pr-reviewer-7b-instruct-awq

pr-reviewer-7b-instruct-awq

A 7B code review model distilled from GPT-4o, fine-tuned on 73 PRs from popular Python repositories (fastapi, pydantic, flask, httpx, sqlmodel).

Model Details

• Base: unsloth/Qwen2.5-Coder-7B-Instruct
• Fine-tuning: QLoRA (rank=32, alpha=32, dropout=0.05, all linear layers)
• Quantization: AWQ 4-bit (~4 GB)
• Training data: 73 PRs, GPT-4o-generated reviews
• Eval: 13 hand-labeled PRs, GPT-4o-as-judge

Eval Results (GPT-4o judge, scale 1-5)

Usage with vLLM

vllm serve Hardik55/pr-reviewer-7b-instruct-awq --quantization awq

Lessons Learned

v1 vs v2 documents a real distillation lesson: with only 73 examples, fine-tuning a base model cannot teach instruction-following AND review style simultaneously. Switching to the Instruct base variant improved overall score 19% (actionability 85%) with no other changes.

Limitations

• 73 training examples is small; model often hedges
• Distillation ceiling: cannot exceed teacher quality
• AWQ calibrated on default wikitext-2 (not domain-specific)

Source

https://github.com/hardik5520/pr-reviewer-distillation