arXiv
Language Drift in RLVR Post-Training of Language Models
This preprint examines why RLVR on novel reasoning tasks may move language-model chains of thought away…
Sara Kiani
arXivResearch previewSara Kiani
Language Drift in RLVR Post-Training of Language Models
Sara KianiThis preprint examines why RLVR on novel reasoning tasks may move language-model chains of thought away from human-readable language.
This is a short research preview. Full results and limitations are available in the complete article.
Read full research ←