Dataset bug: Regex error creates 'rsthon' in multiPL-E
code_star · x · 2026-08-22
A regex error in the MBPP subset of multiPL-E caused 'py' to be replaced with short language names instead of 'python'. This resulted in invalid function names like 'rsthon', 'swthon', and 'gothon'.
More from Research
- Arcee open-sources nac agent harness for long-running tasks with intent alignment — latkins · 2026-08-22
- By-Its-Cover: A CLIP-Only Book Recommendation System — LaidbyKool-aid · 2026-08-22
- Math formalization tool 'span' integrates LLM for adversarial audits — lpachter · 2026-08-22
- NLA results surprising, token-level attribution may help — aryaman2020 · 2026-08-22
- Open-weight models can't write long stories: quality degrades or length undershoots — mrdrozdov · 2026-08-22
- Paper questions interpretability: Black-box prompting often beats white-box tools — aryaman2020 · 2026-08-22