Instructions to use tomasmcm/QwQ-Coder-R1-Distill-32B with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use tomasmcm/QwQ-Coder-R1-Distill-32B with Transformers:

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

pipe = pipeline("text-generation", model="tomasmcm/QwQ-Coder-R1-Distill-32B")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("tomasmcm/QwQ-Coder-R1-Distill-32B")
model = AutoModelForCausalLM.from_pretrained("tomasmcm/QwQ-Coder-R1-Distill-32B")
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]:]))

Notebooks
Google Colab
Kaggle
Local Apps

vLLM

How to use tomasmcm/QwQ-Coder-R1-Distill-32B with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "tomasmcm/QwQ-Coder-R1-Distill-32B"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "tomasmcm/QwQ-Coder-R1-Distill-32B",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/tomasmcm/QwQ-Coder-R1-Distill-32B

SGLang

How to use tomasmcm/QwQ-Coder-R1-Distill-32B 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 "tomasmcm/QwQ-Coder-R1-Distill-32B" \
    --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": "tomasmcm/QwQ-Coder-R1-Distill-32B",
		"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 "tomasmcm/QwQ-Coder-R1-Distill-32B" \
        --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": "tomasmcm/QwQ-Coder-R1-Distill-32B",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use tomasmcm/QwQ-Coder-R1-Distill-32B with Docker Model Runner:
```
docker model run hf.co/tomasmcm/QwQ-Coder-R1-Distill-32B
```
Browse Quantizations to use this model in llama.cpp, Ollama, LM Studio, or any compatible app.

tomasmcm/QwQ-Coder-R1-Distill-32B

This is a merge of pre-trained language models created using mergekit.

Following up on tomasmcm/sky-t1-coder-32b-flash, this experiment tries to merge 2 reasoning models based on Qwen 32B with a Coder model. But it seems to have caused the model to loose it's thinking abilities, even when adding <think> to the prompt.

Merge Details

Merge Method

This model was merged using the SCE merge method using Qwen/Qwen2.5-32B as a base.

Models Merged

The following models were included in the merge:

Configuration

The following YAML configuration was used to produce this model:

models:
  # Pivot model
  - model: Qwen/Qwen2.5-32B
  # Target models
  - model: Qwen/Qwen2.5-Coder-32B-Instruct
  - model: deepseek-ai/DeepSeek-R1-Distill-Qwen-32B
  - model: Qwen/QwQ-32B
merge_method: sce
base_model: Qwen/Qwen2.5-32B
parameters:
  select_topk: 1.0
dtype: bfloat16