deepseek-ai
/

DeepSeek-V3

Text Generation

text-generation-inference

Model card Files Files and versions

Instructions to use deepseek-ai/DeepSeek-V3 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use deepseek-ai/DeepSeek-V3 with Transformers:

# Use a pipeline as a high-level helper
from transformers import pipeline

pipe = pipeline("text-generation", model="deepseek-ai/DeepSeek-V3", trust_remote_code=True)
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("deepseek-ai/DeepSeek-V3", trust_remote_code=True)
model = AutoModelForCausalLM.from_pretrained("deepseek-ai/DeepSeek-V3", trust_remote_code=True)
messages = [
    {"role": "user", "content": "Who are you?"},
]
inputs = tokenizer.apply_chat_template(
	messages,
	add_generation_prompt=True,
	tokenize=True,
	return_dict=True,
	return_tensors="pt",
).to(model.device)

outputs = model.generate(**inputs, max_new_tokens=40)
print(tokenizer.decode(outputs[0][inputs["input_ids"].shape[-1]:]))

Inference
HuggingChat
Notebooks
Google Colab
Kaggle
Local Apps

How to use deepseek-ai/DeepSeek-V3 with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "deepseek-ai/DeepSeek-V3"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "deepseek-ai/DeepSeek-V3",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/deepseek-ai/DeepSeek-V3

How to use deepseek-ai/DeepSeek-V3 with SGLang:

Install from pip and serve model

# Install SGLang from pip:
pip install sglang
# Start the SGLang server:
python3 -m sglang.launch_server \
    --model-path "deepseek-ai/DeepSeek-V3" \
    --host 0.0.0.0 \
    --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "deepseek-ai/DeepSeek-V3",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker images

docker run --gpus all \
    --shm-size 32g \
    -p 30000:30000 \
    -v ~/.cache/huggingface:/root/.cache/huggingface \
    --env "HF_TOKEN=<secret>" \
    --ipc=host \
    lmsysorg/sglang:latest \
    python3 -m sglang.launch_server \
        --model-path "deepseek-ai/DeepSeek-V3" \
        --host 0.0.0.0 \
        --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "deepseek-ai/DeepSeek-V3",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use deepseek-ai/DeepSeek-V3 with Docker Model Runner:
```
docker model run hf.co/deepseek-ai/DeepSeek-V3
```

Resources

View closed (26)

[Proposal] A Blueprint for Persistent AI Sentience: SSD Memory & Self-Evolving Code

#123 opened 7 days ago by

Fix chat_template crash when assistant message omits the `content` key

#122 opened 13 days ago by

Built a free hosted demo of V3 — feedback welcome

#121 opened 28 days ago by

Add GSM8K eval result (89.3)

#120 opened about 2 months ago by

Please take a look at some interesting new "AHA" moments

#119 opened 2 months ago by

hejun0180-pixel

Install & run deepseek-ai/DeepSeek-V3 easily using llmpm

#118 opened 2 months ago by

Benchmark: DeepSeek V3 vs GPT-4o vs Claude for coding tasks

#117 opened 3 months ago by

Add MMLU-Pro evaluation result (64.4)

#116 opened 4 months ago by

Add GSM8K evaluation result

#115 opened 4 months ago by

Add GSM8K evaluation result (89.3%)

#114 opened 4 months ago by

Add GSM8K evaluation result

#113 opened 4 months ago by

Add GSM8K evaluation result

#112 opened 4 months ago by

Production deployment considerations

#111 opened 5 months ago by

dememe4301

#110 opened 6 months ago by

Update inference/model.py

#109 opened 6 months ago by

Update README.md

#107 opened 10 months ago by

Remove redundant code

#106 opened 11 months ago by

MTP Integration: Unexpectedly High Loss with Loaded Weights

#105 opened 11 months ago by

add AIBOM

#104 opened 11 months ago by

deep

#103 opened 11 months ago by

sunde

#102 opened 12 months ago by

Update tokenizer_config.json

#101 opened about 1 year ago by

DeepSeek V3 model Bad Cases Genuine User Open Reviews and Comments Collection

#99 opened about 1 year ago by

Make config params float to avoid warning in Transformers

#97 opened about 1 year ago by

Point to latest checkpoint

#96 opened about 1 year ago by

how to convert model to bf16

#95 opened about 1 year ago by

Update README.md

#94 opened about 1 year ago by

Alirezaaa123456

Deepseek V3

#93 opened about 1 year ago by

【Q】shared_head weights of MTP

#92 opened about 1 year ago by

fix for transformers 4.49 compatibility.

#91 opened about 1 year ago by

Update README.md

#90 opened about 1 year ago by

无辅助损失专家偏置代码实现的小问题 A Small Issue in the Code Implementation of Auxiliary-Loss-Free Load Balancing Expert Bias

#89 opened about 1 year ago by

Fix generation with latest transformers

#88 opened about 1 year ago by

Add pipeline tag

#86 opened over 1 year ago by

Some of the safetensor files are not marked as safe

#85 opened over 1 year ago by

Update README.md

#84 opened over 1 year ago by

ValueError: Must flatten tensors with uniform dtype but got torch.bfloat16 and torch.float8_e4m3fn

#82 opened over 1 year ago by

Update README.md

#81 opened over 1 year ago by deleted

Update README.md

#80 opened over 1 year ago by

Update README.md

#79 opened over 1 year ago by

chat

#77 opened over 1 year ago by

DeepSeek-V3-lite naming conventions?

#76 opened over 1 year ago by

torch.distributed.DistNetworkError

#75 opened over 1 year ago by

remove reference to deprecated transformers code

#74 opened over 1 year ago by

Update README.md

#73 opened over 1 year ago by

DeepSeek R1 answer ChatGPT ??

#72 opened over 1 year ago by

ValueError: Unrecognized configuration class <class 'transformers_modules.configuration_deepseek.DeepseekV3Config'> to build an AutoTokenizer.

#69 opened over 1 year ago by

Paralelized script

#67 opened over 1 year ago by

I am getting an error message while executing pip install - r requirements. txt

#64 opened over 1 year ago by

`aux_loss_alpha` should be 1e-4 instead of 1e-3?

#61 opened over 1 year ago by