TechniqueRLHF / Alignment8 recent entries30 Sept 2026Regime Boundary Alignment for Evidence-Gated Question AnsweringarXiv:2609.37491v1 Announce Type: cross Abstract: Retrieval-augmented language models are expected to answer from the retrieved evidence, but in practice they often keep answering when that evidence i→2 Oct 2026Pre-training interventions, ex post facto: Grafting model beliefs across checkpointsarXiv:2610.00767v1 Announce Type: cross Abstract: Pre-training interventions are critical to alignment research, since beliefs formed during pre-training shape how a model generalizes from later train
TechniqueRAG8 recent entries7 Oct 2026Agentic AutoRAG: RAG Pipeline Optimization through Reasoning-Driven AgentsarXiv:2610.08452v1 Announce Type: new Abstract: Retrieval-augmented generation (RAG) is a widely used approach for grounding large language models (LLMs) in external knowledge. However, configuring a →8 Oct 2026TopoGraphRAG-Bench: Evaluating Multimodal GraphRAG on Layout-Grounded Evidence ReasoningarXiv:2610.09360v1 Announce Type: cross Abstract: Real-world documents distribute evidence across text, tables, figures, and captions within complex page layouts. Answering complex questions over such→8 Oct 2026PHRBench: A Behavioral Evaluation of Post-Hallucination Reasoning in LLMsarXiv:2610.10455v1 Announce Type: new Abstract: Hallucinated information can propagate through multi-stage LLM systems and become part of the context for subsequent reasoning. Existing studies of post→8 Oct 2026Beyond Outcome Rewards: Constructing and Assigning Retrieval Credit for Search AgentsarXiv:2610.10179v1 Announce Type: new Abstract: Search agents enable Large Language Models (LLMs) to iteratively retrieve and use information for complex multi-hop questions. Reinforcement Learning wi→8 Oct 2026BEACON-SP: Ontology-Grounded GraphRAG Framework for Clinical Suicide Risk AssessmentarXiv:2610.09026v1 Announce Type: cross Abstract: We present BEACON-SP, an ontology-grounded Graph Retrieval-Augmented Generation (GraphRAG) framework for clinician-facing decision support in behavior→9 Oct 2026From Retrieval to Reconstruction: Constructing Evolvable Cognitive Memory for Long-Term DialoguearXiv:2610.11314v1 Announce Type: new Abstract: Large Language Models (LLMs) serving as long-term dialogue agents require memory systems that support reliable reasoning over extended interactions. How→9 Oct 2026Event-Centric Memory with Query-Aware Graph Augmentation for Long-Term Conversational AgentsarXiv:2610.11920v1 Announce Type: new Abstract: For persistent and personalized conversational agents, memory systems can enable them to remember, update, and reason over long histories by storing pas→12 Oct 2026SEEDS: Preview-Guided Evidence Admission for Search AgentsarXiv:2610.12932v1 Announce Type: new Abstract: Search agents extend large reasoning models with multi-round access to external knowledge, yet most existing systems largely conflate retrieving evidenc
TechniqueAgents8 recent entries8 Oct 2026Bookkeeping, Composition, or Unreachable Gold? Reading MemoryAgentBench's Conflict-Resolution Scores Against a Frozen Last-Write ResolverarXiv:2610.09193v1 Announce Type: cross Abstract: MemoryAgentBench's Conflict Resolution split is read as measuring 'selective forgetting'. We execute the benchmark's own rule - the newest statement a→8 Oct 2026Beyond Outcome Rewards: Constructing and Assigning Retrieval Credit for Search AgentsarXiv:2610.10179v1 Announce Type: new Abstract: Search agents enable Large Language Models (LLMs) to iteratively retrieve and use information for complex multi-hop questions. Reinforcement Learning wi→9 Oct 2026From Retrieval to Reconstruction: Constructing Evolvable Cognitive Memory for Long-Term DialoguearXiv:2610.11314v1 Announce Type: new Abstract: Large Language Models (LLMs) serving as long-term dialogue agents require memory systems that support reliable reasoning over extended interactions. How→9 Oct 2026Event-Centric Memory with Query-Aware Graph Augmentation for Long-Term Conversational AgentsarXiv:2610.11920v1 Announce Type: new Abstract: For persistent and personalized conversational agents, memory systems can enable them to remember, update, and reason over long histories by storing pas→9 Oct 2026Chronos Enables Code Agents to Reason over Software EvolutionarXiv:2610.11578v1 Announce Type: cross Abstract: Historical pull requests record the design decisions, compatibility constraints, and implementation patterns behind a codebase's current state. Experi→12 Oct 2026SEEDS: Preview-Guided Evidence Admission for Search AgentsarXiv:2610.12932v1 Announce Type: new Abstract: Search agents extend large reasoning models with multi-round access to external knowledge, yet most existing systems largely conflate retrieving evidenc→12 Oct 2026Let Large Language Models Judge Each Other: Multi-Agent Peer-Reviewed Reasoning for Medical Question AnsweringarXiv:2606.15419v2 Announce Type: replace-cross Abstract: Objective: To enhance the accuracy, interpretability, and robustness of large language models (LLMs) in medical question answering (MedQA). Me→12 Oct 2026Collaborative Learning and Reasoning over Model SlicesarXiv:2610.12927v1 Announce Type: new Abstract: Large language models (LLMs) excel across diverse tasks, benefiting from knowledge priors, in-context reasoning, and test-time computation. Despite thes
TechniqueFine-tuning8 recent entries30 Sept 2026Regime Boundary Alignment for Evidence-Gated Question AnsweringarXiv:2609.37491v1 Announce Type: cross Abstract: Retrieval-augmented language models are expected to answer from the retrieved evidence, but in practice they often keep answering when that evidence i→1 Oct 2026GraphForge: Training Working Agents with Graph-Anchored Workspace SynthesisarXiv:2609.38923v1 Announce Type: new Abstract: Working agents need to read diverse files, coordinate tools, and produce deliverables. Training such agents requires tasks built on many real files with→1 Oct 2026Concept-Grounded Attention: A Controlled Evaluation of Graph-Injected Attention, Temporal Versioning, and Epistemic StatusarXiv:2609.38684v1 Announce Type: new Abstract: Knowledge-intensive language-model systems typically represent external knowledge as text chunks or static graphs, with limited support for concept evol→1 Oct 2026A helps B while B hurts A: directed transfer in instruction-tuning mixturearXiv:2609.39702v1 Announce Type: new Abstract: Adapting a language model to a specialized corpus means choosing which instruction-tuning tasks to train on under a fixed budget, and testing one choice→2 Oct 2026Pre-training interventions, ex post facto: Grafting model beliefs across checkpointsarXiv:2610.00767v1 Announce Type: cross Abstract: Pre-training interventions are critical to alignment research, since beliefs formed during pre-training shape how a model generalizes from later train→4 Oct 2026I Distilled an LLM into two 287M encoders (GLiNER + multiple choice) for document extraction, can't match teacher. did i do something wrong?A while ago I asked here how to turn ~5 million court decisions into structured graphs without running an expensive LLM on every document thanks for the advice . I went with the 'small extractor + cla→5 Oct 2026Text-Centric Post-Training for Omni-Modal ReasoningarXiv:2610.02819v1 Announce Type: new Abstract: Improving joint audio-visual reasoning in Omni Large Language Models typically incurs substantial data construction and training costs. Our diagnostics →7 Oct 2026Turnslide: Scalable Multi-Turn Data Synthesis by Walking a Finite-State MachinearXiv:2610.07070v1 Announce Type: new Abstract: Small language models are inexpensive to serve and can run on private infrastructure, but base models are often not good enough at multi-turn tool calli
TechniqueMultimodal8 recent entries6 Oct 2026Render to Reason: Novel-View Semantic Prediction Improves Spatial Understanding in VLMsarXiv:2610.05417v1 Announce Type: new Abstract: Recent works augment Vision-Language Models with geometry features from pretrained 3D models, expecting that the geometric signal will boost spatial rea→7 Oct 2026Natural Language Questions as an Interface for Knowledge Graphs: QRAKEN Graph Distillation and Semantic Self-HealingarXiv:2610.08095v1 Announce Type: new Abstract: Natural-language access to RDF knowledge graphs is a core Semantic Web ambition. Large language models (LLMs) have advanced Text-to-SPARQL, yet on unfam→7 Oct 2026MedHorizon: Towards Long-context Medical Video Understanding in the WildarXiv:2605.06537v2 Announce Type: replace Abstract: Medical multimodal large language models (MLLMs) have advanced image understanding and short-video analysis, but real clinical review often requires→7 Oct 2026How Well Do LLMs Reason with Noisy Evidence? An Active Visual Reasoning BenchmarkarXiv:2610.07751v1 Announce Type: new Abstract: Real-world reasoning rarely reduces to static question answering: agents must actively gather information from tools and sensors that are often noisy an→7 Oct 2026Agentic AI with Structured CoT for Enhancing AI's Spatial Intelligence: Visualization and Reasoning of RotationarXiv:2610.04188v2 Announce Type: replace Abstract: Recent studies show that artificial intelligence (AI) with language and vision capabilities still experiences limitations in spatial reasoning. In t→8 Oct 2026TopoGraphRAG-Bench: Evaluating Multimodal GraphRAG on Layout-Grounded Evidence ReasoningarXiv:2610.09360v1 Announce Type: cross Abstract: Real-world documents distribute evidence across text, tables, figures, and captions within complex page layouts. Answering complex questions over such→8 Oct 2026ExperienceIndex: Artifact-Grounded MemoryarXiv:2610.10091v1 Announce Type: new Abstract: Knowledge-intensive tasks require answering many questions by reasoning about a shared corpus of artifacts (e.g., court cases, or scientific literature)→8 Oct 2026Bookkeeping, Composition, or Unreachable Gold? Reading MemoryAgentBench's Conflict-Resolution Scores Against a Frozen Last-Write ResolverarXiv:2610.09193v1 Announce Type: cross Abstract: MemoryAgentBench's Conflict Resolution split is read as measuring 'selective forgetting'. We execute the benchmark's own rule - the newest statement a
TechniqueSafety8 recent entries30 Sept 2026BRIDGE: Bilevel Retrieval-Credit-Aware Agentic Reinforcement LearningarXiv:2609.36505v1 Announce Type: new Abstract: Agentic reinforcement learning (ARL) with verifiable rewards improves the ability of large language models (LLMs) to tackle knowledge-intensive tasks by→1 Oct 2026Concept-Grounded Attention: A Controlled Evaluation of Graph-Injected Attention, Temporal Versioning, and Epistemic StatusarXiv:2609.38684v1 Announce Type: new Abstract: Knowledge-intensive language-model systems typically represent external knowledge as text chunks or static graphs, with limited support for concept evol→2 Oct 2026Pre-training interventions, ex post facto: Grafting model beliefs across checkpointsarXiv:2610.00767v1 Announce Type: cross Abstract: Pre-training interventions are critical to alignment research, since beliefs formed during pre-training shape how a model generalizes from later train→2 Oct 2026OmniSeek: Native Tool Integration for Multi-turn Audio-Visual ReasoningarXiv:2610.02181v1 Announce Type: new Abstract: We present OmniSeek, an agentic framework that transforms an Omni Large Language Model (Omni-LLM) into an active, multi-turn reasoning agent with native→2 Oct 2026Causal Memory Policy: Making Memory Utility Identifiable by Intervening on RetrievalarXiv:2610.02070v1 Announce Type: new Abstract: Memory-augmented large language models must decide which memories to retain, and recent systems do so by estimating each memory's effect on task perform→5 Oct 2026WebFovea: When the Model Is Right but the Click Is Wrong -- Reliable Round Trips for Vision-Based Web Agents on Live WebsitesarXiv:2610.03036v1 Announce Type: cross Abstract: We present WebFovea, a vision-based web agent that placed 2nd in the WebRetriever Challenge 2026 with a final score of 57.0 out of 100. The challenge →7 Oct 2026Stabilizing Off-Policy Training for Long-Horizon LLM Agent via Turn-Level Importance Sampling and Clipping-Triggered NormalizationarXiv:2511.20718v3 Announce Type: replace-cross Abstract: Reinforcement learning (RL) algorithms such as PPO and GRPO are widely used to train large language models (LLMs) for multi-turn agentic tasks→7 Oct 2026Natural Language Questions as an Interface for Knowledge Graphs: QRAKEN Graph Distillation and Semantic Self-HealingarXiv:2610.08095v1 Announce Type: new Abstract: Natural-language access to RDF knowledge graphs is a core Semantic Web ambition. Large language models (LLMs) have advanced Text-to-SPARQL, yet on unfam