Qwen3-0.6B-Sushi-Coder

A 0.6B code generation model fine-tuned from Qwen3-0.6B for Python code generation.

Training

This model was trained in two stages:

1. GRPO using TRL with reward model based on test execution and formatting
2. SFT on microsoft/rStar-Coder and open-r1/codeforces-cots

Training was done on HuggingFace Jobs infrastructure using TRL.

Usage

from transformers import AutoModelForCausalLM, AutoTokenizer

model_id = "bigatuna/Qwen3-0.6B-Sushi-Coder"
tokenizer = AutoTokenizer.from_pretrained(model_id)
model = AutoModelForCausalLM.from_pretrained(model_id, torch_dtype="auto", device_map="auto")

messages = [
    {"role": "user", "content": "Write a Python function to check if a number is prime."}
]

text = tokenizer.apply_chat_template(messages, tokenize=False, add_generation_prompt=True)
inputs = tokenizer(text, return_tensors="pt").to(model.device)

outputs = model.generate(
    **inputs,
    max_new_tokens=512,
    temperature=0.6,
    top_p=0.95,
    do_sample=True
)
print(tokenizer.decode(outputs[0][inputs.input_ids.shape[1]:], skip_special_tokens=True))

Recommended Parameters

Note: Avoid greedy decoding (temperature=0) as it can cause repetition issues with Qwen3 models.

Evaluation

Note: The model scores slightly higher with prompt tuning and max_model_len=4000, but the results above represent baseline settings.

The evaluation can be reproduced with the following command:

lm_eval \
  --model vllm \
  --model_args pretrained=bigatuna/Qwen3-0.6B-Sushi-Coder,tensor_parallel_size=1,dtype=float16,gpu_memory_utilization=0.8,max_model_len=2048,trust_remote_code=True \
  --tasks humaneval \
  --batch_size 1 \
  --confirm_run_unsafe_code

Baseline comparison:

lm_eval \
  --model vllm \
  --model_args pretrained=Qwen/Qwen3-0.6B,tensor_parallel_size=1,dtype=float16,gpu_memory_utilization=0.8,max_model_len=2048,trust_remote_code=True \
  --tasks humaneval \
  --batch_size 1 \
  --confirm_run_unsafe_code

Limitations

• Optimized for Python; other languages may have reduced quality
• Small model size limits complex reasoning
• May generate plausible but incorrect code for edge cases

License

Apache 2.0