---
license: apache-2.0
datasets:
- alexjercan/bugnet
language:
- en
metrics:
- code_eval
base_model:
- Qwen/Qwen2.5-Coder-0.5B-Instruct
pipeline_tag: text-generation
---
# Qwen BugFix LoRA

LoRA fine-tuned version of Qwen2.5-Coder-0.5B-Instruct
for Python bug fixing.

## Dataset
- alexjercan/bugnet

## Training
- 2000 steps
- ~3.1 epochs
- effective batch size 4

## Results
Training loss reduced:
0.45 → 0.23