metadata
license: apache-2.0
base_model: Qwen/Qwen3.5-4B
tags:
- debugging
- code
- reinforcement-learning
- grpo
- precisecoder
PreciseCoder-4B
馃搫 Paper (arXiv:2610.18755) 路 馃寪 Project page 路 馃捇 Code
Qwen3.5-4B post-trained with reinforcement learning for precise debugging: given buggy Python code, produce a minimal, correct fix (PreciseCoder project).
Recipe. GDPO with two rewards, unit-test pass and unit-gated F1 (edit-block F1 against the ground-truth diff), weighted 2:1.
Eval:
| eval | result |
|---|---|
| PDB test (1205 BigCodeBench + LiveCodeBench debugging tasks) | unit 0.554 路 precision 0.638 路 recall 0.728 路 F1 0.650 路 edit lines 3.0 |
| PDB-Wild (474 tasks: BCB 37 + LCB 209 + SWE-smith 228) | unit 0.276 路 precision 0.382 路 recall 0.410 路 F1 0.377 路 edit lines 10.8 |
| GT Debug (1512 clean inputs) | no-op 54.7% 路 damage 12.1% 路 unit pass 87.9% |
no-op = share of clean (bug-free) inputs returned unchanged; damage = share broken by an edit.
Load with transformers / vLLM (>=0.17) as a standard Qwen3.5 model (thinking enabled by default).
Citation
@article{precisecoder,
title = {{PreciseCoder}: Faithful and Generalizable Debugging Beyond Unit-Test Rewards},
author = {Zhan, Yunkai and Chai, Miaosen and Wang, Shangshang and Zhong, Jike and Yang, Yuqing and Fu, Deqing and Chang, Serina and Zhu, Wang Bill},
journal = {arXiv preprint arXiv:2610.18755},
year = {2026}
}