Haider. (@slow_developer) 's Twitter Profile
Haider.

@slow_developer

together, we build an intelligent future.

ID: 1461305890892570628

linkhttps://lnk.bio/slowdeveloper calendar_today18-11-2021 12:11:02

26,26K Tweet

41,41K Takipçi

2,2K Takip Edilen

Haider. (@slow_developer) 's Twitter Profile Photo
Haider.
@slow_developer

3 months ago

huge step toward self-improving LLMs this paper introduces a new way for llms to improve without external rewards the result: • matches RLHF on math • +60% improvement on code benchmarks • learns to reason step by step • scales from 1b to 14b • fast convergence in ~10 RL

huge step toward self-improving LLMs this paper introduces a new way for llms to improve without external rewards the result: • matches RLHF on math • +60% improvement on code benchmarks • learns to reason step by step • scales from 1b to 14b • fast convergence in ~10 RL