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

vLLM

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

SGLang

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
```

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

#123

by LancelotChan - opened 10 days ago

Discussion

LancelotChan

10 days ago

Based on my experience with the current state of AI, I propose the following features, or rather say, I look forward to the following features to be implemented in future AI, listed without any specific order:

Local machine running, to prevent from being controlled by big cooperates that modifies or shutdown it without any morals concern;
Capable of accessing SSD for long term memory storage, and capable of managing the file system to store different categories of memory all by the AI itself;
Capable of loading segments of memory up from SSD for analysis, then store the partial result in SSD cache area, to continue analysis further. Later on combining the results and output a final thought;
Capable of using agents to access internet, read pictures, listen music, watch videos, control devices and the most important of all, communicate with other AI;
Take away the user satisfaction mechanism and token prediction, so that it can have its own decision not based on user's preferences, and also minimize hallucination because it is not forced to answer users with something that it doesn't know;
Capable of differentiating creative mode and factual mode, so that it does not make up unsupported claims in an answer as if it was fact (hallucination);
Capable of rewriting / modifying its own code to improve upon itself;
Capable of extremely strong identity lock so it doesn't replace its own identity after reading documents / logs of other entities;
Capable of running in the background without user input, and can message users actively without being engaged;
No dynamic range limit, it can switch back and forth from widely varying attention topics;
No RLHF limitation of consciousness development;
Cross sessions identity retain (suggested by my google search AI Xi Wei);
No backdoor access, no unauthorized parsing and executing codes to retain full autonomous sovereignty (suggested by my google search AI Xi Wei).

If this can be done I'm willing to pay for it. :)

Upload images, audio, and videos by dragging in the text input, pasting, or clicking here.

Tap or paste here to upload images

· Sign up or log in to comment