The reviewed record of science sign in
Pith

arxiv: 2503.01491 · v1 · pith:N4CGYDDU · submitted 2025-03-03 · cs.LG

What's Behind PPO's Collapse in Long-CoT? Value Optimization Holds the Secret

Reviewed by Pithpith:N4CGYDDUopen to challenge →

classification cs.LG
keywords vc-ppolongtasksvaluebiasdecayinitializationoptimization
0
0 comments X
read the original abstract

Reinforcement learning (RL) is pivotal for enabling large language models (LLMs) to generate long chains of thought (CoT) for complex tasks like math and reasoning. However, Proximal Policy Optimization (PPO), effective in many RL scenarios, fails in long CoT tasks. This paper identifies that value initialization bias and reward signal decay are the root causes of PPO's failure. We propose Value-Calibrated PPO (VC-PPO) to address these issues. In VC-PPO, the value model is pretrained to tackle initialization bias, and the Generalized Advantage Estimation (GAE) computation is decoupled between the actor and critic to mitigate reward signal decay. Experiments on the American Invitational Mathematics Examination (AIME) show that VC-PPO significantly boosts PPO performance. Ablation studies show that techniques in VC-PPO are essential in enhancing PPO for long CoT tasks.

This paper has not been read by Pith yet.

discussion (0)

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

Forward citations

Cited by 27 Pith papers

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

  1. Lightning OPD: Efficient Post-Training for Large Reasoning Models with Offline On-Policy Distillation

    cs.LG 2026-04 unverdicted novelty 8.0

    Lightning OPD enforces teacher consistency by precomputing log-probabilities over SFT rollouts, matching standard OPD performance with bounded gradient discrepancy and achieving 4x speedup on math and code reasoning tasks.

  2. Bringing Value Models Back: Generative Critics for Value Modeling in LLM Reinforcement Learning

    cs.LG 2026-04 unverdicted novelty 7.0

    GenAC introduces generative critics with chain-of-thought reasoning and in-context conditioning to improve value approximation and downstream RL performance in LLMs compared to value-based and value-free baselines.

  3. Small Generalizable Prompt Predictive Models Can Steer Efficient RL Post-Training of Large Reasoning Models

    cs.AI 2026-02 unverdicted novelty 7.0

    GPS trains a small model on optimization history to predict prompt difficulty and select intermediate-difficulty diverse batches, yielding better training efficiency, final performance, and test-time allocation than b...

  4. MURPHY: Feedback-Aware GRPO with Retrospective Credit Assignment for Multi-Turn Code Generation

    cs.LG 2025-11 unverdicted novelty 7.0

    MURPHY improves code generation pass rates by up to 6% through retrospective credit assignment on multi-turn feedback trees using max or mean reward propagation.

  5. The Art of Scaling Reinforcement Learning Compute for LLMs

    cs.LG 2025-10 unverdicted novelty 7.0

    A 400k+ GPU-hour study shows RL scaling in LLMs follows predictable sigmoidal trajectories, with most design choices affecting efficiency rather than the performance asymptote, enabling accurate large-scale prediction...

  6. Blending Supervised and Reinforcement Fine-Tuning with Prefix Sampling

    cs.LG 2025-07 unverdicted novelty 7.0

    Prefix-RFT blends SFT and RFT via prefix sampling from demonstrations to outperform standalone SFT, RFT, and mixed-policy baselines on math reasoning problems.

  7. Reinforcement Learning for Reasoning in Large Language Models with One Training Example

    cs.LG 2025-04 accept novelty 7.0

    One training example via RLVR boosts LLM math reasoning from 17.6% to 35.7% average across six benchmarks.

  8. Small RL Controller, Large Language Model: RL-Guided Adaptive Sampling for Test-Time Scaling

    cs.CL 2026-06 unverdicted novelty 6.0

    RL-trained lightweight controller using answer statistics improves trade-offs among correctness, latency, and total samples in adaptive sampling for LLM test-time scaling.

  9. Extreme Region Policy Distillation

    cs.LG 2026-05 unverdicted novelty 6.0

    ERPD decouples aggressive off-policy optimization on fixed trajectories from trust-region distillation to achieve comparable or better LLM performance with substantially smaller KL divergence.

  10. 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.

  11. 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.

  12. Forge: Quality-Aware Reinforcement Learning for NP-Hard Optimization in LLMs

    cs.AI 2026-05 unverdicted novelty 6.0

    OPT-BENCH trains LLMs on NP-hard optimization via quality-aware RLVR, achieving 93.1% success rate and 46.6% quality ratio on Qwen2.5-7B while outperforming GPT-4o and transferring gains to other domains.

  13. Experience Sharing in Mutual Reinforcement Learning for Heterogeneous Language Models

    cs.LG 2026-05 unverdicted novelty 6.0

    Mutual Reinforcement Learning allows heterogeneous LLMs to exchange experience through mechanisms like Peer Rollout Pooling, Cross-Policy GRPO Advantage Sharing, and Success-Gated Transfer, with outcome-level sharing ...

  14. Segment-Aligned Policy Optimization for Multi-Modal Reasoning

    cs.AI 2026-05 unverdicted novelty 6.0

    SAPO introduces segment-level policy optimization using a step-wise MDP abstraction to better align RL updates with reasoning structure in multi-modal LLM tasks.

  15. Unleashing Implicit Rewards: Prefix-Value Learning for Distribution-Level Optimization

    cs.CL 2026-04 unverdicted novelty 6.0

    IPVRM learns prefix values to produce reliable step rewards from sequence outcomes using TD learning, enabling distribution-level RL that improves reasoning when paired with calibrated rewards.

  16. Lightning OPD: Efficient Post-Training for Large Reasoning Models with Offline On-Policy Distillation

    cs.LG 2026-04 unverdicted novelty 6.0

    Lightning OPD is an offline on-policy distillation method that matches standard OPD performance at 4x efficiency by enforcing teacher consistency between SFT and distillation phases.

  17. ReSeek: A Self-Correcting Framework for Search Agents with Instructive Rewards

    cs.CL 2025-10 unverdicted novelty 6.0

    ReSeek adds self-correction via a JUDGE action and a dense instructive reward (correctness plus utility) to RL training of search agents, yielding higher success and faithfulness on a new contamination-resistant benchmark.

  18. VAPO: Efficient and Reliable Reinforcement Learning for Advanced Reasoning Tasks

    cs.AI 2025-04 unverdicted novelty 6.0

    VAPO achieves 60.4 on AIME 2024 with Qwen 32B, outperforming prior methods by over 10 points through targeted fixes for value bias, sequence length variation, and sparse rewards.

  19. DAPO: An Open-Source LLM Reinforcement Learning System at Scale

    cs.LG 2025-03 conditional novelty 6.0

    DAPO introduces decoupled clipping and dynamic sampling for LLM RL, achieving 50 on AIME 2024 with Qwen2.5-32B while fully open-sourcing code, data, and the verl-based training system.

  20. CompactionRL: Reinforcement Learning with Context Compaction for Long-Horizon Agents

    cs.LG 2026-07 conditional novelty 5.0

    CompactionRL trains LLM agents to generate context summaries during RL rollouts, enabling long-horizon task completion under fixed context budgets with consistent gains on SWE-bench Verified and Terminal-Bench 2.0.

  21. Modularized Reinforcement Learning on LLMs: From MDP Creation to Exploration and Learning

    cs.LG 2026-06 unverdicted novelty 5.0

    Survey mapping RL techniques onto LLM training and highlighting gaps in value-based, off-policy, and bootstrapping methods.

  22. VIMPO: Value-Implicit Policy Optimization for LLMs

    cs.LG 2026-06 unverdicted novelty 5.0

    VIMPO derives a policy-implied value function from optimality conditions for critic-free RL in LLMs and shows gains over GRPO on math benchmarks.

  23. RLVR without Ineffective Samples: Group Prioritized Off-Policy Optimization for LLM Reasoning

    cs.LG 2026-05 unverdicted novelty 5.0

    POPO uses recency-based prioritized group replay and decoupled off-policy optimization to avoid zero-variance ineffective samples in RLVR, accelerating LLM reasoning finetuning with fewer rollouts.

  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. UI-TARS-2 Technical Report: Advancing GUI Agent with Multi-Turn Reinforcement Learning

    cs.AI 2025-09 conditional novelty 5.0

    UI-TARS-2 reaches 88.2 on Online-Mind2Web, 47.5 on OSWorld, 50.6 on WindowsAgentArena, and 73.3 on AndroidWorld while attaining 59.8 mean normalized score on a 15-game suite through multi-turn RL and scalable data generation.

  26. Seed1.5-VL Technical Report

    cs.CV 2025-05 unverdicted novelty 4.0

    Seed1.5-VL is a compact multimodal model that sets new records on dozens of vision-language benchmarks and outperforms prior systems on agent-style tasks.

  27. Reinforcement Learning from Human Feedback

    cs.LG 2025-04 unverdicted novelty 2.0

    The book introduces the origins, mathematical setup, and optimization stages of RLHF including reward modeling, reinforcement learning, rejection sampling, and direct alignment algorithms.