Re:How Deepseek R1 Was Trained
by okdood64 on Monday January 27, 2025 @07:23PM
(#42847487)
Can someone dumb to me, a generalist engineer who has a very surface level knowledge of how training LLMs work: what people were doing before and what GRPO is doing different?
[ Parent
| on HN ]
Re:How Deepseek R1 Was Trained
by anitakirkovska on Tuesday January 28, 2025 @02:31AM
(#42849790)
They were using techniques like PPO that have a model (like a critic!) that evaluates whether the new model gives accurate responses. With GRPO, the don't have that and instead evaluate the answers based on predefined rules like coherence/ formatting. For example, for math problems, these rules will check if the answer adheres to math principles or logic!
I wrote more here (lmk if this is useful): https://www.vellum.ai/blog/the-training-of-deepseek-r1-and-w...