When Less Language is More: Language-Reasoning Disentanglement Makes LLMs Better Multilingual Reasoners

Devs

When Less Language is More: Language-Reasoning Disentanglement Makes LLMs Better Multilingual Reasoners | Read Paper on Bytez