arXiv AI
Jun 15
SkillAudit: Ground-Truth-Free Skill Evolution via Paired Trajectory Auditing
arXiv:2606. 14239v1 Announce Type: new Abstract: Agent skills are structured procedural packages that guide frozen LLM agents in specialized workflows.
Luck Is Not Skill: When Do Paired Rollouts Help Group-Relative RL of LLM Agents?
Related stories
arXiv Machine Learning
Oct 15
When Non-Adversarial Noise Hurts Learning: Benchmarking Skill Self-Evolution in Agents
arXiv:2610.19910v1 Announce Type: new Abstract: Large language model agents increasingly improve by distilling experience into reusable skills, but errors in that experience can persist in learned ru...
arXiv Machine Learning
Sep 10
DataFlex-RL: An Evaluation Platform for RLVR Data Policies
arXiv:2609.06107v1 Announce Type: new Abstract: Data policies for reinforcement learning with verifiable rewards (RLVR) determine which rollouts are used, how strongly they are weighted, and which do...
arXiv AI
Sep 10
ExecCritic: Learn to Test, Test to Improve for Coding Agents
arXiv:2609.09133v1 Announce Type: new Abstract: Execution feedback can guide coding agents toward correct repository repairs, but only when the tests capture the behavior requested by the issue. Agen...
arXiv AI
Jun 3
When RLHF Fails: A Mechanistic Taxonomy of Reward Hacking, Collapse, and Evaluator Gaming
arXiv:2606. 03238v1 Announce Type: cross Abstract: Reinforcement learning from human feedback (RLHF) makes large-scale post-training possible by replacing an underspecified human objective with learned and scalable proxies.
arXiv Machine Learning
Jun 8
Leveraging Error Diversity in Group Rollouts for Reinforcement Learning
arXiv:2605. 17333v2 Announce Type: replace Abstract: Reinforcement Learning from Verifiable Rewards (RLVR) typically samples multiple responses per prompt and assigns binary rewards based on individual correctness, yet the collective structure of the group output, specifically the distribution of errors, is largely discarded.
arXiv Machine Learning
Oct 7
FC-SWE: Failure-Conditioned RL for Long-Horizon Software Engineering Agents
arXiv:2610.07898v1 Announce Type: new Abstract: Repository-level software engineering (SWE) is a challenging long-horizon setting: agents must reason over extended interactions, use tools, and adapt...
arXiv Computation and Language
1d ago
Improving Reasoning with Rollout Agreement for Group-Based Policy Optimization
arXiv:2610.34951v1 Announce Type: cross Abstract: Recent methods for Reinforcement Learning with Verifiable Rewards (RLVR) use rollout diversity to improve reasoning performance or training efficienc...
arXiv Machine Learning
Sep 10
CircuitLens: Reasoning Circuits as Data Selection Signals for Reinforcement Learning with Verifiable Rewards
The paper introduces Circuit Reasoning Score (CRS), a data‑selection signal for reinforcement learning with verifiable rewards that uses attention‑head activity from a frozen base model to gauge reasoning engagement. CRS is computed in a single forward pass without reward labels or rollouts, and it shows that selecting problems with the lowest reasoning‑circuit engagement can outperform random selection on several medium‑difficulty benchmarks. However, the benefit depends on domain, model scale, and reward conditions, indicating that data selection in this setting is regime‑dependent rather than a fixed ranking of problem quality.
arXiv AI
Jul 21
From Multi-Agent to Single-Agent: When Is Skill Distillation Beneficial?
arXiv:2604. 01608v4 Announce Type: replace Abstract: Multi-agent systems (MAS) tackle complex tasks by distributing expertise, though this often comes at the cost of heavy coordination overhead, context fragmentation, and brittle phase ordering.
arXiv AI
Jun 30
CRAFT: Counterfactual Credit Assignment from Free Sibling Rollouts for Self-Distilled Agentic Reinforcement Learning
arXiv:2606. 29476v1 Announce Type: cross Abstract: Self-distilled agentic reinforcement learning augments trajectory-level reward with a token-level distillation loss, using as its teacher the same policy conditioned on privileged context.
arXiv AI
Aug 13
Rubric Dropout: A Simple Way to Mitigate Reward Hacking in Rubric-as-Reward RL
arXiv:2608. 11669v1 Announce Type: cross Abstract: Reinforcement learning against rubrics, lists of criteria graded by an LLM judge, has become a standard way to post-train language models on tasks with no deterministic answer.