Not All Rollouts Are Worth Learning: On Trajectory Valuation for Post-Training Reinforcement Learning
cs.LG
Submitted: 2026-09-28
Updated: 2026-09-28
Terminology
Sources
- Prioritized Experience Replay
- Learning from the Right Rollouts: Data Attribution for PPO-based LLM Post-Training
- DataInf: Efficiently Estimating Data Influence in LoRA-tuned LLMs and Diffusion Models
- TRAK: Attributing Model Behavior at Scale
- Studying Large Language Model Generalization with Influence Functions
- Revisit, Extend, and Enhance Hessian-Free Influence Functions
- A Snapshot of Influence: A Local Data Attribution Framework for Online Reinforcement Learning
- LearnAlign: Data Selection for LLM Reinforcement Learning with Improved Gradient Alignment
- GradAlign: Gradient-Aligned Data Selection for LLM Reinforcement Learning
- Revisiting Prioritized Experience Replay: A Value Perspective
- Proximal Policy Optimization Algorithms
- DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models
- Training a Helpful and Harmless Assistant with Reinforcement Learning from Human Feedback
- WebGPT: Browser-assisted question-answering with human feedback
- Z0-Inf: Zeroth Order Approximation for Data Influence
- Data Shapley in One Training Run
- Layer-Aware Influence for Online Data Valuation Estimation
- Outlier Gradient Analysis: Efficiently Identifying Detrimental Training Samples for Deep Learning Models
- Training Verifiers to Solve Math Word Problems
- Qwen2 Technical Report
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: The Predictive Cost of ELBO Binding
- Understanding Diffusion Models via Ratio-Based Function Approximation with SignReLU Networks