OpenAI's Perplexity Narrative Questioned: RL, Not Pretraining, Drove Math Breakthrough

A user challenged OpenAI's claim that a model with log-perplexity of 0.76 enabled breakthrough math results, arguing the gains stem from RL rather than pretraining and that perplexity numbers are meaningless without dataset and context details.

2026-11-23 ~ 2026-11-23 · 2 related posts