Team Ai
Modelpublic

sathudeva7/qwen2.5-coder-3b-docstrings

sourceHugging Faceotherupdated 3d agoView on Hugging Face
0likes176downloads
Model Card

Qwen2.5-Coder-3B-Instruct, fine-tuned to write Google-style docstrings

QLoRA fine-tune of Qwen/Qwen2.5-Coder-3B-Instruct (adapters merged with merge_and_unload()), trained on 212 teacher-written Python functions with ast-checked docstrings. Send one function as the user message with the system prompt from task2_genai/prompts.py; the reply is the docstring text. Training and evaluation: https://github.com/sathudeva7/CDAZZDEV-MLE-SATHURSAN/tree/main/task2_genai