Instructions to use ryota39/Qwen3-8B-math-RL-en with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use ryota39/Qwen3-8B-math-RL-en with Transformers:

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

pipe = pipeline("text-generation", model="ryota39/Qwen3-8B-math-RL-en")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("ryota39/Qwen3-8B-math-RL-en")
model = AutoModelForCausalLM.from_pretrained("ryota39/Qwen3-8B-math-RL-en", 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=40)
print(tokenizer.decode(outputs[0][inputs["input_ids"].shape[-1]:]))

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use ryota39/Qwen3-8B-math-RL-en with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "ryota39/Qwen3-8B-math-RL-en"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "ryota39/Qwen3-8B-math-RL-en",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/ryota39/Qwen3-8B-math-RL-en

SGLang

How to use ryota39/Qwen3-8B-math-RL-en 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 "ryota39/Qwen3-8B-math-RL-en" \
    --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": "ryota39/Qwen3-8B-math-RL-en",
		"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 "ryota39/Qwen3-8B-math-RL-en" \
        --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": "ryota39/Qwen3-8B-math-RL-en",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use ryota39/Qwen3-8B-math-RL-en with Docker Model Runner:
```
docker model run hf.co/ryota39/Qwen3-8B-math-RL-en
```

ryota39 commited on Dec 9, 2025

Commit

a6b68ce

verified ·

1 Parent(s): cb7dfaa

Update README.md

Browse files

Files changed (1) hide show

README.md +1 -1

README.md CHANGED Viewed

@@ -11,7 +11,7 @@ base_model:
 # Qwen3-8B-math-RL-en
-- [Qwen/Qwen3-8B](https://huggingface.co/Qwen/Qwen3-8B)を少ないトークン予算の元`max\tokens=256`で、[openai/gsm8k](https://huggingface.co/datasets/openai/gsm8k)を使って事後学習したモデルです。
 - [tinker_cookbook](https://tinker-docs.thinkingmachines.ai/rl/rl-basic)の[rl_basic.py](https://github.com/thinking-machines-lab/tinker-cookbook/blob/main/tinker_cookbook/recipes/rl_basic.py)に従い、accuracy_rewardとformat_rewardを最大化するようなモデルの重みを学習しています。
 - 詳細は[Tinker APIと限られたトークン長でのLLMの強化学習](https://zenn.dev/kaeru39/articles/e4c7d19ab13bea)に記載しています。

 # Qwen3-8B-math-RL-en
+- [Qwen/Qwen3-8B](https://huggingface.co/Qwen/Qwen3-8B)を少ないトークン予算の元`max_new_tokens=256`で、[openai/gsm8k](https://huggingface.co/datasets/openai/gsm8k)を使って事後学習したモデルです。
 - [tinker_cookbook](https://tinker-docs.thinkingmachines.ai/rl/rl-basic)の[rl_basic.py](https://github.com/thinking-machines-lab/tinker-cookbook/blob/main/tinker_cookbook/recipes/rl_basic.py)に従い、accuracy_rewardとformat_rewardを最大化するようなモデルの重みを学習しています。
 - 詳細は[Tinker APIと限られたトークン長でのLLMの強化学習](https://zenn.dev/kaeru39/articles/e4c7d19ab13bea)に記載しています。