ProRL: Prolonged Reinforcement Learning Expands Reasoning Boundaries in Large Language Models — mentioned by t55

Mon, 02 Jun 2025 14:07:36 GMT

yeah, RLVR is still nascent and hence there's lots of noise. > How can these spurious rewards possibly work? Can we get similar gains on other models with broken rewards? it's because in those cases, RLVR merely elicits the reasoning strategies already contained in the model through pre-training this paper, which uses Reasoning gym, shows that you need to train for way longer than those papers you mentioned to actually uncover novel reasoning strategies: https://arxiv.org/abs/2505.24864

Paper and discussion history. A mention is not an endorsement.

ProRL: Prolonged Reinforcement Learning Expands Reasoning Boundaries in Large Language Models — mentioned by ninakostoska

Mon, 02 Jun 2025 11:03:01 GMT

Cool to see NVIDIA’s most recent reasoning model [1] already uses Reasoning Gymas a large part of their data mixture [1] https://arxiv.org/abs/2505.24864

Paper and discussion history. A mention is not an endorsement.

ProRL: Prolonged Reinforcement Learning Expands Reasoning Boundaries in Large Language Models · HackerPapers discussions

ProRL: Prolonged Reinforcement Learning Expands Reasoning Boundaries in Large Language Models — mentioned by t55

ProRL: Prolonged Reinforcement Learning Expands Reasoning Boundaries in Large Language Models — mentioned by ninakostoska