I built code-repair training data and shipped the eval so you can rerun it
1 points
6 hours ago
| 1 comment
| huggingface.co
| HN
user-
6 hours ago
[-]
> Your 7B fixes 17% of our hardest faults. Trained on TrueSET: 40% — more than double. Check it yourself for under $0.50.

> (Hard tier = multi-file faults base models genuinely fail, and every one of its verification scripts survived adversarial attack-testing — wrong-but-plausible code cannot pass them.

Your LLM condensed this to the point where it hurts my brain to read.

reply
TrueSET_Data
6 hours ago
[-]
you are Right let me rewrite it so it don't sound so try hard
reply