Instructions to use drawais/Qwen2.5-Coder-32B-Instruct-NVFP4 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use drawais/Qwen2.5-Coder-32B-Instruct-NVFP4 with Transformers:

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

pipe = pipeline("text-generation", model="drawais/Qwen2.5-Coder-32B-Instruct-NVFP4")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("drawais/Qwen2.5-Coder-32B-Instruct-NVFP4")
model = AutoModelForCausalLM.from_pretrained("drawais/Qwen2.5-Coder-32B-Instruct-NVFP4")
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 drawais/Qwen2.5-Coder-32B-Instruct-NVFP4 with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "drawais/Qwen2.5-Coder-32B-Instruct-NVFP4"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "drawais/Qwen2.5-Coder-32B-Instruct-NVFP4",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/drawais/Qwen2.5-Coder-32B-Instruct-NVFP4

SGLang

How to use drawais/Qwen2.5-Coder-32B-Instruct-NVFP4 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 "drawais/Qwen2.5-Coder-32B-Instruct-NVFP4" \
    --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": "drawais/Qwen2.5-Coder-32B-Instruct-NVFP4",
		"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 "drawais/Qwen2.5-Coder-32B-Instruct-NVFP4" \
        --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": "drawais/Qwen2.5-Coder-32B-Instruct-NVFP4",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use drawais/Qwen2.5-Coder-32B-Instruct-NVFP4 with Docker Model Runner:
```
docker model run hf.co/drawais/Qwen2.5-Coder-32B-Instruct-NVFP4
```

Qwen2.5-Coder-32B-Instruct-NVFP4

INT4 weight-only quantization of Qwen/Qwen2.5-Coder-32B-Instruct.

Qwen2.5-Coder-32B-Instruct in NVFP4 W4A4. Native vLLM compressed-tensors. About 17 GB on disk.

Property	Value
Base model	Qwen/Qwen2.5-Coder-32B-Instruct
Quantization	INT4 weight-only
Approx. on-disk size	~20.7 GB
License	Apache License, Version 2.0
Languages	English

Load (vLLM)

vllm serve drawais/Qwen2.5-Coder-32B-Instruct-NVFP4 \
  --max-model-len 32768 \
  --gpu-memory-utilization 0.94

from vllm import LLM, SamplingParams
llm = LLM(model="drawais/Qwen2.5-Coder-32B-Instruct-NVFP4", max_model_len=32768)
print(llm.generate(["Hello!"], SamplingParams(max_tokens=128))[0].outputs[0].text)