moofeez

We actually tried this and (for us) it doesn't lead to enough of a gain that's worth the extra complexity.

I post-trained Qwen3-Coder to effectively use a real debugger harness (stepping through code, inspecting runtime state, and reasoning from execution traces) instead of reasoning only over the static code.

Results on 27 held-out bug tasks:

- Base model + debugger harness (no RL): 67% solve rate, only ~5% fewer turns

- After post-training with RL: 70% → 89% solve rate, 46 → 19 median turns (−59%)

The key was teaching the model how to use the debugger effectively. Just giving the base model the harness did not help.

More detailed results and a full blog post coming soon. Happy to answer questions in the meantime!