSegmentation` (UperNetConfig model) - **attn_implementation** (`str`, *optional*) -- The attention implementation to use in the model (if relevant). Can be any of `"eager"` (manual implementation of the attention), `"sdpa"` (using [`F.scaled_dot_product_attention`](https://pytorch.org/docs/master/generated/torch.nn.functional.scaled_dot_product_attention.html)), `"flash_attention_2"` (using [Dao-AILab/flash-attention](https://github.com/Dao-AILab/flash-attention)), or `"flash_attention_3"` (using [Dao-AILab/flash-attention/hopper](https://github.com/Dao-AILab/flash-attention/tree/main/hopper)). By default, if available, SDPA will be used for torch>=2.1.1. The default is otherwise the manual `"eager"` implementation.0 Instantiates one of the model classes of the library (with a semantic segmentation head) from a configuration. Note: Loading a model from its configuration file does **not** load the model weights. It only affects the model's configuration. Use [from_pretrained()](/docs/transformers/v5.6.0/ja/model_doc/auto#transformers.AutoModel.from_pretrained) to load the model weights. Examples: ```python >>> from transformers import AutoConfig, AutoModelForSemanticSegmentation >>> # Download configuration from huggingface.co and cache. >>> config = AutoConfig.from_pretrained("google-bert/bert-base-cased") >>> model = AutoModelForSemanticSegmentation.from_config(config) ``` **Parameters:** config ([PreTrainedConfig](/docs/transformers/v5.6.0/ja/main_classes/configuration#transformers.PreTrainedConfig)) : The model class to instantiate is selected based on the configuration class: - [BeitConfig](/docs/transformers/v5.6.0/ja/model_doc/beit#transformers.BeitConfig) configuration class: [BeitForSemanticSegmentation](/docs/transformers/v5.6.0/ja/model_doc/beit#transformers.BeitForSemanticSegmentation) (BeitConfig model) - `DPTConfig` configuration class: `DPTForSemanticSegmentation` (DPTConfig model) - [Data2VecVisionConfig](/docs/transformers/v5.6.0/ja/model_doc/data2vec#transformers.Data2VecVisionConfig) configuration class: [Data2VecVisionForSemanticSegmentation](/docs/transformers/v5.6.0/ja/model_doc/data2vec#transformers.Data2VecVisionForSemanticSegmentation) (Data2VecVisionConfig model) - `MobileNetV2Config` configuration class: `MobileNetV2ForSemanticSegmentation` (MobileNetV2Config model) - `MobileViTConfig` configuration class: `MobileViTForSemanticSegmentation` (MobileViTConfig model) - `MobileViTV2Config` configuration class: `MobileViTV2ForSemanticSegmentation` (MobileViTV2Config model) - `SegformerConfig` configuration class: `SegformerForSemanticSegmentation` (SegformerConfig model) - `UperNetConfig` configuration class: `UperNetForSemanticSegmentation` (UperNetConfig model) attn_implementation (`str`, *optional*) : The attention implementation to use in the model (if relevant). Can be any of `"eager"` (manual implementation of the attention), `"sdpa"` (using [`F.scaled_dot_product_attention`](https://pytorch.org/docs/master/generated/torch.nn.functional.scaled_dot_product_attention.html)), `"flash_attention_2"` (using [Dao-AILab/flash-attention](https://github.com/Dao-AILab/flash-attention)), or `"flash_attention_3"` (using [Dao-AILab/flash-attention/hopper](https://github.com/Dao-AILab/flash-attention/tree/main/hopper)). By default, if available, SDPA will be used for torch>=2.1.1. The default is otherwise the manual `"eager"` implementation. #### from_pretrained[[transformers.AutoModelForSemanticSegmentation.from_pretrained]] [Source](https://github.com/huggingface/transformers/blob/v5.6.0/src/transformers/models/auto/auto_factory.py#L253) Instantiate one of the model classes of the library (with a semantic segmentation head) from a pretrained model. The model class to instantiate is selected based on the `model_type` property of the config object (either passed as an argument or loaded from `pretrained_model_name_or_path` if possible), or when it's missing, by falling back to using pattern matching on `pretrained_model_name_or_path`: - **beit** -- [BeitForSemanticSegmentation](/docs/transformers/v5.6.0/ja/model_doc/beit#transformers.BeitForSemanticSegmentation) (BeitConfig model) - **data2vec-vision** -- [Data2VecVisionForSemanticSegmentation](/docs/transformers/v5.6.0/ja/model_doc/data2vec#transformers.Data2VecVisionForSemanticSegmentation) (Data2VecVisionConfig model) - **dpt** -- `DPTForSemanticSegmentation` (DPTConfig model) - **mobilenet_v2** -- `MobileNetV2ForSemanticSegmentation` (MobileNetV2Config model) - **mobilevit** -- `MobileViTForSemanticSegmentation` (MobileViTConfig model) - **mobilevitv2** -- `MobileViTV2ForSemanticSegmentation` (MobileViTV2Config model) - **segformer** -- `SegformerForSemanticSegmentation` (SegformerConfig model) - **upernet** -- `UperNetForSemanticSegmentation` (UperNetConfig model) The model is set in evaluation mode by default using `model.eval()` (so for instance, dropout modules are deactivated). To train the model, you should first set it back in training mode with `model.train()` Examples: ```python >>> from transformers import AutoConfig, AutoModelForSemanticSegmentation >>> # Download model and configuration from huggingface.co and cache. >>> model = AutoModelForSemanticSegmentation.from_pretrained("google-bert/bert-base-cased") >>> # Update configuration during loading >>> model = AutoModelForSemanticSegmentation.from_pretrained("google-bert/bert-base-cased", output_attentions=True) >>> model.config.output_attentions True ``` **Parameters:** pretrained_model_name_or_path (`str` or `os.PathLike`) : Can be either: - A string, the *model id* of a pretrained model hosted inside a model repo on huggingface.co. - A path to a *directory* containing model weights saved using [save_pretrained()](/docs/transformers/v5.6.0/ja/main_classes/model#transformers.PreTrainedModel.save_pretrained), e.g., `./my_model_directory/`. model_args (additional positional arguments, *optional*) : Will be passed along to the underlying model `__init__()` method. config ([PreTrainedConfig](/docs/transformers/v5.6.0/ja/main_classes/configuration#transformers.PreTrainedConfig), *optional*) : Configuration for the model to use instead of an automatically loaded configuration. Configuration can be automatically loaded when: - The model is a model provided by the library (loaded with the *model id* string of a pretrained model). - The model was saved using [save_pretrained()](/docs/transformers/v5.6.0/ja/main_classes/model#transformers.PreTrainedModel.save_pretrained) and is reloaded by supplying the save directory. - The model is loaded by supplying a local directory as `pretrained_model_name_or_path` and a configuration JSON file named *config.json* is found in the directory. state_dict (*dict[str, torch.Tensor]*, *optional*) : A state dictionary to use instead of a state dictionary loaded from saved weights file. This option can be used if you want to create a model from a pretrained configuration but load your own weights. In this case though, you should check if using [save_pretrained()](/docs/transformers/v5.6.0/ja/main_classes/model#transformers.PreTrainedModel.save_pretrained) and [from_pretrained()](/docs/transformers/v5.6.0/ja/main_classes/model#transformers.PreTrainedModel.from_pretrained) is not a simpler option. cache_dir (`str` or `os.PathLike`, *optional*) : Path to a directory in which a downloaded pretrained model configuration should be cached if the standard cache should not be used. force_download (`bool`, *optional*, defaults to `False`) : Whether or not to force the (re-)download of the model weights and configuration files, overriding the cached versions if they exist. proxies (`dict[str, str]`, *optional*) : A dictionary of proxy servers to use by protocol or endpoint, e.g., `{'http': 'foo.bar:3128', 'http://hostname': 'foo.bar:4012'}`. The proxies are used on each request. output_loading_info(`bool`, *optional*, defaults to `False`) : Whether or not to also return a dictionary containing missing keys, unexpected keys and error messages. local_files_only(`bool`, *optional*, defaults to `False`) : Whether or not to only look at local files (e.g., not try downloading the model). revision (`str`, *optional*, defaults to `"main"`) : The specific model version to use. It can be a branch name, a tag name, or a commit id, since we use a git-based system for storing models and other artifacts on huggingface.co, so `revision` can be any identifier allowed by git. trust_remote_code (`bool`, *optional*, defaults to `False`) : Whether or not to allow for custom models defined on the Hub in their own modeling files. This option should only be set to `True` for repositories you trust and in which you have read the code, as it will execute code present on the Hub on your local machine. code_revision (`str`, *optional*, defaults to `"main"`) : The specific revision to use for the code on the Hub, if the code leaves in a different repository than the rest of the model. It can be a branch name, a tag name, or a commit id, since we use a git-based system for storing models and other artifacts on huggingface.co, so `revision` can be any identifier allowed by git. kwargs (additional keyword arguments, *optional*) : Can be used to update the configuration object (after it being loaded) and initiate the model (e.g., `output_attentions=True`). Behaves differently depending on whether a `config` is provided or automatically loaded: - If a configuration is provided with `config`, `**kwargs` will be directly passed to the underlying model's `__init__` method (we assume all relevant updates to the configuration have already been done) - If a configuration is not provided, `kwargs` will be first passed to the configuration class initialization function ([from_pretrained()](/docs/transformers/v5.6.0/ja/main_classes/configuration#transformers.PreTrainedConfig.from_pretrained)). Each key of `kwargs` that corresponds to a configuration attribute will be used to override said attribute with the supplied `kwargs` value. Remaining keys that do not correspond to any configuration attribute will be passed to the underlying model's `__init__` function. ### AutoModelForInstanceSegmentation[[transformers.AutoModelForInstanceSegmentation]] #### transformers.AutoModelForInstanceSegmentation[[transformers.AutoModelForInstanceSegmentation]] [Source](https://github.com/huggingface/transformers/blob/v5.6.0/src/transformers/models/auto/modeling_auto.py#L2150) This is a generic model class that will be instantiated as one of the model classes of the library (with a instance segmentation head) when created with the [from_pretrained()](/docs/transformers/v5.6.0/ja/model_doc/auto#transformers.AutoModel.from_pretrained) class method or the [from_config()](/docs/transformers/v5.6.0/ja/model_doc/auto#transformers.AutoModel.from_config) class method. This class cannot be instantiated directly using `__init__()` (throws an error). from_configtransformers.AutoModelForInstanceSegmentation.from_confighttps://github.com/huggingface/transformers/blob/v5.6.0/src/transformers/models/auto/auto_factory.py#L206[{"name": "**kwargs", "val": ""}]- **config** ([PreTrainedConfig](/docs/transformers/v5.6.0/ja/main_classes/configuration#transformers.PreTrainedConfig)) -- The model class to instantiate is selected based on the configuration class: - `MaskFormerConfig` configuration class: `MaskFormerForInstanceSegmentation` (MaskFormerConfig model) - **attn_implementation** (`str`, *optional*) -- The attention implementation to use in the model (if relevant). Can be any of `"eager"` (manual implementation of the attention), `"sdpa"` (using [`F.scaled_dot_product_attention`](https://pytorch.org/docs/master/generated/torch.nn.functional.scaled_dot_product_attention.html)), `"flash_attention_2"` (using [Dao-AILab/flash-attention](https://github.com/Dao-AILab/flash-attention)), or `"flash_attention_3"` (using [Dao-AILab/flash-attention/hopper](https://github.com/Dao-AILab/flash-attention/tree/main/hopper)). By default, if available, SDPA will be used for torch>=2.1.1. The default is otherwise the manual `"eager"` implementation.0 Instantiates one of the model classes of the library (with a instance segmentation head) from a configuration. Note: Loading a model from its configuration file does **not** load the model weights. It only affects the model's configuration. Use [from_pretrained()](/docs/transformers/v5.6.0/ja/model_doc/auto#transformers.AutoModel.from_pretrained) to load the model weights. Examples: ```python >>> from transformers import AutoConfig, AutoModelForInstanceSegmentation >>> # Download configuration from huggingface.co and cache. >>> config = AutoConfig.from_pretrained("google-bert/bert-base-cased") >>> model = AutoModelForInstanceSegmentation.from_config(config) ``` **Parameters:** config ([PreTrainedConfig](/docs/transformers/v5.6.0/ja/main_classes/configuration#transformers.PreTrainedConfig)) : The model class to instantiate is selected based on the configuration class: - `MaskFormerConfig` configuration class: `MaskFormerForInstanceSegmentation` (MaskFormerConfig model) attn_implementation (`str`, *optional*) : The attention implementation to use in the model (if relevant). Can be any of `"eager"` (manual implementation of the attention), `"sdpa"` (using [`F.scaled_dot_product_attention`](https://pytorch.org/docs/master/generated/torch.nn.functional.scaled_dot_product_attention.html)), `"flash_attention_2"` (using [Dao-AILab/flash-attention](https://github.com/Dao-AILab/flash-attention)), or `"flash_attention_3"` (using [Dao-AILab/flash-attention/hopper](https://github.com/Dao-AILab/flash-attention/tree/main/hopper)). By default, if available, SDPA will be used for torch>=2.1.1. The default is otherwise the manual `"eager"` implementation. #### from_pretrained[[transformers.AutoModelForInstanceSegmentation.from_pretrained]] [Source](https://github.com/huggingface/transformers/blob/v5.6.0/src/transformers/models/auto/auto_factory.py#L253) Instantiate one of the model classes of the library (with a instance segmentation head) from a pretrained model. The model class to instantiate is selected based on the `model_type` property of the config object (either passed as an argument or loaded from `pretrained_model_name_or_path` if possible), or when it's missing, by falling back to using pattern matching on `pretrained_model_name_or_path`: - **maskformer** -- `MaskFormerForInstanceSegmentation` (MaskFormerConfig model) The model is set in evaluation mode by default using `model.eval()` (so for instance, dropout modules are deactivated). To train the model, you should first set it back in training mode with `model.train()` Examples: ```python >>> from transformers import AutoConfig, AutoModelForInstanceSegmentation >>> # Download model and configuration from huggingface.co and cache. >>> model = AutoModelForInstanceSegmentation.from_pretrained("google-bert/bert-base-cased") >>> # Update configuration during loading >>> model = AutoModelForInstanceSegmentation.from_pretrained("google-bert/bert-base-cased", output_attentions=True) >>> model.config.output_attentions True ``` **Parameters:** pretrained_model_name_or_path (`str` or `os.PathLike`) : Can be either: - A string, the *model id* of a pretrained model hosted inside a model repo on huggingface.co. - A path to a *directory* containing model weights saved using [save_pretrained()](/docs/transformers/v5.6.0/ja/main_classes/model#transformers.PreTrainedModel.save_pretrained), e.g., `./my_model_directory/`. model_args (additional positional arguments, *optional*) : Will be passed along to the underlying model `__init__()` method. config ([PreTrainedConfig](/docs/transformers/v5.6.0/ja/main_classes/configuration#transformers.PreTrainedConfig), *optional*) : Configuration for the model to use instead of an automatically loaded configuration. Configuration can be automatically loaded when: - The model is a model provided by the library (loaded with the *model id* string of a pretrained model). - The model was saved using [save_pretrained()](/docs/transformers/v5.6.0/ja/main_classes/model#transformers.PreTrainedModel.save_pretrained) and is reloaded by supplying the save directory. - The model is loaded by supplying a local directory as `pretrained_model_name_or_path` and a configuration JSON file named *config.json* is found in the directory. state_dict (*dict[str, torch.Tensor]*, *optional*) : A state dictionary to use instead of a state dictionary loaded from saved weights file. This option can be used if you want to create a model from a pretrained configuration but load your own weights. In this case though, you should check if using [save_pretrained()](/docs/transformers/v5.6.0/ja/main_classes/model#transformers.PreTrainedModel.save_pretrained) and [from_pretrained()](/docs/transformers/v5.6.0/ja/main_classes/model#transformers.PreTrainedModel.from_pretrained) is not a simpler option. cache_dir (`str` or `os.PathLike`, *optional*) : Path to a directory in which a downloaded pretrained model configuration should be cached if the standard cache should not be used. force_download (`bool`, *optional*, defaults to `False`) : Whether or not to force the (re-)download of the model weights and configuration files, overriding the cached versions if they exist. proxies (`dict[str, str]`, *optional*) : A dictionary of proxy servers to use by protocol or endpoint, e.g., `{'http': 'foo.bar:3128', 'http://hostname': 'foo.bar:4012'}`. The proxies are used on each request. output_loading_info(`bool`, *optional*, defaults to `False`) : Whether or not to also return a dictionary containing missing keys, unexpected keys and error messages. local_files_only(`bool`, *optional*, defaults to `False`) : Whether or not to only look at local files (e.g., not try downloading the model). revision (`str`, *optional*, defaults to `"main"`) : The specific model version to use. It can be a branch name, a tag name, or a commit id, since we use a git-based system for storing models and other artifacts on huggingface.co, so `revision` can be any identifier allowed by git. trust_remote_code (`bool`, *optional*, defaults to `False`) : Whether or not to allow for custom models defined on the Hub in their own modeling files. This option should only be set to `True` for repositories you trust and in which you have read the code, as it will execute code present on the Hub on your local machine. code_revision (`str`, *optional*, defaults to `"main"`) : The specific revision to use for the code on the Hub, if the code leaves in a different repository than the rest of the model. It can be a branch name, a tag name, or a commit id, since we use a git-based system for storing models and other artifacts on huggingface.co, so `revision` can be any identifier allowed by git. kwargs (additional keyword arguments, *optional*) : Can be used to update the configuration object (after it being loaded) and initiate the model (e.g., `output_attentions=True`). Behaves differently depending on whether a `config` is provided or automatically loaded: - If a configuration is provided with `config`, `**kwargs` will be directly passed to the underlying model's `__init__` method (we assume all relevant updates to the configuration have already been done) - If a configuration is not provided, `kwargs` will be first passed to the configuration class initialization function ([from_pretrained()](/docs/transformers/v5.6.0/ja/main_classes/configuration#transformers.PreTrainedConfig.from_pretrained)). Each key of `kwargs` that corresponds to a configuration attribute will be used to override said attribute with the supplied `kwargs` value. Remaining keys that do not correspond to any configuration attribute will be passed to the underlying model's `__init__` function. ### AutoModelForUniversalSegmentation[[transformers.AutoModelForUniversalSegmentation]] #### transformers.AutoModelForUniversalSegmentation[[transformers.AutoModelForUniversalSegmentation]] [Source](https://github.com/huggingface/transformers/blob/v5.6.0/src/transformers/models/auto/modeling_auto.py#L2141) This is a generic model class that will be instantiated as one of the model classes of the library (with a universal image segmentation head) when created with the [from_pretrained()](/docs/transformers/v5.6.0/ja/model_doc/auto#transformers.AutoModel.from_pretrained) class method or the [from_config()](/docs/transformers/v5.6.0/ja/model_doc/auto#transformers.AutoModel.from_config) class method. This class cannot be instantiated directly using `__init__()` (throws an error). from_configtransformers.AutoModelForUniversalSegmentation.from_confighttps://github.com/huggingface/transformers/blob/v5.6.0/src/transformers/models/auto/auto_factory.py#L206[{"name": "**kwargs", "val": ""}]- **config** ([PreTrainedConfig](/docs/transformers/v5.6.0/ja/main_classes/configuration#transformers.PreTrainedConfig)) -- The model class to instantiate is selected based on the configuration class: - [DetrConfig](/docs/transformers/v5.6.0/ja/model_doc/detr#transformers.DetrConfig) configuration class: [DetrForSegmentation](/docs/transformers/v5.6.0/ja/model_doc/detr#transformers.DetrForSegmentation) (DetrConfig model) - `EomtConfig` configuration class: `EomtForUniversalSegmentation` (EomtConfig model) - `EomtDinov3Config` configuration class: `EomtDinov3ForUniversalSegmentation` (EomtDinov3Config model) - `Mask2FormerConfig` configuration class: `Mask2FormerForUniversalSegmentation` (Mask2FormerConfig model) - `MaskFormerConfig` configuration class: `MaskFormerForInstanceSegmentation` (MaskFormerConfig model) - `OneFormerConfig` configuration class: `OneFormerForUniversalSegmentation` (OneFormerConfig model) - `VideomtConfig` configuration class: `VideomtForUniversalSegmentation` (VideomtConfig model) - **attn_implementation** (`str`, *optional*) -- The attention implementation to use in the model (if relevant). Can be any of `"eager"` (manual implementation of the attention), `"sdpa"` (using [`F.scaled_dot_product_attention`](https://pytorch.org/docs/master/generated/torch.nn.functional.scaled_dot_product_attention.html)), `"flash_attention_2"` (using [Dao-AILab/flash-attention](https://github.com/Dao-AILab/flash-attention)), or `"flash_attention_3"` (using [Dao-AILab/flash-attention/hopper](https://github.com/Dao-AILab/flash-attention/tree/main/hopper)). By default, if available, SDPA will be used for torch>=2.1.1. The default is otherwise the manual `"eager"` implementation.0 Instantiates one of the model classes of the library (with a universal image segmentation head) from a configuration. Note: Loading a model from its configuration file does **not** load the model weights. It only affects the model's configuration. Use [from_pretrained()](/docs/transformers/v5.6.0/ja/model_doc/auto#transformers.AutoModel.from_pretrained) to load the model weights. Examples: ```python >>> from transformers import AutoConfig, AutoModelForUniversalSegmentation >>> # Download configuration from huggingface.co and cache. >>> config = AutoConfig.from_pretrained("google-bert/bert-base-cased") >>> model = AutoModelForUniversalSegmentation.from_config(config) ``` **Parameters:** config ([PreTrainedConfig](/docs/transformers/v5.6.0/ja/main_classes/configuration#transformers.PreTrainedConfig)) : The model class to instantiate is selected based on the configuration class: - [DetrConfig](/docs/transformers/v5.6.0/ja/model_doc/detr#transformers.DetrConfig) configuration class: [DetrForSegmentation](/docs/transformers/v5.6.0/ja/model_doc/detr#transformers.DetrForSegmentation) (DetrConfig model) - `EomtConfig` configuration class: `EomtForUniversalSegmentation` (EomtConfig model) - `EomtDinov3Config` configuration class: `EomtDinov3ForUniversalSegmentation` (EomtDinov3Config model) - `Mask2FormerConfig` configuration class: `Mask2FormerForUniversalSegmentation` (Mask2FormerConfig model) - `MaskFormerConfig` configuration class: `MaskFormerForInstanceSegmentation` (MaskFormerConfig model) - `OneFormerConfig` configuration class: `OneFormerForUniversalSegmentation` (OneFormerConfig model) - `VideomtConfig` configuration class: `VideomtForUniversalSegmentation` (VideomtConfig model) attn_implementation (`str`, *optional*) : The attention implementation to use in the model (if relevant). Can be any of `"eager"` (manual implementation of the attention), `"sdpa"` (using [`F.scaled_dot_product_attention`](https://pytorch.org/docs/master/generated/torch.nn.functional.scaled_dot_product_attention.html)), `"flash_attention_2"` (using [Dao-AILab/flash-attention](https://github.com/Dao-AILab/flash-attention)), or `"flash_attention_3"` (using [Dao-AILab/flash-attention/hopper](https://github.com/Dao-AILab/flash-attention/tree/main/hopper)). By default, if available, SDPA will be used for torch>=2.1.1. The default is otherwise the manual `"eager"` implementation. #### from_pretrained[[transformers.AutoModelForUniversalSegmentation.from_pretrained]] [Source](https://github.com/huggingface/transformers/blob/v5.6.0/src/transformers/models/auto/auto_factory.py#L253) Instantiate one of the model classes of the library (with a universal image segmentation head) from a pretrained model. The model class to instantiate is selected based on the `model_type` property of the config object (either passed as an argument or loaded from `pretrained_model_name_or_path` if possible), or when it's missing, by falling back to using pattern matching on `pretrained_model_name_or_path`: - **detr** -- [DetrForSegmentation](/docs/transformers/v5.6.0/ja/model_doc/detr#transformers.DetrForSegmentation) (DetrConfig model) - **eomt** -- `EomtForUniversalSegmentation` (EomtConfig model) - **eomt_dinov3** -- `EomtDinov3ForUniversalSegmentation` (EomtDinov3Config model) - **mask2former** -- `Mask2FormerForUniversalSegmentation` (Mask2FormerConfig model) - **maskformer** -- `MaskFormerForInstanceSegmentation` (MaskFormerConfig model) - **oneformer** -- `OneFormerForUniversalSegmentation` (OneFormerConfig model) - **videomt** -- `VideomtForUniversalSegmentation` (VideomtConfig model) The model is set in evaluation mode by default using `model.eval()` (so for instance, dropout modules are deactivated). To train the model, you should first set it back in training mode with `model.train()` Examples: ```python >>> from transformers import AutoConfig, AutoModelForUniversalSegmentation >>> # Download model and configuration from huggingface.co and cache. >>> model = AutoModelForUniversalSegmentation.from_pretrained("google-bert/bert-base-cased") >>> # Update configuration during loading >>> model = AutoModelForUniversalSegmentation.from_pretrained("google-bert/bert-base-cased", output_attentions=True) >>> model.config.output_attentions True ``` **Parameters:** pretrained_model_name_or_path (`str` or `os.PathLike`) : Can be either: - A string, the *model id* of a pretrained model hosted inside a model repo on huggingface.co. - A path to a *directory* containing model weights saved using [save_pretrained()](/docs/transformers/v5.6.0/ja/main_classes/model#transformers.PreTrainedModel.save_pretrained), e.g., `./my_model_directory/`. model_args (additional positional arguments, *optional*) : Will be passed along to the underlying model `__init__()` method. config ([PreTrainedConfig](/docs/transformers/v5.6.0/ja/main_classes/configuration#transformers.PreTrainedConfig), *optional*) : Configuration for the model to use instead of an automatically loaded configuration. Configuration can be automatically loaded when: - The model is a model provided by the library (loaded with the *model id* string of a pretrained model). - The model was saved using [save_pretrained()](/docs/transformers/v5.6.0/ja/main_classes/model#transformers.PreTrainedModel.save_pretrained) and is reloaded by supplying the save directory. - The model is loaded by supplying a local directory as `pretrained_model_name_or_path` and a configuration JSON file named *config.json* is found in the directory. state_dict (*dict[str, torch.Tensor]*, *optional*) : A state dictionary to use instead of a state dictionary loaded from saved weights file. This option can be used if you want to create a model from a pretrained configuration but load your own weights. In this case though, you should check if using [save_pretrained()](/docs/transformers/v5.6.0/ja/main_classes/model#transformers.PreTrainedModel.save_pretrained) and [from_pretrained()](/docs/transformers/v5.6.0/ja/main_classes/model#transformers.PreTrainedModel.from_pretrained) is not a simpler option. cache_dir (`str` or `os.PathLike`, *optional*) : Path to a directory in which a downloaded pretrained model configuration should be cached if the standard cache should not be used. force_download (`bool`, *optional*, defaults to `False`) : Whether or not to force the (re-)download of the model weights and configuration files, overriding the cached versions if they exist. proxies (`dict[str, str]`, *optional*) : A dictionary of proxy servers to use by protocol or endpoint, e.g., `{'http': 'foo.bar:3128', 'http://hostname': 'foo.bar:4012'}`. The proxies are used on each request. output_loading_info(`bool`, *optional*, defaults to `False`) : Whether or not to also return a dictionary containing missing keys, unexpected keys and error messages. local_files_only(`bool`, *optional*, defaults to `False`) : Whether or not to only look at local files (e.g., not try downloading the model). revision (`str`, *optional*, defaults to `"main"`) : The specific model version to use. It can be a branch name, a tag name, or a commit id, since we use a git-based system for storing models and other artifacts on huggingface.co, so `revision` can be any identifier allowed by git. trust_remote_code (`bool`, *optional*, defaults to `False`) : Whether or not to allow for custom models defined on the Hub in their own modeling files. This option should only be set to `True` for repositories you trust and in which you have read the code, as it will execute code present on the Hub on your local machine. code_revision (`str`, *optional*, defaults to `"main"`) : The specific revision to use for the code on the Hub, if the code leaves in a different repository than the rest of the model. It can be a branch name, a tag name, or a commit id, since we use a git-based system for storing models and other artifacts on huggingface.co, so `revision` can be any identifier allowed by git. kwargs (additional keyword arguments, *optional*) : Can be used to update the configuration object (after it being loaded) and initiate the model (e.g., `output_attentions=True`). Behaves differently depending on whether a `config` is provided or automatically loaded: - If a configuration is provided with `config`, `**kwargs` will be directly passed to the underlying model's `__init__` method (we assume all relevant updates to the configuration have already been done) - If a configuration is not provided, `kwargs` will be first passed to the configuration class initialization function ([from_pretrained()](/docs/transformers/v5.6.0/ja/main_classes/configuration#transformers.PreTrainedConfig.from_pretrained)). Each key of `kwargs` that corresponds to a configuration attribute will be used to override said attribute with the supplied `kwargs` value. Remaining keys that do not correspond to any configuration attribute will be passed to the underlying model's `__init__` function. ### AutoModelForZeroShotImageClassification[[transformers.AutoModelForZeroShotImageClassification]] #### transformers.AutoModelForZeroShotImageClassification[[transformers.AutoModelForZeroShotImageClassification]] [Source](https://github.com/huggingface/transformers/blob/v5.6.0/src/transformers/models/auto/modeling_auto.py#L2107) This is a generic model class that will be instantiated as one of the model classes of the library (with a zero-shot image classification head) when created with the [from_pretrained()](/docs/transformers/v5.6.0/ja/model_doc/auto#transformers.AutoModel.from_pretrained) class method or the [from_config()](/docs/transformers/v5.6.0/ja/model_doc/auto#transformers.AutoModel.from_config) class method. This class cannot be instantiated directly using `__init__()` (throws an error). from_configtransformers.AutoModelForZeroShotImageClassification.from_confighttps://github.com/huggingface/transformers/blob/v5.6.0/src/transformers/models/auto/auto_factory.py#L206[{"name": "**kwargs", "val": ""}]- **config** ([PreTrainedConfig](/docs/transformers/v5.6.0/ja/main_classes/configuration#transformers.PreTrainedConfig)) -- The model class to instantiate is selected based on the configuration class: - [AlignConfig](/docs/transformers/v5.6.0/ja/model_doc/align#transformers.AlignConfig) configuration class: [AlignModel](/docs/transformers/v5.6.0/ja/model_doc/align#transformers.AlignModel) (AlignConfig model) - [AltCLIPConfig](/docs/transformers/v5.6.0/ja/model_doc/altclip#transformers.AltCLIPConfig) configuration class: [AltCLIPModel](/docs/transformers/v5.6.0/ja/model_doc/altclip#transformers.AltCLIPModel) (AltCLIPConfig model) - [Blip2Config](/docs/transformers/v5.6.0/ja/model_doc/blip-2#transformers.Blip2Config) configuration class: `Blip2ForImageTextRetrieval` (Blip2Config model) - [BlipConfig](/docs/transformers/v5.6.0/ja/model_doc/blip#transformers.BlipConfig) configuration class: [BlipModel](/docs/transformers/v5.6.0/ja/model_doc/blip#transformers.BlipModel) (BlipConfig model) - [CLIPConfig](/docs/transformers/v5.6.0/ja/model_doc/clip#transformers.CLIPConfig) configuration class: [CLIPModel](/docs/transformers/v5.6.0/ja/model_doc/clip#transformers.CLIPModel) (CLIPConfig model) - [CLIPSegConfig](/docs/transformers/v5.6.0/ja/model_doc/clipseg#transformers.CLIPSegConfig) configuration class: [CLIPSegModel](/docs/transformers/v5.6.0/ja/model_doc/clipseg#transformers.CLIPSegModel) (CLIPSegConfig model) - [ChineseCLIPConfig](/docs/transformers/v5.6.0/ja/model_doc/chinese_clip#transformers.ChineseCLIPConfig) configuration class: [ChineseCLIPModel](/docs/transformers/v5.6.0/ja/model_doc/chinese_clip#transformers.ChineseCLIPModel) (ChineseCLIPConfig model) - `MetaClip2Config` configuration class: `MetaClip2Model` (MetaClip2Config model) - `Siglip2Config` configuration class: `Siglip2Model` (Siglip2Config model) - `SiglipConfig` configuration class: `SiglipModel` (SiglipConfig model) - **attn_implementation** (`str`, *optional*) -- The attention implementation to use in the model (if relevant). Can be any of `"eager"` (manual implementation of the attention), `"sdpa"` (using [`F.scaled_dot_product_attention`](https://pytorch.org/docs/master/generated/torch.nn.functional.scaled_dot_product_attention.html)), `"flash_attention_2"` (using [Dao-AILab/flash-attention](https://github.com/Dao-AILab/flash-attention)), or `"flash_attention_3"` (using [Dao-AILab/flash-attention/hopper](https://github.com/Dao-AILab/flash-attention/tree/main/hopper)). By default, if available, SDPA will be used for torch>=2.1.1. The default is otherwise the manual `"eager"` implementation.0 Instantiates one of the model classes of the library (with a zero-shot image classification head) from a configuration. Note: Loading a model from its configuration file does **not** load the model weights. It only affects the model's configuration. Use [from_pretrained()](/docs/transformers/v5.6.0/ja/model_doc/auto#transformers.AutoModel.from_pretrained) to load the model weights. Examples: ```python >>> from transformers import AutoConfig, AutoModelForZeroShotImageClassification >>> # Download configuration from huggingface.co and cache. >>> config = AutoConfig.from_pretrained("google-bert/bert-base-cased") >>> model = AutoModelForZeroShotImageClassification.from_config(config) ``` **Parameters:** config ([PreTrainedConfig](/docs/transformers/v5.6.0/ja/main_classes/configuration#transformers.PreTrainedConfig)) : The model class to instantiate is selected based on the configuration class: - [AlignConfig](/docs/transformers/v5.6.0/ja/model_doc/align#transformers.AlignConfig) configuration class: [AlignModel](/docs/transformers/v5.6.0/ja/model_doc/align#transformers.AlignModel) (AlignConfig model) - [AltCLIPConfig](/docs/transformers/v5.6.0/ja/model_doc/altclip#transformers.AltCLIPConfig) configuration class: [AltCLIPModel](/docs/transformers/v5.6.0/ja/model_doc/altclip#transformers.AltCLIPModel) (AltCLIPConfig model) - [Blip2Config](/docs/transformers/v5.6.0/ja/model_doc/blip-2#transformers.Blip2Config) configuration class: `Blip2ForImageTextRetrieval` (Blip2Config model) - [BlipConfig](/docs/transformers/v5.6.0/ja/model_doc/blip#transformers.BlipConfig) configuration class: [BlipModel](/docs/transformers/v5.6.0/ja/model_doc/blip#transformers.BlipModel) (BlipConfig model) - [CLIPConfig](/docs/transformers/v5.6.0/ja/model_doc/clip#transformers.CLIPConfig) configuration class: [CLIPModel](/docs/transformers/v5.6.0/ja/model_doc/clip#transformers.CLIPModel) (CLIPConfig model) - [CLIPSegConfig](/docs/transformers/v5.6.0/ja/model_doc/clipseg#transformers.CLIPSegConfig) configuration class: [CLIPSegModel](/docs/transformers/v5.6.0/ja/model_doc/clipseg#transformers.CLIPSegModel) (CLIPSegConfig model) - [ChineseCLIPConfig](/docs/transformers/v5.6.0/ja/model_doc/chinese_clip#transformers.ChineseCLIPConfig) configuration class: [ChineseCLIPModel](/docs/transformers/v5.6.0/ja/model_doc/chinese_clip#transformers.ChineseCLIPModel) (ChineseCLIPConfig model) - `MetaClip2Config` configuration class: `MetaClip2Model` (MetaClip2Config model) - `Siglip2Config` configuration class: `Siglip2Model` (Siglip2Config model) - `SiglipConfig` configuration class: `SiglipModel` (SiglipConfig model) attn_implementation (`str`, *optional*) : The attention implementation to use in the model (if relevant). Can be any of `"eager"` (manual implementation of the attention), `"sdpa"` (using [`F.scaled_dot_product_attention`](https://pytorch.org/docs/master/generated/torch.nn.functional.scaled_dot_product_attention.html)), `"flash_attention_2"` (using [Dao-AILab/flash-attention](https://github.com/Dao-AILab/flash-attention)), or `"flash_attention_3"` (using [Dao-AILab/flash-attention/hopper](https://github.com/Dao-AILab/flash-attention/tree/main/hopper)). By default, if available, SDPA will be used for torch>=2.1.1. The default is otherwise the manual `"eager"` implementation. #### from_pretrained[[transformers.AutoModelForZeroShotImageClassification.from_pretrained]] [Source](https://github.com/huggingface/transformers/blob/v5.6.0/src/transformers/models/auto/auto_factory.py#L253) Instantiate one of the model classes of the library (with a zero-shot image classification head) from a pretrained model. The model class to instantiate is selected based on the `model_type` property of the config object (either passed as an argument or loaded from `pretrained_model_name_or_path` if possible), or when it's missing, by falling back to using pattern matching on `pretrained_model_name_or_path`: - **align** -- [AlignModel](/docs/transformers/v5.6.0/ja/model_doc/align#transformers.AlignModel) (AlignConfig model) - **altclip** -- [AltCLIPModel](/docs/transformers/v5.6.0/ja/model_doc/altclip#transformers.AltCLIPModel) (AltCLIPConfig model) - **blip** -- [BlipModel](/docs/transformers/v5.6.0/ja/model_doc/blip#transformers.BlipModel) (BlipConfig model) - **blip-2** -- `Blip2ForImageTextRetrieval` (Blip2Config model) - **chinese_clip** -- [ChineseCLIPModel](/docs/transformers/v5.6.0/ja/model_doc/chinese_clip#transformers.ChineseCLIPModel) (ChineseCLIPConfig model) - **clip** -- [CLIPModel](/docs/transformers/v5.6.0/ja/model_doc/clip#transformers.CLIPModel) (CLIPConfig model) - **clipseg** -- [CLIPSegModel](/docs/transformers/v5.6.0/ja/model_doc/clipseg#transformers.CLIPSegModel) (CLIPSegConfig model) - **metaclip_2** -- `MetaClip2Model` (MetaClip2Config model) - **siglip** -- `SiglipModel` (SiglipConfig model) - **siglip2** -- `Siglip2Model` (Siglip2Config model) The model is set in evaluation mode by default using `model.eval()` (so for instance, dropout modules are deactivated). To train the model, you should first set it back in training mode with `model.train()` Examples: ```python >>> from transformers import AutoConfig, AutoModelForZeroShotImageClassification >>> # Download model and configuration from huggingface.co and cache. >>> model = AutoModelForZeroShotImageClassification.from_pretrained("google-bert/bert-base-cased") >>> # Update configuration during loading >>> model = AutoModelForZeroShotImageClassification.from_pretrained("google-bert/bert-base-cased", output_attentions=True) >>> model.config.output_attentions True ``` **Parameters:** pretrained_model_name_or_path (`str` or `os.PathLike`) : Can be either: - A string, the *model id* of a pretrained model hosted inside a model repo on huggingface.co. - A path to a *directory* containing model weights saved using [save_pretrained()](/docs/transformers/v5.6.0/ja/main_classes/model#transformers.PreTrainedModel.save_pretrained), e.g., `./my_model_directory/`. model_args (additional positional arguments, *optional*) : Will be passed along to the underlying model `__init__()` method. config ([PreTrainedConfig](/docs/transformers/v5.6.0/ja/main_classes/configuration#transformers.PreTrainedConfig), *optional*) : Configuration for the model to use instead of an automatically loaded configuration. Configuration can be automatically loaded when: - The model is a model provided by the library (loaded with the *model id* string of a pretrained model). - The model was saved using [save_pretrained()](/docs/transformers/v5.6.0/ja/main_classes/model#transformers.PreTrainedModel.save_pretrained) and is reloaded by supplying the save directory. - The model is loaded by supplying a local directory as `pretrained_model_name_or_path` and a configuration JSON file named *config.json* is found in the directory. state_dict (*dict[str, torch.Tensor]*, *optional*) : A state dictionary to use instead of a state dictionary loaded from saved weights file. This option can be used if you want to create a model from a pretrained configuration but load your own weights. In this case though, you should check if using [save_pretrained()](/docs/transformers/v5.6.0/ja/main_classes/model#transformers.PreTrainedModel.save_pretrained) and [from_pretrained()](/docs/transformers/v5.6.0/ja/main_classes/model#transformers.PreTrainedModel.from_pretrained) is not a simpler option. cache_dir (`str` or `os.PathLike`, *optional*) : Path to a directory in which a downloaded pretrained model configuration should be cached if the standard cache should not be used. force_download (`bool`, *optional*, defaults to `False`) : Whether or not to force the (re-)download of the model weights and configuration files, overriding the cached versions if they exist. proxies (`dict[str, str]`, *optional*) : A dictionary of proxy servers to use by protocol or endpoint, e.g., `{'http': 'foo.bar:3128', 'http://hostname': 'foo.bar:4012'}`. The proxies are used on each request. output_loading_info(`bool`, *optional*, defaults to `False`) : Whether or not to also return a dictionary containing missing keys, unexpected keys and error messages. local_files_only(`bool`, *optional*, defaults to `False`) : Whether or not to only look at local files (e.g., not try downloading the model). revision (`str`, *optional*, defaults to `"main"`) : The specific model version to use. It can be a branch name, a tag name, or a commit id, since we use a git-based system for storing models and other artifacts on huggingface.co, so `revision` can be any identifier allowed by git. trust_remote_code (`bool`, *optional*, defaults to `False`) : Whether or not to allow for custom models defined on the Hub in their own modeling files. This option should only be set to `True` for repositories you trust and in which you have read the code, as it will execute code present on the Hub on your local machine. code_revision (`str`, *optional*, defaults to `"main"`) : The specific revision to use for the code on the Hub, if the code leaves in a different repository than the rest of the model. It can be a branch name, a tag name, or a commit id, since we use a git-based system for storing models and other artifacts on huggingface.co, so `revision` can be any identifier allowed by git. kwargs (additional keyword arguments, *optional*) : Can be used to update the configuration object (after it being loaded) and initiate the model (e.g., `output_attentions=True`). Behaves differently depending on whether a `config` is provided or automatically loaded: - If a configuration is provided with `config`, `**kwargs` will be directly passed to the underlying model's `__init__` method (we assume all relevant updates to the configuration have already been done) - If a configuration is not provided, `kwargs` will be first passed to the configuration class initialization function ([from_pretrained()](/docs/transformers/v5.6.0/ja/main_classes/configuration#transformers.PreTrainedConfig.from_pretrained)). Each key of `kwargs` that corresponds to a configuration attribute will be used to override said attribute with the supplied `kwargs` value. Remaining keys that do not correspond to any configuration attribute will be passed to the underlying model's `__init__` function. ### AutoModelForZeroShotObjectDetection[[transformers.AutoModelForZeroShotObjectDetection]] #### transformers.AutoModelForZeroShotObjectDetection[[transformers.AutoModelForZeroShotObjectDetection]] [Source](https://github.com/huggingface/transformers/blob/v5.6.0/src/transformers/models/auto/modeling_auto.py#L2166) This is a generic model class that will be instantiated as one of the model classes of the library (with a zero-shot object detection head) when created with the [from_pretrained()](/docs/transformers/v5.6.0/ja/model_doc/auto#transformers.AutoModel.from_pretrained) class method or the [from_config()](/docs/transformers/v5.6.0/ja/model_doc/auto#transformers.AutoModel.from_config) class method. This class cannot be instantiated directly using `__init__()` (throws an error). from_configtransformers.AutoModelForZeroShotObjectDetection.from_confighttps://github.com/huggingface/transformers/blob/v5.6.0/src/transformers/models/auto/auto_factory.py#L206[{"name": "**kwargs", "val": ""}]- **config** ([PreTrainedConfig](/docs/transformers/v5.6.0/ja/main_classes/configuration#transformers.PreTrainedConfig)) -- The model class to instantiate is selected based on the configuration class: - `GroundingDinoConfig` configuration class: `GroundingDinoForObjectDetection` (GroundingDinoConfig model) - `MMGroundingDinoConfig` configuration class: `MMGroundingDinoForObjectDetection` (MMGroundingDinoConfig model) - `OmDetTurboConfig` configuration class: `OmDetTurboForObjectDetection` (OmDetTurboConfig model) - `OwlViTConfig` configuration class: `OwlViTForObjectDetection` (OwlViTConfig model) - `Owlv2Config` configuration class: `Owlv2ForObjectDetection` (Owlv2Config model) - **attn_implementation** (`str`, *optional*) -- The attention implementation to use in the model (if relevant). Can be any of `"eager"` (manual implementation of the attention), `"sdpa"` (using [`F.scaled_dot_product_attention`](https://pytorch.org/docs/master/generated/torch.nn.functional.scaled_dot_product_attention.html)), `"flash_attention_2"` (using [Dao-AILab/flash-attention](https://github.com/Dao-AILab/flash-attention)), or `"flash_attention_3"` (using [Dao-AILab/flash-attention/hopper](https://github.com/Dao-AILab/flash-attention/tree/main/hopper)). By default, if available, SDPA will be used for torch>=2.1.1. The default is otherwise the manual `"eager"` implementation.0 Instantiates one of the model classes of the library (with a zero-shot object detection head) from a configuration. Note: Loading a model from its configuration file does **not** load the model weights. It only affects the model's configuration. Use [from_pretrained()](/docs/transformers/v5.6.0/ja/model_doc/auto#transformers.AutoModel.from_pretrained) to load the model weights. Examples: ```python >>> from transformers import AutoConfig, AutoModelForZeroShotObjectDetection >>> # Download configuration from huggingface.co and cache. >>> config = AutoConfig.from_pretrained("google-bert/bert-base-cased") >>> model = AutoModelForZeroShotObjectDetection.from_config(config) ``` **Parameters:** config ([PreTrainedConfig](/docs/transformers/v5.6.0/ja/main_classes/configuration#transformers.PreTrainedConfig)) : The model class to instantiate is selected based on the configuration class: - `GroundingDinoConfig` configuration class: `GroundingDinoForObjectDetection` (GroundingDinoConfig model) - `MMGroundingDinoConfig` configuration class: `MMGroundingDinoForObjectDetection` (MMGroundingDinoConfig model) - `OmDetTurboConfig` configuration class: `OmDetTurboForObjectDetection` (OmDetTurboConfig model) - `OwlViTConfig` configuration class: `OwlViTForObjectDetection` (OwlViTConfig model) - `Owlv2Config` configuration class: `Owlv2ForObjectDetection` (Owlv2Config model) attn_implementation (`str`, *optional*) : The attention implementation to use in the model (if relevant). Can be any of `"eager"` (manual implementation of the attention), `"sdpa"` (using [`F.scaled_dot_product_attention`](https://pytorch.org/docs/master/generated/torch.nn.functional.scaled_dot_product_attention.html)), `"flash_attention_2"` (using [Dao-AILab/flash-attention](https://github.com/Dao-AILab/flash-attention)), or `"flash_attention_3"` (using [Dao-AILab/flash-attention/hopper](https://github.com/Dao-AILab/flash-attention/tree/main/hopper)). By default, if available, SDPA will be used for torch>=2.1.1. The default is otherwise the manual `"eager"` implementation. #### from_pretrained[[transformers.AutoModelForZeroShotObjectDetection.from_pretrained]] [Source](https://github.com/huggingface/transformers/blob/v5.6.0/src/transformers/models/auto/auto_factory.py#L253) Instantiate one of the model classes of the library (with a zero-shot object detection head) from a pretrained model. The model class to instantiate is selected based on the `model_type` property of the config object (either passed as an argument or loaded from `pretrained_model_name_or_path` if possible), or when it's missing, by falling back to using pattern matching on `pretrained_model_name_or_path`: - **grounding-dino** -- `GroundingDinoForObjectDetection` (GroundingDinoConfig model) - **mm-grounding-dino** -- `MMGroundingDinoForObjectDetection` (MMGroundingDinoConfig model) - **omdet-turbo** -- `OmDetTurboForObjectDetection` (OmDetTurboConfig model) - **owlv2** -- `Owlv2ForObjectDetection` (Owlv2Config model) - **owlvit** -- `OwlViTForObjectDetection` (OwlViTConfig model) The model is set in evaluation mode by default using `model.eval()` (so for instance, dropout modules are deactivated). To train the model, you should first set it back in training mode with `model.train()` Examples: ```python >>> from transformers import AutoConfig, AutoModelForZeroShotObjectDetection >>> # Download model and configuration from huggingface.co and cache. >>> model = AutoModelForZeroShotObjectDetection.from_pretrained("google-bert/bert-base-cased") >>> # Update configuration during loading >>> model = AutoModelForZeroShotObjectDetection.from_pretrained("google-bert/bert-base-cased", output_attentions=True) >>> model.config.output_attentions True ``` **Parameters:** pretrained_model_name_or_path (`str` or `os.PathLike`) : Can be either: - A string, the *model id* of a pretrained model hosted inside a model repo on huggingface.co. - A path to a *directory* containing model weights saved using [save_pretrained()](/docs/transformers/v5.6.0/ja/main_classes/model#transformers.PreTrainedModel.save_pretrained), e.g., `./my_model_directory/`. model_args (additional positional arguments, *optional*) : Will be passed along to the underlying model `__init__()` method. config ([PreTrainedConfig](/docs/transformers/v5.6.0/ja/main_classes/configuration#transformers.PreTrainedConfig), *optional*) : Configuration for the model to use instead of an automatically loaded configuration. Configuration can be automatically loaded when: - The model is a model provided by the library (loaded with the *model id* string of a pretrained model). - The model was saved using [save_pretrained()](/docs/transformers/v5.6.0/ja/main_classes/model#transformers.PreTrainedModel.save_pretrained) and is reloaded by supplying the save directory. - The model is loaded by supplying a local directory as `pretrained_model_name_or_path` and a configuration JSON file named *config.json* is found in the directory. state_dict (*dict[str, torch.Tensor]*, *optional*) : A state dictionary to use instead of a state dictionary loaded from saved weights file. This option can be used if you want to create a model from a pretrained configuration but load your own weights. In this case though, you should check if using [save_pretrained()](/docs/transformers/v5.6.0/ja/main_classes/model#transformers.PreTrainedModel.save_pretrained) and [from_pretrained()](/docs/transformers/v5.6.0/ja/main_classes/model#transformers.PreTrainedModel.from_pretrained) is not a simpler option. cache_dir (`str` or `os.PathLike`, *optional*) : Path to a directory in which a downloaded pretrained model configuration should be cached if the standard cache should not be used. force_download (`bool`, *optional*, defaults to `False`) : Whether or not to force the (re-)download of the model weights and configuration files, overriding the cached versions if they exist. proxies (`dict[str, str]`, *optional*) : A dictionary of proxy servers to use by protocol or endpoint, e.g., `{'http': 'foo.bar:3128', 'http://hostname': 'foo.bar:4012'}`. The proxies are used on each request. output_loading_info(`bool`, *optional*, defaults to `False`) : Whether or not to also return a dictionary containing missing keys, unexpected keys and error messages. local_files_only(`bool`, *optional*, defaults to `False`) : Whether or not to only look at local files (e.g., not try downloading the model). revision (`str`, *optional*, defaults to `"main"`) : The specific model version to use. It can be a branch name, a tag name, or a commit id, since we use a git-based system for storing models and other artifacts on huggingface.co, so `revision` can be any identifier allowed by git. trust_remote_code (`bool`, *optional*, defaults to `False`) : Whether or not to allow for custom models defined on the Hub in their own modeling files. This option should only be set to `True` for repositories you trust and in which you have read the code, as it will execute code present on the Hub on your local machine. code_revision (`str`, *optional*, defaults to `"main"`) : The specific revision to use for the code on the Hub, if the code leaves in a different repository than the rest of the model. It can be a branch name, a tag name, or a commit id, since we use a git-based system for storing models and other artifacts on huggingface.co, so `revision` can be any identifier allowed by git. kwargs (additional keyword arguments, *optional*) : Can be used to update the configuration object (after it being loaded) and initiate the model (e.g., `output_attentions=True`). Behaves differently depending on whether a `config` is provided or automatically loaded: - If a configuration is provided with `config`, `**kwargs` will be directly passed to the underlying model's `__init__` method (we assume all relevant updates to the configuration have already been done) - If a configuration is not provided, `kwargs` will be first passed to the configuration class initialization function ([from_pretrained()](/docs/transformers/v5.6.0/ja/main_classes/configuration#transformers.PreTrainedConfig.from_pretrained)). Each key of `kwargs` that corresponds to a configuration attribute will be used to override said attribute with the supplied `kwargs` value. Remaining keys that do not correspond to any configuration attribute will be passed to the underlying model's `__init__` function. ## Audio 以下の自動クラスは、次の音声タスクに利用可能です。 ### AutoModelForAudioClassification[[transformers.AutoModelForAudioClassification]] #### transformers.AutoModelForAudioClassification[[transformers.AutoModelForAudioClassification]] [Source](https://github.com/huggingface/transformers/blob/v5.6.0/src/transformers/models/auto/modeling_auto.py#L2227) This is a generic model class that will be instantiated as one of the model classes of the library (with a audio classification head) when created with the [from_pretrained()](/docs/transformers/v5.6.0/ja/model_doc/auto#transformers.AutoModel.from_pretrained) class method or the [from_config()](/docs/transformers/v5.6.0/ja/model_doc/auto#transformers.AutoModel.from_config) class method. This class cannot be instantiated directly using `__init__()` (throws an error). from_configtransformers.AutoModelForAudioClassification.from_confighttps://github.com/huggingface/transformers/blob/v5.6.0/src/transformers/models/auto/auto_factory.py#L206[{"name": "**kwargs", "val": ""}]- **config** ([PreTrainedConfig](/docs/transformers/v5.6.0/ja/main_classes/configuration#transformers.PreTrainedConfig)) -- The model class to instantiate is selected based on the configuration class: - [ASTConfig](/docs/transformers/v5.6.0/ja/model_doc/audio-spectrogram-transformer#transformers.ASTConfig) configuration class: [ASTForAudioClassification](/docs/transformers/v5.6.0/ja/model_doc/audio-spectrogram-transformer#transformers.ASTForAudioClassification) (ASTConfig model) - [Data2VecAudioConfig](/docs/transformers/v5.6.0/ja/model_doc/data2vec#transformers.Data2VecAudioConfig) configuration class: [Data2VecAudioForSequenceClassification](/docs/transformers/v5.6.0/ja/model_doc/data2vec#transformers.Data2VecAudioForSequenceClassification) (Data2VecAudioConfig model) - `HubertConfig` configuration class: `HubertForSequenceClassification` (HubertConfig model) - `SEWConfig` configuration class: `SEWForSequenceClassification` (SEWConfig model) - `SEWDConfig` configuration class: `SEWDForSequenceClassification` (SEWDConfig model) - `UniSpeechConfig` configuration class: `UniSpeechForSequenceClassification` (UniSpeechConfig model) - `UniSpeechSatConfig` configuration class: `UniSpeechSatForSequenceClassification` (UniSpeechSatConfig model) - `Wav2Vec2BertConfig` configuration class: `Wav2Vec2BertForSequenceClassification` (Wav2Vec2BertConfig model) - `Wav2Vec2Config` configuration class: `Wav2Vec2ForSequenceClassification` (Wav2Vec2Config model) - `Wav2Vec2ConformerConfig` configuration class: `Wav2Vec2ConformerForSequenceClassification` (Wav2Vec2ConformerConfig model) - `WavLMConfig` configuration class: `WavLMForSequenceClassification` (WavLMConfig model) - `WhisperConfig` configuration class: `WhisperForAudioClassification` (WhisperConfig model) - **attn_implementation** (`str`, *optional*) -- The attention implementation to use in the model (if relevant). Can be any of `"eager"` (manual implementation of the attention), `"sdpa"` (using [`F.scaled_dot_product_attention`](https://pytorch.org/docs/master/generated/torch.nn.functional.scaled_dot_product_attention.html)), `"flash_attention_2"` (using [Dao-AILab/flash-attention](https://github.com/Dao-AILab/flash-attention)), or `"flash_attention_3"` (using [Dao-AILab/flash-attention/hopper](https://github.com/Dao-AILab/flash-attention/tree/main/hopper)). By default, if available, SDPA will be used for torch>=2.1.1. The default is otherwise the manual `"eager"` implementation.0 Instantiates one of the model classes of the library (with a audio classification head) from a configuration. Note: Loading a model from its configuration file does **not** load the model weights. It only affects the model's configuration. Use [from_pretrained()](/docs/transformers/v5.6.0/ja/model_doc/auto#transformers.AutoModel.from_pretrained) to load the model weights. Examples: ```python >>> from transformers import AutoConfig, AutoModelForAudioClassification >>> # Download configuration from huggingface.co and cache. >>> config = AutoConfig.from_pretrained("google-bert/bert-base-cased") >>> model = AutoModelForAudioClassification.from_config(config) ``` **Parameters:** config ([PreTrainedConfig](/docs/transformers/v5.6.0/ja/main_classes/configuration#transformers.PreTrainedConfig)) : The model class to instantiate is selected based on the configuration class: - [ASTConfig](/docs/transformers/v5.6.0/ja/model_doc/audio-spectrogram-transformer#transformers.ASTConfig) configuration class: [ASTForAudioClassification](/docs/transformers/v5.6.0/ja/model_doc/audio-spectrogram-transformer#transformers.ASTForAudioClassification) (ASTConfig model) - [Data2VecAudioConfig](/docs/transformers/v5.6.0/ja/model_doc/data2vec#transformers.Data2VecAudioConfig) configuration class: [Data2VecAudioForSequenceClassification](/docs/transformers/v5.6.0/ja/model_doc/data2vec#transformers.Data2VecAudioForSequenceClassification) (Data2VecAudioConfig model) - `HubertConfig` configuration class: `HubertForSequenceClassification` (HubertConfig model) - `SEWConfig` configuration class: `SEWForSequenceClassification` (SEWConfig model) - `SEWDConfig` configuration class: `SEWDForSequenceClassification` (SEWDConfig model) - `UniSpeechConfig` configuration class: `UniSpeechForSequenceClassification` (UniSpeechConfig model) - `UniSpeechSatConfig` configuration class: `UniSpeechSatForSequenceClassification` (UniSpeechSatConfig model) - `Wav2Vec2BertConfig` configuration class: `Wav2Vec2BertForSequenceClassification` (Wav2Vec2BertConfig model) - `Wav2Vec2Config` configuration class: `Wav2Vec2ForSequenceClassification` (Wav2Vec2Config model) - `Wav2Vec2ConformerConfig` configuration class: `Wav2Vec2ConformerForSequenceClassification` (Wav2Vec2ConformerConfig model) - `WavLMConfig` configuration class: `WavLMForSequenceClassification` (WavLMConfig model) - `WhisperConfig` configuration class: `WhisperForAudioClassification` (WhisperConfig model) attn_implementation (`str`, *optional*) : The attention implementation to use in the model (if relevant). Can be any of `"eager"` (manual implementation of the attention), `"sdpa"` (using [`F.scaled_dot_product_attention`](https://pytorch.org/docs/master/generated/torch.nn.functional.scaled_dot_product_attention.html)), `"flash_attention_2"` (using [Dao-AILab/flash-attention](https://github.com/Dao-AILab/flash-attention)), or `"flash_attention_3"` (using [Dao-AILab/flash-attention/hopper](https://github.com/Dao-AILab/flash-attention/tree/main/hopper)). By default, if available, SDPA will be used for torch>=2.1.1. The default is otherwise the manual `"eager"` implementation. #### from_pretrained[[transformers.AutoModelForAudioClassification.from_pretrained]] [Source](https://github.com/huggingface/transformers/blob/v5.6.0/src/transformers/models/auto/auto_factory.py#L253) Instantiate one of the model classes of the library (with a audio classification head) from a pretrained model. The model class to instantiate is selected based on the `model_type` property of the config object (either passed as an argument or loaded from `pretrained_model_name_or_path` if possible), or when it's missing, by falling back to using pattern matching on `pretrained_model_name_or_path`: - **audio-spectrogram-transformer** -- [ASTForAudioClassification](/docs/transformers/v5.6.0/ja/model_doc/audio-spectrogram-transformer#transformers.ASTForAudioClassification) (ASTConfig model) - **data2vec-audio** -- [Data2VecAudioForSequenceClassification](/docs/transformers/v5.6.0/ja/model_doc/data2vec#transformers.Data2VecAudioForSequenceClassification) (Data2VecAudioConfig model) - **hubert** -- `HubertForSequenceClassification` (HubertConfig model) - **sew** -- `SEWForSequenceClassification` (SEWConfig model) - **sew-d** -- `SEWDForSequenceClassification` (SEWDConfig model) - **unispeech** -- `UniSpeechForSequenceClassification` (UniSpeechConfig model) - **unispeech-sat** -- `UniSpeechSatForSequenceClassification` (UniSpeechSatConfig model) - **wav2vec2** -- `Wav2Vec2ForSequenceClassification` (Wav2Vec2Config model) - **wav2vec2-bert** -- `Wav2Vec2BertForSequenceClassification` (Wav2Vec2BertConfig model) - **wav2vec2-conformer** -- `Wav2Vec2ConformerForSequenceClassification` (Wav2Vec2ConformerConfig model) - **wavlm** -- `WavLMForSequenceClassification` (WavLMConfig model) - **whisper** -- `WhisperForAudioClassification` (WhisperConfig model) The model is set in evaluation mode by default using `model.eval()` (so for instance, dropout modules are deactivated). To train the model, you should first set it back in training mode with `model.train()` Examples: ```python >>> from transformers import AutoConfig, AutoModelForAudioClassification >>> # Download model and configuration from huggingface.co and cache. >>> model = AutoModelForAudioClassification.from_pretrained("google-bert/bert-base-cased") >>> # Update configuration during loading >>> model = AutoModelForAudioClassification.from_pretrained("google-bert/bert-base-cased", output_attentions=True) >>> model.config.output_attentions True ``` **Parameters:** pretrained_model_name_or_path (`str` or `os.PathLike`) : Can be either: - A string, the *model id* of a pretrained model hosted inside a model repo on huggingface.co. - A path to a *directory* containing model weights saved using [save_pretrained()](/docs/transformers/v5.6.0/ja/main_classes/model#transformers.PreTrainedModel.save_pretrained), e.g., `./my_model_directory/`. model_args (additional positional arguments, *optional*) : Will be passed along to the underlying model `__init__()` method. config ([PreTrainedConfig](/docs/transformers/v5.6.0/ja/main_classes/configuration#transformers.PreTrainedConfig), *optional*) : Configuration for the model to use instead of an automatically loaded configuration. Configuration can be automatically loaded when: - The model is a model provided by the library (loaded with the *model id* string of a pretrained model). - The model was saved using [save_pretrained()](/docs/transformers/v5.6.0/ja/main_classes/model#transformers.PreTrainedModel.save_pretrained) and is reloaded by supplying the save directory. - The model is loaded by supplying a local directory as `pretrained_model_name_or_path` and a configuration JSON file named *config.json* is found in the directory. state_dict (*dict[str, torch.Tensor]*, *optional*) : A state dictionary to use instead of a state dictionary loaded from saved weights file. This option can be used if you want to create a model from a pretrained configuration but load your own weights. In this case though, you should check if using [save_pretrained()](/docs/transformers/v5.6.0/ja/main_classes/model#transformers.PreTrainedModel.save_pretrained) and [from_pretrained()](/docs/transformers/v5.6.0/ja/main_classes/model#transformers.PreTrainedModel.from_pretrained) is not a simpler option. cache_dir (`str` or `os.PathLike`, *optional*) : Path to a directory in which a downloaded pretrained model configuration should be cached if the standard cache should not be used. force_download (`bool`, *optional*, defaults to `False`) : Whether or not to force the (re-)download of the model weights and configuration files, overriding the cached versions if they exist. proxies (`dict[str, str]`, *optional*) : A dictionary of proxy servers to use by protocol or endpoint, e.g., `{'http': 'foo.bar:3128', 'http://hostname': 'foo.bar:4012'}`. The proxies are used on each request. output_loading_info(`bool`, *optional*, defaults to `False`) : Whether or not to also return a dictionary containing missing keys, unexpected keys and error messages. local_files_only(`bool`, *optional*, defaults to `False`) : Whether or not to only look at local files (e.g., not try downloading the model). revision (`str`, *optional*, defaults to `"main"`) : The specific model version to use. It can be a branch name, a tag name, or a commit id, since we use a git-based system for storing models and other artifacts on huggingface.co, so `revision` can be any identifier allowed by git. trust_remote_code (`bool`, *optional*, defaults to `False`) : Whether or not to allow for custom models defined on the Hub in their own modeling files. This option should only be set to `True` for repositories you trust and in which you have read the code, as it will execute code present on the Hub on your local machine. code_revision (`str`, *optional*, defaults to `"main"`) : The specific revision to use for the code on the Hub, if the code leaves in a different repository than the rest of the model. It can be a branch name, a tag name, or a commit id, since we use a git-based system for storing models and other artifacts on huggingface.co, so `revision` can be any identifier allowed by git. kwargs (additional keyword arguments, *optional*) : Can be used to update the configuration object (after it being loaded) and initiate the model (e.g., `output_attentions=True`). Behaves differently depending on whether a `config` is provided or automatically loaded: - If a configuration is provided with `config`, `**kwargs` will be directly passed to the underlying model's `__init__` method (we assume all relevant updates to the configuration have already been done) - If a configuration is not provided, `kwargs` will be first passed to the configuration class initialization function ([from_pretrained()](/docs/transformers/v5.6.0/ja/main_classes/configuration#transformers.PreTrainedConfig.from_pretrained)). Each key of `kwargs` that corresponds to a configuration attribute will be used to override said attribute with the supplied `kwargs` value. Remaining keys that do not correspond to any configuration attribute will be passed to the underlying model's `__init__` function. ### AutoModelForAudioFrameClassification[[transformers.AutoModelForAudioFrameClassification]] #### transformers.AutoModelForAudioFrameClassification[[transformers.AutoModelForAudioFrameClassification]] [Source](https://github.com/huggingface/transformers/blob/v5.6.0/src/transformers/models/auto/modeling_auto.py#L2250) This is a generic model class that will be instantiated as one of the model classes of the library (with a audio frame (token) classification head) when created with the [from_pretrained()](/docs/transformers/v5.6.0/ja/model_doc/auto#transformers.AutoModel.from_pretrained) class method or the [from_config()](/docs/transformers/v5.6.0/ja/model_doc/auto#transformers.AutoModel.from_config) class method. This class cannot be instantiated directly using `__init__()` (throws an error). from_configtransformers.AutoModelForAudioFrameClassification.from_confighttps://github.com/huggingface/transformers/blob/v5.6.0/src/transformers/models/auto/auto_factory.py#L206[{"name": "**kwargs", "val": ""}]- **config** ([PreTrainedConfig](/docs/transformers/v5.6.0/ja/main_classes/configuration#transformers.PreTrainedConfig)) -- The model class to instantiate is selected based on the configuration class: - [Data2VecAudioConfig](/docs/transformers/v5.6.0/ja/model_doc/data2vec#transformers.Data2VecAudioConfig) configuration class: [Data2VecAudioForAudioFrameClassification](/docs/transformers/v5.6.0/ja/model_doc/data2vec#transformers.Data2VecAudioForAudioFrameClassification) (Data2VecAudioConfig model) - `UniSpeechSatConfig` configuration class: `UniSpeechSatForAudioFrameClassification` (UniSpeechSatConfig model) - `Wav2Vec2BertConfig` configuration class: `Wav2Vec2BertForAudioFrameClassification` (Wav2Vec2BertConfig model) - `Wav2Vec2Config` configuration class: `Wav2Vec2ForAudioFrameClassification` (Wav2Vec2Config model) - `Wav2Vec2ConformerConfig` configuration class: `Wav2Vec2ConformerForAudioFrameClassification` (Wav2Vec2ConformerConfig model) - `WavLMConfig` configuration class: `WavLMForAudioFrameClassification` (WavLMConfig model) - **attn_implementation** (`str`, *optional*) -- The attention implementation to use in the model (if relevant). Can be any of `"eager"` (manual implementation of the attention), `"sdpa"` (using [`F.scaled_dot_product_attention`](https://pytorch.org/docs/master/generated/torch.nn.functional.scaled_dot_product_attention.html)), `"flash_attention_2"` (using [Dao-AILab/flash-attention](https://github.com/Dao-AILab/flash-attention)), or `"flash_attention_3"` (using [Dao-AILab/flash-attention/hopper](https://github.com/Dao-AILab/flash-attention/tree/main/hopper)). By default, if available, SDPA will be used for torch>=2.1.1. The default is otherwise the manual `"eager"` implementation.0 Instantiates one of the model classes of the library (with a audio frame (token) classification head) from a configuration. Note: Loading a model from its configuration file does **not** load the model weights. It only affects the model's configuration. Use [from_pretrained()](/docs/transformers/v5.6.0/ja/model_doc/auto#transformers.AutoModel.from_pretrained) to load the model weights. Examples: ```python >>> from transformers import AutoConfig, AutoModelForAudioFrameClassification >>> # Download configuration from huggingface.co and cache. >>> config = AutoConfig.from_pretrained("google-bert/bert-base-cased") >>> model = AutoModelForAudioFrameClassification.from_config(config) ``` **Parameters:** config ([PreTrainedConfig](/docs/transformers/v5.6.0/ja/main_classes/configuration#transformers.PreTrainedConfig)) : The model class to instantiate is selected based on the configuration class: - [Data2VecAudioConfig](/docs/transformers/v5.6.0/ja/model_doc/data2vec#transformers.Data2VecAudioConfig) configuration class: [Data2VecAudioForAudioFrameClassification](/docs/transformers/v5.6.0/ja/model_doc/data2vec#transformers.Data2VecAudioForAudioFrameClassification) (Data2VecAudioConfig model) - `UniSpeechSatConfig` configuration class: `UniSpeechSatForAudioFrameClassification` (UniSpeechSatConfig model) - `Wav2Vec2BertConfig` configuration class: `Wav2Vec2BertForAudioFrameClassification` (Wav2Vec2BertConfig model) - `Wav2Vec2Config` configuration class: `Wav2Vec2ForAudioFrameClassification` (Wav2Vec2Config model) - `Wav2Vec2ConformerConfig` configuration class: `Wav2Vec2ConformerForAudioFrameClassification` (Wav2Vec2ConformerConfig model) - `WavLMConfig` configuration class: `WavLMForAudioFrameClassification` (WavLMConfig model) attn_implementation (`str`, *optional*) : The attention implementation to use in the model (if relevant). Can be any of `"eager"` (manual implementation of the attention), `"sdpa"` (using [`F.scaled_dot_product_attention`](https://pytorch.org/docs/master/generated/torch.nn.functional.scaled_dot_product_attention.html)), `"flash_attention_2"` (using [Dao-AILab/flash-attention](https://github.com/Dao-AILab/flash-attention)), or `"flash_attention_3"` (using [Dao-AILab/flash-attention/hopper](https://github.com/Dao-AILab/flash-attention/tree/main/hopper)). By default, if available, SDPA will be used for torch>=2.1.1. The default is otherwise the manual `"eager"` implementation. #### from_pretrained[[transformers.AutoModelForAudioFrameClassification.from_pretrained]] [Source](https://github.com/huggingface/transformers/blob/v5.6.0/src/transformers/models/auto/auto_factory.py#L253) Instantiate one of the model classes of the library (with a audio frame (token) classification head) from a pretrained model. The model class to instantiate is selected based on the `model_type` property of the config object (either passed as an argument or loaded from `pretrained_model_name_or_path` if possible), or when it's missing, by falling back to using pattern matching on `pretrained_model_name_or_path`: - **data2vec-audio** -- [Data2VecAudioForAudioFrameClassification](/docs/transformers/v5.6.0/ja/model_doc/data2vec#transformers.Data2VecAudioForAudioFrameClassification) (Data2VecAudioConfig model) - **unispeech-sat** -- `UniSpeechSatForAudioFrameClassification` (UniSpeechSatConfig model) - **wav2vec2** -- `Wav2Vec2ForAudioFrameClassification` (Wav2Vec2Config model) - **wav2vec2-bert** -- `Wav2Vec2BertForAudioFrameClassification` (Wav2Vec2BertConfig model) - **wav2vec2-conformer** -- `Wav2Vec2ConformerForAudioFrameClassification` (Wav2Vec2ConformerConfig model) - **wavlm** -- `WavLMForAudioFrameClassification` (WavLMConfig model) The model is set in evaluation mode by default using `model.eval()` (so for instance, dropout modules are deactivated). To train the model, you should first set it back in training mode with `model.train()` Examples: ```python >>> from transformers import AutoConfig, AutoModelForAudioFrameClassification >>> # Download model and configuration from huggingface.co and cache. >>> model = AutoModelForAudioFrameClassification.from_pretrained("google-bert/bert-base-cased") >>> # Update configuration during loading >>> model = AutoModelForAudioFrameClassification.from_pretrained("google-bert/bert-base-cased", output_attentions=True) >>> model.config.output_attentions True ``` **Parameters:** pretrained_model_name_or_path (`str` or `os.PathLike`) : Can be either: - A string, the *model id* of a pretrained model hosted inside a model repo on huggingface.co. - A path to a *directory* containing model weights saved using [save_pretrained()](/docs/transformers/v5.6.0/ja/main_classes/model#transformers.PreTrainedModel.save_pretrained), e.g., `./my_model_directory/`. model_args (additional positional arguments, *optional*) : Will be passed along to the underlying model `__init__()` method. config ([PreTrainedConfig](/docs/transformers/v5.6.0/ja/main_classes/configuration#transformers.PreTrainedConfig), *optional*) : Configuration for the model to use instead of an automatically loaded configuration. Configuration can be automatically loaded when: - The model is a model provided by the library (loaded with the *model id* string of a pretrained model). - The model was saved using [save_pretrained()](/docs/transformers/v5.6.0/ja/main_classes/model#transformers.PreTrainedModel.save_pretrained) and is reloaded by supplying the save directory. - The model is loaded by supplying a local directory as `pretrained_model_name_or_path` and a configuration JSON file named *config.json* is found in the directory. state_dict (*dict[str, torch.Tensor]*, *optional*) : A state dictionary to use instead of a state dictionary loaded from saved weights file. This option can be used if you want to create a model from a pretrained configuration but load your own weights. In this case though, you should check if using [save_pretrained()](/docs/transformers/v5.6.0/ja/main_classes/model#transformers.PreTrainedModel.save_pretrained) and [from_pretrained()](/docs/transformers/v5.6.0/ja/main_classes/model#transformers.PreTrainedModel.from_pretrained) is not a simpler option. cache_dir (`str` or `os.PathLike`, *optional*) : Path to a directory in which a downloaded pretrained model configuration should be cached if the standard cache should not be used. force_download (`bool`, *optional*, defaults to `False`) : Whether or not to force the (re-)download of the model weights and configuration files, overriding the cached versions if they exist. proxies (`dict[str, str]`, *optional*) : A dictionary of proxy servers to use by protocol or endpoint, e.g., `{'http': 'foo.bar:3128', 'http://hostname': 'foo.bar:4012'}`. The proxies are used on each request. output_loading_info(`bool`, *optional*, defaults to `False`) : Whether or not to also return a dictionary containing missing keys, unexpected keys and error messages. local_files_only(`bool`, *optional*, defaults to `False`) : Whether or not to only look at local files (e.g., not try downloading the model). revision (`str`, *optional*, defaults to `"main"`) : The specific model version to use. It can be a branch name, a tag name, or a commit id, since we use a git-based system for storing models and other artifacts on huggingface.co, so `revision` can be any identifier allowed by git. trust_remote_code (`bool`, *optional*, defaults to `False`) : Whether or not to allow for custom models defined on the Hub in their own modeling files. This option should only be set to `True` for repositories you trust and in which you have read the code, as it will execute code present on the Hub on your local machine. code_revision (`str`, *optional*, defaults to `"main"`) : The specific revision to use for the code on the Hub, if the code leaves in a different repository than the rest of the model. It can be a branch name, a tag name, or a commit id, since we use a git-based system for storing models and other artifacts on huggingface.co, so `revision` can be any identifier allowed by git. kwargs (additional keyword arguments, *optional*) : Can be used to update the configuration object (after it being loaded) and initiate the model (e.g., `output_attentions=True`). Behaves differently depending on whether a `config` is provided or automatically loaded: - If a configuration is provided with `config`, `**kwargs` will be directly passed to the underlying model's `__init__` method (we assume all relevant updates to the configuration have already been done) - If a configuration is not provided, `kwargs` will be first passed to the configuration class initialization function ([from_pretrained()](/docs/transformers/v5.6.0/ja/main_classes/configuration#transformers.PreTrainedConfig.from_pretrained)). Each key of `kwargs` that corresponds to a configuration attribute will be used to override said attribute with the supplied `kwargs` value. Remaining keys that do not correspond to any configuration attribute will be passed to the underlying model's `__init__` function. ### AutoModelForCTC[[transformers.AutoModelForCTC]] #### transformers.AutoModelForCTC[[transformers.AutoModelForCTC]] [Source](https://github.com/huggingface/transformers/blob/v5.6.0/src/transformers/models/auto/modeling_auto.py#L2234) This is a generic model class that will be instantiated as one of the model classes of the library (with a connectionist temporal classification head) when created with the [from_pretrained()](/docs/transformers/v5.6.0/ja/model_doc/auto#transformers.AutoModel.from_pretrained) class method or the [from_config()](/docs/transformers/v5.6.0/ja/model_doc/auto#transformers.AutoModel.from_config) class method. This class cannot be instantiated directly using `__init__()` (throws an error). from_configtransformers.AutoModelForCTC.from_confighttps://github.com/huggingface/transformers/blob/v5.6.0/src/transformers/models/auto/auto_factory.py#L206[{"name": "**kwargs", "val": ""}]- **config** ([PreTrainedConfig](/docs/transformers/v5.6.0/ja/main_classes/configuration#transformers.PreTrainedConfig)) -- The model class to instantiate is selected based on the configuration class: - [Data2VecAudioConfig](/docs/transformers/v5.6.0/ja/model_doc/data2vec#transformers.Data2VecAudioConfig) configuration class: [Data2VecAudioForCTC](/docs/transformers/v5.6.0/ja/model_doc/data2vec#transformers.Data2VecAudioForCTC) (Data2VecAudioConfig model) - `HubertConfig` configuration class: `HubertForCTC` (HubertConfig model) - `LasrCTCConfig` configuration class: `LasrForCTC` (LasrCTCConfig model) - `ParakeetCTCConfig` configuration class: `ParakeetForCTC` (ParakeetCTCConfig model) - `SEWConfig` configuration class: `SEWForCTC` (SEWConfig model) - `SEWDConfig` configuration class: `SEWDForCTC` (SEWDConfig model) - `UniSpeechConfig` configuration class: `UniSpeechForCTC` (UniSpeechConfig model) - `UniSpeechSatConfig` configuration class: `UniSpeechSatForCTC` (UniSpeechSatConfig model) - `Wav2Vec2BertConfig` configuration class: `Wav2Vec2BertForCTC` (Wav2Vec2BertConfig model) - `Wav2Vec2Config` configuration class: `Wav2Vec2ForCTC` (Wav2Vec2Config model) - `Wav2Vec2ConformerConfig` configuration class: `Wav2Vec2ConformerForCTC` (Wav2Vec2ConformerConfig model) - `WavLMConfig` configuration class: `WavLMForCTC` (WavLMConfig model) - **attn_implementation** (`str`, *optional*) -- The attention implementation to use in the model (if relevant). Can be any of `"eager"` (manual implementation of the attention), `"sdpa"` (using [`F.scaled_dot_product_attention`](https://pytorch.org/docs/master/generated/torch.nn.functional.scaled_dot_product_attention.html)), `"flash_attention_2"` (using [Dao-AILab/flash-attention](https://github.com/Dao-AILab/flash-attention)), or `"flash_attention_3"` (using [Dao-AILab/flash-attention/hopper](https://github.com/Dao-AILab/flash-attention/tree/main/hopper)). By default, if available, SDPA will be used for torch>=2.1.1. The default is otherwise the manual `"eager"` implementation.0 Instantiates one of the model classes of the library (with a connectionist temporal classification head) from a configuration. Note: Loading a model from its configuration file does **not** load the model weights. It only affects the model's configuration. Use [from_pretrained()](/docs/transformers/v5.6.0/ja/model_doc/auto#transformers.AutoModel.from_pretrained) to load the model weights. Examples: ```python >>> from transformers import AutoConfig, AutoModelForCTC >>> # Download configuration from huggingface.co and cache. >>> config = AutoConfig.from_pretrained("google-bert/bert-base-cased") >>> model = AutoModelForCTC.from_config(config) ``` **Parameters:** config ([PreTrainedConfig](/docs/transformers/v5.6.0/ja/main_classes/configuration#transformers.PreTrainedConfig)) : The model class to instantiate is selected based on the configuration class: - [Data2VecAudioConfig](/docs/transformers/v5.6.0/ja/model_doc/data2vec#transformers.Data2VecAudioConfig) configuration class: [Data2VecAudioForCTC](/docs/transformers/v5.6.0/ja/model_doc/data2vec#transformers.Data2VecAudioForCTC) (Data2VecAudioConfig model) - `HubertConfig` configuration class: `HubertForCTC` (HubertConfig model) - `LasrCTCConfig` configuration class: `LasrForCTC` (LasrCTCConfig model) - `ParakeetCTCConfig` configuration class: `ParakeetForCTC` (ParakeetCTCConfig model) - `SEWConfig` configuration class: `SEWForCTC` (SEWConfig model) - `SEWDConfig` configuration class: `SEWDForCTC` (SEWDConfig model) - `UniSpeechConfig` configuration class: `UniSpeechForCTC` (UniSpeechConfig model) - `UniSpeechSatConfig` configuration class: `UniSpeechSatForCTC` (UniSpeechSatConfig model) - `Wav2Vec2BertConfig` configuration class: `Wav2Vec2BertForCTC` (Wav2Vec2BertConfig model) - `Wav2Vec2Config` configuration class: `Wav2Vec2ForCTC` (Wav2Vec2Config model) - `Wav2Vec2ConformerConfig` configuration class: `Wav2Vec2ConformerForCTC` (Wav2Vec2ConformerConfig model) - `WavLMConfig` configuration class: `WavLMForCTC` (WavLMConfig model) attn_implementation (`str`, *optional*) : The attention implementation to use in the model (if relevant). Can be any of `"eager"` (manual implementation of the attention), `"sdpa"` (using [`F.scaled_dot_product_attention`](https://pytorch.org/docs/master/generated/torch.nn.functional.scaled_dot_product_attention.html)), `"flash_attention_2"` (using [Dao-AILab/flash-attention](https://github.com/Dao-AILab/flash-attention)), or `"flash_attention_3"` (using [Dao-AILab/flash-attention/hopper](https://github.com/Dao-AILab/flash-attention/tree/main/hopper)). By default, if available, SDPA will be used for torch>=2.1.1. The default is otherwise the manual `"eager"` implementation. #### from_pretrained[[transformers.AutoModelForCTC.from_pretrained]] [Source](https://github.com/huggingface/transformers/blob/v5.6.0/src/transformers/models/auto/auto_factory.py#L253) Instantiate one of the model classes of the library (with a connectionist temporal classification head) from a pretrained model. The model class to instantiate is selected based on the `model_type` property of the config object (either passed as an argument or loaded from `pretrained_model_name_or_path` if possible), or when it's missing, by falling back to using pattern matching on `pretrained_model_name_or_path`: - **data2vec-audio** -- [Data2VecAudioForCTC](/docs/transformers/v5.6.0/ja/model_doc/data2vec#transformers.Data2VecAudioForCTC) (Data2VecAudioConfig model) - **hubert** -- `HubertForCTC` (HubertConfig model) - **lasr_ctc** -- `LasrForCTC` (LasrCTCConfig model) - **parakeet_ctc** -- `ParakeetForCTC` (ParakeetCTCConfig model) - **sew** -- `SEWForCTC` (SEWConfig model) - **sew-d** -- `SEWDForCTC` (SEWDConfig model) - **unispeech** -- `UniSpeechForCTC` (UniSpeechConfig model) - **unispeech-sat** -- `UniSpeechSatForCTC` (UniSpeechSatConfig model) - **wav2vec2** -- `Wav2Vec2ForCTC` (Wav2Vec2Config model) - **wav2vec2-bert** -- `Wav2Vec2BertForCTC` (Wav2Vec2BertConfig model) - **wav2vec2-conformer** -- `Wav2Vec2ConformerForCTC` (Wav2Vec2ConformerConfig model) - **wavlm** -- `WavLMForCTC` (WavLMConfig model) The model is set in evaluation mode by default using `model.eval()` (so for instance, dropout modules are deactivated). To train the model, you should first set it back in training mode with `model.train()` Examples: ```python >>> from transformers import AutoConfig, AutoModelForCTC >>> # Download model and configuration from huggingface.co and cache. >>> model = AutoModelForCTC.from_pretrained("google-bert/bert-base-cased") >>> # Update configuration during loading >>> model = AutoModelForCTC.from_pretrained("google-bert/bert-base-cased", output_attentions=True) >>> model.config.output_attentions True ``` **Parameters:** pretrained_model_name_or_path (`str` or `os.PathLike`) : Can be either: - A string, the *model id* of a pretrained model hosted inside a model repo on huggingface.co. - A path to a *directory* containing model weights saved using [save_pretrained()](/docs/transformers/v5.6.0/ja/main_classes/model#transformers.PreTrainedModel.save_pretrained), e.g., `./my_model_directory/`. model_args (additional positional arguments, *optional*) : Will be passed along to the underlying model `__init__()` method. config ([PreTrainedConfig](/docs/transformers/v5.6.0/ja/main_classes/configuration#transformers.PreTrainedConfig), *optional*) : Configuration for the model to use instead of an automatically loaded configuration. Configuration can be automatically loaded when: - The model is a model provided by the library (loaded with the *model id* string of a pretrained model). - The model was saved using [save_pretrained()](/docs/transformers/v5.6.0/ja/main_classes/model#transformers.PreTrainedModel.save_pretrained) and is reloaded by supplying the save directory. - The model is loaded by supplying a local directory as `pretrained_model_name_or_path` and a configuration JSON file named *config.json* is found in the directory. state_dict (*dict[str, torch.Tensor]*, *optional*) : A state dictionary to use instead of a state dictionary loaded from saved weights file. This option can be used if you want to create a model from a pretrained configuration but load your own weights. In this case though, you should check if using [save_pretrained()](/docs/transformers/v5.6.0/ja/main_classes/model#transformers.PreTrainedModel.save_pretrained) and [from_pretrained()](/docs/transformers/v5.6.0/ja/main_classes/model#transformers.PreTrainedModel.from_pretrained) is not a simpler option. cache_dir (`str` or `os.PathLike`, *optional*) : Path to a directory in which a downloaded pretrained model configuration should be cached if the standard cache should not be used. force_download (`bool`, *optional*, defaults to `False`) : Whether or not to force the (re-)download of the model weights and configuration files, overriding the cached versions if they exist. proxies (`dict[str, str]`, *optional*) : A dictionary of proxy servers to use by protocol or endpoint, e.g., `{'http': 'foo.bar:3128', 'http://hostname': 'foo.bar:4012'}`. The proxies are used on each request. output_loading_info(`bool`, *optional*, defaults to `False`) : Whether or not to also return a dictionary containing missing keys, unexpected keys and error messages. local_files_only(`bool`, *optional*, defaults to `False`) : Whether or not to only look at local files (e.g., not try downloading the model). revision (`str`, *optional*, defaults to `"main"`) : The specific model version to use. It can be a branch name, a tag name, or a commit id, since we use a git-based system for storing models and other artifacts on huggingface.co, so `revision` can be any identifier allowed by git. trust_remote_code (`bool`, *optional*, defaults to `False`) : Whether or not to allow for custom models defined on the Hub in their own modeling files. This option should only be set to `True` for repositories you trust and in which you have read the code, as it will execute code present on the Hub on your local machine. code_revision (`str`, *optional*, defaults to `"main"`) : The specific revision to use for the code on the Hub, if the code leaves in a different repository than the rest of the model. It can be a branch name, a tag name, or a commit id, since we use a git-based system for storing models and other artifacts on huggingface.co, so `revision` can be any identifier allowed by git. kwargs (additional keyword arguments, *optional*) : Can be used to update the configuration object (after it being loaded) and initiate the model (e.g., `output_attentions=True`). Behaves differently depending on whether a `config` is provided or automatically loaded: - If a configuration is provided with `config`, `**kwargs` will be directly passed to the underlying model's `__init__` method (we assume all relevant updates to the configuration have already been done) - If a configuration is not provided, `kwargs` will be first passed to the configuration class initialization function ([from_pretrained()](/docs/transformers/v5.6.0/ja/main_classes/configuration#transformers.PreTrainedConfig.from_pretrained)). Each key of `kwargs` that corresponds to a configuration attribute will be used to override said attribute with the supplied `kwargs` value. Remaining keys that do not correspond to any configuration attribute will be passed to the underlying model's `__init__` function. ### AutoModelForSpeechSeq2Seq[[transformers.AutoModelForSpeechSeq2Seq]] #### transformers.AutoModelForSpeechSeq2Seq[[transformers.AutoModelForSpeechSeq2Seq]] [Source](https://github.com/huggingface/transformers/blob/v5.6.0/src/transformers/models/auto/modeling_auto.py#L2241) This is a generic model class that will be instantiated as one of the model classes of the library (with a sequence-to-sequence speech-to-text modeling head) when created with the [from_pretrained()](/docs/transformers/v5.6.0/ja/model_doc/auto#transformers.AutoModel.from_pretrained) class method or the [from_config()](/docs/transformers/v5.6.0/ja/model_doc/auto#transformers.AutoModel.from_config) class method. This class cannot be instantiated directly using `__init__()` (throws an error). from_configtransformers.AutoModelForSpeechSeq2Seq.from_confighttps://github.com/huggingface/transformers/blob/v5.6.0/src/transformers/models/auto/auto_factory.py#L206[{"name": "**kwargs", "val": ""}]- **config** ([PreTrainedConfig](/docs/transformers/v5.6.0/ja/main_classes/configuration#transformers.PreTrainedConfig)) -- The model class to instantiate is selected based on the configuration class: - `CohereAsrConfig` configuration class: `CohereAsrForConditionalGeneration` (CohereAsrConfig model) - `DiaConfig` configuration class: `DiaForConditionalGeneration` (DiaConfig model) - `GraniteSpeechConfig` configuration class: `GraniteSpeechForConditionalGeneration` (GraniteSpeechConfig model) - `KyutaiSpeechToTextConfig` configuration class: `KyutaiSpeechToTextForConditionalGeneration` (KyutaiSpeechToTextConfig model) - `MoonshineConfig` configuration class: `MoonshineForConditionalGeneration` (MoonshineConfig model) - `MoonshineStreamingConfig` configuration class: `MoonshineStreamingForConditionalGeneration` (MoonshineStreamingConfig model) - `Pop2PianoConfig` configuration class: `Pop2PianoForConditionalGeneration` (Pop2PianoConfig model) - `SeamlessM4TConfig` configuration class: `SeamlessM4TForSpeechToText` (SeamlessM4TConfig model) - `SeamlessM4Tv2Config` configuration class: `SeamlessM4Tv2ForSpeechToText` (SeamlessM4Tv2Config model) - `Speech2TextConfig` configuration class: `Speech2TextForConditionalGeneration` (Speech2TextConfig model) - `SpeechEncoderDecoderConfig` configuration class: `SpeechEncoderDecoderModel` (SpeechEncoderDecoderConfig model) - `SpeechT5Config` configuration class: `SpeechT5ForSpeechToText` (SpeechT5Config model) - `VibeVoiceAsrConfig` configuration class: `VibeVoiceAsrForConditionalGeneration` (VibeVoiceAsrConfig model) - `VoxtralConfig` configuration class: `VoxtralForConditionalGeneration` (VoxtralConfig model) - `VoxtralRealtimeConfig` configuration class: `VoxtralRealtimeForConditionalGeneration` (VoxtralRealtimeConfig model) - `WhisperConfig` configuration class: `WhisperForConditionalGeneration` (WhisperConfig model) - **attn_implementation** (`str`, *optional*) -- The attention implementation to use in the model (if relevant). Can be any of `"eager"` (manual implementation of the attention), `"sdpa"` (using [`F.scaled_dot_product_attention`](https://pytorch.org/docs/master/generated/torch.nn.functional.scaled_dot_product_attention.html)), `"flash_attention_2"` (using [Dao-AILab/flash-attention](https://github.com/Dao-AILab/flash-attention)), or `"flash_attention_3"` (using [Dao-AILab/flash-attention/hopper](https://github.com/Dao-AILab/flash-attention/tree/main/hopper)). By default, if available, SDPA will be used for torch>=2.1.1. The default is otherwise the manual `"eager"` implementation.0 Instantiates one of the model classes of the library (with a sequence-to-sequence speech-to-text modeling head) from a configuration. Note: Loading a model from its configuration file does **not** load the model weights. It only affects the model's configuration. Use [from_pretrained()](/docs/transformers/v5.6.0/ja/model_doc/auto#transformers.AutoModel.from_pretrained) to load the model weights. Examples: ```python >>> from transformers import AutoConfig, AutoModelForSpeechSeq2Seq >>> # Download configuration from huggingface.co and cache. >>> config = AutoConfig.from_pretrained("google-bert/bert-base-cased") >>> model = AutoModelForSpeechSeq2Seq.from_config(config) ``` **Parameters:** config ([PreTrainedConfig](/docs/transformers/v5.6.0/ja/main_classes/configuration#transformers.PreTrainedConfig)) : The model class to instantiate is selected based on the configuration class: - `CohereAsrConfig` configuration class: `CohereAsrForConditionalGeneration` (CohereAsrConfig model) - `DiaConfig` configuration class: `DiaForConditionalGeneration` (DiaConfig model) - `GraniteSpeechConfig` configuration class: `GraniteSpeechForConditionalGeneration` (GraniteSpeechConfig model) - `KyutaiSpeechToTextConfig` configuration class: `KyutaiSpeechToTextForConditionalGeneration` (KyutaiSpeechToTextConfig model) - `MoonshineConfig` configuration class: `MoonshineForConditionalGeneration` (MoonshineConfig model) - `MoonshineStreamingConfig` configuration class: `MoonshineStreamingForConditionalGeneration` (MoonshineStreamingConfig model) - `Pop2PianoConfig` configuration class: `Pop2PianoForConditionalGeneration` (Pop2PianoConfig model) - `SeamlessM4TConfig` configuration class: `SeamlessM4TForSpeechToText` (SeamlessM4TConfig model) - `SeamlessM4Tv2Config` configuration class: `SeamlessM4Tv2ForSpeechToText` (SeamlessM4Tv2Config model) - `Speech2TextConfig` configuration class: `Speech2TextForConditionalGeneration` (Speech2TextConfig model) - `SpeechEncoderDecoderConfig` configuration class: `SpeechEncoderDecoderModel` (SpeechEncoderDecoderConfig model) - `SpeechT5Config` configuration class: `SpeechT5ForSpeechToText` (SpeechT5Config model) - `VibeVoiceAsrConfig` configuration class: `VibeVoiceAsrForConditionalGeneration` (VibeVoiceAsrConfig model) - `VoxtralConfig` configuration class: `VoxtralForConditionalGeneration` (VoxtralConfig model) - `VoxtralRealtimeConfig` configuration class: `VoxtralRealtimeForConditionalGeneration` (VoxtralRealtimeConfig model) - `WhisperConfig` configuration class: `WhisperForConditionalGeneration` (WhisperConfig model) attn_implementation (`str`, *optional*) : The attention implementation to use in the model (if relevant). Can be any of `"eager"` (manual implementation of the attention), `"sdpa"` (using [`F.scaled_dot_product_attention`](https://pytorch.org/docs/master/generated/torch.nn.functional.scaled_dot_product_attention.html)), `"flash_attention_2"` (using [Dao-AILab/flash-attention](https://github.com/Dao-AILab/flash-attention)), or `"flash_attention_3"` (using [Dao-AILab/flash-attention/hopper](https://github.com/Dao-AILab/flash-attention/tree/main/hopper)). By default, if available, SDPA will be used for torch>=2.1.1. The default is otherwise the manual `"eager"` implementation. #### from_pretrained[[transformers.AutoModelForSpeechSeq2Seq.from_pretrained]] [Source](https://github.com/huggingface/transformers/blob/v5.6.0/src/transformers/models/auto/auto_factory.py#L253) Instantiate one of the model classes of the library (with a sequence-to-sequence speech-to-text modeling head) from a pretrained model. The model class to instantiate is selected based on the `model_type` property of the config object (either passed as an argument or loaded from `pretrained_model_name_or_path` if possible), or when it's missing, by falling back to using pattern matching on `pretrained_model_name_or_path`: - **cohere_asr** -- `CohereAsrForConditionalGeneration` (CohereAsrConfig model) - **dia** -- `DiaForConditionalGeneration` (DiaConfig model) - **granite_speech** -- `GraniteSpeechForConditionalGeneration` (GraniteSpeechConfig model) - **kyutai_speech_to_text** -- `KyutaiSpeechToTextForConditionalGeneration` (KyutaiSpeechToTextConfig model) - **moonshine** -- `MoonshineForConditionalGeneration` (MoonshineConfig model) - **moonshine_streaming** -- `MoonshineStreamingForConditionalGeneration` (MoonshineStreamingConfig model) - **pop2piano** -- `Pop2PianoForConditionalGeneration` (Pop2PianoConfig model) - **seamless_m4t** -- `SeamlessM4TForSpeechToText` (SeamlessM4TConfig model) - **seamless_m4t_v2** -- `SeamlessM4Tv2ForSpeechToText` (SeamlessM4Tv2Config model) - **speech-encoder-decoder** -- `SpeechEncoderDecoderModel` (SpeechEncoderDecoderConfig model) - **speech_to_text** -- `Speech2TextForConditionalGeneration` (Speech2TextConfig model) - **speecht5** -- `SpeechT5ForSpeechToText` (SpeechT5Config model) - **vibevoice_asr** -- `VibeVoiceAsrForConditionalGeneration` (VibeVoiceAsrConfig model) - **voxtral** -- `VoxtralForConditionalGeneration` (VoxtralConfig model) - **voxtral_realtime** -- `VoxtralRealtimeForConditionalGeneration` (VoxtralRealtimeConfig model) - **whisper** -- `WhisperForConditionalGeneration` (WhisperConfig model) The model is set in evaluation mode by default using `model.eval()` (so for instance, dropout modules are deactivated). To train the model, you should first set it back in training mode with `model.train()` Examples: ```python >>> from transformers import AutoConfig, AutoModelForSpeechSeq2Seq >>> # Download model and configuration from huggingface.co and cache. >>> model = AutoModelForSpeechSeq2Seq.from_pretrained("google-bert/bert-base-cased") >>> # Update configuration during loading >>> model = AutoModelForSpeechSeq2Seq.from_pretrained("google-bert/bert-base-cased", output_attentions=True) >>> model.config.output_attentions True ``` **Parameters:** pretrained_model_name_or_path (`str` or `os.PathLike`) : Can be either: - A string, the *model id* of a pretrained model hosted inside a model repo on huggingface.co. - A path to a *directory* containing model weights saved using [save_pretrained()](/docs/transformers/v5.6.0/ja/main_classes/model#transformers.PreTrainedModel.save_pretrained), e.g., `./my_model_directory/`. model_args (additional positional arguments, *optional*) : Will be passed along to the underlying model `__init__()` method. config ([PreTrainedConfig](/docs/transformers/v5.6.0/ja/main_classes/configuration#transformers.PreTrainedConfig), *optional*) : Configuration for the model to use instead of an automatically loaded configuration. Configuration can be automatically loaded when: - The model is a model provided by the library (loaded with the *model id* string of a pretrained model). - The model was saved using [save_pretrained()](/docs/transformers/v5.6.0/ja/main_classes/model#transformers.PreTrainedModel.save_pretrained) and is reloaded by supplying the save directory. - The model is loaded by supplying a local directory as `pretrained_model_name_or_path` and a configuration JSON file named *config.json* is found in the directory. state_dict (*dict[str, torch.Tensor]*, *optional*) : A state dictionary to use instead of a state dictionary loaded from saved weights file. This option can be used if you want to create a model from a pretrained configuration but load your own weights. In this case though, you should check if using [save_pretrained()](/docs/transformers/v5.6.0/ja/main_classes/model#transformers.PreTrainedModel.save_pretrained) and [from_pretrained()](/docs/transformers/v5.6.0/ja/main_classes/model#transformers.PreTrainedModel.from_pretrained) is not a simpler option. cache_dir (`str` or `os.PathLike`, *optional*) : Path to a directory in which a downloaded pretrained model configuration should be cached if the standard cache should not be used. force_download (`bool`, *optional*, defaults to `False`) : Whether or not to force the (re-)download of the model weights and configuration files, overriding the cached versions if they exist. proxies (`dict[str, str]`, *optional*) : A dictionary of proxy servers to use by protocol or endpoint, e.g., `{'http': 'foo.bar:3128', 'http://hostname': 'foo.bar:4012'}`. The proxies are used on each request. output_loading_info(`bool`, *optional*, defaults to `False`) : Whether or not to also return a dictionary containing missing keys, unexpected keys and error messages. local_files_only(`bool`, *optional*, defaults to `False`) : Whether or not to only look at local files (e.g., not try downloading the model). revision (`str`, *optional*, defaults to `"main"`) : The specific model version to use. It can be a branch name, a tag name, or a commit id, since we use a git-based system for storing models and other artifacts on huggingface.co, so `revision` can be any identifier allowed by git. trust_remote_code (`bool`, *optional*, defaults to `False`) : Whether or not to allow for custom models defined on the Hub in their own modeling files. This option should only be set to `True` for repositories you trust and in which you have read the code, as it will execute code present on the Hub on your local machine. code_revision (`str`, *optional*, defaults to `"main"`) : The specific revision to use for the code on the Hub, if the code leaves in a different repository than the rest of the model. It can be a branch name, a tag name, or a commit id, since we use a git-based system for storing models and other artifacts on huggingface.co, so `revision` can be any identifier allowed by git. kwargs (additional keyword arguments, *optional*) : Can be used to update the configuration object (after it being loaded) and initiate the model (e.g., `output_attentions=True`). Behaves differently depending on whether a `config` is provided or automatically loaded: - If a configuration is provided with `config`, `**kwargs` will be directly passed to the underlying model's `__init__` method (we assume all relevant updates to the configuration have already been done) - If a configuration is not provided, `kwargs` will be first passed to the configuration class initialization function ([from_pretrained()](/docs/transformers/v5.6.0/ja/main_classes/configuration#transformers.PreTrainedConfig.from_pretrained)). Each key of `kwargs` that corresponds to a configuration attribute will be used to override said attribute with the supplied `kwargs` value. Remaining keys that do not correspond to any configuration attribute will be passed to the underlying model's `__init__` function. ### AutoModelForAudioXVector[[transformers.AutoModelForAudioXVector]] #### transformers.AutoModelForAudioXVector[[transformers.AutoModelForAudioXVector]] [Source](https://github.com/huggingface/transformers/blob/v5.6.0/src/transformers/models/auto/modeling_auto.py#L2259) This is a generic model class that will be instantiated as one of the model classes of the library (with a audio retrieval via x-vector head) when created with the [from_pretrained()](/docs/transformers/v5.6.0/ja/model_doc/auto#transformers.AutoModel.from_pretrained) class method or the [from_config()](/docs/transformers/v5.6.0/ja/model_doc/auto#transformers.AutoModel.from_config) class method. This class cannot be instantiated directly using `__init__()` (throws an error). from_configtransformers.AutoModelForAudioXVector.from_confighttps://github.com/huggingface/transformers/blob/v5.6.0/src/transformers/models/auto/auto_factory.py#L206[{"name": "**kwargs", "val": ""}]- **config** ([PreTrainedConfig](/docs/transformers/v5.6.0/ja/main_classes/configuration#transformers.PreTrainedConfig)) -- The model class to instantiate is selected based on the configuration class: - [Data2VecAudioConfig](/docs/transformers/v5.6.0/ja/model_doc/data2vec#transformers.Data2VecAudioConfig) configuration class: [Data2VecAudioForXVector](/docs/transformers/v5.6.0/ja/model_doc/data2vec#transformers.Data2VecAudioForXVector) (Data2VecAudioConfig model) - `UniSpeechSatConfig` configuration class: `UniSpeechSatForXVector` (UniSpeechSatConfig model) - `Wav2Vec2BertConfig` configuration class: `Wav2Vec2BertForXVector` (Wav2Vec2BertConfig model) - `Wav2Vec2Config` configuration class: `Wav2Vec2ForXVector` (Wav2Vec2Config model) - `Wav2Vec2ConformerConfig` configuration class: `Wav2Vec2ConformerForXVector` (Wav2Vec2ConformerConfig model) - `WavLMConfig` configuration class: `WavLMForXVector` (WavLMConfig model) - **attn_implementation** (`str`, *optional*) -- The attention implementation to use in the model (if relevant). Can be any of `"eager"` (manual implementation of the attention), `"sdpa"` (using [`F.scaled_dot_product_attention`](https://pytorch.org/docs/master/generated/torch.nn.functional.scaled_dot_product_attention.html)), `"flash_attention_2"` (using [Dao-AILab/flash-attention](https://github.com/Dao-AILab/flash-attention)), or `"flash_attention_3"` (using [Dao-AILab/flash-attention/hopper](https://github.com/Dao-AILab/flash-attention/tree/main/hopper)). By default, if available, SDPA will be used for torch>=2.1.1. The default is otherwise the manual `"eager"` implementation.0 Instantiates one of the model classes of the library (with a audio retrieval via x-vector head) from a configuration. Note: Loading a model from its configuration file does **not** load the model weights. It only affects the model's configuration. Use [from_pretrained()](/docs/transformers/v5.6.0/ja/model_doc/auto#transformers.AutoModel.from_pretrained) to load the model weights. Examples: ```python >>> from transformers import AutoConfig, AutoModelForAudioXVector >>> # Download configuration from huggingface.co and cache. >>> config = AutoConfig.from_pretrained("google-bert/bert-base-cased") >>> model = AutoModelForAudioXVector.from_config(config) ``` **Parameters:** config ([PreTrainedConfig](/docs/transformers/v5.6.0/ja/main_classes/configuration#transformers.PreTrainedConfig)) : The model class to instantiate is selected based on the configuration class: - [Data2VecAudioConfig](/docs/transformers/v5.6.0/ja/model_doc/data2vec#transformers.Data2VecAudioConfig) configuration class: [Data2VecAudioForXVector](/docs/transformers/v5.6.0/ja/model_doc/data2vec#transformers.Data2VecAudioForXVector) (Data2VecAudioConfig model) - `UniSpeechSatConfig` configuration class: `UniSpeechSatForXVector` (UniSpeechSatConfig model) - `Wav2Vec2BertConfig` configuration class: `Wav2Vec2BertForXVector` (Wav2Vec2BertConfig model) - `Wav2Vec2Config` configuration class: `Wav2Vec2ForXVector` (Wav2Vec2Config model) - `Wav2Vec2ConformerConfig` configuration class: `Wav2Vec2ConformerForXVector` (Wav2Vec2ConformerConfig model) - `WavLMConfig` configuration class: `WavLMForXVector` (WavLMConfig model) attn_implementation (`str`, *optional*) : The attention implementation to use in the model (if relevant). Can be any of `"eager"` (manual implementation of the attention), `"sdpa"` (using [`F.scaled_dot_product_attention`](https://pytorch.org/docs/master/generated/torch.nn.functional.scaled_dot_product_attention.html)), `"flash_attention_2"` (using [Dao-AILab/flash-attention](https://github.com/Dao-AILab/flash-attention)), or `"flash_attention_3"` (using [Dao-AILab/flash-attention/hopper](https://github.com/Dao-AILab/flash-attention/tree/main/hopper)). By default, if available, SDPA will be used for torch>=2.1.1. The default is otherwise the manual `"eager"` implementation. #### from_pretrained[[transformers.AutoModelForAudioXVector.from_pretrained]] [Source](https://github.com/huggingface/transformers/blob/v5.6.0/src/transformers/models/auto/auto_factory.py#L253) Instantiate one of the model classes of the library (with a audio retrieval via x-vector head) from a pretrained model. The model class to instantiate is selected based on the `model_type` property of the config object (either passed as an argument or loaded from `pretrained_model_name_or_path` if possible), or when it's missing, by falling back to using pattern matching on `pretrained_model_name_or_path`: - **data2vec-audio** -- [Data2VecAudioForXVector](/docs/transformers/v5.6.0/ja/model_doc/data2vec#transformers.Data2VecAudioForXVector) (Data2VecAudioConfig model) - **unispeech-sat** -- `UniSpeechSatForXVector` (UniSpeechSatConfig model) - **wav2vec2** -- `Wav2Vec2ForXVector` (Wav2Vec2Config model) - **wav2vec2-bert** -- `Wav2Vec2BertForXVector` (Wav2Vec2BertConfig model) - **wav2vec2-conformer** -- `Wav2Vec2ConformerForXVector` (Wav2Vec2ConformerConfig model) - **wavlm** -- `WavLMForXVector` (WavLMConfig model) The model is set in evaluation mode by default using `model.eval()` (so for instance, dropout modules are deactivated). To train the model, you should first set it back in training mode with `model.train()` Examples: ```python >>> from transformers import AutoConfig, AutoModelForAudioXVector >>> # Download model and configuration from huggingface.co and cache. >>> model = AutoModelForAudioXVector.from_pretrained("google-bert/bert-base-cased") >>> # Update configuration during loading >>> model = AutoModelForAudioXVector.from_pretrained("google-bert/bert-base-cased", output_attentions=True) >>> model.config.output_attentions True ``` **Parameters:** pretrained_model_name_or_path (`str` or `os.PathLike`) : Can be either: - A string, the *model id* of a pretrained model hosted inside a model repo on huggingface.co. - A path to a *directory* containing model weights saved using [save_pretrained()](/docs/transformers/v5.6.0/ja/main_classes/model#transformers.PreTrainedModel.save_pretrained), e.g., `./my_model_directory/`. model_args (additional positional arguments, *optional*) : Will be passed along to the underlying model `__init__()` method. config ([PreTrainedConfig](/docs/transformers/v5.6.0/ja/main_classes/configuration#transformers.PreTrainedConfig), *optional*) : Configuration for the model to use instead of an automatically loaded configuration. Configuration can be automatically loaded when: - The model is a model provided by the library (loaded with the *model id* string of a pretrained model). - The model was saved using [save_pretrained()](/docs/transformers/v5.6.0/ja/main_classes/model#transformers.PreTrainedModel.save_pretrained) and is reloaded by supplying the save directory. - The model is loaded by supplying a local directory as `pretrained_model_name_or_path` and a configuration JSON file named *config.json* is found in the directory. state_dict (*dict[str, torch.Tensor]*, *optional*) : A state dictionary to use instead of a state dictionary loaded from saved weights file. This option can be used if you want to create a model from a pretrained configuration but load your own weights. In this case though, you should check if using [save_pretrained()](/docs/transformers/v5.6.0/ja/main_classes/model#transformers.PreTrainedModel.save_pretrained) and [from_pretrained()](/docs/transformers/v5.6.0/ja/main_classes/model#transformers.PreTrainedModel.from_pretrained) is not a simpler option. cache_dir (`str` or `os.PathLike`, *optional*) : Path to a directory in which a downloaded pretrained model configuration should be cached if the standard cache should not be used. force_download (`bool`, *optional*, defaults to `False`) : Whether or not to force the (re-)download of the model weights and configuration files, overriding the cached versions if they exist. proxies (`dict[str, str]`, *optional*) : A dictionary of proxy servers to use by protocol or endpoint, e.g., `{'http': 'foo.bar:3128', 'http://hostname': 'foo.bar:4012'}`. The proxies are used on each request. output_loading_info(`bool`, *optional*, defaults to `False`) : Whether or not to also return a dictionary containing missing keys, unexpected keys and error messages. local_files_only(`bool`, *optional*, defaults to `False`) : Whether or not to only look at local files (e.g., not try downloading the model). revision (`str`, *optional*, defaults to `"main"`) : The specific model version to use. It can be a branch name, a tag name, or a commit id, since we use a git-based system for storing models and other artifacts on huggingface.co, so `revision` can be any identifier allowed by git. trust_remote_code (`bool`, *optional*, defaults to `False`) : Whether or not to allow for custom models defined on the Hub in their own modeling files. This option should only be set to `True` for repositories you trust and in which you have read the code, as it will execute code present on the Hub on your local machine. code_revision (`str`, *optional*, defaults to `"main"`) : The specific revision to use for the code on the Hub, if the code leaves in a different repository than the rest of the model. It can be a branch name, a tag name, or a commit id, since we use a git-based system for storing models and other artifacts on huggingface.co, so `revision` can be any identifier allowed by git. kwargs (additional keyword arguments, *optional*) : Can be used to update the configuration object (after it being loaded) and initiate the model (e.g., `output_attentions=True`). Behaves differently depending on whether a `config` is provided or automatically loaded: - If a configuration is provided with `config`, `**kwargs` will be directly passed to the underlying model's `__init__` method (we assume all relevant updates to the configuration have already been done) - If a configuration is not provided, `kwargs` will be first passed to the configuration class initialization function ([from_pretrained()](/docs/transformers/v5.6.0/ja/main_classes/configuration#transformers.PreTrainedConfig.from_pretrained)). Each key of `kwargs` that corresponds to a configuration attribute will be used to override said attribute with the supplied `kwargs` value. Remaining keys that do not correspond to any configuration attribute will be passed to the underlying model's `__init__` function. ### AutoModelForTextToSpectrogram[[transformers.AutoModelForTextToSpectrogram]] #### transformers.AutoModelForTextToSpectrogram[[transformers.AutoModelForTextToSpectrogram]] [Source](https://github.com/huggingface/transformers/blob/v5.6.0/src/transformers/models/auto/modeling_auto.py#L2263) ### AutoModelForTextToWaveform[[transformers.AutoModelForTextToWaveform]] #### transformers.AutoModelForTextToWaveform[[transformers.AutoModelForTextToWaveform]] [Source](https://github.com/huggingface/transformers/blob/v5.6.0/src/transformers/models/auto/modeling_auto.py#L2267) ## Multimodal 以下の自動クラスは、次のマルチモーダルタスクに利用可能です。 ### AutoModelForTableQuestionAnswering[[transformers.AutoModelForTableQuestionAnswering]] #### transformers.AutoModelForTableQuestionAnswering[[transformers.AutoModelForTableQuestionAnswering]] [Source](https://github.com/huggingface/transformers/blob/v5.6.0/src/transformers/models/auto/modeling_auto.py#L2044) This is a generic model class that will be instantiated as one of the model classes of the library (with a table question answering head) when created with the [from_pretrained()](/docs/transformers/v5.6.0/ja/model_doc/auto#transformers.AutoModel.from_pretrained) class method or the [from_config()](/docs/transformers/v5.6.0/ja/model_doc/auto#transformers.AutoModel.from_config) class method. This class cannot be instantiated directly using `__init__()` (throws an error). from_configtransformers.AutoModelForTableQuestionAnswering.from_confighttps://github.com/huggingface/transformers/blob/v5.6.0/src/transformers/models/auto/auto_factory.py#L206[{"name": "**kwargs", "val": ""}]- **config** ([PreTrainedConfig](/docs/transformers/v5.6.0/ja/main_classes/configuration#transformers.PreTrainedConfig)) -- The model class to instantiate is selected based on the configuration class: - `TapasConfig` configuration class: `TapasForQuestionAnswering` (TapasConfig model) - **attn_implementation** (`str`, *optional*) -- The attention implementation to use in the model (if relevant). Can be any of `"eager"` (manual implementation of the attention), `"sdpa"` (using [`F.scaled_dot_product_attention`](https://pytorch.org/docs/master/generated/torch.nn.functional.scaled_dot_product_attention.html)), `"flash_attention_2"` (using [Dao-AILab/flash-attention](https://github.com/Dao-AILab/flash-attention)), or `"flash_attention_3"` (using [Dao-AILab/flash-attention/hopper](https://github.com/Dao-AILab/flash-attention/tree/main/hopper)). By default, if available, SDPA will be used for torch>=2.1.1. The default is otherwise the manual `"eager"` implementation.0 Instantiates one of the model classes of the library (with a table question answering head) from a configuration. Note: Loading a model from its configuration file does **not** load the model weights. It only affects the model's configuration. Use [from_pretrained()](/docs/transformers/v5.6.0/ja/model_doc/auto#transformers.AutoModel.from_pretrained) to load the model weights. Examples: ```python >>> from transformers import AutoConfig, AutoModelForTableQuestionAnswering >>> # Download configuration from huggingface.co and cache. >>> config = AutoConfig.from_pretrained("google/tapas-base-finetuned-wtq") >>> model = AutoModelForTableQuestionAnswering.from_config(config) ``` **Parameters:** config ([PreTrainedConfig](/docs/transformers/v5.6.0/ja/main_classes/configuration#transformers.PreTrainedConfig)) : The model class to instantiate is selected based on the configuration class: - `TapasConfig` configuration class: `TapasForQuestionAnswering` (TapasConfig model) attn_implementation (`str`, *optional*) : The attention implementation to use in the model (if relevant). Can be any of `"eager"` (manual implementation of the attention), `"sdpa"` (using [`F.scaled_dot_product_attention`](https://pytorch.org/docs/master/generated/torch.nn.functional.scaled_dot_product_attention.html)), `"flash_attention_2"` (using [Dao-AILab/flash-attention](https://github.com/Dao-AILab/flash-attention)), or `"flash_attention_3"` (using [Dao-AILab/flash-attention/hopper](https://github.com/Dao-AILab/flash-attention/tree/main/hopper)). By default, if available, SDPA will be used for torch>=2.1.1. The default is otherwise the manual `"eager"` implementation. #### from_pretrained[[transformers.AutoModelForTableQuestionAnswering.from_pretrained]] [Source](https://github.com/huggingface/transformers/blob/v5.6.0/src/transformers/models/auto/auto_factory.py#L253) Instantiate one of the model classes of the library (with a table question answering head) from a pretrained model. The model class to instantiate is selected based on the `model_type` property of the config object (either passed as an argument or loaded from `pretrained_model_name_or_path` if possible), or when it's missing, by falling back to using pattern matching on `pretrained_model_name_or_path`: - **tapas** -- `TapasForQuestionAnswering` (TapasConfig model) The model is set in evaluation mode by default using `model.eval()` (so for instance, dropout modules are deactivated). To train the model, you should first set it back in training mode with `model.train()` Examples: ```python >>> from transformers import AutoConfig, AutoModelForTableQuestionAnswering >>> # Download model and configuration from huggingface.co and cache. >>> model = AutoModelForTableQuestionAnswering.from_pretrained("google/tapas-base-finetuned-wtq") >>> # Update configuration during loading >>> model = AutoModelForTableQuestionAnswering.from_pretrained("google/tapas-base-finetuned-wtq", output_attentions=True) >>> model.config.output_attentions True ``` **Parameters:** pretrained_model_name_or_path (`str` or `os.PathLike`) : Can be either: - A string, the *model id* of a pretrained model hosted inside a model repo on huggingface.co. - A path to a *directory* containing model weights saved using [save_pretrained()](/docs/transformers/v5.6.0/ja/main_classes/model#transformers.PreTrainedModel.save_pretrained), e.g., `./my_model_directory/`. model_args (additional positional arguments, *optional*) : Will be passed along to the underlying model `__init__()` method. config ([PreTrainedConfig](/docs/transformers/v5.6.0/ja/main_classes/configuration#transformers.PreTrainedConfig), *optional*) : Configuration for the model to use instead of an automatically loaded configuration. Configuration can be automatically loaded when: - The model is a model provided by the library (loaded with the *model id* string of a pretrained model). - The model was saved using [save_pretrained()](/docs/transformers/v5.6.0/ja/main_classes/model#transformers.PreTrainedModel.save_pretrained) and is reloaded by supplying the save directory. - The model is loaded by supplying a local directory as `pretrained_model_name_or_path` and a configuration JSON file named *config.json* is found in the directory. state_dict (*dict[str, torch.Tensor]*, *optional*) : A state dictionary to use instead of a state dictionary loaded from saved weights file. This option can be used if you want to create a model from a pretrained configuration but load your own weights. In this case though, you should check if using [save_pretrained()](/docs/transformers/v5.6.0/ja/main_classes/model#transformers.PreTrainedModel.save_pretrained) and [from_pretrained()](/docs/transformers/v5.6.0/ja/main_classes/model#transformers.PreTrainedModel.from_pretrained) is not a simpler option. cache_dir (`str` or `os.PathLike`, *optional*) : Path to a directory in which a downloaded pretrained model configuration should be cached if the standard cache should not be used. force_download (`bool`, *optional*, defaults to `False`) : Whether or not to force the (re-)download of the model weights and configuration files, overriding the cached versions if they exist. proxies (`dict[str, str]`, *optional*) : A dictionary of proxy servers to use by protocol or endpoint, e.g., `{'http': 'foo.bar:3128', 'http://hostname': 'foo.bar:4012'}`. The proxies are used on each request. output_loading_info(`bool`, *optional*, defaults to `False`) : Whether or not to also return a dictionary containing missing keys, unexpected keys and error messages. local_files_only(`bool`, *optional*, defaults to `False`) : Whether or not to only look at local files (e.g., not try downloading the model). revision (`str`, *optional*, defaults to `"main"`) : The specific model version to use. It can be a branch name, a tag name, or a commit id, since we use a git-based system for storing models and other artifacts on huggingface.co, so `revision` can be any identifier allowed by git. trust_remote_code (`bool`, *optional*, defaults to `False`) : Whether or not to allow for custom models defined on the Hub in their own modeling files. This option should only be set to `True` for repositories you trust and in which you have read the code, as it will execute code present on the Hub on your local machine. code_revision (`str`, *optional*, defaults to `"main"`) : The specific revision to use for the code on the Hub, if the code leaves in a different repository than the rest of the model. It can be a branch name, a tag name, or a commit id, since we use a git-based system for storing models and other artifacts on huggingface.co, so `revision` can be any identifier allowed by git. kwargs (additional keyword arguments, *optional*) : Can be used to update the configuration object (after it being loaded) and initiate the model (e.g., `output_attentions=True`). Behaves differently depending on whether a `config` is provided or automatically loaded: - If a configuration is provided with `config`, `**kwargs` will be directly passed to the underlying model's `__init__` method (we assume all relevant updates to the configuration have already been done) - If a configuration is not provided, `kwargs` will be first passed to the configuration class initialization function ([from_pretrained()](/docs/transformers/v5.6.0/ja/main_classes/configuration#transformers.PreTrainedConfig.from_pretrained)). Each key of `kwargs` that corresponds to a configuration attribute will be used to override said attribute with the supplied `kwargs` value. Remaining keys that do not correspond to any configuration attribute will be passed to the underlying model's `__init__` function. ### AutoModelForDocumentQuestionAnswering[[transformers.AutoModelForDocumentQuestionAnswering]] #### transformers.AutoModelForDocumentQuestionAnswering[[transformers.AutoModelForDocumentQuestionAnswering]] [Source](https://github.com/huggingface/transformers/blob/v5.6.0/src/transformers/models/auto/modeling_auto.py#L2066) This is a generic model class that will be instantiated as one of the model classes of the library (with a document question answering head) when created with the [from_pretrained()](/docs/transformers/v5.6.0/ja/model_doc/auto#transformers.AutoModel.from_pretrained) class method or the [from_config()](/docs/transformers/v5.6.0/ja/model_doc/auto#transformers.AutoModel.from_config) class method. This class cannot be instantiated directly using `__init__()` (throws an error). from_configtransformers.AutoModelForDocumentQuestionAnswering.from_confighttps://github.com/huggingface/transformers/blob/v5.6.0/src/transformers/models/auto/auto_factory.py#L206[{"name": "**kwargs", "val": ""}]- **config** ([PreTrainedConfig](/docs/transformers/v5.6.0/ja/main_classes/configuration#transformers.PreTrainedConfig)) -- The model class to instantiate is selected based on the configuration class: - `LayoutLMConfig` configuration class: `LayoutLMForQuestionAnswering` (LayoutLMConfig model) - `LayoutLMv2Config` configuration class: `LayoutLMv2ForQuestionAnswering` (LayoutLMv2Config model) - `LayoutLMv3Config` configuration class: `LayoutLMv3ForQuestionAnswering` (LayoutLMv3Config model) - **attn_implementation** (`str`, *optional*) -- The attention implementation to use in the model (if relevant). Can be any of `"eager"` (manual implementation of the attention), `"sdpa"` (using [`F.scaled_dot_product_attention`](https://pytorch.org/docs/master/generated/torch.nn.functional.scaled_dot_product_attention.html)), `"flash_attention_2"` (using [Dao-AILab/flash-attention](https://github.com/Dao-AILab/flash-attention)), or `"flash_attention_3"` (using [Dao-AILab/flash-attention/hopper](https://github.com/Dao-AILab/flash-attention/tree/main/hopper)). By default, if available, SDPA will be used for torch>=2.1.1. The default is otherwise the manual `"eager"` implementation.0 Instantiates one of the model classes of the library (with a document question answering head) from a configuration. Note: Loading a model from its configuration file does **not** load the model weights. It only affects the model's configuration. Use [from_pretrained()](/docs/transformers/v5.6.0/ja/model_doc/auto#transformers.AutoModel.from_pretrained) to load the model weights. Examples: ```python >>> from transformers import AutoConfig, AutoModelForDocumentQuestionAnswering >>> # Download configuration from huggingface.co and cache. >>> config = AutoConfig.from_pretrained("impira/layoutlm-document-qa", revision="52e01b3") >>> model = AutoModelForDocumentQuestionAnswering.from_config(config) ``` **Parameters:** config ([PreTrainedConfig](/docs/transformers/v5.6.0/ja/main_classes/configuration#transformers.PreTrainedConfig)) : The model class to instantiate is selected based on the configuration class: - `LayoutLMConfig` configuration class: `LayoutLMForQuestionAnswering` (LayoutLMConfig model) - `LayoutLMv2Config` configuration class: `LayoutLMv2ForQuestionAnswering` (LayoutLMv2Config model) - `LayoutLMv3Config` configuration class: `LayoutLMv3ForQuestionAnswering` (LayoutLMv3Config model) attn_implementation (`str`, *optional*) : The attention implementation to use in the model (if relevant). Can be any of `"eager"` (manual implementation of the attention), `"sdpa"` (using [`F.scaled_dot_product_attention`](https://pytorch.org/docs/master/generated/torch.nn.functional.scaled_dot_product_attention.html)), `"flash_attention_2"` (using [Dao-AILab/flash-attention](https://github.com/Dao-AILab/flash-attention)), or `"flash_attention_3"` (using [Dao-AILab/flash-attention/hopper](https://github.com/Dao-AILab/flash-attention/tree/main/hopper)). By default, if available, SDPA will be used for torch>=2.1.1. The default is otherwise the manual `"eager"` implementation. #### from_pretrained[[transformers.AutoModelForDocumentQuestionAnswering.from_pretrained]] [Source](https://github.com/huggingface/transformers/blob/v5.6.0/src/transformers/models/auto/auto_factory.py#L253) Instantiate one of the model classes of the library (with a document question answering head) from a pretrained model. The model class to instantiate is selected based on the `model_type` property of the config object (either passed as an argument or loaded from `pretrained_model_name_or_path` if possible), or when it's missing, by falling back to using pattern matching on `pretrained_model_name_or_path`: - **layoutlm** -- `LayoutLMForQuestionAnswering` (LayoutLMConfig model) - **layoutlmv2** -- `LayoutLMv2ForQuestionAnswering` (LayoutLMv2Config model) - **layoutlmv3** -- `LayoutLMv3ForQuestionAnswering` (LayoutLMv3Config model) The model is set in evaluation mode by default using `model.eval()` (so for instance, dropout modules are deactivated). To train the model, you should first set it back in training mode with `model.train()` Examples: ```python >>> from transformers import AutoConfig, AutoModelForDocumentQuestionAnswering >>> # Download model and configuration from huggingface.co and cache. >>> model = AutoModelForDocumentQuestionAnswering.from_pretrained("impira/layoutlm-document-qa", revision="52e01b3") >>> # Update configuration during loading >>> model = AutoModelForDocumentQuestionAnswering.from_pretrained("impira/layoutlm-document-qa", revision="52e01b3", output_attentions=True) >>> model.config.output_attentions True ``` **Parameters:** pretrained_model_name_or_path (`str` or `os.PathLike`) : Can be either: - A string, the *model id* of a pretrained model hosted inside a model repo on huggingface.co. - A path to a *directory* containing model weights saved using [save_pretrained()](/docs/transformers/v5.6.0/ja/main_classes/model#transformers.PreTrainedModel.save_pretrained), e.g., `./my_model_directory/`. model_args (additional positional arguments, *optional*) : Will be passed along to the underlying model `__init__()` method. config ([PreTrainedConfig](/docs/transformers/v5.6.0/ja/main_classes/configuration#transformers.PreTrainedConfig), *optional*) : Configuration for the model to use instead of an automatically loaded configuration. Configuration can be automatically loaded when: - The model is a model provided by the library (loaded with the *model id* string of a pretrained model). - The model was saved using [save_pretrained()](/docs/transformers/v5.6.0/ja/main_classes/model#transformers.PreTrainedModel.save_pretrained) and is reloaded by supplying the save directory. - The model is loaded by supplying a local directory as `pretrained_model_name_or_path` and a configuration JSON file named *config.json* is found in the directory. state_dict (*dict[str, torch.Tensor]*, *optional*) : A state dictionary to use instead of a state dictionary loaded from saved weights file. This option can be used if you want to create a model from a pretrained configuration but load your own weights. In this case though, you should check if using [save_pretrained()](/docs/transformers/v5.6.0/ja/main_classes/model#transformers.PreTrainedModel.save_pretrained) and [from_pretrained()](/docs/transformers/v5.6.0/ja/main_classes/model#transformers.PreTrainedModel.from_pretrained) is not a simpler option. cache_dir (`str` or `os.PathLike`, *optional*) : Path to a directory in which a downloaded pretrained model configuration should be cached if the standard cache should not be used. force_download (`bool`, *optional*, defaults to `False`) : Whether or not to force the (re-)download of the model weights and configuration files, overriding the cached versions if they exist. proxies (`dict[str, str]`, *optional*) : A dictionary of proxy servers to use by protocol or endpoint, e.g., `{'http': 'foo.bar:3128', 'http://hostname': 'foo.bar:4012'}`. The proxies are used on each request. output_loading_info(`bool`, *optional*, defaults to `False`) : Whether or not to also return a dictionary containing missing keys, unexpected keys and error messages. local_files_only(`bool`, *optional*, defaults to `False`) : Whether or not to only look at local files (e.g., not try downloading the model). revision (`str`, *optional*, defaults to `"main"`) : The specific model version to use. It can be a branch name, a tag name, or a commit id, since we use a git-based system for storing models and other artifacts on huggingface.co, so `revision` can be any identifier allowed by git. trust_remote_code (`bool`, *optional*, defaults to `False`) : Whether or not to allow for custom models defined on the Hub in their own modeling files. This option should only be set to `True` for repositories you trust and in which you have read the code, as it will execute code present on the Hub on your local machine. code_revision (`str`, *optional*, defaults to `"main"`) : The specific revision to use for the code on the Hub, if the code leaves in a different repository than the rest of the model. It can be a branch name, a tag name, or a commit id, since we use a git-based system for storing models and other artifacts on huggingface.co, so `revision` can be any identifier allowed by git. kwargs (additional keyword arguments, *optional*) : Can be used to update the configuration object (after it being loaded) and initiate the model (e.g., `output_attentions=True`). Behaves differently depending on whether a `config` is provided or automatically loaded: - If a configuration is provided with `config`, `**kwargs` will be directly passed to the underlying model's `__init__` method (we assume all relevant updates to the configuration have already been done) - If a configuration is not provided, `kwargs` will be first passed to the configuration class initialization function ([from_pretrained()](/docs/transformers/v5.6.0/ja/main_classes/configuration#transformers.PreTrainedConfig.from_pretrained)). Each key of `kwargs` that corresponds to a configuration attribute will be used to override said attribute with the supplied `kwargs` value. Remaining keys that do not correspond to any configuration attribute will be passed to the underlying model's `__init__` function. ### AutoModelForVisualQuestionAnswering[[transformers.AutoModelForVisualQuestionAnswering]] #### transformers.AutoModelForVisualQuestionAnswering[[transformers.AutoModelForVisualQuestionAnswering]] [Source](https://github.com/huggingface/transformers/blob/v5.6.0/src/transformers/models/auto/modeling_auto.py#L2055) This is a generic model class that will be instantiated as one of the model classes of the library (with a visual question answering head) when created with the [from_pretrained()](/docs/transformers/v5.6.0/ja/model_doc/auto#transformers.AutoModel.from_pretrained) class method or the [from_config()](/docs/transformers/v5.6.0/ja/model_doc/auto#transformers.AutoModel.from_config) class method. This class cannot be instantiated directly using `__init__()` (throws an error). from_configtransformers.AutoModelForVisualQuestionAnswering.from_confighttps://github.com/huggingface/transformers/blob/v5.6.0/src/transformers/models/auto/auto_factory.py#L206[{"name": "**kwargs", "val": ""}]- **config** ([PreTrainedConfig](/docs/transformers/v5.6.0/ja/main_classes/configuration#transformers.PreTrainedConfig)) -- The model class to instantiate is selected based on the configuration class: - [Blip2Config](/docs/transformers/v5.6.0/ja/model_doc/blip-2#transformers.Blip2Config) configuration class: [Blip2ForConditionalGeneration](/docs/transformers/v5.6.0/ja/model_doc/blip-2#transformers.Blip2ForConditionalGeneration) (Blip2Config model) - [BlipConfig](/docs/transformers/v5.6.0/ja/model_doc/blip#transformers.BlipConfig) configuration class: [BlipForQuestionAnswering](/docs/transformers/v5.6.0/ja/model_doc/blip#transformers.BlipForQuestionAnswering) (BlipConfig model) - `ViltConfig` configuration class: `ViltForQuestionAnswering` (ViltConfig model) - **attn_implementation** (`str`, *optional*) -- The attention implementation to use in the model (if relevant). Can be any of `"eager"` (manual implementation of the attention), `"sdpa"` (using [`F.scaled_dot_product_attention`](https://pytorch.org/docs/master/generated/torch.nn.functional.scaled_dot_product_attention.html)), `"flash_attention_2"` (using [Dao-AILab/flash-attention](https://github.com/Dao-AILab/flash-attention)), or `"flash_attention_3"` (using [Dao-AILab/flash-attention/hopper](https://github.com/Dao-AILab/flash-attention/tree/main/hopper)). By default, if available, SDPA will be used for torch>=2.1.1. The default is otherwise the manual `"eager"` implementation.0 Instantiates one of the model classes of the library (with a visual question answering head) from a configuration. Note: Loading a model from its configuration file does **not** load the model weights. It only affects the model's configuration. Use [from_pretrained()](/docs/transformers/v5.6.0/ja/model_doc/auto#transformers.AutoModel.from_pretrained) to load the model weights. Examples: ```python >>> from transformers import AutoConfig, AutoModelForVisualQuestionAnswering >>> # Download configuration from huggingface.co and cache. >>> config = AutoConfig.from_pretrained("dandelin/vilt-b32-finetuned-vqa") >>> model = AutoModelForVisualQuestionAnswering.from_config(config) ``` **Parameters:** config ([PreTrainedConfig](/docs/transformers/v5.6.0/ja/main_classes/configuration#transformers.PreTrainedConfig)) : The model class to instantiate is selected based on the configuration class: - [Blip2Config](/docs/transformers/v5.6.0/ja/model_doc/blip-2#transformers.Blip2Config) configuration class: [Blip2ForConditionalGeneration](/docs/transformers/v5.6.0/ja/model_doc/blip-2#transformers.Blip2ForConditionalGeneration) (Blip2Config model) - [BlipConfig](/docs/transformers/v5.6.0/ja/model_doc/blip#transformers.BlipConfig) configuration class: [BlipForQuestionAnswering](/docs/transformers/v5.6.0/ja/model_doc/blip#transformers.BlipForQuestionAnswering) (BlipConfig model) - `ViltConfig` configuration class: `ViltForQuestionAnswering` (ViltConfig model) attn_implementation (`str`, *optional*) : The attention implementation to use in the model (if relevant). Can be any of `"eager"` (manual implementation of the attention), `"sdpa"` (using [`F.scaled_dot_product_attention`](https://pytorch.org/docs/master/generated/torch.nn.functional.scaled_dot_product_attention.html)), `"flash_attention_2"` (using [Dao-AILab/flash-attention](https://github.com/Dao-AILab/flash-attention)), or `"flash_attention_3"` (using [Dao-AILab/flash-attention/hopper](https://github.com/Dao-AILab/flash-attention/tree/main/hopper)). By default, if available, SDPA will be used for torch>=2.1.1. The default is otherwise the manual `"eager"` implementation. #### from_pretrained[[transformers.AutoModelForVisualQuestionAnswering.from_pretrained]] [Source](https://github.com/huggingface/transformers/blob/v5.6.0/src/transformers/models/auto/auto_factory.py#L253) Instantiate one of the model classes of the library (with a visual question answering head) from a pretrained model. The model class to instantiate is selected based on the `model_type` property of the config object (either passed as an argument or loaded from `pretrained_model_name_or_path` if possible), or when it's missing, by falling back to using pattern matching on `pretrained_model_name_or_path`: - **blip** -- [BlipForQuestionAnswering](/docs/transformers/v5.6.0/ja/model_doc/blip#transformers.BlipForQuestionAnswering) (BlipConfig model) - **blip-2** -- [Blip2ForConditionalGeneration](/docs/transformers/v5.6.0/ja/model_doc/blip-2#transformers.Blip2ForConditionalGeneration) (Blip2Config model) - **vilt** -- `ViltForQuestionAnswering` (ViltConfig model) The model is set in evaluation mode by default using `model.eval()` (so for instance, dropout modules are deactivated). To train the model, you should first set it back in training mode with `model.train()` Examples: ```python >>> from transformers import AutoConfig, AutoModelForVisualQuestionAnswering >>> # Download model and configuration from huggingface.co and cache. >>> model = AutoModelForVisualQuestionAnswering.from_pretrained("dandelin/vilt-b32-finetuned-vqa") >>> # Update configuration during loading >>> model = AutoModelForVisualQuestionAnswering.from_pretrained("dandelin/vilt-b32-finetuned-vqa", output_attentions=True) >>> model.config.output_attentions True ``` **Parameters:** pretrained_model_name_or_path (`str` or `os.PathLike`) : Can be either: - A string, the *model id* of a pretrained model hosted inside a model repo on huggingface.co. - A path to a *directory* containing model weights saved using [save_pretrained()](/docs/transformers/v5.6.0/ja/main_classes/model#transformers.PreTrainedModel.save_pretrained), e.g., `./my_model_directory/`. model_args (additional positional arguments, *optional*) : Will be passed along to the underlying model `__init__()` method. config ([PreTrainedConfig](/docs/transformers/v5.6.0/ja/main_classes/configuration#transformers.PreTrainedConfig), *optional*) : Configuration for the model to use instead of an automatically loaded configuration. Configuration can be automatically loaded when: - The model is a model provided by the library (loaded with the *model id* string of a pretrained model). - The model was saved using [save_pretrained()](/docs/transformers/v5.6.0/ja/main_classes/model#transformers.PreTrainedModel.save_pretrained) and is reloaded by supplying the save directory. - The model is loaded by supplying a local directory as `pretrained_model_name_or_path` and a configuration JSON file named *config.json* is found in the directory. state_dict (*dict[str, torch.Tensor]*, *optional*) : A state dictionary to use instead of a state dictionary loaded from saved weights file. This option can be used if you want to create a model from a pretrained configuration but load your own weights. In this case though, you should check if using [save_pretrained()](/docs/transformers/v5.6.0/ja/main_classes/model#transformers.PreTrainedModel.save_pretrained) and [from_pretrained()](/docs/transformers/v5.6.0/ja/main_classes/model#transformers.PreTrainedModel.from_pretrained) is not a simpler option. cache_dir (`str` or `os.PathLike`, *optional*) : Path to a directory in which a downloaded pretrained model configuration should be cached if the standard cache should not be used. force_download (`bool`, *optional*, defaults to `False`) : Whether or not to force the (re-)download of the model weights and configuration files, overriding the cached versions if they exist. proxies (`dict[str, str]`, *optional*) : A dictionary of proxy servers to use by protocol or endpoint, e.g., `{'http': 'foo.bar:3128', 'http://hostname': 'foo.bar:4012'}`. The proxies are used on each request. output_loading_info(`bool`, *optional*, defaults to `False`) : Whether or not to also return a dictionary containing missing keys, unexpected keys and error messages. local_files_only(`bool`, *optional*, defaults to `False`) : Whether or not to only look at local files (e.g., not try downloading the model). revision (`str`, *optional*, defaults to `"main"`) : The specific model version to use. It can be a branch name, a tag name, or a commit id, since we use a git-based system for storing models and other artifacts on huggingface.co, so `revision` can be any identifier allowed by git. trust_remote_code (`bool`, *optional*, defaults to `False`) : Whether or not to allow for custom models defined on the Hub in their own modeling files. This option should only be set to `True` for repositories you trust and in which you have read the code, as it will execute code present on the Hub on your local machine. code_revision (`str`, *optional*, defaults to `"main"`) : The specific revision to use for the code on the Hub, if the code leaves in a different repository than the rest of the model. It can be a branch name, a tag name, or a commit id, since we use a git-based system for storing models and other artifacts on huggingface.co, so `revision` can be any identifier allowed by git. kwargs (additional keyword arguments, *optional*) : Can be used to update the configuration object (after it being loaded) and initiate the model (e.g., `output_attentions=True`). Behaves differently depending on whether a `config` is provided or automatically loaded: - If a configuration is provided with `config`, `**kwargs` will be directly passed to the underlying model's `__init__` method (we assume all relevant updates to the configuration have already been done) - If a configuration is not provided, `kwargs` will be first passed to the configuration class initialization function ([from_pretrained()](/docs/transformers/v5.6.0/ja/main_classes/configuration#transformers.PreTrainedConfig.from_pretrained)). Each key of `kwargs` that corresponds to a configuration attribute will be used to override said attribute with the supplied `kwargs` value. Remaining keys that do not correspond to any configuration attribute will be passed to the underlying model's `__init__` function. ### AutoModelForImageTextToText[[transformers.AutoModelForImageTextToText]] #### transformers.AutoModelForImageTextToText[[transformers.AutoModelForImageTextToText]] [Source](https://github.com/huggingface/transformers/blob/v5.6.0/src/transformers/models/auto/modeling_auto.py#L2203) This is a generic model class that will be instantiated as one of the model classes of the library (with a image-text-to-text modeling head) when created with the [from_pretrained()](/docs/transformers/v5.6.0/ja/model_doc/auto#transformers.AutoModel.from_pretrained) class method or the [from_config()](/docs/transformers/v5.6.0/ja/model_doc/auto#transformers.AutoModel.from_config) class method. This class cannot be instantiated directly using `__init__()` (throws an error). from_configtransformers.AutoModelForImageTextToText.from_confighttps://github.com/huggingface/transformers/blob/v5.6.0/src/transformers/models/auto/auto_factory.py#L206[{"name": "**kwargs", "val": ""}]- **config** ([PreTrainedConfig](/docs/transformers/v5.6.0/ja/main_classes/configuration#transformers.PreTrainedConfig)) -- The model class to instantiate is selected based on the configuration class: - `AriaConfig` configuration class: `AriaForConditionalGeneration` (AriaConfig model) - `AyaVisionConfig` configuration class: `AyaVisionForConditionalGeneration` (AyaVisionConfig model) - [Blip2Config](/docs/transformers/v5.6.0/ja/model_doc/blip-2#transformers.Blip2Config) configuration class: [Blip2ForConditionalGeneration](/docs/transformers/v5.6.0/ja/model_doc/blip-2#transformers.Blip2ForConditionalGeneration) (Blip2Config model) - [BlipConfig](/docs/transformers/v5.6.0/ja/model_doc/blip#transformers.BlipConfig) configuration class: [BlipForConditionalGeneration](/docs/transformers/v5.6.0/ja/model_doc/blip#transformers.BlipForConditionalGeneration) (BlipConfig model) - `ChameleonConfig` configuration class: `ChameleonForConditionalGeneration` (ChameleonConfig model) - `Cohere2VisionConfig` configuration class: `Cohere2VisionForConditionalGeneration` (Cohere2VisionConfig model) - `DeepseekVLConfig` configuration class: `DeepseekVLForConditionalGeneration` (DeepseekVLConfig model) - `DeepseekVLHybridConfig` configuration class: `DeepseekVLHybridForConditionalGeneration` (DeepseekVLHybridConfig model) - `Emu3Config` configuration class: `Emu3ForConditionalGeneration` (Emu3Config model) - `Ernie4_5_VLMoeConfig` configuration class: `Ernie4_5_VLMoeForConditionalGeneration` (Ernie4_5_VLMoeConfig model) - `EvollaConfig` configuration class: `EvollaForProteinText2Text` (EvollaConfig model) - `FastVlmConfig` configuration class: `FastVlmForConditionalGeneration` (FastVlmConfig model) - `Florence2Config` configuration class: `Florence2ForConditionalGeneration` (Florence2Config model) - `FuyuConfig` configuration class: `FuyuForCausalLM` (FuyuConfig model) - `Gemma3Config` configuration class: `Gemma3ForConditionalGeneration` (Gemma3Config model) - `Gemma3nConfig` configuration class: `Gemma3nForConditionalGeneration` (Gemma3nConfig model) - `Gemma4Config` configuration class: `Gemma4ForConditionalGeneration` (Gemma4Config model) - `GitConfig` configuration class: `GitForCausalLM` (GitConfig model) - `Glm46VConfig` configuration class: `Glm46VForConditionalGeneration` (Glm46VConfig model) - `Glm4vConfig` configuration class: `Glm4vForConditionalGeneration` (Glm4vConfig model) - `Glm4vMoeConfig` configuration class: `Glm4vMoeForConditionalGeneration` (Glm4vMoeConfig model) - `GlmOcrConfig` configuration class: `GlmOcrForConditionalGeneration` (GlmOcrConfig model) - `GotOcr2Config` configuration class: `GotOcr2ForConditionalGeneration` (GotOcr2Config model) - `Idefics2Config` configuration class: `Idefics2ForConditionalGeneration` (Idefics2Config model) - `Idefics3Config` configuration class: `Idefics3ForConditionalGeneration` (Idefics3Config model) - `IdeficsConfig` configuration class: `IdeficsForVisionText2Text` (IdeficsConfig model) - `InstructBlipConfig` configuration class: `InstructBlipForConditionalGeneration` (InstructBlipConfig model) - `InstructBlipVideoConfig` configuration class: `InstructBlipVideoForConditionalGeneration` (InstructBlipVideoConfig model) - `InternVLConfig` configuration class: `InternVLForConditionalGeneration` (InternVLConfig model) - `JanusConfig` configuration class: `JanusForConditionalGeneration` (JanusConfig model) - `Kosmos2Config` configuration class: `Kosmos2ForConditionalGeneration` (Kosmos2Config model) - `Kosmos2_5Config` configuration class: `Kosmos2_5ForConditionalGeneration` (Kosmos2_5Config model) - `Lfm2VlConfig` configuration class: `Lfm2VlForConditionalGeneration` (Lfm2VlConfig model) - `LightOnOcrConfig` configuration class: `LightOnOcrForConditionalGeneration` (LightOnOcrConfig model) - `Llama4Config` configuration class: `Llama4ForConditionalGeneration` (Llama4Config model) - `LlavaConfig` configuration class: `LlavaForConditionalGeneration` (LlavaConfig model) - `LlavaNextConfig` configuration class: `LlavaNextForConditionalGeneration` (LlavaNextConfig model) - `LlavaNextVideoConfig` configuration class: `LlavaNextVideoForConditionalGeneration` (LlavaNextVideoConfig model) - `LlavaOnevisionConfig` configuration class: `LlavaOnevisionForConditionalGeneration` (LlavaOnevisionConfig model) - `Mistral3Config` configuration class: `Mistral3ForConditionalGeneration` (Mistral3Config model) - `Mistral4Config` configuration class: `Mistral4ForCausalLM` (Mistral4Config model) - `MllamaConfig` configuration class: `MllamaForConditionalGeneration` (MllamaConfig model) - `Ovis2Config` configuration class: `Ovis2ForConditionalGeneration` (Ovis2Config model) - `PI0Config` configuration class: `PI0ForConditionalGeneration` (PI0Config model) - `PPChart2TableConfig` configuration class: `GotOcr2ForConditionalGeneration` (PPChart2TableConfig model) - `PaddleOCRVLConfig` configuration class: `PaddleOCRVLForConditionalGeneration` (PaddleOCRVLConfig model) - `PaliGemmaConfig` configuration class: `PaliGemmaForConditionalGeneration` (PaliGemmaConfig model) - `PerceptionLMConfig` configuration class: `PerceptionLMForConditionalGeneration` (PerceptionLMConfig model) - `Pix2StructConfig` configuration class: `Pix2StructForConditionalGeneration` (Pix2StructConfig model) - `QianfanOCRConfig` configuration class: `QianfanOCRForConditionalGeneration` (QianfanOCRConfig model) - `Qwen2VLConfig` configuration class: `Qwen2VLForConditionalGeneration` (Qwen2VLConfig model) - `Qwen2_5OmniThinkerConfig` configuration class: `Qwen2_5OmniThinkerForConditionalGeneration` (Qwen2_5OmniThinkerConfig model) - `Qwen2_5_VLConfig` configuration class: `Qwen2_5_VLForConditionalGeneration` (Qwen2_5_VLConfig model) - `Qwen3OmniMoeThinkerConfig` configuration class: `Qwen3OmniMoeThinkerForConditionalGeneration` (Qwen3OmniMoeThinkerConfig model) - `Qwen3VLConfig` configuration class: `Qwen3VLForConditionalGeneration` (Qwen3VLConfig model) - `Qwen3VLMoeConfig` configuration class: `Qwen3VLMoeForConditionalGeneration` (Qwen3VLMoeConfig model) - `Qwen3_5Config` configuration class: `Qwen3_5ForConditionalGeneration` (Qwen3_5Config model) - `Qwen3_5MoeConfig` configuration class: `Qwen3_5MoeForConditionalGeneration` (Qwen3_5MoeConfig model) - `ShieldGemma2Config` configuration class: `Gemma3ForConditionalGeneration` (ShieldGemma2Config model) - `SmolVLMConfig` configuration class: `SmolVLMForConditionalGeneration` (SmolVLMConfig model) - `T5Gemma2Config` configuration class: `T5Gemma2ForConditionalGeneration` (T5Gemma2Config model) - `UdopConfig` configuration class: `UdopForConditionalGeneration` (UdopConfig model) - `VideoLlama3Config` configuration class: `VideoLlama3ForConditionalGeneration` (VideoLlama3Config model) - `VideoLlavaConfig` configuration class: `VideoLlavaForConditionalGeneration` (VideoLlavaConfig model) - `VipLlavaConfig` configuration class: `VipLlavaForConditionalGeneration` (VipLlavaConfig model) - `VisionEncoderDecoderConfig` configuration class: `VisionEncoderDecoderModel` (VisionEncoderDecoderConfig model) - **attn_implementation** (`str`, *optional*) -- The attention implementation to use in the model (if relevant). Can be any of `"eager"` (manual implementation of the attention), `"sdpa"` (using [`F.scaled_dot_product_attention`](https://pytorch.org/docs/master/generated/torch.nn.functional.scaled_dot_product_attention.html)), `"flash_attention_2"` (using [Dao-AILab/flash-attention](https://github.com/Dao-AILab/flash-attention)), or `"flash_attention_3"` (using [Dao-AILab/flash-attention/hopper](https://github.com/Dao-AILab/flash-attention/tree/main/hopper)). By default, if available, SDPA will be used for torch>=2.1.1. The default is otherwise the manual `"eager"` implementation.0 Instantiates one of the model classes of the library (with a image-text-to-text modeling head) from a configuration. Note: Loading a model from its configuration file does **not** load the model weights. It only affects the model's configuration. Use [from_pretrained()](/docs/transformers/v5.6.0/ja/model_doc/auto#transformers.AutoModel.from_pretrained) to load the model weights. Examples: ```python >>> from transformers import AutoConfig, AutoModelForImageTextToText >>> # Download configuration from huggingface.co and cache. >>> config = AutoConfig.from_pretrained("google-bert/bert-base-cased") >>> model = AutoModelForImageTextToText.from_config(config) ``` **Parameters:** config ([PreTrainedConfig](/docs/transformers/v5.6.0/ja/main_classes/configuration#transformers.PreTrainedConfig)) : The model class to instantiate is selected based on the configuration class: - `AriaConfig` configuration class: `AriaForConditionalGeneration` (AriaConfig model) - `AyaVisionConfig` configuration class: `AyaVisionForConditionalGeneration` (AyaVisionConfig model) - [Blip2Config](/docs/transformers/v5.6.0/ja/model_doc/blip-2#transformers.Blip2Config) configuration class: [Blip2ForConditionalGeneration](/docs/transformers/v5.6.0/ja/model_doc/blip-2#transformers.Blip2ForConditionalGeneration) (Blip2Config model) - [BlipConfig](/docs/transformers/v5.6.0/ja/model_doc/blip#transformers.BlipConfig) configuration class: [BlipForConditionalGeneration](/docs/transformers/v5.6.0/ja/model_doc/blip#transformers.BlipForConditionalGeneration) (BlipConfig model) - `ChameleonConfig` configuration class: `ChameleonForConditionalGeneration` (ChameleonConfig model) - `Cohere2VisionConfig` configuration class: `Cohere2VisionForConditionalGeneration` (Cohere2VisionConfig model) - `DeepseekVLConfig` configuration class: `DeepseekVLForConditionalGeneration` (DeepseekVLConfig model) - `DeepseekVLHybridConfig` configuration class: `DeepseekVLHybridForConditionalGeneration` (DeepseekVLHybridConfig model) - `Emu3Config` configuration class: `Emu3ForConditionalGeneration` (Emu3Config model) - `Ernie4_5_VLMoeConfig` configuration class: `Ernie4_5_VLMoeForConditionalGeneration` (Ernie4_5_VLMoeConfig model) - `EvollaConfig` configuration class: `EvollaForProteinText2Text` (EvollaConfig model) - `FastVlmConfig` configuration class: `FastVlmForConditionalGeneration` (FastVlmConfig model) - `Florence2Config` configuration class: `Florence2ForConditionalGeneration` (Florence2Config model) - `FuyuConfig` configuration class: `FuyuForCausalLM` (FuyuConfig model) - `Gemma3Config` configuration class: `Gemma3ForConditionalGeneration` (Gemma3Config model) - `Gemma3nConfig` configuration class: `Gemma3nForConditionalGeneration` (Gemma3nConfig model) - `Gemma4Config` configuration class: `Gemma4ForConditionalGeneration` (Gemma4Config model) - `GitConfig` configuration class: `GitForCausalLM` (GitConfig model) - `Glm46VConfig` configuration class: `Glm46VForConditionalGeneration` (Glm46VConfig model) - `Glm4vConfig` configuration class: `Glm4vForConditionalGeneration` (Glm4vConfig model) - `Glm4vMoeConfig` configuration class: `Glm4vMoeForConditionalGeneration` (Glm4vMoeConfig model) - `GlmOcrConfig` configuration class: `GlmOcrForConditionalGeneration` (GlmOcrConfig model) - `GotOcr2Config` configuration class: `GotOcr2ForConditionalGeneration` (GotOcr2Config model) - `Idefics2Config` configuration class: `Idefics2ForConditionalGeneration` (Idefics2Config model) - `Idefics3Config` configuration class: `Idefics3ForConditionalGeneration` (Idefics3Config model) - `IdeficsConfig` configuration class: `IdeficsForVisionText2Text` (IdeficsConfig model) - `InstructBlipConfig` configuration class: `InstructBlipForConditionalGeneration` (InstructBlipConfig model) - `InstructBlipVideoConfig` configuration class: `InstructBlipVideoForConditionalGeneration` (InstructBlipVideoConfig model) - `InternVLConfig` configuration class: `InternVLForConditionalGeneration` (InternVLConfig model) - `JanusConfig` configuration class: `JanusForConditionalGeneration` (JanusConfig model) - `Kosmos2Config` configuration class: `Kosmos2ForConditionalGeneration` (Kosmos2Config model) - `Kosmos2_5Config` configuration class: `Kosmos2_5ForConditionalGeneration` (Kosmos2_5Config model) - `Lfm2VlConfig` configuration class: `Lfm2VlForConditionalGeneration` (Lfm2VlConfig model) - `LightOnOcrConfig` configuration class: `LightOnOcrForConditionalGeneration` (LightOnOcrConfig model) - `Llama4Config` configuration class: `Llama4ForConditionalGeneration` (Llama4Config model) - `LlavaConfig` configuration class: `LlavaForConditionalGeneration` (LlavaConfig model) - `LlavaNextConfig` configuration class: `LlavaNextForConditionalGeneration` (LlavaNextConfig model) - `LlavaNextVideoConfig` configuration class: `LlavaNextVideoForConditionalGeneration` (LlavaNextVideoConfig model) - `LlavaOnevisionConfig` configuration class: `LlavaOnevisionForConditionalGeneration` (LlavaOnevisionConfig model) - `Mistral3Config` configuration class: `Mistral3ForConditionalGeneration` (Mistral3Config model) - `Mistral4Config` configuration class: `Mistral4ForCausalLM` (Mistral4Config model) - `MllamaConfig` configuration class: `MllamaForConditionalGeneration` (MllamaConfig model) - `Ovis2Config` configuration class: `Ovis2ForConditionalGeneration` (Ovis2Config model) - `PI0Config` configuration class: `PI0ForConditionalGeneration` (PI0Config model) - `PPChart2TableConfig` configuration class: `GotOcr2ForConditionalGeneration` (PPChart2TableConfig model) - `PaddleOCRVLConfig` configuration class: `PaddleOCRVLForConditionalGeneration` (PaddleOCRVLConfig model) - `PaliGemmaConfig` configuration class: `PaliGemmaForConditionalGeneration` (PaliGemmaConfig model) - `PerceptionLMConfig` configuration class: `PerceptionLMForConditionalGeneration` (PerceptionLMConfig model) - `Pix2StructConfig` configuration class: `Pix2StructForConditionalGeneration` (Pix2StructConfig model) - `QianfanOCRConfig` configuration class: `QianfanOCRForConditionalGeneration` (QianfanOCRConfig model) - `Qwen2VLConfig` configuration class: `Qwen2VLForConditionalGeneration` (Qwen2VLConfig model) - `Qwen2_5OmniThinkerConfig` configuration class: `Qwen2_5OmniThinkerForConditionalGeneration` (Qwen2_5OmniThinkerConfig model) - `Qwen2_5_VLConfig` configuration class: `Qwen2_5_VLForConditionalGeneration` (Qwen2_5_VLConfig model) - `Qwen3OmniMoeThinkerConfig` configuration class: `Qwen3OmniMoeThinkerForConditionalGeneration` (Qwen3OmniMoeThinkerConfig model) - `Qwen3VLConfig` configuration class: `Qwen3VLForConditionalGeneration` (Qwen3VLConfig model) - `Qwen3VLMoeConfig` configuration class: `Qwen3VLMoeForConditionalGeneration` (Qwen3VLMoeConfig model) - `Qwen3_5Config` configuration class: `Qwen3_5ForConditionalGeneration` (Qwen3_5Config model) - `Qwen3_5MoeConfig` configuration class: `Qwen3_5MoeForConditionalGeneration` (Qwen3_5MoeConfig model) - `ShieldGemma2Config` configuration class: `Gemma3ForConditionalGeneration` (ShieldGemma2Config model) - `SmolVLMConfig` configuration class: `SmolVLMForConditionalGeneration` (SmolVLMConfig model) - `T5Gemma2Config` configuration class: `T5Gemma2ForConditionalGeneration` (T5Gemma2Config model) - `UdopConfig` configuration class: `UdopForConditionalGeneration` (UdopConfig model) - `VideoLlama3Config` configuration class: `VideoLlama3ForConditionalGeneration` (VideoLlama3Config model) - `VideoLlavaConfig` configuration class: `VideoLlavaForConditionalGeneration` (VideoLlavaConfig model) - `VipLlavaConfig` configuration class: `VipLlavaForConditionalGeneration` (VipLlavaConfig model) - `VisionEncoderDecoderConfig` configuration class: `VisionEncoderDecoderModel` (VisionEncoderDecoderConfig model) attn_implementation (`str`, *optional*) : The attention implementation to use in the model (if relevant). Can be any of `"eager"` (manual implementation of the attention), `"sdpa"` (using [`F.scaled_dot_product_attention`](https://pytorch.org/docs/master/generated/torch.nn.functional.scaled_dot_product_attention.html)), `"flash_attention_2"` (using [Dao-AILab/flash-attention](https://github.com/Dao-AILab/flash-attention)), or `"flash_attention_3"` (using [Dao-AILab/flash-attention/hopper](https://github.com/Dao-AILab/flash-attention/tree/main/hopper)). By default, if available, SDPA will be used for torch>=2.1.1. The default is otherwise the manual `"eager"` implementation. #### from_pretrained[[transformers.AutoModelForImageTextToText.from_pretrained]] [Source](https://github.com/huggingface/transformers/blob/v5.6.0/src/transformers/models/auto/auto_factory.py#L253) Instantiate one of the model classes of the library (with a image-text-to-text modeling head) from a pretrained model. The model class to instantiate is selected based on the `model_type` property of the config object (either passed as an argument or loaded from `pretrained_model_name_or_path` if possible), or when it's missing, by falling back to using pattern matching on `pretrained_model_name_or_path`: - **aria** -- `AriaForConditionalGeneration` (AriaConfig model) - **aya_vision** -- `AyaVisionForConditionalGeneration` (AyaVisionConfig model) - **blip** -- [BlipForConditionalGeneration](/docs/transformers/v5.6.0/ja/model_doc/blip#transformers.BlipForConditionalGeneration) (BlipConfig model) - **blip-2** -- [Blip2ForConditionalGeneration](/docs/transformers/v5.6.0/ja/model_doc/blip-2#transformers.Blip2ForConditionalGeneration) (Blip2Config model) - **chameleon** -- `ChameleonForConditionalGeneration` (ChameleonConfig model) - **cohere2_vision** -- `Cohere2VisionForConditionalGeneration` (Cohere2VisionConfig model) - **deepseek_vl** -- `DeepseekVLForConditionalGeneration` (DeepseekVLConfig model) - **deepseek_vl_hybrid** -- `DeepseekVLHybridForConditionalGeneration` (DeepseekVLHybridConfig model) - **emu3** -- `Emu3ForConditionalGeneration` (Emu3Config model) - **ernie4_5_vl_moe** -- `Ernie4_5_VLMoeForConditionalGeneration` (Ernie4_5_VLMoeConfig model) - **evolla** -- `EvollaForProteinText2Text` (EvollaConfig model) - **fast_vlm** -- `FastVlmForConditionalGeneration` (FastVlmConfig model) - **florence2** -- `Florence2ForConditionalGeneration` (Florence2Config model) - **fuyu** -- `FuyuForCausalLM` (FuyuConfig model) - **gemma3** -- `Gemma3ForConditionalGeneration` (Gemma3Config model) - **gemma3n** -- `Gemma3nForConditionalGeneration` (Gemma3nConfig model) - **gemma4** -- `Gemma4ForConditionalGeneration` (Gemma4Config model) - **git** -- `GitForCausalLM` (GitConfig model) - **glm46v** -- `Glm46VForConditionalGeneration` (Glm46VConfig model) - **glm4v** -- `Glm4vForConditionalGeneration` (Glm4vConfig model) - **glm4v_moe** -- `Glm4vMoeForConditionalGeneration` (Glm4vMoeConfig model) - **glm_ocr** -- `GlmOcrForConditionalGeneration` (GlmOcrConfig model) - **got_ocr2** -- `GotOcr2ForConditionalGeneration` (GotOcr2Config model) - **idefics** -- `IdeficsForVisionText2Text` (IdeficsConfig model) - **idefics2** -- `Idefics2ForConditionalGeneration` (Idefics2Config model) - **idefics3** -- `Idefics3ForConditionalGeneration` (Idefics3Config model) - **instructblip** -- `InstructBlipForConditionalGeneration` (InstructBlipConfig model) - **instructblipvideo** -- `InstructBlipVideoForConditionalGeneration` (InstructBlipVideoConfig model) - **internvl** -- `InternVLForConditionalGeneration` (InternVLConfig model) - **janus** -- `JanusForConditionalGeneration` (JanusConfig model) - **kosmos-2** -- `Kosmos2ForConditionalGeneration` (Kosmos2Config model) - **kosmos-2.5** -- `Kosmos2_5ForConditionalGeneration` (Kosmos2_5Config model) - **lfm2_vl** -- `Lfm2VlForConditionalGeneration` (Lfm2VlConfig model) - **lighton_ocr** -- `LightOnOcrForConditionalGeneration` (LightOnOcrConfig model) - **llama4** -- `Llama4ForConditionalGeneration` (Llama4Config model) - **llava** -- `LlavaForConditionalGeneration` (LlavaConfig model) - **llava_next** -- `LlavaNextForConditionalGeneration` (LlavaNextConfig model) - **llava_next_video** -- `LlavaNextVideoForConditionalGeneration` (LlavaNextVideoConfig model) - **llava_onevision** -- `LlavaOnevisionForConditionalGeneration` (LlavaOnevisionConfig model) - **mistral3** -- `Mistral3ForConditionalGeneration` (Mistral3Config model) - **mistral4** -- `Mistral4ForCausalLM` (Mistral4Config model) - **mllama** -- `MllamaForConditionalGeneration` (MllamaConfig model) - **ovis2** -- `Ovis2ForConditionalGeneration` (Ovis2Config model) - **paddleocr_vl** -- `PaddleOCRVLForConditionalGeneration` (PaddleOCRVLConfig model) - **paligemma** -- `PaliGemmaForConditionalGeneration` (PaliGemmaConfig model) - **perception_lm** -- `PerceptionLMForConditionalGeneration` (PerceptionLMConfig model) - **pi0** -- `PI0ForConditionalGeneration` (PI0Config model) - **pix2struct** -- `Pix2StructForConditionalGeneration` (Pix2StructConfig model) - **pp_chart2table** -- `GotOcr2ForConditionalGeneration` (PPChart2TableConfig model) - **qianfan_ocr** -- `QianfanOCRForConditionalGeneration` (QianfanOCRConfig model) - **qwen2_5_omni_thinker** -- `Qwen2_5OmniThinkerForConditionalGeneration` (Qwen2_5OmniThinkerConfig model) - **qwen2_5_vl** -- `Qwen2_5_VLForConditionalGeneration` (Qwen2_5_VLConfig model) - **qwen2_vl** -- `Qwen2VLForConditionalGeneration` (Qwen2VLConfig model) - **qwen3_5** -- `Qwen3_5ForConditionalGeneration` (Qwen3_5Config model) - **qwen3_5_moe** -- `Qwen3_5MoeForConditionalGeneration` (Qwen3_5MoeConfig model) - **qwen3_omni_moe_thinker** -- `Qwen3OmniMoeThinkerForConditionalGeneration` (Qwen3OmniMoeThinkerConfig model) - **qwen3_vl** -- `Qwen3VLForConditionalGeneration` (Qwen3VLConfig model) - **qwen3_vl_moe** -- `Qwen3VLMoeForConditionalGeneration` (Qwen3VLMoeConfig model) - **shieldgemma2** -- `Gemma3ForConditionalGeneration` (ShieldGemma2Config model) - **smolvlm** -- `SmolVLMForConditionalGeneration` (SmolVLMConfig model) - **t5gemma2** -- `T5Gemma2ForConditionalGeneration` (T5Gemma2Config model) - **udop** -- `UdopForConditionalGeneration` (UdopConfig model) - **video_llama_3** -- `VideoLlama3ForConditionalGeneration` (VideoLlama3Config model) - **video_llava** -- `VideoLlavaForConditionalGeneration` (VideoLlavaConfig model) - **vipllava** -- `VipLlavaForConditionalGeneration` (VipLlavaConfig model) - **vision-encoder-decoder** -- `VisionEncoderDecoderModel` (VisionEncoderDecoderConfig model) The model is set in evaluation mode by default using `model.eval()` (so for instance, dropout modules are deactivated). To train the model, you should first set it back in training mode with `model.train()` Examples: ```python >>> from transformers import AutoConfig, AutoModelForImageTextToText >>> # Download model and configuration from huggingface.co and cache. >>> model = AutoModelForImageTextToText.from_pretrained("google-bert/bert-base-cased") >>> # Update configuration during loading >>> model = AutoModelForImageTextToText.from_pretrained("google-bert/bert-base-cased", output_attentions=True) >>> model.config.output_attentions True ``` **Parameters:** pretrained_model_name_or_path (`str` or `os.PathLike`) : Can be either: - A string, the *model id* of a pretrained model hosted inside a model repo on huggingface.co. - A path to a *directory* containing model weights saved using [save_pretrained()](/docs/transformers/v5.6.0/ja/main_classes/model#transformers.PreTrainedModel.save_pretrained), e.g., `./my_model_directory/`. model_args (additional positional arguments, *optional*) : Will be passed along to the underlying model `__init__()` method. config ([PreTrainedConfig](/docs/transformers/v5.6.0/ja/main_classes/configuration#transformers.PreTrainedConfig), *optional*) : Configuration for the model to use instead of an automatically loaded configuration. Configuration can be automatically loaded when: - The model is a model provided by the library (loaded with the *model id* string of a pretrained model). - The model was saved using [save_pretrained()](/docs/transformers/v5.6.0/ja/main_classes/model#transformers.PreTrainedModel.save_pretrained) and is reloaded by supplying the save directory. - The model is loaded by supplying a local directory as `pretrained_model_name_or_path` and a configuration JSON file named *config.json* is found in the directory. state_dict (*dict[str, torch.Tensor]*, *optional*) : A state dictionary to use instead of a state dictionary loaded from saved weights file. This option can be used if you want to create a model from a pretrained configuration but load your own weights. In this case though, you should check if using [save_pretrained()](/docs/transformers/v5.6.0/ja/main_classes/model#transformers.PreTrainedModel.save_pretrained) and [from_pretrained()](/docs/transformers/v5.6.0/ja/main_classes/model#transformers.PreTrainedModel.from_pretrained) is not a simpler option. cache_dir (`str` or `os.PathLike`, *optional*) : Path to a directory in which a downloaded pretrained model configuration should be cached if the standard cache should not be used. force_download (`bool`, *optional*, defaults to `False`) : Whether or not to force the (re-)download of the model weights and configuration files, overriding the cached versions if they exist. proxies (`dict[str, str]`, *optional*) : A dictionary of proxy servers to use by protocol or endpoint, e.g., `{'http': 'foo.bar:3128', 'http://hostname': 'foo.bar:4012'}`. The proxies are used on each request. output_loading_info(`bool`, *optional*, defaults to `False`) : Whether or not to also return a dictionary containing missing keys, unexpected keys and error messages. local_files_only(`bool`, *optional*, defaults to `False`) : Whether or not to only look at local files (e.g., not try downloading the model). revision (`str`, *optional*, defaults to `"main"`) : The specific model version to use. It can be a branch name, a tag name, or a commit id, since we use a git-based system for storing models and other artifacts on huggingface.co, so `revision` can be any identifier allowed by git. trust_remote_code (`bool`, *optional*, defaults to `False`) : Whether or not to allow for custom models defined on the Hub in their own modeling files. This option should only be set to `True` for repositories you trust and in which you have read the code, as it will execute code present on the Hub on your local machine. code_revision (`str`, *optional*, defaults to `"main"`) : The specific revision to use for the code on the Hub, if the code leaves in a different repository than the rest of the model. It can be a branch name, a tag name, or a commit id, since we use a git-based system for storing models and other artifacts on huggingface.co, so `revision` can be any identifier allowed by git. kwargs (additional keyword arguments, *optional*) : Can be used to update the configuration object (after it being loaded) and initiate the model (e.g., `output_attentions=True`). Behaves differently depending on whether a `config` is provided or automatically loaded: - If a configuration is provided with `config`, `**kwargs` will be directly passed to the underlying model's `__init__` method (we assume all relevant updates to the configuration have already been done) - If a configuration is not provided, `kwargs` will be first passed to the configuration class initialization function ([from_pretrained()](/docs/transformers/v5.6.0/ja/main_classes/configuration#transformers.PreTrainedConfig.from_pretrained)). Each key of `kwargs` that corresponds to a configuration attribute will be used to override said attribute with the supplied `kwargs` value. Remaining keys that do not correspond to any configuration attribute will be passed to the underlying model's `__init__` function. ## Time Series ### AutoModelForTimeSeriesPrediction[[transformers.AutoModelForTimeSeriesPrediction]] #### transformers.AutoModelForTimeSeriesPrediction[[transformers.AutoModelForTimeSeriesPrediction]] [Source](https://github.com/huggingface/transformers/blob/v5.6.0/src/transformers/models/auto/modeling_auto.py#L2132) This is a generic model class that will be instantiated as one of the model classes of the library (with a time-series prediction head) when created with the [from_pretrained()](/docs/transformers/v5.6.0/ja/model_doc/auto#transformers.AutoModel.from_pretrained) class method or the [from_config()](/docs/transformers/v5.6.0/ja/model_doc/auto#transformers.AutoModel.from_config) class method. This class cannot be instantiated directly using `__init__()` (throws an error). from_configtransformers.AutoModelForTimeSeriesPrediction.from_confighttps://github.com/huggingface/transformers/blob/v5.6.0/src/transformers/models/auto/auto_factory.py#L206[{"name": "**kwargs", "val": ""}]- **config** ([PreTrainedConfig](/docs/transformers/v5.6.0/ja/main_classes/configuration#transformers.PreTrainedConfig)) -- The model class to instantiate is selected based on the configuration class: - `TimesFm2_5Config` configuration class: `TimesFm2_5ModelForPrediction` (TimesFm2_5Config model) - `TimesFmConfig` configuration class: `TimesFmModelForPrediction` (TimesFmConfig model) - **attn_implementation** (`str`, *optional*) -- The attention implementation to use in the model (if relevant). Can be any of `"eager"` (manual implementation of the attention), `"sdpa"` (using [`F.scaled_dot_product_attention`](https://pytorch.org/docs/master/generated/torch.nn.functional.scaled_dot_product_attention.html)), `"flash_attention_2"` (using [Dao-AILab/flash-attention](https://github.com/Dao-AILab/flash-attention)), or `"flash_attention_3"` (using [Dao-AILab/flash-attention/hopper](https://github.com/Dao-AILab/flash-attention/tree/main/hopper)). By default, if available, SDPA will be used for torch>=2.1.1. The default is otherwise the manual `"eager"` implementation.0 Instantiates one of the model classes of the library (with a time-series prediction head) from a configuration. Note: Loading a model from its configuration file does **not** load the model weights. It only affects the model's configuration. Use [from_pretrained()](/docs/transformers/v5.6.0/ja/model_doc/auto#transformers.AutoModel.from_pretrained) to load the model weights. Examples: ```python >>> from transformers import AutoConfig, AutoModelForTimeSeriesPrediction >>> # Download configuration from huggingface.co and cache. >>> config = AutoConfig.from_pretrained("google-bert/bert-base-cased") >>> model = AutoModelForTimeSeriesPrediction.from_config(config) ``` **Parameters:** config ([PreTrainedConfig](/docs/transformers/v5.6.0/ja/main_classes/configuration#transformers.PreTrainedConfig)) : The model class to instantiate is selected based on the configuration class: - `TimesFm2_5Config` configuration class: `TimesFm2_5ModelForPrediction` (TimesFm2_5Config model) - `TimesFmConfig` configuration class: `TimesFmModelForPrediction` (TimesFmConfig model) attn_implementation (`str`, *optional*) : The attention implementation to use in the model (if relevant). Can be any of `"eager"` (manual implementation of the attention), `"sdpa"` (using [`F.scaled_dot_product_attention`](https://pytorch.org/docs/master/generated/torch.nn.functional.scaled_dot_product_attention.html)), `"flash_attention_2"` (using [Dao-AILab/flash-attention](https://github.com/Dao-AILab/flash-attention)), or `"flash_attention_3"` (using [Dao-AILab/flash-attention/hopper](https://github.com/Dao-AILab/flash-attention/tree/main/hopper)). By default, if available, SDPA will be used for torch>=2.1.1. The default is otherwise the manual `"eager"` implementation. #### from_pretrained[[transformers.AutoModelForTimeSeriesPrediction.from_pretrained]] [Source](https://github.com/huggingface/transformers/blob/v5.6.0/src/transformers/models/auto/auto_factory.py#L253) Instantiate one of the model classes of the library (with a time-series prediction head) from a pretrained model. The model class to instantiate is selected based on the `model_type` property of the config object (either passed as an argument or loaded from `pretrained_model_name_or_path` if possible), or when it's missing, by falling back to using pattern matching on `pretrained_model_name_or_path`: - **timesfm** -- `TimesFmModelForPrediction` (TimesFmConfig model) - **timesfm2_5** -- `TimesFm2_5ModelForPrediction` (TimesFm2_5Config model) The model is set in evaluation mode by default using `model.eval()` (so for instance, dropout modules are deactivated). To train the model, you should first set it back in training mode with `model.train()` Examples: ```python >>> from transformers import AutoConfig, AutoModelForTimeSeriesPrediction >>> # Download model and configuration from huggingface.co and cache. >>> model = AutoModelForTimeSeriesPrediction.from_pretrained("google-bert/bert-base-cased") >>> # Update configuration during loading >>> model = AutoModelForTimeSeriesPrediction.from_pretrained("google-bert/bert-base-cased", output_attentions=True) >>> model.config.output_attentions True ``` **Parameters:** pretrained_model_name_or_path (`str` or `os.PathLike`) : Can be either: - A string, the *model id* of a pretrained model hosted inside a model repo on huggingface.co. - A path to a *directory* containing model weights saved using [save_pretrained()](/docs/transformers/v5.6.0/ja/main_classes/model#transformers.PreTrainedModel.save_pretrained), e.g., `./my_model_directory/`. model_args (additional positional arguments, *optional*) : Will be passed along to the underlying model `__init__()` method. config ([PreTrainedConfig](/docs/transformers/v5.6.0/ja/main_classes/configuration#transformers.PreTrainedConfig), *optional*) : Configuration for the model to use instead of an automatically loaded configuration. Configuration can be automatically loaded when: - The model is a model provided by the library (loaded with the *model id* string of a pretrained model). - The model was saved using [save_pretrained()](/docs/transformers/v5.6.0/ja/main_classes/model#transformers.PreTrainedModel.save_pretrained) and is reloaded by supplying the save directory. - The model is loaded by supplying a local directory as `pretrained_model_name_or_path` and a configuration JSON file named *config.json* is found in the directory. state_dict (*dict[str, torch.Tensor]*, *optional*) : A state dictionary to use instead of a state dictionary loaded from saved weights file. This option can be used if you want to create a model from a pretrained configuration but load your own weights. In this case though, you should check if using [save_pretrained()](/docs/transformers/v5.6.0/ja/main_classes/model#transformers.PreTrainedModel.save_pretrained) and [from_pretrained()](/docs/transformers/v5.6.0/ja/main_classes/model#transformers.PreTrainedModel.from_pretrained) is not a simpler option. cache_dir (`str` or `os.PathLike`, *optional*) : Path to a directory in which a downloaded pretrained model configuration should be cached if the standard cache should not be used. force_download (`bool`, *optional*, defaults to `False`) : Whether or not to force the (re-)download of the model weights and configuration files, overriding the cached versions if they exist. proxies (`dict[str, str]`, *optional*) : A dictionary of proxy servers to use by protocol or endpoint, e.g., `{'http': 'foo.bar:3128', 'http://hostname': 'foo.bar:4012'}`. The proxies are used on each request. output_loading_info(`bool`, *optional*, defaults to `False`) : Whether or not to also return a dictionary containing missing keys, unexpected keys and error messages. local_files_only(`bool`, *optional*, defaults to `False`) : Whether or not to only look at local files (e.g., not try downloading the model). revision (`str`, *optional*, defaults to `"main"`) : The specific model version to use. It can be a branch name, a tag name, or a commit id, since we use a git-based system for storing models and other artifacts on huggingface.co, so `revision` can be any identifier allowed by git. trust_remote_code (`bool`, *optional*, defaults to `False`) : Whether or not to allow for custom models defined on the Hub in their own modeling files. This option should only be set to `True` for repositories you trust and in which you have read the code, as it will execute code present on the Hub on your local machine. code_revision (`str`, *optional*, defaults to `"main"`) : The specific revision to use for the code on the Hub, if the code leaves in a different repository than the rest of the model. It can be a branch name, a tag name, or a commit id, since we use a git-based system for storing models and other artifacts on huggingface.co, so `revision` can be any identifier allowed by git. kwargs (additional keyword arguments, *optional*) : Can be used to update the configuration object (after it being loaded) and initiate the model (e.g., `output_attentions=True`). Behaves differently depending on whether a `config` is provided or automatically loaded: - If a configuration is provided with `config`, `**kwargs` will be directly passed to the underlying model's `__init__` method (we assume all relevant updates to the configuration have already been done) - If a configuration is not provided, `kwargs` will be first passed to the configuration class initialization function ([from_pretrained()](/docs/transformers/v5.6.0/ja/main_classes/configuration#transformers.PreTrainedConfig.from_pretrained)). Each key of `kwargs` that corresponds to a configuration attribute will be used to override said attribute with the supplied `kwargs` value. Remaining keys that do not correspond to any configuration attribute will be passed to the underlying model's `__init__` function.