TechniqueRLHF / Alignment8 recent entries29 Sept 2026Clarify the User or Verify the World? Uncertainty Routing for Proactive AgentsarXiv:2609.32255v1 Announce Type: cross Abstract: Tool-using LLM agents must decide not only whether additional information is needed, but also which source can resolve the uncertainty. Existing proac→30 Sept 2026The Default Trap: Rethinking Plan Evaluation in Tool-Using LLM AgentsarXiv:2609.36829v1 Announce Type: new Abstract: An executor can respond strongly to a change in a supplied plan's priority while showing a small change in the same information-selection probability wh
TechniqueRAG8 recent entries17 Sept 2026Selection Is Retrieval, Abstention Is Not: On-Device Tool Routing over 70 Korean-English ActionsarXiv:2609.18672v1 Announce Type: new Abstract: An AI assistant that calls tools makes two decisions on every request: which tool to invoke, and whether any available tool applies. In the usual design→17 Sept 2026M-SQE: Multilingual Skill Quality Estimation for Enhancing Language Equality in Agentic Skill UsearXiv:2609.18445v1 Announce Type: new Abstract: Agent skills, reusable procedural documents that extend LLM agents beyond their parametric memory, have become an important interface for deploying agen→28 Sept 2026ToolSearcher: Optimizing Tool Selection at Scale via Reinforcement LearningarXiv:2609.30906v1 Announce Type: new Abstract: Large language models (LLMs) excel at natural language processing but struggle to interact with external environments. Tool learning provides a promisin→28 Sept 2026FlyAOC: Evaluating Agentic Ontology Curation of Drosophila Scientific Knowledge BasesarXiv:2602.09163v2 Announce Type: replace Abstract: Scientific knowledge bases accelerate discovery by curating findings from primary literature into structured, queryable formats for both human resea→5 Oct 2026Lost in the Request: How Communication Variation Disrupts Retrieval and Action in Email AgentsarXiv:2610.02627v1 Announce Type: new Abstract: An email assistant should not complete less work simply because a user phrases the same request differently. Yet most benchmarks test each task with onl→12 Oct 2026Web$hell: A Unified Interface for Agentic Web AccessarXiv:2610.13416v1 Announce Type: new Abstract: Large language model (LLM) agents are becoming direct users of the open web. As they take on increasingly complex and information-intensive tasks, effec→12 Oct 2026SmartSalesBench: A Resistance-Aware Multi-Turn Benchmark for Evaluating LLM Sales AgentsarXiv:2610.12975v1 Announce Type: new Abstract: We present SmartSalesBench, the first multi-turn conversational benchmark explicitly designed to evaluate large language model (LLM) sales agents agains→12 Oct 2026AgentTune: A Unified Training Lifecycle for Tool-Using LLM AgentsarXiv:2610.13335v1 Announce Type: new Abstract: Tool-using agents increasingly require a common substrate for building, training, evaluating, distilling, and maintaining agentic systems, yet existing
TechniqueAgents8 recent entries12 Oct 2026PACT: Privileged Trace Co-Training for Multi-Turn Tool-Use AgentsarXiv:2606.16215v2 Announce Type: replace-cross Abstract: Multi-turn tool-use agents must reason, call tools, and adapt to observations across several interaction turns. Post-training such agents is c→12 Oct 2026Nice paper on optimizing agents and their graders together. The key finding is that you want to update your LLM grader alongside the agent i…Nice paper on optimizing agents and their graders together. The key finding is that you want to update your LLM grader alongside the agent instead of keeping it fixed, because the agent's failure mode→12 Oct 2026Masked Diffusion Language Agents for Tool-Integrated Chemical ReasoningarXiv:2610.13330v1 Announce Type: new Abstract: Designing tool-integrated reasoning agent systems for complex chemical tasks remains a fundamental challenge. Although recent chemical agents have demon→12 Oct 2026Introducing DigitalOcean Signals and Simulations: see how your agents and models really behave in production, then test the fix before it shipsYou ship an agent. It mostly works. Then the feedback loop becomes a support ticket: a user got the wrong answer, a tool call looped four times, a refund went out that shouldn’t have. Standard observa→12 Oct 2026How distributed inference keeps large AI models movingDistributed inference splits a large AI model into sub‑components that run concurrently on multiple compute nodes, allowing each node to process only part of the workload. This parallelism boosts thro→12 Oct 2026How Anthropic’s Dynamic Workflows Coordinate Claude AgentsAt AI Engineer New York 2026, Anthropic introduced **dynamic workflows** (also called “flows”), a new type of multi‑agent orchestration designed for large, ambitious workloads using Claude Managed Age→12 Oct 2026AgentTune: A Unified Training Lifecycle for Tool-Using LLM AgentsarXiv:2610.13335v1 Announce Type: new Abstract: Tool-using agents increasingly require a common substrate for building, training, evaluating, distilling, and maintaining agentic systems, yet existing →13 Oct 2026LiquidAI/LFM2.5-24B-A2BGguf's: https://huggingface.co/LiquidAI/LFM2.5-24B-A2B-GGUF Free Demo: https://playground.liquid.ai 'LFM2.5-24B-A2B is a 24B-parameter reasoning model with only 2B active parameters, built for fast, p
TechniqueFine-tuning8 recent entries2 Oct 2026CineMR: Tool-Integrated Vision-Language Reasoning for Quantitative Cardiac MRI AssessmentarXiv:2610.01166v1 Announce Type: cross Abstract: Cardiovascular magnetic resonance (CMR), including cine imaging, is a reference standard for the noninvasive assessment of cardiac morphology and vent→7 Oct 2026POLAR: Ontology-Guided Risk Prevention for Tool-Calling LLM AgentsarXiv:2610.08082v1 Announce Type: new Abstract: LLM tool-use agents operate in dynamic environments where many actions carry operational risk. However, most safety mechanisms react only after errors m→7 Oct 2026AdvSim2Real : Training Web Agents Against Adaptive Prompt Injection in a Web World ModelarXiv:2610.08773v1 Announce Type: cross Abstract: Web agents complete user requests by reading and acting on pages that third parties write, so an instruction planted on a page can redirect the agent →9 Oct 2026VAMR: Multi-Question Agentic Reasoning for Efficient Long-Form Video UnderstandingarXiv:2610.11171v1 Announce Type: cross Abstract: Long-form video understanding often involves multiple questions about different aspects of the same recording. Yet existing video agents typically pro→9 Oct 2026DataSense-Bench: The First Step Toward an AI ScientistarXiv:2610.12190v1 Announce Type: new Abstract: As claims about recursive self-improvement (RSI) and artificial general intelligence (AGI) proliferate, we ask a simple question: do frontier AI models →12 Oct 2026PACT: Privileged Trace Co-Training for Multi-Turn Tool-Use AgentsarXiv:2606.16215v2 Announce Type: replace-cross Abstract: Multi-turn tool-use agents must reason, call tools, and adapt to observations across several interaction turns. Post-training such agents is c→12 Oct 2026Masked Diffusion Language Agents for Tool-Integrated Chemical ReasoningarXiv:2610.13330v1 Announce Type: new Abstract: Designing tool-integrated reasoning agent systems for complex chemical tasks remains a fundamental challenge. Although recent chemical agents have demon→12 Oct 2026Introducing DigitalOcean Signals and Simulations: see how your agents and models really behave in production, then test the fix before it shipsYou ship an agent. It mostly works. Then the feedback loop becomes a support ticket: a user got the wrong answer, a tool call looped four times, a refund went out that shouldn’t have. Standard observa
TechniqueMultimodal8 recent entries2 Oct 2026Kepler: Auditable World Models for ARC-AGI-3arXiv:2610.00834v1 Announce Type: new Abstract: ARC-AGI-3 evaluates agents in interactive environments whose rules and objectives must be inferred from observation. We present Kepler, an open-source h→2 Oct 2026CineMR: Tool-Integrated Vision-Language Reasoning for Quantitative Cardiac MRI AssessmentarXiv:2610.01166v1 Announce Type: cross Abstract: Cardiovascular magnetic resonance (CMR), including cine imaging, is a reference standard for the noninvasive assessment of cardiac morphology and vent→5 Oct 2026RoboBridge: A Self-Evolving Embodied Agent Framework for Sim-to-Real TransferarXiv:2610.02717v1 Announce Type: new Abstract: A key challenge in bringing embodied intelligence into the real world is transferring capabilities from simulation to reality and enabling agents to con→5 Oct 2026HyperBrowseComp: A Multilingual and Multimodal Stress Test for Web-Browsing AgentsarXiv:2610.03574v1 Announce Type: new Abstract: We introduce HyperBrowseComp, a multilingual and multimodal browsing benchmark comprising 423 manually authored and human-validated questions across 13 →6 Oct 2026Recursive Self-Improvement of Visuomotor Policies through Local Recovery SupervisionarXiv:2610.05151v1 Announce Type: new Abstract: Visuomotor policies can execute familiar tasks yet lack the corrective behavior needed after their own mistakes. We present a framework for recursive se→6 Oct 2026G-CARB: Graph-Localized Conformal Agent Risk Budget for Compositional HarmarXiv:2610.05563v1 Announce Type: cross Abstract: Small language model (SLM) agents need safety controls that track consequences across tool calls with little monitoring overhead. A private read, for →7 Oct 2026Visual Orchestration Tax in Agentic VLM Pipelines: Auditing and Certifying Visual Evidence ReusearXiv:2610.08170v1 Announce Type: cross Abstract: Agentic VLM pipelines increasingly pass the same static visual evidence through multiple specialist agents and tools. This design creates an orchestra→12 Oct 2026When Charts Deceive Web Agents: A Paired Benchmark for Misleading Visualization-Grounded TasksarXiv:2610.12823v1 Announce Type: new Abstract: Web agents increasingly browse, interpret, and act on web content on behalf of users. In many realistic workflows, however, the visual evidence they con
TechniqueSafety8 recent entries6 Oct 2026G-CARB: Graph-Localized Conformal Agent Risk Budget for Compositional HarmarXiv:2610.05563v1 Announce Type: cross Abstract: Small language model (SLM) agents need safety controls that track consequences across tool calls with little monitoring overhead. A private read, for →7 Oct 2026POLAR: Ontology-Guided Risk Prevention for Tool-Calling LLM AgentsarXiv:2610.08082v1 Announce Type: new Abstract: LLM tool-use agents operate in dynamic environments where many actions carry operational risk. However, most safety mechanisms react only after errors m→7 Oct 2026FC-SWE: Failure-Conditioned RL for Long-Horizon Software Engineering AgentsarXiv: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 t→9 Oct 2026VAMR: Multi-Question Agentic Reasoning for Efficient Long-Form Video UnderstandingarXiv:2610.11171v1 Announce Type: cross Abstract: Long-form video understanding often involves multiple questions about different aspects of the same recording. Yet existing video agents typically pro→9 Oct 2026Dex-One2Many: Learning Dexterous Manipulation from a Single Human DemonstrationarXiv:2610.12470v1 Announce Type: cross Abstract: While learning dexterous manipulation from a single human video offers a promising alternative to costly robot demonstrations, many recent methods pre→12 Oct 2026When Charts Deceive Web Agents: A Paired Benchmark for Misleading Visualization-Grounded TasksarXiv:2610.12823v1 Announce Type: new Abstract: Web agents increasingly browse, interpret, and act on web content on behalf of users. In many realistic workflows, however, the visual evidence they con→12 Oct 2026VEPA: Verified Evolving Formal Policy for Lifelong LLM Agent GuardrailsarXiv:2610.12813v1 Announce Type: new Abstract: LLM agents increasingly operate over sensitive resources, such as private documents and mutable system states, which can create serious safety and secur→12 Oct 2026Rollout-Time Intervention: Unlocking Agentic Reinforcement Learning Beyond Policy LimitsarXiv:2610.13438v1 Announce Type: new Abstract: Reinforcement learning (RL) has emerged as a key paradigm for improving large language model (LLM) agents, enabling stronger reasoning, planning, and to