logoalt Hacker News

npntoday at 3:24 AM0 repliesview on HN

openai did human crafted chain of thought dataset training. deepseek didn't have the resources so they attempted RL. doing RL correctly is hard because of the risk of model collapsing.