Search: “routing”

The Teacher Is a Direction, Not a Destination: Extrapolating RL-Induced Representation Residuals in On-Policy Distillation

arXiv:2609.36484v1 Announce Type: new Abstract: On-policy distillation (OPD) trains a student to match the teacher's next-token distributions on the student's own trajectories and has yielded substant

30 Sept 2026

Content type

All Blog X Post Paper YouTube Reddit GitHub

Clear filters

Hardware

Tracing Agent Harness Behavior with NVIDIA NeMo Relay

DGX agent

NVIDIA NeMo Relay supplies a common observability layer that captures ordered lifecycle events and structured trajectories—ATOF event streams, ATIF step‑by‑step footprints, and OpenTelemetry spans wit

hardwarenvidia-developer

30 Sept 2026

Beyond Teacher Assignment: Domain-Normalized Multi-Teacher On-Policy Distillation

arXiv:2609.35347v1 Announce Type: new Abstract: Reinforcement learning can turn one language model into several specialists, each excellent at a single skill such as mathematics, coding or following i

ChemOPD: Multi-Teacher On-Policy Distillation for Multi-Task Chemical Reasoning

arXiv:2609.33838v1 Announce Type: new Abstract: Large language models are increasingly expected to support diverse chemical reasoning capabilities within a unified model. One approach is to develop sp

Copy the Same, Distill the Difference: Initializing Linear Vision Transformers

arXiv:2609.35745v1 Announce Type: new Abstract: Linear Vision Transformers (ViTs) are designed to replace the attention in Softmax ViTs with the linear-complexity attention operator for more efficient

tutorialsarxiv-cs-cv

Do Latent-CoT Models Think Step-by-Step? A Mechanistic Study on Sequential Reasoning Tasks

arXiv:2602.00449v2 Announce Type: replace-cross Abstract: Latent Chain-of-Thought aims to enable step-by-step computation without emitting long rationales, yet its internal mechanisms remain unclear.

tutorialsarxiv-cs-lg

DPAMixerSR: An Efficient Degradation-Pattern-Aware Model for Image Super-Resolution

arXiv:2609.32705v1 Announce Type: new Abstract: While content-adaptive schemes have delivered notable advances in image super-resolution (SR), existing approaches typically focus on texture complexity

researcharxiv-cs-cv

MemTransfer: Benchmarking Memory Beyond Matched Experience in Embodied Decision-Making

model-releasesarxiv-cs-ro

arXiv:2609.32313v1 Announce Type: new Abstract: Memory lets an embodied agent reuse past experience, yet retaining useful information does not ensure that the agent can apply it when conditions change

MM-VeriRec: Failure-Guided Fusion for Verifiable Agentic Multimodal Recommendation

arXiv:2609.31718v1 Announce Type: cross Abstract: Images often carry recommendation constraints that text metadata only hints at. A movie may need to look dark, a product may need a minimal style, and

[Release] GSQ-RCO GGUFs for Qwen3.8-Flash-Next, plus a 50% expert-pruned Coder build at ~1.89 bpw

model-releasesr-localllama

We have released Qwen3.8-Flash-Next quantized with GSQ and RCO, together with a second, capability-targeted build in which half of the model's experts have been removed. Flash-Next is a sparse mixture

ReSight-SMC: Two-Stage Power Sampling via Island SMC with Visual Scouts

arXiv:2609.34905v1 Announce Type: new Abstract: Power sampling has emerged as a training-free approach to LLM reasoning, eliciting capabilities comparable to reinforcement learning by sharpening the m

Resource-Aware Federated Mixture-of-Experts with Adaptive Pruning for Onboard Learning in LEO Satellite Constellations

arXiv:2609.31932v1 Announce Type: cross Abstract: Low-Earth-orbit (LEO) satellites are increasingly expected to perform onboard learning for applications such as disaster response and environmental mo

local-aiarxiv-cs-cv

Hardware

SparSP: Exploiting Communication Sparsity for Sequence-Parallel Video DiTs

arXiv:2609.32197v1 Announce Type: cross Abstract: Diffusion Transformers have become the dominant architecture for video generation. Their substantial computational cost motivates scaling inference ac

hardwarearxiv-cs-lg

Applications

Task-Oriented Communications for Edge-Assisted Multi-View Localization

arXiv:2609.35173v1 Announce Type: cross Abstract: Unmanned aerial vehicles (UAVs) and unmanned ground vehicles (UGVs) often lose satellite positioning in urban canyons, indoor facilities, and jammed o

applicationsarxiv-cs-ro

Temporal-Attention Head Specialization During Video Diffusion Training

arXiv:2609.31654v1 Announce Type: new Abstract: Video diffusion transformers depend on temporal attention to coordinate information across frames, yet nearly everything known about this mechanism come

researcharxiv-cs-cv

Token-Disentangled Latent Test-Time Scaling for Vision-Language Reasoning

arXiv:2609.35228v1 Announce Type: new Abstract: Latent test-time scaling improves reasoning by refining hidden states during inference, but existing methods typically apply a single scalar reward to a

Applications

Using the biggest model for every task can work in a demo. In production, cost, latency, and task-specific quality can tell a different stor…

applicationsfireworks-ai--x

Using the biggest model for every task can work in a demo. In production, cost, latency, and task-specific quality can tell a different story. At ODSC AI West 2026, Rob Ferguson, VP of Technology and

What Should Data Teach? Moving Bottlenecks Across Circuit, Store, and Use

arXiv:2609.32991v1 Announce Type: new Abstract: What should data teach a language model at a particular point in training? A circuit view reveals three distinct bottlenecks: forming a computation, mak

local-aiarxiv-cs-lg

ConsultMind:Towards Automated Diagnostic Consultation via Uncertainty-Aware Reasoning

arXiv:2609.30796v1 Announce Type: new Abstract: Diagnostic consultation is an online sequential decision-making process in which clinicians gather evidence through patient interaction until a diagnosi

Cross-Backend QIEO: Universal Runtime Portability across OpenMP5, CUDA, HIP, and Multi-Language Interfaces

model-releasesarxiv-cs-cl

arXiv:2609.30914v1 Announce Type: cross Abstract: Quantum-inspired algorithms emulate quantum mechanical principles, such as, superposition, interference, and probabilistic amplitude evolution, on cla

In-Context Binding Capacity in Language Models

arXiv:2609.30634v1 Announce Type: new Abstract: How many assignments can a language model recall before it loses track of which value belongs to which entity? We measure this limit using continuous re

Intent2Tc: Automated Intent-to-Traffic Control Translation with Language Models

arXiv:2609.31397v1 Announce Type: cross Abstract: Automated and highly usable Quality-of-Service (QoS) enforcement requires translating high-level service intents into deployable traffic-management po

LLM Parkinsonism: Executive-Control Failure, Token-Inefficient Persistence, and an Uncertainty-Aware Global Executive Control Architecture for Autonomous Language-Model Agents

arXiv:2609.30662v1 Announce Type: new Abstract: Large language models (LLMs) can plan, use tools, write code, and execute long-horizon workflows, yet strong local competence does not guarantee project

Muslim: A Deployed Arabic Voice AI Platform for Grounded Islamic Knowledge

model-releasesarxiv-cs-cl

arXiv:2609.31511v1 Announce Type: new Abstract: We present Muslim, a production Arabic voice AI platform serving grounded, sourced Islamic knowledge to real users. Beyond a real-time voice pipeline (N

MVVBench: Benchmarking 4D Reasoning in Vision-Language Models

arXiv:2609.30952v1 Announce Type: cross Abstract: Multi-view video understanding requires integrating spatial and temporal evidence across multiple, often non-overlapping camera streams: tracking enti

New LoRA Skills Should Read but Never Write

model-releasesarxiv-cs-lg

arXiv:2609.31600v1 Announce Type: new Abstract: Low-rank adapters (LoRA) make it cheap to fine-tune a large language model once per task, but combining several independently trained adapters into one

Nice Fold or Hero Call: Learning Budget-Efficient Thinking under Policy-Dependent Solvability

arXiv:2605.11625v2 Announce Type: replace Abstract: Large reasoning models (LRMs) improve problem solving through extended reasoning, but often misallocate test-time compute. Existing efficiency metho

safetyarxiv-cs-ai

SatNav: A Scalable Benchmark for Long-Horizon UAV Vision-Language Navigation from Satellite Imagery

arXiv:2609.31507v1 Announce Type: new Abstract: Urban uncrewed aerial vehicle (UAV) vision-language navigation (VLN) requires agents to follow instructions across extended urban spaces, inherently dem

The Linear Representation Hypothesis for Vision-Language-Action Models

arXiv:2609.30996v1 Announce Type: cross Abstract: The linear representation hypothesis (LRH) has become a standard lens for measuring and intervening on semantic information through the internal repre

TrafficImag: A Benchmark for Counterfactual Roadside Traffic Video Generation

arXiv:2609.30722v1 Announce Type: cross Abstract: Existing roadside traffic datasets support perception, forecasting, and visual question answering, but they do not evaluate counterfactual video gener

Syntheses

Wiki Lint Report — 2026-09-27

linthealth-checkautomated

Automated lint: 17 errors, 59669 warnings, 3 info

27 Sept 2026

A Field-Deployable GNSS-based Navigation Stack for Outdoor Mobile Robots

arXiv:2609.28933v1 Announce Type: new Abstract: Outdoor robots require more than an accurate receiver and a path-tracking law: the navigation system must preserve geometric consistency from geographic

local-aiarxiv-cs-ro

CataOPD: Catalytic On-Policy Distillation for Large Language Model Reasoning

arXiv:2609.29518v1 Announce Type: new Abstract: Reinforcement learning (RL) and on-policy distillation (OPD) are two representative paradigms for improving large language model reasoning. However, whe

Qwengram-0.8B: I transferred Qwen3.8 Flash-Next’s n-gram memory into Qwen3.5-0.8B — 5.05% lower validation perplexity

model-releasesr-localllama

I’ve been experimenting with whether Qwen3.8-Flash-Next’s pretrained PLE n-gram memory can improve a much smaller Qwen3.5-0.8B model. I trained the 0.8B setup with limited resources, mostly using free

Requirement-Bound Verified Commissioning: A Frozen Four-Billion-Parameter Local Model as a Candidate Generator under an External Acceptance Layer with Verification and Release Authority

arXiv:2609.30219v1 Announce Type: cross Abstract: An acceptance protocol is developed for sensor-coordinate and polarity binding in mechatronic commissioning. Candidate generation is separated from re

Task-Aware Spectral Pruning: A Mixture-of-Masks Framework for Efficient LLM Inference

model-releasesarxiv-cs-lg

arXiv:2609.29499v1 Announce Type: new Abstract: Static pruning imposes one sparse structure on every prompt, even though reasoning, retrieval, generation, coding, and translation can depend on differe

Where Hallucinations Live: A Cross-Architecture Circuit in VQ-Tokenized Vision-Language Models

arXiv:2609.29048v1 Announce Type: new Abstract: Unified vision-language models (VLMs) that tokenize images through a vector-quantized (VQ) codebook routinely hallucinate objects on grounded yes/no ben

researcharxiv-cs-cv

Applications

Brain-to-Language Decoding: Tasks, Signals, Methods, Evaluation, Practical Use and Beyond

arXiv:2609.27650v1 Announce Type: new Abstract: Brain-to-language decoding translates neural activity associated with language production, internal speech and perception into linguistic or expressive

applicationsarxiv-cs-cl

DRSR: Learning Set-Level Deletion Risk for Efficient Long-Horizon Agents

arXiv:2609.27276v1 Announce Type: new Abstract: Long-horizon language-model agents accumulate reasoning traces, tool exchanges, and observations whose relevance changes with the current decision. Exis

safetyarxiv-cs-ai

Everytime AI spend grows where I flirt with token budgeting, Anthropic + OpenAI save me Astra had a spike then settled down lower Opus 5.5 i…

Everytime AI spend grows where I flirt with token budgeting, Anthropic + OpenAI save me Astra had a spike then settled down lower Opus 5.5 it looks like spend is lowered for current usecases Of course

tutorialsdylan-patel--x

How Google Cloud Networking Supports Your Fluid Compute Choices for AI Workloads

model-releasesgoogle-cloud-ai

The availability of resources for AI workloads can be challenging across the industry, especially accelerators. This can slow your AI workload deployment if it’s built around a specific type of accele

Pay attention to this new wave of System One models if you are building custom harnesses. First Jev. Now, Contrastive Language Model (CLM). …

Pay attention to this new wave of System One models if you are building custom harnesses. First Jev. Now, Contrastive Language Model (CLM). CLM is 9x faster than Jev. CLM seems to be a better verifier

tutorialsdair-ai--x

RoadOcc Learns When to Persist, Transport, or Refresh Memory for Roadside Occupancy Prediction

arXiv:2609.27677v1 Announce Type: new Abstract: Fixed roadside cameras repeatedly observe a stable scene overlaid by sparse moving traffic. Temporal memory can recover weak observations, but reusing m

Learning to Defer with Guidance on Real World Medical Data

arXiv:2609.26384v1 Announce Type: new Abstract: Medical image interpretation is high-volume and time-consuming, and while AI interpretation can reduce workload, fully autonomous deployment carries pot

23 Sept 2026

Agents

Real-time autonomous magnetic microrobot navigation across dynamic and biologically relevant environments

arXiv:2509.05500v2 Announce Type: replace Abstract: Autonomous microrobots could enable minimally invasive interventions in confined biological environments, but their operation requires real-time nav

agentsarxiv-cs-ro

23 Sept 2026

3D-MoE: Towards Spatial Intelligence with Mixture-of-Experts for 3D Reasoning and Action Generation

arXiv:2501.16698v2 Announce Type: replace Abstract: Spatial intelligence, encompassing 3D perception and reasoning, is the essential next frontier of AI. Scaling current 3D vision-language models (VLM

researcharxiv-cs-cl

22 Sept 2026

ARID: A Deployable Edge AI System for Structured Information Extraction from Industrial Maintenance Work Orders

arXiv:2609.23582v1 Announce Type: new Abstract: Maintenance work orders must often be processed offline on embedded hardware, yet downstream software requires predictable structured output. We present

local-aiarxiv-cs-cl

22 Sept 2026

Beyond Final-Token Classification: Heterogeneous Readouts for Evidence-Grounded Suicide Risk Detection