Instructions to use WeiboAI/VibeThinker-3B with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use WeiboAI/VibeThinker-3B with Transformers:

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

pipe = pipeline("text-generation", model="WeiboAI/VibeThinker-3B")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("WeiboAI/VibeThinker-3B")
model = AutoModelForCausalLM.from_pretrained("WeiboAI/VibeThinker-3B")
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
Local Apps Settings

vLLM

How to use WeiboAI/VibeThinker-3B with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "WeiboAI/VibeThinker-3B"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "WeiboAI/VibeThinker-3B",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/WeiboAI/VibeThinker-3B

SGLang

How to use WeiboAI/VibeThinker-3B 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 "WeiboAI/VibeThinker-3B" \
    --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": "WeiboAI/VibeThinker-3B",
		"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 "WeiboAI/VibeThinker-3B" \
        --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": "WeiboAI/VibeThinker-3B",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use WeiboAI/VibeThinker-3B with Docker Model Runner:
```
docker model run hf.co/WeiboAI/VibeThinker-3B
```

Really looking forward for 9B or 12B variants

#20

by perelmanych - opened 3 days ago

Discussion

perelmanych

3 days ago

Thank you very much for the wonderful model! Do you guys have any plans to scale it to 9B or 12B sizes?

junlinzhang

WeiboAI org 3 days ago

Thanks for your attention. We haven’t finalized our next steps yet, and will set future goals according to community feedback on this version.

ramaerik97

3 days ago

thanks for uploading the model, can I added your model to my side project ? I'll credit and mention you if you have x(twitter), my x account @erik_es2030, my side project is focus running llm on cpu only so my side project is still on research phase thanks.

junlinzhang

WeiboAI org 3 days ago

Of course, feel free to use it for your side projects. VibeThinker adopts the MIT open-source license, which has extremely loose restrictions. You can do almost anything with it, even commercialize it if you want.

MartialTerran

2 days ago

ramaerik97 said: "my side project is focus running llm on cpu only".
ramaerik97, what opesource LLM models have you run inference mode in "cpu only"? What is your target CPU RAM and SIMD bit range? Intel Core Ultra with SIMD 256? SIMD 512? The Pi02W and related ARM processers are said to have a decent SIMD/Neon instruction set (small RAM and slow however https://www.linkedin.com/posts/ayush440v_techexperiments-llm-edgeai-activity-7301867880336211968-DFI4 ). Do you have a table of performance/results for models that you evaluated on "cpu only"?

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