SRPO: Kuaishou's New RL Method 10x More Efficient Than GRPO for LLM Training

Researchers from Kuaishou's Kwaipilot team have introduced SRPO, a two-stage reinforcement learning framework that achieves DeepSeek-R1-Zero-level performance in math and code domains with only one-tenth of the training steps. SRPO addresses cross-domain optimization conflicts, low sample efficiency, and premature saturation through staged training and history resampling.

DeepSeek releases DeepSeek-Prover-V2 with recursive proof search and new ProverBench benchmark

DeepSeek AI has released DeepSeek-Prover-V2, an open-source large language model for formal theorem proving in Lean 4. It uses a recursive pipeline to generate training data via DeepSeek-V3, achieves state-of-the-art results on MiniF2F and PutnamBench, and introduces ProverBench, a new benchmark with 325 problems.

DeepSeek