News without the noise
Reinforcement learning for agentic behaviour
22 articles
Liu et al. replace sampled reinforcement-learning updates with exact optimization over the complete, enumerable set of genomic tool combinations. Across five frozen reasoners and three benchmarks, their method outperformed GRPO in all 15 experimental settings, by 6.75 percentage points on average.
11 Sept 2026The authors built a digital-advertising simulator in which injected interventions provide known labels for otherwise ambiguous diagnostic problems. On 235 held-out episodes, reinforcement learning with these synthesized rewards improved a Qwen3.5-35B-A3B agent’s full top-ranked attribution accuracy to 0.757, above all evaluated prompted baselines.
11 Sept 2026Yu, Jang and Yu study how the structure of reinforcement-learning rollouts affects exploration in mathematical reasoning models. They report that their difficulty-adaptive, sentence-level branching method improves pass@k—the probability of finding a correct answer within k attempts—and consequently benefits test-time scaling.
10 Sept 2026Yang et al. introduce FactoSR, a reinforcement-learning framework designed to help vision-language models recover spatial and temporal structure from 2D visual inputs. The authors report improvements of 5.9% on VSI-Bench and 4.5% on All-Angles-Bench.
6 Sept 2026Wang et al. show that Group Relative Policy Optimization (GRPO) can assign large advantages to rollouts that reach correct answers by guessing rather than reasoning. Their alternative, SignBalance, removes dependence on the reward composition of each rollout group and reportedly improves bounded-answer mathematics and search-agent performance.
6 Sept 2026Gandhi et al. introduce DRACO, a reinforcement-learning method for training agents when task success cannot be checked programmatically. On AppWorld and the out-of-domain τ-bench, its dynamically generated rubrics and step-level credit redistribution outperformed the base model and several sparse-reward or rubric-based alternatives.
6 Sept 2026Li and colleagues compare ways to combine on-policy distillation (OPD), which supplies dense teacher feedback at the token level, with reinforcement learning from verifiable rewards (RLVR), which uses sparse outcome-based signals. Across logic and mathematics benchmarks, they find that completing OPD before switching to RLVR works consistently better than pure OPD, pure RLVR, or joint optimization.
5 Sept 2026Zhou et al. study how reinforcement-learning credit should be distributed across multi-turn agent trajectories when only the final outcome is verified. They find that the decisive factor is verifier information density—the share of prerequisite steps whose correctness is observable—and that broad coverage outperforms targeted credit in the low-information regime typical of terminal-state evaluation.
4 Sept 2026The authors introduce SAGE, a reinforcement-learning framework that uses an expensive vision-language model as a temporary teacher rather than a permanent policy. Across sparse-reward visual reasoning and navigation tasks, selectively requested guidance improved learning in several environments, sometimes producing policies that surpassed the teacher and required no model calls at evaluation time.
3 Sept 2026Yang and colleagues present a conceptual framework for training large reasoning models as direct human supervision recedes. Their L0–L4 ladder tracks the shift from human-provided judgments and tasks toward autonomous rewards, self-generated curricula, constructed environments, and co-evolving learning systems.
2 Sept 2026Yang et al. address a credit-assignment problem in reinforcement learning for agents: a verified task outcome is typically applied uniformly to every decision in a long interaction. Their TASPO method uses privileged information from successful experiences to weight individual actions, improving performance over GRPO by 10.6% across three agentic benchmarks and transferring better to unseen tasks.
2 Sept 2026The authors turn 20,000 scientific papers into training environments for research-plan generation, deriving prompts from each paper’s goals and background while extracting evaluation criteria from its methods and experiments. Across three Qwen3 model sizes, their rubric-centered training schedule produced higher benchmark scores than supervised fine-tuning, either training stage alone, or the same stages in reverse order.
2 Sept 2026Meng et al. introduce HARTS, a training system that compresses shared prefixes in agentic reinforcement-learning rollout trees and schedules the remaining work across microbatches and data-parallel replicas. On rollouts generated from SWE-bench tasks, it delivered 4.81–4.87× speedups for forward, backward, and gradient computation while maintaining numerical and early-training behavior close to trajectory-wise training.
1 Sept 2026Wang et al. introduce Contrastive Branch Policy Optimization (CBPO), a reinforcement-learning method that separates where to sample alternative continuations from how to assign credit to tokens. Across ten mathematical-reasoning and knowledge-search benchmarks, the authors report the highest macro-average accuracy among the evaluated policy-optimization and branch-based methods at two model scales.
27 Aug 2026The authors introduce CAFE, a framework in which one shared-parameter model alternates between searching and providing corrective feedback during a search trajectory. Combining online reinforcement learning with offline preference optimization produced gains across in-domain and out-of-domain evaluations, while training only the agent or only the critic eventually plateaued.
27 Aug 2026Qian et al. propose MileGPO, a policy-optimization method for agents that receive supervision mainly from final task outcomes. On ALFWorld and WebShop, the authors report that its locally calibrated intermediate credits improve performance without auxiliary models or additional environment interactions.
23 Aug 2026The authors present OpenForgeRL, an open-source framework that trains stateful, harness-based AI agents end-to-end using standard reinforcement learning libraries. It wraps the harness's model calls via a proxy for data recording and orchestrates rollouts in remote Kubernetes containers, yielding competitive results on tool-use, browser, and computer-use benchmarks with relatively few training tasks.
24 July 2026Hu et al. introduce Molt, a PyTorch-native training framework for agentic reinforcement learning that prioritizes researcher readability and end-to-end algorithm traceability. In benchmarked asynchronous training runs, Molt achieved statistically comparable performance to a state-of-the-art Megatron-based stack, demonstrating that leanness does not compromise scalability.
23 July 2026The authors propose TRACE, a credit-assignment method for long-horizon agentic RL that derives per-action rewards from the log-probability of the gold answer under a frozen reference model, using temporal-difference learning. On the closed-web BrowseComp-Plus benchmark, TRACE raises Qwen3-4B from 7.2 to 35.6 and Qwen3-30B-A3B from 8.4 to 42.6, and enables pure RL training without cold-start supervised fine-tuning or live-web data.
16 July 2026Li et al. propose CompactionRL, a reinforcement learning method that trains LLM agents to perform long-horizon tasks by compressing interaction histories. Applied to open models, it yields consistent gains on code agent benchmarks, with GLM-4.5-Air reaching 66.8% Pass@1 on SWE-bench Verified.
7 July 2026The authors present PROVE, a framework for reinforcement learning (RL) on multi-step tool use that overcomes three key obstacles: lack of realistic execution environments, ungrounded synthetic queries, and reward-induced verbosity. Training four different LLMs with PROVE yields significant gains on three multi-turn tool-use benchmarks, with improvements up to +10.2 points on BFCL Multi-Turn.
3 June 2026The authors introduce EvoTrainer, a framework that treats the training harness itself as an evolvable component in LLM reinforcement learning. By co-evolving policies and harness heuristics through empirical feedback, EvoTrainer matches or exceeds human-engineered RL baselines across three domains, with the largest gains in long-horizon agentic software engineering.
2 June 2026