Understanding Evolution Strategies for LLM Reasoning: Broader Reasoning Coverage than GRPO

• URL: http://arxiv.org/abs/2608.27351v2
• Date: Fri, 28 Aug 2026 12:17:58 GMT
• Title: Understanding Evolution Strategies for LLM Reasoning: Broader Reasoning Coverage than GRPO
• Abstract summary: Evolution Strategies (ES) have emerged as a memory-efficient post-training paradigm for reasoning.<n>This paper first identifies a performance advantage of ES over Group Relative Policy Optimization ( GRPO)<n>We develop a sequential GRPO-ES training strategy that combines GRPO's strength in Pass@1 with ES's gains in Pass@K.
• Score: 29.055105879127904
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: Evolution Strategies (ES) have recently emerged as a memory-efficient post-training paradigm for LLM reasoning. However, the optimization behavior of ES remains understudied, making it hard to define its advantage scope compared to mainstream post-training paradigms (e.g., Group Relative Policy Optimization (GRPO)). By systematically investigating ES dynamics and mechanisms, this paper first identifies a performance advantage of ES over GRPO, theoretically and empirically showing that ES can lead to broader reasoning coverage, thereby better exploiting the reasoning capabilities of pretrained LLMs. Theoretically, we show that verifier-projected Jensen-Shannon diversity across the ES population is helpful to higher Pass@K performances. Empirically, unlike GRPO, which exhibits entropy collapse, ES improves Pass@1 while attaining higher Pass@K than GRPO. We further develop a sequential GRPO-ES training strategy that combines GRPO's strength in Pass@1 with ES's gains in Pass@K. Second, we find that despite substantial whole-model parameter drift, the task-performance gains of ES are only contributed to a sparse subset of larger-magnitude updates. This functional sparsity suggests that large parameter movement need not imply widespread functional change, and held-out evaluations further show that it does not necessarily lead to catastrophic forgetting. Finally, we study how hyperparameter design affects the effectiveness of ES, demonstrating that ES requires a smaller population size in a larger LLM. These findings position ES as a distinct reasoning post-training paradigm rather than a less effective, memory-efficient alternative to GRPO.

Related papers

• AdaGRPO: A Capability-Aware Adaptive Enhancement for Flow-based GRPO [78.36537400975298]
  Group Relative Policy Optimization (GRPO) has demonstrated remarkable success in aligning text-to-image (T2I) flow models with human preferences.<n>We have identified that the learning loop of current flow-based GRPO is fundamentally decoupled from the learner's current capability.<n>We propose Adaptive GRPO (AdaGRPO), a novel capability-aware RL algorithm tailored for flow models.
  arXiv  Detail & Related papers  (2026-06-05T02:07:08Z)
• GAGPO: Generalized Advantage Grouped Policy Optimization [13.085463559931156]
  We propose a critic-free reinforcement learning method for precise, step-aligned temporal credit assignment.<n>Experiments on ALFWorld and WebShop show that GAGPO outperforms strong reinforcement learning baselines.
  arXiv  Detail & Related papers  (2026-05-13T09:10:03Z)
• AGPO: Asymmetric Group Policy Optimization for Verifiable Reasoning and Search Ads Relevance at JD [12.270495684580546]
  We propose Asymmetric Group Policy Optimization (AGPO) to counteract this boundary shrinkage.<n>AGPO adopts a negative-dominant reinforcement strategy to suppress incorrect reasoning paths, maintaining the base model's exploration capacity.<n>Our experiments on five mathematical benchmarks demonstrate that AGPO achieves state-of-the-art accuracy while consistently improving pass@$k$ performance at scale.
  arXiv  Detail & Related papers  (2026-05-07T08:00:48Z)
• iGRPO: Self-Feedback-Driven LLM Reasoning [88.83313431248473]
  Large Language Models (LLMs) have shown promise in solving complex mathematical problems, yet they still fall short of producing accurate and consistent solutions.<n>We introduce Iterative Group Relative Policy Optimization (iGRPO), a two-stage extension of GRPO that adds dynamic self-conditioning through model-generated drafts.<n>Under matched rollout budgets, iGRPO consistently outperforms GRPO across base models.
  arXiv  Detail & Related papers  (2026-02-09T18:45:11Z)
• MAESTRO: Meta-learning Adaptive Estimation of Scalarization Trade-offs for Reward Optimization [56.074760766965085]
  Group-Relative Policy Optimization has emerged as an efficient paradigm for aligning Large Language Models (LLMs)<n>We propose MAESTRO, which treats reward scalarization as a dynamic latent policy, leveraging the model's terminal hidden states as a semantic bottleneck.<n>We formulate this as a contextual bandit problem within a bi-level optimization framework, where a lightweight Conductor network co-evolves with the policy by utilizing group-relative advantages as a meta-reward signal.
  arXiv  Detail & Related papers  (2026-01-12T05:02:48Z)
• Expand and Prune: Maximizing Trajectory Diversity for Effective GRPO in Generative Models [37.48289959306949]
  Group Relative Policy Optimization is a powerful technique for aligning generative models.<n>But its effectiveness is bottlenecked by the conflict between large group sizes and prohibitive computational costs.<n>We propose Pro-GRPO, a novel dynamic framework that integrates latent feature-based trajectory pruning into the sampling process.
  arXiv  Detail & Related papers  (2025-12-17T11:44:34Z)
• GRPO-MA: Multi-Answer Generation in GRPO for Stable and Efficient Chain-of-Thought Training [13.218620894749067]
  GRPO, a Reinforcement Learning (RL) approach, can effectively train Chain-of-Thought (CoT) reasoning in Large Language Models (LLMs) and Vision-Language Models (VLMs)<n>In this paper, we analyze three challenges of GRPO: gradient coupling between thoughts and answers, sparse reward signals caused by limited parallel sampling, and unstable advantage estimation.<n>We propose GRPO-MA, a simple yet theoretically grounded method that leverages multi-answer generation from each thought process, enabling more robust and efficient optimization.
  arXiv  Detail & Related papers  (2025-09-29T09:07:45Z)
• On the Effect of Negative Gradient in Group Relative Deep Reinforcement Optimization [52.76330545825083]
  Reinforcement learning (RL) has become popular in enhancing the reasoning capabilities of large language models (LLMs)<n>We identify a previously unrecognized phenomenon we term Lazy Likelihood Displacement (LLD), wherein the likelihood of correct responses marginally increases or even decreases during training.<n>We develop a method called NTHR, which downweights penalties on tokens contributing to the LLD. Unlike prior DPO-based approaches, NTHR takes advantage of GRPO's group-based structure, using correct responses as anchors to identify influential tokens.
  arXiv  Detail & Related papers  (2025-05-24T18:58:51Z)
• DisCO: Reinforcing Large Reasoning Models with Discriminative Constrained Optimization [50.91849555841057]
  Group Relative Policy Optimization is a reinforcement learning method for large reasoning models (LRMs)<n>We introduce a new Discriminative Constrained Optimization framework for reinforcing LRMs, grounded in the principle of discriminative learning.<n>DisCO significantly outperforms GRPO and its improved variants such as DAPO, achieving average gains of 7% over GRPO and 6% over DAPO.
  arXiv  Detail & Related papers  (2025-05-18T11:08:32Z)
• Stepwise Guided Policy Optimization: Coloring your Incorrect Reasoning in GRPO [22.00487909203855]
  Group Relative Policy Optimization fails to update a policy when all responses within a group are incorrect.<n>This limitation underscores a key gap between artificial and human intelligence.<n>We introduce a simple framework that mitigates the all-negative-sample issue by incorporating response diversity within groups.
  arXiv  Detail & Related papers  (2025-05-16T18:02:05Z)
• A Minimalist Approach to LLM Reasoning: from Rejection Sampling to Reinforce [68.99924691391048]
  We revisit GRPO from a reinforce-like algorithm perspective and analyze its core components.<n>We find that a simple rejection sampling baseline, RAFT, yields competitive performance than GRPO and PPO.<n>Motivated by this insight, we propose Reinforce-Rej, a minimal extension of policy gradient that filters both entirely incorrect and entirely correct samples.
  arXiv  Detail & Related papers  (2025-04-15T16:15:02Z)

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.