DeepConf: Scaling LLM reasoning with confidence, not just compute

98 points | by che_shr_cat a day ago

36 comments