Pith. sign in

REVIEW 29 cited by

SEED-GRPO: Semantic Entropy Enhanced GRPO for Uncertainty-Aware Policy Optimization

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

arxiv 2505.12346 v1 pith:ZZC4PPGQ submitted 2025-05-18 cs.AI

SEED-GRPO: Semantic Entropy Enhanced GRPO for Uncertainty-Aware Policy Optimization

classification cs.AI
keywords policyentropysemanticgrpoinputoptimizationpromptsseed-grpo
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved
0 comments
read the original abstract

Large language models (LLMs) exhibit varying levels of confidence across input prompts (questions): some lead to consistent, semantically similar answers, while others yield diverse or contradictory outputs. This variation reflects LLM's uncertainty about the input prompt, a signal of how confidently the model understands a given problem. However, vanilla Group Relative Policy Optimization (GRPO) treats all prompts equally during policy updates, ignoring this important information about the model's knowledge boundaries. To address this limitation, we propose SEED-GRPO (Semantic Entropy EnhanceD GRPO), which explicitly measures LLMs' uncertainty of the input prompts semantic entropy. Semantic entropy measures the diversity of meaning in multiple generated answers given a prompt and uses this to modulate the magnitude of policy updates. This uncertainty-aware training mechanism enables dynamic adjustment of policy update magnitudes based on question uncertainty. It allows more conservative updates on high-uncertainty questions while maintaining the original learning signal on confident ones. Experimental results on five mathematical reasoning benchmarks (AIME24 56.7, AMC 68.7, MATH 83.4, Minerva 34.2, and OlympiadBench 48.0) demonstrate that SEED-GRPO achieves new state-of-the-art performance in average accuracy, validating the effectiveness of uncertainty-aware policy optimization.

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Forward citations

Cited by 29 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score.

  1. Reinforcement Learning from Rich Feedback with Distributional DAgger

    cs.LG 2026-06 unverdicted novelty 7.0

    DistIL applies distributional DAgger with forward cross-entropy to achieve monotonic policy improvement and better Pass@N from rich feedback in RL for reasoning tasks.

  2. DUET: Optimize Token-Budget Allocation for Reinforcement Learning with Verifiable Rewards

    cs.LG 2026-05 unverdicted novelty 7.0

    DUET improves RLVR by allocating tokens across both prompt selection and rollout length, outperforming full-budget baselines even when using only half the tokens.

  3. Self-Induced Outcome Potential: Turn-Level Credit Assignment for Agents without Verifiers

    cs.LG 2026-05 unverdicted novelty 7.0

    SIOP enables turn-level credit assignment in LLM agents via semantic clustering of final answers as latent outcomes, improving performance on reasoning benchmarks without verifiers.

  4. Rethinking Entropy Minimization in Test-Time Adaptation for Autoregressive Models

    eess.AS 2026-05 unverdicted novelty 7.0

    Derives a rigorous entropy minimization formulation for autoregressive test-time adaptation that decomposes into policy gradient and entropy terms, reinterpreting prior methods and improving Whisper ASR across 20+ domains.

  5. Walking Through Uncertainty: An Empirical Study of Uncertainty Estimation for Audio-Aware Large Language Models

    eess.AS 2026-04 unverdicted novelty 7.0

    Semantic-level and verification-based uncertainty methods outperform token-level baselines for audio reasoning in ALLMs, but their relative performance on hallucination and unanswerable-question benchmarks is model- a...

  6. LASER: A Data-Centric Method for Low-Cost and Efficient SQL Rewriting based on SQL-GRPO

    cs.DB 2026-04 unverdicted novelty 7.0

    LASER generates complex slow-query training data with MCTS and aligns small models via SQL-GRPO to deliver efficient, low-cost SQL rewriting that outperforms rules and large models.

  7. Generate, Filter, Control, Replay: A Comprehensive Survey of Rollout Strategies for LLM Reinforcement Learning

    cs.LG 2026-04 unverdicted novelty 7.0

    This survey introduces the Generate-Filter-Control-Replay (GFCR) taxonomy to structure rollout pipelines for RL-based post-training of reasoning LLMs.

  8. Self-Distilled RLVR

    cs.LG 2026-04 unverdicted novelty 7.0

    RLSD mixes self-distillation for token-level policy difference magnitudes with RLVR for reliable update directions from response correctness to reach higher convergence and better training stability.

  9. Reinforcement Learning with Metacognitive Feedback Elicits Faithful Uncertainty Expression in LLMs

    cs.CL 2026-06 unverdicted novelty 6.0

    RLMF uses quality of model self-judgments to refine RL rankings and select training data, achieving SOTA faithful calibration while preserving accuracy and outperforming standard RL by up to 63%.

  10. Which Tokens Matter? Adaptive Token Selection for RLVR with the Relative Surprisal Index

    cs.AI 2026-06 unverdicted novelty 6.0

    Introduces RSI metric and RSI-S filtering method for adaptive token selection in RLVR, reporting 2-3 point gains over GRPO on AIME/AMC benchmarks.

  11. What are Key Factors for Updates in RL for LLM Reasoning?

    cs.CL 2026-06 unverdicted novelty 6.0

    Theoretical analysis of RLVR update dynamics leads to ACPO, an adaptive clipping method that outperforms DAPO and CISPO on reasoning benchmarks with 3B and 7B models.

  12. Architecture-Aware Reinforcement Learning Makes Sliding-Window Attention Competitive in Math Reasoning

    cs.AI 2026-06 unverdicted novelty 6.0

    Reinforcement learning after SFT conversion narrows the performance gap between sliding-window attention and full self-attention on math reasoning benchmarks while preserving linear complexity.

  13. Back on Track: Aligning Rewards and States for Reasoning in Diffusion Large Language Models

    cs.CL 2026-06 unverdicted novelty 6.0

    PAPO improves reasoning performance in diffusion LLMs by converting sparse terminal rewards into dense step-wise credit and replaying real high-uncertainty trajectories, reporting gains up to 42.2% on Countdown.

  14. FGRPO: Federated GRPO with Adaptive Aggregation on Non-IID Data

    cs.LG 2026-06 unverdicted novelty 6.0

    FGRPO decentralizes GRPO fine-tuning via adaptive aggregation based on relative performance gain to achieve robust convergence on non-IID data while preserving privacy.

  15. What and When to Distill: Selective Hindsight Distillation for Multi-Turn Agents

    cs.AI 2026-05 unverdicted novelty 6.0

    SERL selectively reweights learning using task success and environment feedback to reach 90.0% success on ALFWorld and 80.1% on WebShop, outperforming RL and distillation baselines.

  16. Holder Policy Optimisation

    cs.LG 2026-05 unverdicted novelty 6.0

    HölderPO unifies token aggregation in GRPO via the Hölder mean with dynamic p annealing, reporting 54.9% average math-benchmark accuracy and 93.8% ALFWorld success.

  17. Holder Policy Optimisation

    cs.LG 2026-05 unverdicted novelty 6.0

    HölderPO unifies token-level aggregation in GRPO via the Hölder mean with a tunable p parameter and annealing schedule, delivering 54.9% average accuracy on math benchmarks and 93.8% success on ALFWorld.

  18. T$^2$PO: Uncertainty-Guided Exploration Control for Stable Multi-Turn Agentic Reinforcement Learning

    cs.AI 2026-05 unverdicted novelty 6.0

    T²PO improves stability and performance in multi-turn agentic RL by using uncertainty dynamics at token and turn levels to guide exploration and avoid wasted rollouts.

  19. HEALing Entropy Collapse: Enhancing Exploration in Few-Shot RLVR via Hybrid-Domain Entropy Dynamics Alignment

    cs.LG 2026-04 unverdicted novelty 6.0

    HEAL mitigates entropy collapse in few-shot RLVR by selectively adding general-domain data and aligning trajectory-level entropy dynamics, matching full-shot performance with 32 target samples.

  20. Mimic Human Cognition, Master Multi-Image Reasoning: A Meta-Action Framework for Enhanced Visual Understanding

    cs.CV 2026-01 conditional novelty 6.0

    A structured five-step reasoning template plus diverse-trajectory cold start and diversity-preserving two-stage RL lifts a 7B multimodal model to state-of-the-art multi-image reasoning on several benchmarks.

  21. Unlocking Exploration in RLVR: Uncertainty-aware Advantage Shaping for Deeper Reasoning

    cs.AI 2025-10 unverdicted novelty 6.0

    UCAS refines RLVR advantage signals with a logit-space self-confidence proxy for response-level modulation and asymmetric token-level penalties based on raw logit certainty to boost exploration and reduce entropy collapse.

  22. GEOALIGN: Geometric Rollout Curation for Robust LLM Reinforcement Learning

    cs.LG 2026-06 unverdicted novelty 5.0

    Geoalign curates rollouts in online LLM RL by learning a projector on hidden states to flag and replace directionally inconsistent examples, yielding higher final performance and less oscillation than baselines on dia...

  23. Empowering Economic Simulation Through Situation-Aware Llm-Driven Generative System

    cs.MA 2026-06 unverdicted novelty 5.0

    SAMAS is an LLM-based generative agent system for economic ABM that jointly models macro patterns and micro behaviors to claim better volatility realism and turning point prediction.

  24. Trust Region On-Policy Distillation

    cs.LG 2026-05 unverdicted novelty 5.0

    TrOPD stabilizes on-policy distillation for LLMs with trust-region learning, outlier estimation, and off-policy guidance, outperforming prior OPD methods on reasoning and code benchmarks.

  25. Why Semantic Entropy Fails: Geometry-Aware and Calibrated Uncertainty for Policy Optimization

    cs.LG 2026-05 unverdicted novelty 5.0

    Identifies two gaps in entropy-based uncertainty for LLM post-training and proposes GCPO to align geometry-aware disagreement measures with reward-based calibration for better gradient regulation.

  26. EMO-R3: Reflective Reinforcement Learning for Emotional Reasoning in Multimodal Large Language Models

    cs.AI 2026-02 conditional novelty 5.0

    EMO-R3, which combines a three-step emotional reasoning prompt with a reward for the model agreeing with its own image–emotion judgments, raises visual emotion-recognition accuracy by about one point over plain GRPO.

  27. AQA-TTRL: Self-Adaptation in Audio Question Answering with Test-Time Reinforcement Learning

    eess.AS 2025-10 conditional novelty 5.0

    Test-time reinforcement learning against self-generated majority-vote pseudo-labels improves audio question answering accuracy on MMAU, MMAR, and MMSU.

  28. Self-Distilled Policy Gradient

    cs.LG 2026-06 unverdicted novelty 4.0

    SDPG combines group-relative verifier advantages, normalized standard deviation, full-vocabulary on-policy self-distillation, and reference-policy KL regularization to improve stability and performance over RLVR and s...

  29. A Survey of Reinforcement Learning for Large Reasoning Models

    cs.CL 2025-09 accept novelty 3.0

    A survey compiling RL methods, challenges, data resources, and applications for enhancing reasoning in large language models and large reasoning models since DeepSeek-R1.