Papers

Filtered to policy optimization · clear filter

Browse by term

continual learning 83reinforcement learning 50large language models 12benchmarking 11benchmarks 11language models 11vision-language models 11robotics 7natural language processing 6world models 6generative models 5recursive self-improvement 5attention mechanisms 4diffusion Transformers 4multi-agent systems 4multimodal learning 4multimodal models 4on-policy distillation 4self-distillation 4self-supervised learning 4transformers 4video generation 4vision-language-action models 4agent-based systems 3agentic models 3agentic search 3autonomous systems 3coding agents 3diffusion models 3image generation 3

Matching papers

Learn What's Left, Not What's Mastered: Saturation Aware Advantage Reweighting for Multi-Reward Policy Optimization

142 upvotes · 17 AUG 2026 · Yixuan Wang, Yifei Chen, Haichao Zhang et al.

This paper improves reinforcement learning for post-training language model reasoners by addressing two problems in existing methods: identical advantages for distinct reward profiles and fixed relative weights for all objectives. A new method, SA-MRPO, dynamically reallocates optimization effort toward under-optimized objectives while maintaining performance on well-satisfied objectives.

Eliciting Weak-to-Strong Generalization with On-Policy Reverse Distillation

78 upvotes · 8 SEP 2026 · Youngrok Park, Sangmin Bae, Hojung Jung et al.

This paper introduces a method called On-Policy Reverse Distillation that helps stronger models learn from weaker supervisors by selectively amplifying the parts of the weaker model's guidance that are most useful to the stronger model. This can lead to faster and more efficient learning, especially in situations where it's expensive to retrain models from scratch.

CoRT: Counterfactual Replay for Token-Level Rubric-Guided Policy Optimization

73 upvotes · 28 JUL 2026 · Bo-Wen Zhang, Junwei He, Wen Wang et al.

This paper proposes a method to improve language model training by allocating credit to individual tokens within a response, allowing for more nuanced evaluation of model performance. Practitioners may care about this method as it can lead to better language model performance, especially in tasks that require specific formatting or semantic choices.