SkillReason: Reasoning-Enhanced Agent Skill Retrieval for Implicit User Requests

Large language model agents increasingly rely on reusable skills to extend their capabilities beyond parametric knowl- edge. However, retrieving the appropriate skill from a large- scale library remains challenging because realistic user re- quests are often concise and underspecified, stating only the task goal while leaving the required capabilities and execu- tion steps implicit. Existing benchmarks provide limited cov- erage of such requests. To address this gap, we introduce SkillReason-Bench, a large-scale cross-domain benchmark containing 3,729 queries and a retrieval corpus of 61,228 skills spanning nine domains. We further propose SkillRea- son, a two-stage framework that uses chain-of-thought rea- soning as training-time supervision for skill retrieval. In Stage I, capability reasoning traces generated by a stronger teacher provide explicit supervision through contrastive learning, re- trieval distribution alignment, and language modeling, en- couraging the retriever to internalize capability reasoning in its query representation. In Stage II, a retrieval-guided GRPO objective encourages the model to explore reasoning trajecto- ries better suited to its own capabilities and more effective for retrieval. At inference, SkillReason directly encodes the orig- inal query without autoregressive CoT generation, preserv- ing efficient query-only retrieval. Extensive experiments on SkillReason-Bench, SkillRet, and SRA-Bench show that Skill- Reason achieves state-of-the-art performance across all three benchmarks, demonstrating that reasoning-enhanced training better bridges the semantic gap between high-level task goals and skill capabilities.

Agentic Reinforcement Learning with Observation-Calibrated Self-Distillation

Large language model agents are commonly trained through reinforcement learning with sparse trajectory-level rewards, which offer limited guidance on how strongly individual tokens should be updated. On-Policy Self-Distillation (OPSD) addresses this by re-scoring generated tokens under a privileged replay view to obtain dense, token-level supervision. However, we identify a confounding issue: the resulting support may reflect both the privileged information contained in the replay view and score shifts induced by the replay scaffold, making it difficult to attribute the support specifically to that information. This issue is especially pronounced when future environment observations serve as privileged information, since replaying them requires reconstructing an extended scaffold that itself perturbs token scores. To resolve this confounding, we propose Observation-Calibrated Self-Distillation (OCSD), which contrasts two structurally matched replay views, Full and Observation-Ablated, differing only in whether the actual future observation is present, to derive an observation residual that discounts score changes shared by the replay scaffold. OCSD then applies this residual to modulate token-level GRPO updates at high-uncertainty steps, while preserving the trajectory-level update direction. Experiments on ALFWorld, WebShop, and Search-QA across three Qwen3 model scales show that OCSD consistently outperforms strong baselines. Diagnostic analyses further confirm that the calibrated residual aligns better with local environment feedback. Our code is publicly available at https://github.com/yiy1x/OCSD.

Hermite Curves as Trajectory Priors for Vision-Language-Action Models

Despite recent progress in Vision-Language-Action (VLA) models for robotic manipulation, the action chunk remains a weakly structured interface. Existing work typically flatten each chunk into per-timestep controls, relying on implicit data learning that manifests as jagged motion and boundary discontinuities during physical execution. To address these limitations, we introduce Hermite trajectory priors, parameterizing the chunk trajectory as a piecewise cubic Hermite curve defined by endpoint positions and velocities to explicitly enforce smoothness and continuity. We instantiate this fixed operator across discrete autoregressive and continuous generative paradigms via three variants: (1) Hermite Tokens, which predict quantized boundary variables autoregressively; (2) Hermite Scaffold, which decomposes clean actions into a base scaffold and residuals; and (3) Hermite Regularization, which applies the prior strictly as an auxiliary training objective. Across simulation benchmarks and real-robot platforms, Hermite Regularization achieves superior performance among these three variants, improving π0.5 baseline success rates from 95.9% to 98.7% on LIBERO, 85.7% to 90.9% on LIBERO-plus, and 63.4% to 90.0% across four real-robot tasks without additional inference overhead. Trajectory analyses reveal that explicitly structuring trajectory priors serves most effectively as a learning inductive bias rather than a runtime constraint.

ChunkFlow: Towards Continuity-Consistent Chunked Policy Learning

Vision-language action (VLA) models increasingly adopt chunked action heads to satisfy real-time constraints; however, this introduces boundary jitter: overlapping regions between consecutive chunks often yield inconsistent predictions, degrading temporal coherence and the task success rate. Existing methods, such as inference-time blending, merely reweight mismatched proposals without correcting underlying errors, leading to residual accumulation under biased or noisy histories. We propose ChunkFlow, a seam-aware training-and-execution framework for chunked policies that aligns chunk structure with boundary execution. It partitions each chunk into frozen, editable, and future zones, applies deterministic overlap blending at execution, and trains raw predictions with seam and first- and second-order continuity losses. History corruption and scheduled sampling improve robustness to executed-history errors, while an AWAC fine-tuning stage adapts the policy without removing these structural regularizers. Under mild smoothness assumptions, pre-blending seam discrepancies provably decay with increasing overlap. Experiments on CALVIN, LIBERO, and real robots show an improved success-stability trade-off with low-latency inference. Project page: https://cytoderm-ai.github.io/chunkflow.

Modularized Reinforcement Learning on LLMs: From MDP Creation to Exploration and Learning

Reinforcement learning (RL) has become central to LLM post-training, yet the methods that dominate current pipelines, PPO and GRPO, represent only a narrow slice of what RL offers. Understanding why these methods prevail, and what alternatives exist, requires a principled examination of the design decisions that underlie any RL algorithm. This survey organizes that examination around three stages of algorithm construction. We begin with MDP creation: how the reward function, state space, action space, termination condition, and discount factor are, or could be, defined for LLM training. We then turn to exploration, covering temperature sampling, entropy regularization, intrinsic motivation, tree search, and curriculum learning. Finally, we address learning along four classical RL dimensions: model-free versus model-based, value-based versus policy-based versus actor-critic, on-policy versus off-policy, and credit assignment, including both Monte Carlo methods, which rely on full return estimates, and bootstrapping methods, which update estimates using other learned predictions. Mapping the LLM literature onto this taxonomy reveals a strikingly non-uniform distribution of research effort. Critic-free policy gradients and Monte Carlo credit assignment are densely populated, while value-based methods, off-policy actor-critic training, and bootstrapping-based credit assignment remain largely unexplored despite well-established counterparts in classical RL. These gaps represent concrete opportunities for transferring proven RL techniques to LLM training. By making these gaps explicit alongside the methods that have proven effective, this survey offers researchers in both RL and LLMs a shared framework for understanding current practice and identifying promising directions for future work.

VeriGraph: Towards Verifiable Data-Analytic Agents

LLM-based agents have demonstrated strong capabilities in data-intensive analytical tasks, yet their outputs are rarely verifiable: a reliance on linear text trajectories makes their reasoning difficult to audit. In particular, deterministic computations over raw data and semantic deductions over natural-language claims are often entangled in an unstructured stream, leaving numerical conclusions hard to reproduce and qualitative judgments hard to inspect. To address this, we propose VeriGraph, a traceable neuro-symbolic reasoning framework that enables agents to construct an explicit heterogeneous evidence directed acyclic graph (DAG) during execution. VeriGraph introduces three evidence-expansion primitives, namely computational, grounding, and derivational expansion, to connect raw data, interpreter variables, computed results, and natural-language claims in a unified graph. Under this formulation, structural traceability is reduced to graph reachability from raw data sources to terminal claims, while semantic support is measured by claim-level evidence evaluation. To improve graph construction, we further design a graph-based policy optimization strategy with a composite reward that jointly supervises answer correctness, computational integrity, and derivational coherence. Experiments on four benchmarks show that VeriGraph-8B achieves the highest overall score among all baselines. More importantly, VeriGraph produces auditable evidence graphs with substantially stronger claim grounding, achieving a 87.61\% Grounding Rate under our claim-level evidence support evaluation. These results suggest that explicit evidence-graph construction is a promising path toward verifiable data-analytic agents. Our code is available at https://github.com/ignorejjj/VeriGraph.

Biological Reasoning-Informed Regression for Interpretable Regulatory DNA Activity Prediction

DNA cis-regulatory elements (CREs) such as enhancers control gene expression levels. Accurately predicting regulatory activity from DNA sequences is valuable but challenging, as it requires understanding complex biological regulatory processes. Existing methods typically regress activity scores from sequences in a black-box manner, limiting both interpretability and regression performance. Meanwhile, large language models (LLMs) benefit from explicit reasoning processes, yet directly applying LLMs to raw DNA sequences performs poorly. In this paper, we bridge this gap by introducing R3LM, a framework that teaches LLMs reasoning-informed regression on regulatory DNA through structured biological knowledge. Specifically, we design a biologically grounded data format that structures DNA's regulatory information for improved LLM understanding, and construct CRE-ReasonBench, the first dataset that associates DNA sequences and activity scores with mechanistic reasoning traces. Through two-stage training that first teaches LLMs reasoning over structured biological information then performs regression, R3LM achieves state-of-the-art performance on enhancer prediction across three cell types, outperforming both LLMs with raw sequence input and specialized DNA models while providing interpretable mechanistic explanations. We expect R3LM as an interpretable reward model that can effectively assist biologists in CRE design. Code is available at https://github.com/DuanYi516/R3LM.

Reinforcement Learning from Cross-domain Videos with Video Prediction Model

Reinforcement learning from expert videos across visually distinct domains is challenging due to the absence of reward signals and the presence of domain gaps. We introduce XIPER (Cross-domain Video Prediction Reward), a reward model for learning from expert videos collected in a visually different domain, where the agent's appearance differs due to factors such as color, morphology, or the sim-to-real gap. More specifically, XIPER trains a cross-domain video prediction model that maps agent observations into the expert domain and uses the prediction likelihood as a reward signal. Experiments on the DMC Color Suite (8 tasks) and DMC Body Suite (3 tasks) show that XIPER consistently outperforms baselines despite domain gaps such as differences in agent color and morphology. We further analyze XIPER on a sim-to-real transfer dataset, demonstrating that it produces meaningful reward signals for real-robot observations given only simulated expert videos. Code, pretrained models, datasets and video demonstrations can be found on our project webpage: https://sites.google.com/view/xiper

CityGen: Structure-Guided City-Style Synthesis for Cross-City Autonomous Driving

Autonomous driving systems are commonly trained and evaluated within limited geographic regions, which hinders their scalability when deployed in new cities. However, significant domain shifts in appearance, road topology, and traffic patterns often cause severe performance degradation under cross-city deployment. Existing approaches based on domain adaptation, data augmentation, or synthetic data generation typically rely on labeled target data, city-specific annotations, or task-specific designs, limiting their scalability and effectiveness for holistic evaluation. In this paper, we introduce CityTransfer-Bench, a geographically disjoint benchmark for evaluating cross-city generalization across perception, segmentation, and planning, and propose CityGen, a diffusion-based generative framework that performs zero-label city adaptation via HD-map-conditioned synthesis guided by city-level visual prompts. Extensive experiments demonstrate that CityGen consistently improves cross-city robustness across multiple tasks, establishing a scalable and label-efficient foundation for generalizable autonomous driving.

A Heterogeneous Temporal Memory Governance Framework for Long-Term LLM Persona Consistency

Large language models often suffer from fact loss, timeline confusion, persona drift, and reduced stability during long-range interaction, especially under high-noise knowledge bases, context clearing, and cross-model transfer. To address these issues, we introduce ARPM, an external temporal memory governance framework for long-term dialogue. ARPM separates static knowledge memory from dynamic dialogue experience memory and combines vector retrieval, BM25, RRF fusion, dual-temporal reranking, chronological evidence reading, and a controlled analysis protocol for evidence verification and answer binding. Unlike approaches that encode persona consistency into model weights or rely only on long context, ARPM treats continuity as a traceable, auditable, and transferable governance problem. Using engineering logs, we conduct three experiments. First, in a 50-round question-answering setting, we compare signal-to-noise ratios of 1:5 and 1:200+, and distinguish CSV auto-judgment from manual review. Under 1:5, CSV recall accuracy is 54.0%, while manual review raises it to 100.0%. Under 1:200+, the values are 44.0% and 80.0%. These results show that automatic rules can underestimate recall after supporting evidence enters the prompt. Second, ablation results show that dialogue history retrieval is necessary for recent continuity: disabling it reduces strict accuracy from 100% to 66.7%, and disabling BM25 reduces it to 80.0%, indicating that pure semantic retrieval is insufficient for correction and tracing. Third, under a 5.1-million-character noise substrate, periodic context clearing, and multi-model handoff, ARPM maintains semantic continuity, boundary continuity, and persona consistency, while exposing limits caused by weak protocol compliance. These findings show that long-term persona consistency can be decomposed into governable components and evaluated in a white-box manner.