Qwen
/

QwQ-32B-Preview

Text Generation

text-generation-inference

Model card Files Files and versions

Instructions to use Qwen/QwQ-32B-Preview with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use Qwen/QwQ-32B-Preview with Transformers:

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

pipe = pipeline("text-generation", model="Qwen/QwQ-32B-Preview")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# pip install -U transformers accelerate
# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("Qwen/QwQ-32B-Preview")
model = AutoModelForCausalLM.from_pretrained("Qwen/QwQ-32B-Preview", device_map="auto")
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=256)
print(tokenizer.decode(outputs[0][inputs["input_ids"].shape[-1]:]))

Inference
Notebooks
Google Colab
Kaggle
Local Apps Settings

How to use Qwen/QwQ-32B-Preview with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "Qwen/QwQ-32B-Preview"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "Qwen/QwQ-32B-Preview",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/Qwen/QwQ-32B-Preview

How to use Qwen/QwQ-32B-Preview 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 "Qwen/QwQ-32B-Preview" \
    --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": "Qwen/QwQ-32B-Preview",
		"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 "Qwen/QwQ-32B-Preview" \
        --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": "Qwen/QwQ-32B-Preview",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use Qwen/QwQ-32B-Preview with Docker Model Runner:
```
docker model run hf.co/Qwen/QwQ-32B-Preview
```

Resources

View closed (15)

Add EvalEval community eval results

#63 opened 5 months ago by

Improve language tag

#61 opened over 1 year ago by

add new model metadata

#60 opened over 1 year ago by

question about output length

#59 opened almost 2 years ago by

Remove CoT prompt from chat template

#58 opened almost 2 years ago by

Why is this not in inference api?

#56 opened almost 2 years ago by

请问你们qwq的prompt是什么？推理有什么特殊设置吗？我这边复现live-codebench只有35～38分。（live-code-bench-0801-1101）

#53 opened almost 2 years ago by

Does QwQ support function calling?

#52 opened almost 2 years ago by

Tried the "strawberry" demo case but got wrong answer😂

#51 opened almost 2 years ago by

HuggingLianWang

🖐️🖐️ Can we please get a 7B reasoning model? 😎🤗🤗😍

#50 opened almost 2 years ago by

Request: DOI

#49 opened almost 2 years ago by

Can't reproduce the evaluation result of GPQA dataset

#47 opened almost 2 years ago by

Adding Evaluation Results

#44 opened almost 2 years ago by

Evaluation Details of QwQ-32B-Preview on LiveCodeBench Benchmark

#43 opened almost 2 years ago by

Independent Evaluation of QwQ-32B-Preview on MMLU-Pro Benchmark

#42 opened almost 2 years ago by

Can't copy the full reasoning

#41 opened almost 2 years ago by

Create bsnwritingservices

#40 opened almost 2 years ago by

https://huggingface.co/Qwen/QwQ-32B-Preview/resolve/main/.gitattributes

#39 opened almost 2 years ago by

Floating point exception

#38 opened almost 2 years ago by

My prompt induced infinite loops with QwQ

#35 opened almost 2 years ago by

docs: update README.md

#34 opened almost 2 years ago by

Which one is more risky, type 1 or type 2?

#33 opened almost 2 years ago by

Silly tavern template?

#32 opened almost 2 years ago by

Update README.md

#31 opened almost 2 years ago by

Adding Evaluation Results

#30 opened almost 2 years ago by

leaderboard-pr-bot

Fine-tuning or system prompt for fiction writing?

#28 opened almost 2 years ago by deleted

Kkk

#27 opened almost 2 years ago by

Update README.md

#26 opened almost 2 years ago by

OMSRI-SRI-RADHAKRSNA-1728

Will the training code and data be released?

#25 opened almost 2 years ago by

early exit reasoning inference

#24 opened almost 2 years ago by

RAG architecture based on the model family qwen2.5

#22 opened almost 2 years ago by

This is way too much... USB? Yes. U SB.

#21 opened almost 2 years ago by

该模型只适合聊天，没有其他能力

#20 opened almost 2 years ago by

trick for less token usage and less halucination

#19 opened almost 2 years ago by

multi GPU inferencing

#18 opened almost 2 years ago by

Install QwQ 32B Locally - Step by Step Video Tutorial

#17 opened almost 2 years ago by

Adds Chinese characters to responses

#16 opened almost 2 years ago by

Problem with specific output format

#15 opened almost 2 years ago by

UwU

#14 opened almost 2 years ago by

About Releasing Some Examples of QWQ's Training Dataset

#13 opened almost 2 years ago by

Nice name QAQ. I'll later upload a 4.7bpw quantized model if it works.

#12 opened almost 2 years ago by

QuietSTAR?

#9 opened almost 2 years ago by

albatrossbirdie

Inference Speed Benchmark and GPU memeory usage

#8 opened almost 2 years ago by

Some Badcases

#7 opened almost 2 years ago by

ways to contain the thought process from users for a cleaner interface

#5 opened almost 2 years ago by

A huge thanks!

#2 opened almost 2 years ago by

ggggggggg

#1 opened almost 2 years ago by