rahilfahim/code-reviewer-lora

Code Reviewer LoRA — Llama 3.2 3B

A LoRA adapter fine-tuned with QLoRA on Llama 3.2 3B Instruct to review Python code with severity levels (Critical, Warning, Info).

📊 Training Summary

Metric	Value
Base model	Llama 3.2 3B Instruct
Method	QLoRA (rank 16, alpha 16)
Training examples	500
Training time	2.5 min (Colab T4)
Final loss	0.11
Adapter size	88 MB

🚀 Usage

from unsloth import FastLanguageModel

model, tokenizer = FastLanguageModel.from_pretrained(
    model_name="unsloth/Llama-3.2-3B-Instruct-bnb-4bit",
    max_seq_length=2048,
    load_in_4bit=True,
)
model.load_adapter("rahilfahim/code-reviewer-lora")
FastLanguageModel.for_inference(model)

prompt = """### Instruction:
You are a Python code reviewer. Review the following code and identify bugs, style issues, and improvements.

### Input:
def add(a,b): return a+b

### Response:
"""

inputs = tokenizer([prompt], return_tensors="pt").to("cuda")
outputs = model.generate(**inputs, max_new_tokens=256, temperature=0.3)
print(tokenizer.batch_decode(outputs, skip_special_tokens=True)[0])

📦 Links

• Full project: GitHub Repo
• Live demo: HF Space
• Benchmark: docs/benchmark_results.json

🙏 Acknowledgments

Trained using Unsloth on Google Colab.