arXiv:2601. 22211v2 Announce Type: replace Abstract: Reinforcement learning (RL) with combinatorial action spaces remains challenging because feasible action sets are exponentially large and governed by complex feasibility constraints, making direct policy parameterization impractical.
arXiv:2608. 16798v1 Announce Type: cross Abstract: Agent harnesses have substantially improved performance on long-horizon tasks by coordinating agent interactions with the environment.
arXiv:2601. 06487v3 Announce Type: replace-cross Abstract: Reinforcement learning has substantially improved the performance of LLM agents on tasks with verifiable outcomes, but it still struggles on open-ended agent tasks with vast solution spaces (e.
arXiv:2603. 02462v2 Announce Type: replace-cross Abstract: A key challenge in developing unified neural solvers for combinatorial optimization (CO) is the efficient generalization of models from a given set of tasks to new tasks unseen during initial training.
Group-Graph Policy Optimization (G2PO) is a novel group-based reinforcement learning algorithm designed for long-horizon agentic tasks. It transforms linear interaction trajectories into a global state‑transition graph, aggregates identical observations across trajectories to reduce sampling variance, and redefines actions as transitions between state nodes. By employing an edge‑centric advantage estimation and globally standardizing TD errors, G2PO identifies and prioritizes critical transitions, achieving up to 22.2% higher success rates than state‑of‑the‑art baselines on benchmarks such as WebShop, ALFWorld, and AppWorld.
arXiv:2509. 18930v3 Announce Type: replace-cross Abstract: Neural algorithmic reasoning (NAR) is a paradigm that trains neural networks to execute classic algorithms by supervised learning.
GeoPAR is a geometry-guided parallel autoregressive reinforcement learning framework designed for large-scale multi-agent combinatorial optimization. It introduces a projection-window sparse geometry mechanism, sparse edge-biased attention, and cache-guided conflict-aware assignment to better model local geometric structures and reduce duplicate task selections. Experiments on heterogeneous vehicle routing and multi-depot pickup-and-delivery problems demonstrate improved zero-shot generalization, fewer rollout steps, and efficient inference.
arXiv:2605. 09382v2 Announce Type: replace Abstract: The Linear Assignment Problem is a fundamental combinatorial optimization task where classical exact solvers ensure optimality but suffer from an $\mathcal{O}(N^{3})$ bottleneck, while recent neural approximations struggle with scalability and exactness.
arXiv:2601. 15158v4 Announce Type: replace-cross Abstract: Transformers trained via Reinforcement Learning (RL) with outcome-based supervision can spontaneously develop the ability to generate intermediate reasoning steps (Chain-of-Thought).
arXiv:2604. 17892v4 Announce Type: replace-cross Abstract: Recently, latent reasoning has been introduced into large language models (LLMs) to leverage rich information within a continuous space.
The paper introduces a method that integrates action abstraction into policy optimization for reinforcement learning and generative flow networks. By iteratively identifying frequently used action subsequences in high‑reward trajectories and treating them as single high‑level actions, the approach expands the action space and improves sample efficiency. Experiments on synthetic and real‑world tasks show that this technique discovers diverse high‑reward states more effectively, especially on challenging exploration problems, and yields interpretable abstract actions that reflect the underlying reward structure.
MiMo‑V2.6 is an omni‑modal reinforcement learning framework that scales compute to advance large foundation models toward self‑improvement. It trains on a broad multimodal corpus before RL, then scales RL along three axes: larger batches and throughput, diverse environments across code, general, visual, and cyber domains, and enhanced grader compute for better reward signals. The system incorporates infrastructure such as a frozen MoE router, multi‑layer defense against reward hacking, unified trajectory representation, high‑concurrency rollout, and training‑inference consistency, and it is open‑source to support reproducibility and further research.