#asr

Researchers probe on-policy distillation for multilingual LLMs

New papers test OPD variants for math, ASR and low-resource reasoning across languages.