TechniqueRLHF / Alignment8 recent entries12 Oct 2026Tracing and Taming Hallucinated Terms via Subsequence-Association Filtering and Breadth-Aware Selection for LLM Query ExpansionarXiv:2610.13810v1 Announce Type: cross Abstract: Large language models (LLMs) are now a standard source of query expansion terms but they hallucinate. They inject non existent entities and off-topic →12 Oct 2026Surfing the Activation Wave: Analyzing the Information and Geometry of LLM Hidden States for Speculative DecodingarXiv:2610.12939v1 Announce Type: cross Abstract: Decoder hidden states from different layers of large language models (LLMs) often encode complementary information. However, existing speculative deco
TechniqueRAG8 recent entries7 Oct 2026Have I Seen Enough? Frozen Video-Language Models Encode Evidence ReadinessarXiv:2610.08560v1 Announce Type: cross Abstract: Streaming video-language models must decide not only what to answer, but whether the evidence needed for the current question has arrived. Existing sy→7 Oct 2026Foveated Compression: Selective High-Resolution Preservation for Token-Efficient VLMsarXiv:2610.07729v1 Announce Type: new Abstract: Visual tokens are a major source of inference cost in vision-language models, yet simple image downsampling remains a surprisingly strong compression ba→7 Oct 2026Evidence-Bound Reasoning: Neuro-Semantic Verification of Biomedical AI in Glioblastoma RadiogenomicsarXiv:2610.08660v1 Announce Type: new Abstract: Background: Biomedical AI can generate plausible explanations without reliably verifying whether each statement is supported by patient-specific evidenc→7 Oct 2026Even Small Reasoners Should Quote Their Sources: Introducing the Pleias-RAG Model FamilyarXiv:2504.18225v2 Announce Type: replace Abstract: We introduce a new generation of small reasoning models for RAG, search, and source summarization. Pleias-RAG-350m and Pleias-RAG-1B are mid-trained→7 Oct 2026Dynamic Budget Allocation for LLM Evaluation under Hard Resource ConstraintsarXiv:2610.07362v1 Announce Type: new Abstract: We evaluate large language models (LLMs) in multi-turn interactions through their time-to-event: the number of interaction steps required to produce an →8 Oct 2026I built a local server that speaks the Ollama API - sessions survive crashes, and your chat doesn't wait behind your agentsDisclosure first: I'm one of the maintainers, this is my project. If you've pointed Open WebUI or a script at a local server, you know the failure modes: the server restarts and every conversation is →12 Oct 2026MHR-Router: Multimodal Heterogeneous Retrieval-Enhanced Adaptive Routing for VLM InferencearXiv:2610.13278v1 Announce Type: new Abstract: Vision-language models (VLMs) exhibit complementary capabilities, yet selecting the most suitable expert for a given image-question query remains challe→12 Oct 2026CopRoute: Dependence-Aware Conformal Routing for Language ModelsarXiv:2610.13276v1 Announce Type: new Abstract: Set-based language model routing aims to increase the coverage of the event where at least one model answers a query correctly while controlling the num
TechniqueAgents8 recent entries12 Oct 2026From Benchmark to Bench: Can Agents Survive Real-World Drug Discovery?arXiv:2610.06411v2 Announce Type: replace Abstract: Agentic systems increasingly coordinate molecular-design tools, but it is unclear which layer of the stack limits outcomes on real projects. We deve→12 Oct 2026Does Knowledge About Perceptual Uncertainty Help an Agent in Automated Driving?arXiv:2502.11864v2 Announce Type: replace-cross Abstract: Agents in real-world scenarios like automated driving deal with uncertainty in their environment, in particular due to perceptual uncertainty.→12 Oct 2026Constructing Pareto Frontiers of Optimal Reasoning Termination Methods for Overthinking in Reasoning ModelsarXiv:2610.13570v1 Announce Type: new Abstract: Reasoning models can benefit from test-time scaling, but continued reasoning may become ineffective or harmful, a phenomenon termed overthinking. Evalua→12 Oct 2026Competence-Aware Orchestration for LLM-Based Multi-Agent SystemsarXiv:2610.13144v1 Announce Type: new Abstract: LLM-based multi-agent systems (MAS) have emerged as a promising paradigm for enhancing complex reasoning through role specialization and collaboration, →12 Oct 2026Collective Autonomous Vehicle Routing Needs Social Awareness for Coexistence with Human DriversarXiv:2610.13439v1 Announce Type: cross Abstract: Collective Autonomous Vehicle (AV) route choices in real-world traffic networks can create externalities for Human-Driven Vehicles (HDVs), as routes t→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→12 Oct 2026Branching On-Policy DistillationarXiv:2610.13481v1 Announce Type: new Abstract: On-policy distillation (OPD) provides teacher supervision on student-generated trajectories, but a single trajectory realizes only one continuation at e→12 Oct 2026Agentic Peer-to-Peer Networks: From Content Distribution to Capability and Action SharingarXiv:2603.03753v2 Announce Type: replace-cross Abstract: The ongoing shift of Artificial Intelligence (AI) from centralized cloud application programming interfaces (APIs) to local inference on edge
TechniqueFine-tuning8 recent entries9 Oct 2026Qwen 3.8 Flash Next-GSQ-RCO-IQ2_XS at ~21 tok/s on just an RTX 3060 12GB + 16GB DDR4 RAM(No gate pruning, 100% bit-exact)(Don't judge by the screenshot, the cache is cold. It hits 24+ tok/s with a warm cache!) About two months ago, I made a post here asking whether predicting which MoE experts would be used on the next →9 Oct 2026DivMoE: Fine-Grained MoE Upcycling via Cross-Domain Expert CompositionarXiv:2610.11317v1 Announce Type: new Abstract: Mixture-of-Experts (MoE) architectures have become essential for scaling large language models, with recent work demonstrating the benefits of fine-grai→9 Oct 2026C-LoRA: Continual Low-Rank Adaptation for Pre-trained Visual ModelsarXiv:2502.17920v2 Announce Type: replace Abstract: Pre-trained visual models have become fundamental in computer vision, but they face challenges in continual learning scenarios where data and tasks →11 Oct 2026New paper from the team at Sakana AI: Recursive Self-Improvement through Multi-Agent Self-Supervision (MASS) How do we keep improving AI whe…New paper from the team at Sakana AI: Recursive Self-Improvement through Multi-Agent Self-Supervision (MASS) How do we keep improving AI when the tasks become too complex for humans to reliably evalua→12 Oct 2026QLoRA on ExLlamaV3 Quants - Train @ 2-8bpwEXL3-QLoRA I've built this over the last few months and I think it's stable enough to share now. A fork of ExLlamaV3, this is a native ExLlamaV3 training implementation with plain PyTorch, without tra→12 Oct 2026Not All Dimensions Are Equal: Outlier-Aware Attention Linearization with Grouped Quadratic KernelsarXiv:2610.13674v1 Announce Type: cross Abstract: Linearizing softmax attention offers a route to efficient Transformer inference, but preserving pretrained attention remains challenging. We identify →12 Oct 2026JoyInside-VLM-Agent: Product-Conditioned Multimodal Policy Learning for Embodied AgentsarXiv:2610.12957v1 Announce Type: cross Abstract: Embodied agents must translate multimodal observations and user requests into executable actions while adapting to product constraints, changing goals→12 Oct 2026Hindsight Structured Advice Distillation: Steering Executor Agents with Advisors Trained on Their Failures and Task CharacteristicsarXiv:2610.13000v1 Announce Type: new Abstract: Language-model agents are used for specialized work such as financial research, healthcare and law. Such agents are often capable enough for a task but
TechniqueMultimodal8 recent entries12 Oct 2026Neural networks are more modular than single neurons suggestarXiv:2610.12842v1 Announce Type: new Abstract: Neural networks routinely develop representations in which individual neurons contribute to multiple distinct functions, leading to the widespread view →12 Oct 2026MHR-Router: Multimodal Heterogeneous Retrieval-Enhanced Adaptive Routing for VLM InferencearXiv:2610.13278v1 Announce Type: new Abstract: Vision-language models (VLMs) exhibit complementary capabilities, yet selecting the most suitable expert for a given image-question query remains challe→12 Oct 2026JoyInside-VLM-Agent: Product-Conditioned Multimodal Policy Learning for Embodied AgentsarXiv:2610.12957v1 Announce Type: cross Abstract: Embodied agents must translate multimodal observations and user requests into executable actions while adapting to product constraints, changing goals→12 Oct 2026Hindsight Structured Advice Distillation: Steering Executor Agents with Advisors Trained on Their Failures and Task CharacteristicsarXiv:2610.13000v1 Announce Type: new Abstract: Language-model agents are used for specialized work such as financial research, healthcare and law. Such agents are often capable enough for a task but →12 Oct 2026GeoGR: Spatial-Semantic Token Compression and Group Routing for Efficient 3D VLMsarXiv:2610.13035v1 Announce Type: cross Abstract: Multi-view representations enable pretrained image and video vision-language models to understand 3D scenes, but their long visual token sequences mak→12 Oct 2026Exploring Attention Sinks in Text-to-Image Transformers from Diffusion and Autoregressive ModelsarXiv:2610.13337v1 Announce Type: cross Abstract: Attention sinks are tokens or positions that consistently receive disproportionate attention mass, often beyond what their semantic content would sugg→12 Oct 2026Collective Autonomous Vehicle Routing Needs Social Awareness for Coexistence with Human DriversarXiv:2610.13439v1 Announce Type: cross Abstract: Collective Autonomous Vehicle (AV) route choices in real-world traffic networks can create externalities for Human-Driven Vehicles (HDVs), as routes t→12 Oct 2026Agentic Peer-to-Peer Networks: From Content Distribution to Capability and Action SharingarXiv:2603.03753v2 Announce Type: replace-cross Abstract: The ongoing shift of Artificial Intelligence (AI) from centralized cloud application programming interfaces (APIs) to local inference on edge
TechniqueSafety8 recent entries12 Oct 2026Playing the network backward: A Game Theoretic Attribution FrameworkarXiv:2605.06212v2 Announce Type: replace Abstract: Attribution methods explain which input features drive a model's prediction, making them central to model debugging and mechanistic interpretability→12 Oct 2026KITE: A Knowledge-Injected Temporal Encoder for Time-Series Event ForecastingarXiv:2610.12738v1 Announce Type: cross Abstract: Forecasting rare, high-cost events in multivariate time-series systems is hard because their precursors are seldom observed, so a purely data-driven m→12 Oct 2026JoyInside-VLM-Agent: Product-Conditioned Multimodal Policy Learning for Embodied AgentsarXiv:2610.12957v1 Announce Type: cross Abstract: Embodied agents must translate multimodal observations and user requests into executable actions while adapting to product constraints, changing goals→12 Oct 2026DistressBench: Evaluating Multi-Dimensional Crisis Support in Large Language ModelsarXiv:2610.12758v1 Announce Type: new Abstract: Language models have been adopted faster than evaluation practices that reliably assess their risks, and a growing share of use is affective rather than→12 Oct 2026CopRoute: Dependence-Aware Conformal Routing for Language ModelsarXiv:2610.13276v1 Announce Type: new Abstract: Set-based language model routing aims to increase the coverage of the event where at least one model answers a query correctly while controlling the num→12 Oct 2026Branching On-Policy DistillationarXiv:2610.13481v1 Announce Type: new Abstract: On-policy distillation (OPD) provides teacher supervision on student-generated trajectories, but a single trajectory realizes only one continuation at e→12 Oct 2026Beyond Domain-Routed Supervision: Cross-Teacher Guarding for Multi-Teacher On-Policy DistillationarXiv:2610.13521v1 Announce Type: new Abstract: Multi-teacher on-policy distillation (MOPD) extends on-policy distillation (OPD) by routing domain-specific data to specialized teachers, integrating th→12 Oct 2026Attention Sees, MLP Refuses: Tracing the Attention-to-MLP Safety Pathway in LLMsarXiv:2610.13078v1 Announce Type: new Abstract: Safety-aligned large language models (LLMs) remain vulnerable to jailbreaks and can over-refuse benign requests. Understanding these failures requires d