REVIEW 11 cited by
Decomposing the Entropy-Performance Exchange: The Missing Keys to Unlocking Effective Reinforcement Learning
Not yet reviewed by Pith; the record is open.
This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.
SPECIMEN: schema-true, not a live event
T0 review · schema-true
One-sentence machine reading of the paper's core claim.
pith:XXXXXXXX · record.json · timestamp
read the original abstract
Recently, reinforcement learning with verifiable rewards (RLVR) has been widely used for enhancing the reasoning abilities of large language models (LLMs). A core challenge in RLVR involves managing the exchange between entropy and performance of policies. Despite the importance of this exchange, a fine-grained understanding of when and how this exchange operates most effectively remains limited. To bridge this gap, we conduct a systematic empirical analysis of the entropy-performance exchange mechanism of RLVR across different levels of granularity. Specifically, we first divide the training process into two distinct stages based on entropy dynamics, i.e., rising stage and plateau stage, and then systematically investigate how this mechanism varies across stage-level, instance-level, and token-level granularitiess. Our analysis reveals that, in the rising stage, entropy reduction in negative samples facilitates the learning of effective reasoning patterns, which in turn drives rapid performance gains. Moreover, in the plateau stage, learning efficiency strongly correlates with high-entropy tokens present in low-perplexity samples and those located at the end of sequences. Motivated by these findings, we propose two methods that dynamically adjust the reward signal using perplexity and positional information to focus RL updates on tokens that exhibit high learning potential, achieving improvements compared to the baseline methods on various LLMs.
Forward citations
Cited by 11 Pith papers
-
Low-rank Optimization Trajectories Modeling for LLM RLVR Acceleration
NExt accelerates RLVR training for LLMs by nonlinearly extrapolating low-rank parameter trajectories extracted from LoRA runs.
-
Beyond Fully Random Masking: Attention-Guided Denoising and Optimization for Diffusion Language Models
AGDO improves dLLM reasoning performance by determining denoising order and emphasizing tokens based on attention-derived dependencies rather than random masking.
-
Advantage Collapse in Group Relative Policy Optimization: Diagnosis and Mitigation
GRPO suffers advantage collapse on uniform-reward groups; ACR quantifies it and AVSPO adds virtual samples to restore gradients, yielding 4-6% accuracy gains on math benchmarks across 0.5B-14B models.
-
Nudging Beyond the Comfort Zone: Efficient Strategy-Guided Exploration for RLVR
NudgeRL conditions RLVR rollouts on strategy-level contexts to drive diverse trajectories and applies an inter/intra-context reward decomposition plus distillation objective, outperforming GRPO and oracle baselines on...
-
SAGE: Shaping Anchors for Guided Exploration in RLVR of LLMs
SAGE reshapes the reverse-KL anchor via guide function q(x,y) for controllable empirical support expansion, yielding gains in both pass@1 and pass@k on math reasoning benchmarks.
-
Understanding and Preventing Entropy Collapse in RLVR with On-Policy Entropy Flow Optimization
OPEFO prevents entropy collapse in RLVR by rescaling token updates according to their entropy change contributions, yielding more stable optimization and better results on math benchmarks.
-
Policy Improvement Reinforcement Learning
PIRL maximizes cumulative policy improvement across iterations instead of surrogate rewards and is proven aligned with final performance; PIPO implements it via retrospective verification for stable closed-loop optimization.
-
Beyond Entropy: Correctness-Aware Advantage Shaping via Contrastive Policy Optimization
CPO uses the log-ratio of reference-guided to vanilla token probabilities as a correctness signal for per-token advantage shaping in RLVR, beating entropy-based methods on math and generalization benchmarks.
-
Advantage Collapse in Group Relative Policy Optimization: Diagnosis and Mitigation
The paper shows that advantage collapse in GRPO causes training stagnation on math reasoning benchmarks and proposes AVSPO, which uses real-time monitoring to inject virtual reward samples and reduces collapse while i...
-
Policy Improvement Reinforcement Learning
PIPO adds closed-loop policy-improvement feedback to RL post-training so updates that raise measured performance are reinforced and those that drop it are suppressed.
-
From Trial-and-Error to Improvement: A Systematic Analysis of LLM Exploration Mechanisms in RLVR
A systematic analysis of LLM exploration in RLVR, introducing capability-boundary metrics and examining entropy-performance exchange across training stages and token levels.
Discussion (0). Sign in to comment.