arXiv Machine Learning
Jul 14
Predictive Divergence Masks for LLM RL
arXiv:2607. 10848v1 Announce Type: new Abstract: Reinforcement learning for large language models (LLMs) typically relies on trust-region masks to stabilize off-policy updates.
Predictive Divergence Masks for LLM RL
Reinforcement learning for large language models (LLMs) typically relies on trust-region masks to stabilize off-policy updates. The dominant PPO-style approach uses the sampled-token importance ratio for two criteria: a proximity criterion, which asks whether the policy has moved too far from the behavior policy, and a direction criterion, which asks whether the update pushes it farther away.
Related stories
arXiv Machine Learning
Jun 9
Rethinking the Divergence Regularization in LLM RL
arXiv:2606. 09821v1 Announce Type: new Abstract: Reinforcement learning (RL) has become a key component of post-training large language models (LLMs).
arXiv AI
Jun 15
Rethinking the Trust Region in LLM Reinforcement Learning
arXiv:2602. 04879v3 Announce Type: replace-cross Abstract: Reinforcement learning (RL) has become a cornerstone for fine-tuning Large Language Models (LLMs), with Proximal Policy Optimization (PPO) serving as the de facto standard algorithm.
arXiv AI
Jul 7
Turning Off-Policy Tokens On-Policy: A Plug-in Approach for Improving LLM Alignment
arXiv:2607. 04728v1 Announce Type: cross Abstract: Reinforcement learning (RL) post-training for large language models (LLMs) follows a efficient paradigm of "rollout then update", which inevitably results in off-policy training data.
Hugging Face Trending Papers
Jul 6
Turning Off-Policy Tokens On-Policy: A Plug-in Approach for Improving LLM Alignment
Reinforcement learning (RL) post-training for large language models (LLMs) follows a efficient paradigm of "rollout then update", which inevitably results in off-policy training data. To resolve this, Importance sampling (IS) is proposed, while the token-level ratios compound over long sequences, causing severe variance exploded.
arXiv Computation and Language
Sep 22
What are Key Factors for Updates in RL for LLM Reasoning?
arXiv:2606.22570v2 Announce Type: replace Abstract: Reinforcement Learning from Verifiable Rewards (RLVR) has emerged as a promising framework for enhancing the reasoning ability of large language mo...
arXiv AI
Jul 1
Which Tokens Matter? Adaptive Token Selection for RLVR with the Relative Surprisal Index
arXiv:2606. 31575v1 Announce Type: new Abstract: Reinforcement learning (RL) has become a powerful tool for propelling Large Language Models (LLMs) beyond imitation-based training towards more robust reasoning capabilities.
arXiv AI
Jun 29
Trust Region Masking for Long-Horizon LLM Reinforcement Learning
arXiv:2512. 23075v5 Announce Type: replace-cross Abstract: Policy gradient methods for Large Language Models optimize a policy $\pi_\theta$ via a surrogate objective computed from samples of a rollout policy $\pi_{\text{roll}}$.
arXiv AI
Jul 7
Gradient Regularization Mitigates Reward Hacking in Reinforcement Learning from Human Feedback and Verifiable Rewards
arXiv:2602. 18037v2 Announce Type: replace-cross Abstract: Reinforcement Learning from Human Feedback (RLHF) or Verifiable Rewards (RLVR) are two key steps in the post-training of modern Language Models (LMs).
arXiv Machine Learning
4d ago
Understanding LLM Parameter Update Sparsity through the Lens of Fisher
arXiv:2609.36262v1 Announce Type: new Abstract: Recent studies have observed that parameter changes during language-model post-training can be concentrated in a small subset of coordinates. This phen...
arXiv Machine Learning
1d ago
Why GRPO Needs Normalization: A Local-Curvature Perspective on Adaptive Gradients
The paper investigates why Group Relative Policy Optimization (GRPO) benefits from per‑prompt normalization by examining the local curvature of the sequence‑level policy gradient. It shows that standard deviation normalization acts as an adaptive gradient, yielding a provably faster convergence rate than unnormalized REINFORCE under mild conditions, with the improvement tied to the average within‑prompt reward standard deviation. The authors also propose IS‑GRPO, an importance‑sampling variant that maintains alignment with the full gradient and offers a tighter convergence guarantee, and empirically validate these theoretical insights on GSM8K and MATH datasets at 1.5B and 7B model scales.
arXiv Machine Learning
Sep 23
PACT: From Credit Assignment to Critic Alignment
arXiv:2609.26355v1 Announce Type: new Abstract: Reinforcement learning has become a central component of large language model (LLM) post-training, yet token-level credit lacks a generally accepted ma...