English(EN)CEPO: RLVR Self-Distillation using Contrastive Evidence Policy Optimization
新的自蒸馏方法提高了大型语言模型在推理任务上的性能
作者PulseAugur 编辑部·[5 个来源]·
研究人员开发了新的大型语言模型自蒸馏技术，可在不依赖外部反馈的情况下提高其性能。AVSD（自适应视图自蒸馏）在多个特权信息视图之间平衡共识信号，并使用视图特定的残差来增强学习。自策略蒸馏（SPD）从梯度中提取能力子空间，以提高性能和泛化能力，尤其是在代码生成和数学推理方面。CEPO（对比证据策略优化）通过对比正确答案和错误答案来锐化关键标记的信用分配，从而提高了多模态数学推理基准的准确性。
AI
arXiv:2605.20643v1 Announce Type: cross Abstract: Self-distillation enables language models to learn on-policy from their own trajectories by using the same model as both student and teacher, with the teacher being conditioned on privileged information unavailable to the student.…
arXiv:2605.22675v1 Announce Type: new Abstract: Self-distillation bootstraps large language models (LLMs) by training on their own generations. However, existing methods either rely on external signals to curate self-generated outputs (e.g., correctness filtering, execution feedb…
Self-distillation bootstraps large language models (LLMs) by training on their own generations. However, existing methods either rely on external signals to curate self-generated outputs (e.g., correctness filtering, execution feedback, and reward search), which are costly and un…
Self-distillation enables language models to learn on-policy from their own trajectories by using the same model as both student and teacher, with the teacher being conditioned on privileged information unavailable to the student. Such information can come in different types or v…
When a model produces a correct solution under reinforcement learning with verifiable rewards (RLVR), every token receives the same reward signal regardless of whether it was a decisive reasoning step or a grammatical filler. A natural fix is to condition the model on the correct…
