4 ms·DeepSeek-R1 incentivizes reasoning in LLMs through reinforcement learning9 points by Anon84 1y ago