Search: “routing”

Running Next Flash IQ3_XXS at ~70 tok/s with 100k context or 2 instances of Qwen 3.6 35B A3B IQ4 at ~145 tok/s with 256k all on $500 of ex mining BC-250 boards

This will be my third update on the bc-250 cluster and for my first forray into local ai I have been having a blast. This has turned from a weekend experiment into something I think I will use in the

11 Oct 2026

Content type

All Blog X Post Paper YouTube Reddit GitHub

Clear filters

vllm-ascend updates (GLM5.3-flash) on dual 310p Ascend cards

Alright, I learend a few valuable lessons since I posted about a week ago, and I've made more strides with the DDR4x Ascend 96GB 310P cards I purchased, so here we go, meet the new friendly and less v

11 Oct 2026

good framing from @ch3nweiii - a lot of agent steps are yes/no calls, not generation Jev answers those as typed questions with calibrated pr…

good framing from @ch3nweiii - a lot of agent steps are yes/no calls, not generation Jev answers those as typed questions with calibrated probabilities, so the frontier model sticks to the hard work w

agentsharrison-chase--x

10 Oct 2026

Tutorials

2DGS-Planner: Rasterization-based Path Planning in 2D Gaussian Splatting Map

arXiv:2610.11752v1 Announce Type: new Abstract: Gaussian splatting provides an explicit and efficiently rasterizable scene representation for robot navigation. However, individual Gaussian primitives

tutorialsarxiv-cs-ro

Tutorials

An Interpretable Approach to PDE Solution Discovery via Structural Experience Distillation

arXiv:2610.12003v1 Announce Type: new Abstract: PDE solution discovery aims to identify explicit symbolic expressions for unknown physical fields from observations under known physical constraints. Ex

tutorialsarxiv-cs-ai

C-LoRA: Continual Low-Rank Adaptation for Pre-trained Visual Models

arXiv: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

ContourVLA: A Closed-Loop Perception-Action Contour Policy for Generalized Referring Expression Segmentation

arXiv:2610.12107v1 Announce Type: new Abstract: Generalized referring expression segmentation (GRES) requires dynamically balancing high-level semantics for identifying a variable number of language-s

safetyarxiv-cs-cv

Controlled Acquisition and Abstention in Three-Channel Score Conflicts

arXiv:2610.10808v1 Announce Type: new Abstract: When audio, video, and text disagree, accuracy alone does not show whether to acquire another source or abstain. We study these choices in a controlled

Tutorials

DisFace3DNet: Explainable Facial Attractiveness Prediction via 3D Component Disentanglement

arXiv:2610.11656v1 Announce Type: new Abstract: Facial attractiveness prediction usually assigns one overall rating, leaving the roles of shape, appearance, and viewing conditions implicit. We propose

tutorialsarxiv-cs-cv

Local Ai

Finsler Flow Matching: Dynamics-Aware Geodesic Interpolation for Single-Snapshot Trajectory Inference

arXiv:2610.11318v1 Announce Type: new Abstract: Single-cell snapshot data can resolve a continuum of cellular states but do not uniquely determine the dynamics governing transitions between them. Howe

local-aiarxiv-cs-ai

From Reactive Containment to Proactive Assurance: Lessons from OpenAI, Anthropic, and Google Agent Security Incidents

arXiv:2610.12463v1 Announce Type: cross Abstract: In 2026, cybersecurity evaluations involving OpenAI, Anthropic, and Google agents reached real systems outside their authorized test scope. The paths

Humanize: Judgement Engineering for Agentic Coding

arXiv:2610.08900v2 Announce Type: replace Abstract: Agentic coding makes code generation cheap, but reliable completion remains difficult: the agent that writes the code is a weak judge of whether it

agentsarxiv-cs-ai

Look Back, Think Ahead: Visual Memory on Demand for Efficient Multimodal Reasoning

arXiv:2610.12060v1 Announce Type: new Abstract: Processing long visual token sequences from high-resolution images makes multi-step reasoning computationally expensive for multimodal Large Language Mo

researcharxiv-cs-cv

Memento 3: Model-Based Recursive Self-Improvement through Reflective Rulebooks

arXiv:2610.11794v1 Announce Type: new Abstract: Learning to act in unfamiliar environments requires agents to infer how the world works and revise that understanding as new evidence arrives. Yet limit

agentsarxiv-cs-ai

Memory Type Varies: Empowering LLM Agents for Long-Term Memory with Diverse Strategies

arXiv:2610.11573v1 Announce Type: new Abstract: The memory capabilities of Large Language Models (LLMs) have garnered increasing attention recently. Despite great success achieved, existing retrieval-

Mine Odyssey: Benchmarking Spatial Agentic Intelligence in the Wild

arXiv:2610.11328v1 Announce Type: new Abstract: Advances in foundation models are driving efforts to introduce agents to assist people in the physical world. Such agents require agentic spatial intell

My 'Anthropic/ChatGPT in a box' is... now open source and publicly hosted on Github.

It's been a long time coming, and I should have done this in the first place. But LumaBrowser is now open source. Since I first released it a year ago, it's had several thousand installs and an absolu

NavGPT-3: Harnessing Context in a Hierarchical Navigation Runtime

arXiv:2610.10787v1 Announce Type: cross Abstract: Language models trained with long-horizon agentic reinforcement learning can generalize knowledge through reasoning, express precise actions, and purs

safetyarxiv-cs-ai

One Block, Multiple Depths: Recurrent Vision Transformers with Depth-Programmed Experts

model-releasesarxiv-cs-cv

arXiv:2610.12448v1 Announce Type: new Abstract: In this work, we show that a single Transformer block, applied recurrently, can match the accuracy of a full-depth vision encoder at comparable inferenc

[Paper] EngramEdit: Decoupled Knowledge Updates in LLMs through Conditional Memory

Conditional memory architectures such as DeepSeek Engram use input n-grams to look up learned embeddings, expanding the capacity of large language models (LLMs) with limited additional computation. Be

PIVOT: Perplexity-Informed KD-to-RL Transition Scheduling for Vertical-Domain Few-Shot Distillation

arXiv:2610.11167v1 Announce Type: cross Abstract: Vertical-domain few-shot classification remains challenging for small language models, as limited supervision makes it difficult to acquire domain-spe

safetyarxiv-cs-ai

Predictive Multiplicity in Cell-Fate Assignment: Label-Free Rashomon Sets and the Limits of Per-Cell Certification

arXiv:2610.11185v1 Announce Type: new Abstract: Single-cell trajectory inference maps transcriptomic measurements onto developmental continua, yet configurations that fit the data equally well can ass

researcharxiv-cs-lg

Qwen 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

Recursive Self-Improvement through Multi-Agent Self-Supervision

arXiv:2610.12176v1 Announce Type: new Abstract: Recursive self-improvement (RSI) of a model on non-verifiable tasks, such as open-ended research, faces a supervision bottleneck when its outputs exceed

agentsarxiv-cs-ai

Rethinking Contrastive Loss in CLIP Post-training: A Complementary Framework with Frozen Text Encoder

arXiv:2610.11374v1 Announce Type: cross Abstract: CLIP serves as a foundational vision-language model and the de facto vision encoder for downstream VLMs such as LLaVA. Post-training offers a lightwei

Rounding in Preconditioner Space: Redesigning 4-bit AdamW Optimizer-State Quantization

arXiv:2610.12444v1 Announce Type: new Abstract: Quantizing AdamW's optimizer states reduces persistent storage, but quantization errors propagate through the moment recurrences and perturb subsequent

Local Ai

Scalable AI Uncertainty Quantification via Generalized Laplace Active Subspaces

arXiv:2610.11738v1 Announce Type: new Abstract: Reliable uncertainty quantification (UQ) is essential for deploying neural networks in scientific and high-stakes applications, but full Bayesian infere

local-aiarxiv-cs-ai

SlopSoup TV. A live, never-ending 24/7 pixel-art TV network inspired by old-school late night Adult Swim. No human makes any of it.

SlopSoup TV is a live, never-ending pixel-art TV network. Every script, character, voice, camera cut and schedule decision is made by models. It's made to be very absurd, dark and strange. Hardware: o

Syn-Omni: Structured Specialization and Progressive Collaboration for Omnimodal Embeddings

arXiv:2610.12256v1 Announce Type: cross Abstract: Omnimodal embeddings naturally involve both shared representations and modality-specific features across heterogeneous inputs. However, existing omnim

Task-Centric Personalized Federated Fine-Tuning of Language Models

arXiv:2604.00050v3 Announce Type: replace-cross Abstract: Federated Learning (FL) has emerged as a promising technique for training language models on distributed and private datasets of diverse tasks

Local Ai

The '10th Juror': Open-Set Standpoint Screening for Bureaucratic Bias Detection

arXiv:2610.11136v1 Announce Type: new Abstract: Presupposing the boundaries of bias is itself a form of bias. We study closed-loop bias governance for Dutch government documents, where a system must d

local-aiarxiv-cs-cl

The Polytopal Neural Network

arXiv:2610.12004v1 Announce Type: cross Abstract: Understanding how deep neural networks process information remains a central challenge. Existing interpretability methods often compromise structural

safetyarxiv-cs-ai

Applications

Timer-M1: A Multivariate Time Series Foundation Model via Learning Primitives

arXiv:2610.11734v1 Announce Type: cross Abstract: We introduce Timer-M1, a pretrained multivariate time series foundation model that learns with primitives for zero-shot forecasting. Across domains, t

applicationsarxiv-cs-ai

Towards Path-Creative Navigation: Robot Navigation through Embodied Interaction

arXiv:2610.11072v1 Announce Type: new Abstract: Autonomous navigation in cluttered and constrained environments typically assumes a fixed environment and searches only for paths within existing free s

agentsarxiv-cs-ro

TypedBench: A Benchmark for Calibration, Framing Sensitivity, and Cost in System One Decision Models

arXiv:2610.11392v1 Announce Type: new Abstract: System One models output calibrated probabilities over typed answers such as categorical choices, ordinal levels, or binary outcomes, via a non-generati

VINCIE-NExT: Unlocking Video Editing from Images via In-Context Modeling

arXiv:2610.12104v1 Announce Type: cross Abstract: Building a capable video editor remains significantly harder than a video generator: editing requires (source, instruction, edited) triplets that are

$2800 rig with 8x Radeon Pro V620 (256 GB VRAM) + custom vLLM fork = Qwen3.8-Flash-Next at 60 to 100 t/s decode and 3000+ t/s prefill

Post title is slightly misleading, I don't think you can get these for $350 each anymore but they're still pretty cheap all things considered. They're Radeon Pro V620's which are older RDNA2 enterpris

AutoAdapt: Automatic Domain Discovery Enables Low-Cost Extensibility

arXiv:2610.10349v1 Announce Type: new Abstract: Instruction-tuned models are deployed into environments where domains are heterogeneous and evolve, yet adding new domains or data typically requires co

Beyond Risk Prediction: Evidence Grounding and Psychosocial Factor Verification for Explainable Suicide Risk Assessment

arXiv:2610.08842v1 Announce Type: new Abstract: Identifying suicide risk from social networking services (SNS) posts is important for detecting suicide-related signals in online environments. However,

researcharxiv-cs-cl

Bounded Autonomy and Verifiable Safety for Agentic AI Enabled Automation

arXiv:2610.08815v1 Announce Type: new Abstract: Agentic AI-enabled automation cannot be safely deployed in high-stakes environments on probabilistic reasoning alone. A recurring risk is epistemic drif

Hardware

Building Reliable Data Analytics Agents: Lessons from the KDD Cup

The NVIDIA KGMON team secured second place in the KDD Cup 2026 Data Agents competition by building a constrained harness around a small fixed LLM, unifying heterogeneous data into an SQLite database w

hardwarenvidia-developer

CoTrace: Data Recipes for Training Terminal Agents with Harness-Model Co-Evolution

arXiv:2610.10426v1 Announce Type: new Abstract: Terminal-agent capability depends jointly on model weights and the runtime harness that formats prompts, binds tools, and handles error recovery. Existi

safetyarxiv-cs-cl

DeepTopoClustering: Unsupervised Derivation of Surface Process Taxonomy from 4D Point Clouds for Topographic Monitoring

arXiv:2610.09860v1 Announce Type: new Abstract: 4D point clouds acquired by permanent laser scanning (PLS) enable accurate high-frequency monitoring of surface change in dynamic topographic environmen

researcharxiv-cs-cv

Design of a Fully Actuated 4-DOF Robotic Finger With Joint-Specific Hybrid Remote Actuation

arXiv:2610.10180v1 Announce Type: new Abstract: This paper presents a fully actuated 4-DOF robotic finger using a joint-specific hybrid remote-actuation architecture. The metacarpophalangeal (MCP) joi

agentsarxiv-cs-ro

Design optimization of tendon-driven robots considering tendon wrapping and shortcut

arXiv:2610.09645v1 Announce Type: new Abstract: This study proposes a multi-objective optimization method for tendon-driven arm design, addressing the inherent trade-off between joint torque and arm t

researcharxiv-cs-ro

EASE: Entropy-Adaptive Distribution Shaping for Evading AI-generated Text Detectors

arXiv:2610.09976v1 Announce Type: new Abstract: AI-generated text (AIGT) detection can be sensitive to the decoding choices of the source large language model (LLM). We observe that perturbing next-to

researcharxiv-cs-cl

Empowering SMBs to do more with Gemini

model-releasesgoogle-cloud-ai

Today, we announced the new Gemini agent, your single, universal agent for work. For small and midsize businesses, with lean teams and margins, Gemini will be a critical tool for you to operate effici

EngramEdit: Decoupled Knowledge Updates in LLMs through Conditional Memory

model-releasesarxiv-cs-cl

arXiv:2610.10533v1 Announce Type: new Abstract: Conditional memory architectures such as DeepSeek Engram use input n-grams to look up learned embeddings, expanding the capacity of large language model