Skill Reuse as Compression in Agentic RL
cs.LG, cs.AI
Submitted: 2026-05-29
Updated: 2026-09-01
Comments: Accepted by EMNLP 2026 Main Conference
License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
Terminology
Sources
- Theoretical Analysis of Byte-Pair Encoding
- Boosting LLM Reasoning via Human-Inspired Reward Shaping
- Qwen2.5 Technical Report
- Realizable Abstractions: Near-Optimal Hierarchical Reinforcement Learning
- Proximal Policy Optimization Algorithms
- DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models
- RAGEN-2: Reasoning Collapse in Agentic RL
- RAGEN: Understanding Self-Evolution in LLM Agents via Multi-Turn Reinforcement Learning
- Experiential Reinforcement Learning
- LAPO: Internalizing Reasoning Efficiency via Length-Adaptive Policy Optimization
- Qwen3 Technical Report
- Shorten After You're Right: Lazy Length Penalties for Reasoning RL
- DYSTIL: Dynamic Strategy Induction with Large Language Models for Reinforcement Learning
- SkillVLA: Tackling Combinatorial Diversity in Dual-Arm Manipulation via Skill Reuse
- Why Reasoning Fails to Plan: A Planning-Centric Analysis of Long-Horizon Decision Making in LLM Agents
Related papers
- Polynomial-Augmented Neural Networks (PANNs) with Weak Orthogonality Constraints for Enhanced Function and PDE Approximation
- AIRL-S: Unifying Reinforcement Learning and Search-Based Test-Time Scaling via Adversarial Inverse Reinforcement Learning
- Transformers as Bayesian In-Context Experimenters: Smoothness-Adaptive Efficient ATE Estimation
- Convergence issues in Relational Concept Analysis based on AOC-posets
- Beliefs Beyond Posteriors: Local-Consistency Optimisation for Bayesian Neural Networks
- Understanding Diffusion Models via Ratio-Based Function Approximation with SignReLU Networks