DeepSeek-Prover-V1.5: Harnessing Proof Assistant Feedback for Reinforcement Learning and Monte-Carlo Tree Search · HackerPapers discussions

DeepSeek-Prover-V1.5: Harnessing Proof Assistant Feedback for Reinforcement Learning and Monte-Carlo Tree Search — mentioned by dvrp

Wed, 30 Apr 2025 09:27:28 GMT

seems like it's not a new 671 B behemoth, just the V3 MoE engine with long-context RoPE + FA2, retargeted at theorem proving. see [1,2]: [1]: https://arxiv.org/abs/2405.14333 [2]: https://arxiv.org/abs/2408.08152

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

DeepSeek-Prover-V1.5: Harnessing Proof Assistant Feedback for Reinforcement Learning and Monte-Carlo Tree Search — mentioned by maxrmk

Mon, 14 Oct 2024 21:59:45 GMT

There's a newer version of this model that takes a really cool RL based approach: https://arxiv.org/pdf/2408.08152

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

DeepSeek-Prover-V1.5: Harnessing Proof Assistant Feedback for Reinforcement Learning and Monte-Carlo Tree Search — mentioned by jumploops

Fri, 11 Oct 2024 20:10:00 GMT

> There is just no way you can build reliable agents on this foundation, where changing a word or two in irrelevant ways or adding a few bits of irrelevant info can give you a different answer. LLMs are not magic bullets for every problem, but that doesn't preclude them from being used to build reliable systems or "agents." It's clear that we don't yet have the all-encompassing AGI architecture, especially with the transformer model alone, but adding steps beyond the transformer leads to interesting results, as we've seen with current coding tools and the new o1-series models by OpenAI. For…

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

DeepSeek-Prover-V1.5: Harnessing Proof Assistant Feedback for Reinforcement Learning and Monte-Carlo Tree Search — HN submission

Sat, 17 Aug 2024 03:34:57 GMT

DeepSeek-Prover-v1.5: Proof Assistant for RL and Monte-Carlo Tree Search

1 point; 0 discussion replies on HN. Paper and source history. A submission is not an endorsement.