Expected Return Causes Outcome-Level Mode Collapse in Reinforcement Learning and How to Fix It with Inverse Probability Scaling

• URL: http://arxiv.org/abs/2601.21669v1
• Date: Thu, 29 Jan 2026 13:03:33 GMT
• Title: Expected Return Causes Outcome-Level Mode Collapse in Reinforcement Learning and How to Fix It with Inverse Probability Scaling
• Authors: Abhijeet Sinha, Sundari Elango, Dianbo Liu,
• Abstract summary: We show that outcome-level mode collapse is a structural consequence of the expected-return itself.<n>We propose a minimal correction: inverse probability scaling, which removes outcome-frequency from the learning signal.
• Score: 2.8825582215429186
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Many reinforcement learning (RL) problems admit multiple terminal solutions of comparable quality, where the goal is not to identify a single optimum but to represent a diverse set of high-quality outcomes. Nevertheless, policies trained by standard expected return maximization routinely collapse onto a small subset of outcomes, a phenomenon commonly attributed to insufficient exploration or weak regularization. We show that this explanation is incomplete: outcome level mode collapse is a structural consequence of the expected-return objective itself. Under idealized learning dynamics, the log-probability ratio between any two outcomes evolves linearly in their reward difference, implying exponential ratio divergence and inevitable collapse independent of the exploration strategy, entropy regularization, or optimization algorithm. We identify the source of this pathology as the probability multiplier inside the expectation and propose a minimal correction: inverse probability scaling, which removes outcome-frequency amplification from the learning signal, fundamentally changes the learning dynamics, and provably yields reward-proportional terminal distributions, preventing collapse in multimodal settings. We instantiate this principle in Group Relative Policy Optimization (GRPO) as a drop-in modification, IPS-GRPO, requiring no auxiliary models or architectural changes. Across different reasoning and molecular generation tasks, IPS-GRPO consistently reduces outcome-level mode collapse while matching or exceeding baseline performance, suggesting that correcting the objective rather than adding exploration heuristics is key to reliable multimodal policy optimization.

Related papers

• Online Learning for Multi-Layer Hierarchical Inference under Partial and Policy-Dependent Feedback [22.44021085629083]
  We study online routing for hierarchical inference under long-term resource constraints and terminal-only feedback.<n>We develop a variance-optimal EXP4-based algorithm integrated with Lyapunov optimization, yielding unbiased loss estimation and stable learning under sparse and policy-dependent feedback.
  arXiv  Detail & Related papers  (2026-03-04T16:35:33Z)
• Difficulty-Estimated Policy Optimization [38.86673795561421]
  We propose Difficulty-Estimated Policy Optimization (DEPO), a novel framework designed to optimize the efficiency and robustness of reasoning alignment.<n>Our approach significantly lowers the computational barrier for training high-performance reasoning models, offering a more sustainable path for reasoning scaling.
  arXiv  Detail & Related papers  (2026-02-06T04:12:23Z)
• Back to Basics: Revisiting Exploration in Reinforcement Learning for LLM Reasoning via Generative Probabilities [10.235183326885794]
  Reinforcement Learning with Verifiable Rewards (RLVR) has emerged as an indispensable paradigm for enhancing reasoning in Large Language Models (LLMs)<n>We analyze this issue from the perspective of sampling probability dynamics, identifying that the standard objective disproportionately reinforces the highest-likelihood paths.<n>We propose a novel Advantage Re-weighting Mechanism (ARM) designed to equilibrate the confidence levels across all correct responses.
  arXiv  Detail & Related papers  (2026-02-05T04:06:55Z)
• Rethinking the Trust Region in LLM Reinforcement Learning [72.25890308541334]
  Proximal Policy Optimization (PPO) serves as the de facto standard algorithm for Large Language Models (LLMs)<n>We propose Divergence Proximal Policy Optimization (DPPO), which substitutes clipping with a more principled constraint.<n>DPPO achieves superior training and efficiency compared to existing methods, offering a more robust foundation for RL-based fine-tuning.
  arXiv  Detail & Related papers  (2026-02-04T18:59:04Z)
• Statistical analysis of Inverse Entropy-regularized Reinforcement Learning [15.054399128586232]
  Inverse reinforcement learning aims to infer the reward function that explains expert behavior observed through trajectories of state--action pairs.<n>Many reward functions can induce the same optimal policy, rendering the inverse problem ill-posed.<n>We develop a statistical framework for Inverse Entropy-regularized Reinforcement Learning.
  arXiv  Detail & Related papers  (2025-12-07T18:26:19Z)
• DaGRPO: Rectifying Gradient Conflict in Reasoning via Distinctiveness-Aware Group Relative Policy Optimization [20.66452395111739]
  We propose Distinctiveness-aware Group Relative Policy Optimization (DaGRPO)<n>DaGRPO incorporates two core mechanisms: (1) Sequence-level Gradient Rectification, which utilizes fine-grained scoring to dynamically mask sample pairs with low distinctiveness; and (2) Off-policy Data Augmentation, which introduces high-quality anchors to recover training signals for challenging tasks.<n>In-depth analysis confirms that DaGRPO effectively mitigates gradient explosion and accelerates the emergence of long-chain reasoning capabilities.
  arXiv  Detail & Related papers  (2025-12-06T07:51:36Z)
• BAPO: Stabilizing Off-Policy Reinforcement Learning for LLMs via Balanced Policy Optimization with Adaptive Clipping [69.74252624161652]
  We propose BAlanced Policy Optimization with Adaptive Clipping (BAPO)<n>BAPO dynamically adjusts clipping bounds to adaptively re-balance positive and negative contributions, preserve entropy, and stabilize RL optimization.<n>On AIME 2024 and AIME 2025 benchmarks, our 7B BAPO model surpasses open-source counterparts such as SkyWork-OR1-7B.
  arXiv  Detail & Related papers  (2025-10-21T12:55:04Z)
• Rediscovering Entropy Regularization: Adaptive Coefficient Unlocks Its Potential for LLM Reinforcement Learning [55.59724323303857]
  We propose a framework that balances exploration and exploitation via three components: difficulty-aware coefficient allocation, initial-anchored target entropy, and dynamic global coefficient adjustment.<n>Experiments on multiple mathematical reasoning benchmarks show that AER consistently outperforms baselines, improving both reasoning accuracy and exploration capability.
  arXiv  Detail & Related papers  (2025-10-13T03:10:26Z)
• Stratified GRPO: Handling Structural Heterogeneity in Reinforcement Learning of LLM Search Agents [90.45197506653341]
  Large language model (LLM) agents increasingly rely on external tools such as search engines to solve complex, multi-step problems.<n>The trajectories of search agents are structurally heterogeneous, where variations in the number, placement, and outcomes of search calls lead to fundamentally different answer directions and reward distributions.<n>Standard policy gradient methods, which use a single global baseline, suffer from what we identify and formalize as cross-stratum bias.<n>We propose Stratified GRPO, whose central component, Stratified Advantage Normalization (SAN), partitions trajectories into homogeneous strata based on their structural properties and computes advantages locally within each
  arXiv  Detail & Related papers  (2025-10-07T17:59:13Z)
• Generalization Bounds of Surrogate Policies for Combinatorial Optimization Problems [53.03951222945921]
  We analyze smoothed (perturbed) policies, adding controlled random perturbations to the direction used by the linear oracle.<n>Our main contribution is a generalization bound that decomposes the excess risk into perturbation bias, statistical estimation error, and optimization error.<n>We illustrate the scope of the results on applications such as vehicle scheduling, highlighting how smoothing enables both tractable training and controlled generalization.
  arXiv  Detail & Related papers  (2024-07-24T12:00:30Z)
• Provably Mitigating Overoptimization in RLHF: Your SFT Loss is Implicitly an Adversarial Regularizer [52.09480867526656]
  We identify the source of misalignment as a form of distributional shift and uncertainty in learning human preferences.<n>To mitigate overoptimization, we first propose a theoretical algorithm that chooses the best policy for an adversarially chosen reward model.<n>Using the equivalence between reward models and the corresponding optimal policy, the algorithm features a simple objective that combines a preference optimization loss and a supervised learning loss.
  arXiv  Detail & Related papers  (2024-05-26T05:38:50Z)
• Mitigating Covariate Shift in Misspecified Regression with Applications to Reinforcement Learning [39.02112341007981]
  We study the effect of distribution shift in the presence of model misspecification. We show that empirical risk minimization, or standard least squares regression, can result in undesirable misspecification amplification. We develop a new algorithm that avoids this undesirable behavior, resulting in no misspecification amplification while still obtaining optimal statistical rates.
  arXiv  Detail & Related papers  (2024-01-22T18:59:12Z)
• Beyond Reverse KL: Generalizing Direct Preference Optimization with Diverse Divergence Constraints [26.274786600234876]
  The increasing capabilities of large language models (LLMs) raise opportunities for artificial general intelligence but amplify safety concerns. RLHF has emerged as a promising pathway towards AI alignment but brings forth challenges due to its complexity and dependence on a separate reward model. DPO has been proposed as an alternative, and it remains equivalent to RLHF under the reverse KL regularization constraint. We show that under certain $f$-divergences, including Jensen-Shannon divergence, forward KL divergences and $alpha$-divergences, the complex relationship between the reward and optimal policy can also be simplified
  arXiv  Detail & Related papers  (2023-09-28T08:29:44Z)
• DROMO: Distributionally Robust Offline Model-based Policy Optimization [0.0]
  We consider the problem of offline reinforcement learning with model-based control. We propose distributionally robust offline model-based policy optimization (DROMO)
  arXiv  Detail & Related papers  (2021-09-15T13:25:14Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.