Chinese content论文提出 Effective Entropy：用有效性感知熵修正 RLVR 探索多样性
做 RLVR 训练的人该看看这篇。它指出 token 熵高其实是模型在乱说话，不是在探索，然后给了 VEPO 这个修法，在 MATH500 等六个数学基准上比 GRPO 准确率更高，还用统计检验验证了策略确实更多样。
#DeepSeek-R1-Distill-Qwen
2 stories · 1 in 7 days · 1 in 30 days
Stories, products and papers tagged "DeepSeek-R1-Distill-Qwen", newest first.
Nov 13
Sep 17
New method improves pruning of large reasoning models
This research provides a more effective method for reducing the cost of large reasoning models, which is important for making them more practical and accessible.