gpt-4 was the last model that improved llms significantly
retarded take4 was simply continuation of work from 3.5o1 was a step changeclosest thing since then is the programming / math RL'd models. not sure which claude model got there first, but i think it was them this time.
>>108205176>o1 was a step changeno not really apart from fake synthetic benchmarks
>>108205186>the first model with reasoning was not a step changesmol brain
>>108205193>thinks reasoning changed anything
>>108205202are you one of those people who has sex with the llms?
>>108205231>faghomineni have sex with your sister tho
>>108205323i'll take that as a yes. i don't sex with llms, but i could see why reasoning may not play a big role there. no judgement, different strokes and all that.
>>108205070idk about that
