Search: “evaluation”

Condition-Contrastive Extrapolation for On-Policy Distillation of Language Models

arXiv:2610.21195v1 Announce Type: new Abstract: On-policy distillation (OPD) improves a student language model by training on its own generated trajectories with supervision from a stronger teacher. H

Content type

All Blog X Post Paper YouTube Reddit GitHub

Clear filters

CoTranslator: Reverse to Forward Chain-of-Thought Translator for Reasoning Distillation

DGX agent

arXiv:2610.20828v1 Announce Type: cross Abstract: Reasoning distillation requires chain-of-thought (CoT) data with correct answers, but obtaining such data efficiently remains challenging. Forward CoT

researcharxiv-cs-ai

16 Oct 2026

Counterfactual Neighborhood Risk Fields: Evidence-Grounded Selective Safety for Vision-Language Models

arXiv:2610.20679v1 Announce Type: new Abstract: Representation-based safety defenses for vision-language models (VLMs) often learn to separate harmful from benign inputs, but separation alone does not

CoViG: Coarsened Vision GNN for Image Classification

arXiv:2610.21321v1 Announce Type: new Abstract: Although Graph Neural Networks are widely used in various domains, they remained mostly unexplored in computer vision. Convolutional Neural Networks and

Cross-Prompt Inference Mitigates Rare-Class Suppression in Imbalanced VLM Finetuning

arXiv:2610.21470v1 Announce Type: new Abstract: When a vision-language model is finetuned on class-imbalanced data, rare-class recall can fall well below the model's own zero-shot performance. In this

Data Strategies for Large Language Models: A Unified Perspective on the Training Lifecycle

arXiv:2610.21681v1 Announce Type: new Abstract: Data strategy plays a central role in training large language models (LLMs). It governs the selection, weighting, and scheduling of training data, there

researcharxiv-cs-cl

Day 0 support for Reflection’s Beam on Google Cloud: 5x faster perf with llm-d and NVIDIA Dynamo

model-releasesgoogle-cloud-ai

Today, Reflection released Beam, its first open-weight model built to give developers, researchers, and enterprises open control and customization for frontier software engineering and long-horizon ag

Decentralized LLM Workflows via Sequential Auctions

arXiv:2610.22201v1 Announce Type: cross Abstract: We present a scalable framework for solving multi-step tasks through a decentralized market of self-interested agents. Prior theoretical work demonstr

Defining and Generating Compositional Structure

arXiv:2610.20922v1 Announce Type: new Abstract: Compositional structure is pervasive in nature and across scales: atoms combine into molecules, molecules into cells, and cells into organisms, while fu

Dense4D: Pixel-Level 4D Human Correspondence

arXiv:2610.22325v1 Announce Type: new Abstract: Reconstructing human motion in 4D requires identifying the same surface points through articulation, deformation and changes in visibility. We introduce

DIAL: Dynamic and Informed Adjustment of the Native Stochastic Law for Reliable Weather Ensembles

arXiv:2610.21639v1 Announce Type: new Abstract: Ensemble weather forecasts can look reliable in global averages while hiding too little spread in some cases and too much in others. These errors vary a

Differentially Private Distribution Release of Gaussian Mixture Models via KL-Divergence Minimization

model-releasesarxiv-cs-lg

arXiv:2506.03467v4 Announce Type: replace-cross Abstract: Gaussian Mixture Models (GMMs) are widely used statistical models for representing multi-modal data distributions, with numerous applications

Digital Typhoon Dataset V3: Benchmarking for Multi-spectral Extension and Analyzing Systematic Bias in Intensity Regression

model-releasesarxiv-cs-cv

arXiv:2610.21654v1 Announce Type: new Abstract: This paper presents the Digital Typhoon Dataset V3, a new version of the longest typhoon satellite image dataset spanning nearly 50 years, designed to b

Dimension-Free PAC-Bayes Generalization for Deep Neural Networks via Dirichlet Processes

model-releasesarxiv-cs-lg

arXiv:2610.21340v1 Announce Type: cross Abstract: Classical PAC-Bayesian generalization theory places prior distributions over Euclidean parameter space Theta subseteq R^D, causing relative entropy to

Diminishing Returns of Data in Sparse Autoencoders: Functional Behavior does not Guarantee Feature Reproducibility

model-releasesarxiv-cs-lg

arXiv:2610.20680v1 Announce Type: new Abstract: Sparse autoencoders (SAEs) are typically trained on hundreds of millions to billions of activation tokens, yet little work has tested whether this scale

Agents

Diviner: Agentic RCA with Human-Governed Adaptation for Large-Scale Infrastructure

arXiv:2610.21281v1 Announce Type: new Abstract: Root-cause analysis (RCA) in large-scale distributed systems is challenging because a single symptom can cross service, data, configuration, compute, an

agentsarxiv-cs-ai

DLOSAC: Joint Spatial-Geometric Image Estimation via Delaunay-Guided Local Optimization RANSAC

arXiv:2610.21215v1 Announce Type: new Abstract: Reliable image geometric estimation from noisy matching points is typically addressed using RANSAC, often combined with local optimization to refine inl

local-aiarxiv-cs-cv

Do Agents Need a New Kind of Paper? SPRA: Structured Paper Representation for Agents

arXiv:2610.21424v1 Announce Type: new Abstract: Scientific agents must connect research problems, methods, and findings dispersed through narrative articles. We introduce SPRA (Structured Paper Repres

Does On-Policy Distillation Need to Complete Every Rollout?

arXiv:2610.21337v1 Announce Type: cross Abstract: On-policy distillation (OPD) effectively transfers capabilities from a teacher to a student through feedback on student-generated rollouts. However, g

Does Personalization Break Trust? Open-Set Rejection Calibration Under Few-Shot On-Device Intent Adaptation

arXiv:2610.20652v1 Announce Type: new Abstract: Always-on virtual assistants increasingly personalize on-device to new user-specific commands via few-shot fine-tuning (e.g., LoRA), but this adaptation

local-aiarxiv-cs-cl

DropKV: Decoupling Residual-Output Perturbation for Near-Optimal KV-Cache Eviction

arXiv:2610.22130v1 Announce Type: new Abstract: Key-value (KV) cache management has become a critical bottleneck for scaling Large Language Models (LLMs) to long-context scenarios due to linear memory

DuET: Note-Frame Dual Encoding for Piano Performance-to-Score Transcription

arXiv:2610.21467v1 Announce Type: cross Abstract: Transcribing expressive piano performances into scores requires interpreting individual note timings within an underlying metrical structure. However,

Efficient Pairwise Difference Learning: A Comparative Study of Encoding Strategies and Training Pair Selection

model-releasesarxiv-cs-ai

arXiv:2610.21644v1 Announce Type: cross Abstract: Pairwise Difference Learning for Classification (PDC) enhances supervised learning in low-data regimes by training a binary similarity classifier on p

Applications

EgoJEPA: Beyond Retargeting for Dexterous Manipulation from Egocentric Videos

arXiv:2610.20904v1 Announce Type: cross Abstract: Egocentric human videos offer a scalable alternative to costly robot demonstra- tions, yet transferring their manipulation priors to robot control rem

applicationsarxiv-cs-cv

Emotion Is Not External Knowledge: Internal Attention Signals for Affective State Reasoning

arXiv:2610.21347v1 Announce Type: new Abstract: Large Language Models (LLMs) increasingly operate in multi-turn interactions that require maintaining latent conversational states, not only explicit fa

local-aiarxiv-cs-cl

Encoded but Exposed: Language Embeddings can Reveal More Than They Should in Recommender Systems

arXiv:2610.21713v1 Announce Type: cross Abstract: Language embeddings are increasingly used in recommenders to represent item content, but their privacy risks remain poorly understood. We study whethe

Enhancing Multimodal RLVR with Pairwise Token Contrast and Adaptive Trajectory Confidence

arXiv:2610.21671v1 Announce Type: new Abstract: Reinforcement learning with verifiable rewards (RLVR) improves multimodal reasoning, but group-relative policy optimization broadcasts a single sequence

safetyarxiv-cs-cl

Entertain, then Cancel: How Thinking Makes Reasoning Models Read Some and Or Literally

model-releasesarxiv-cs-ai

arXiv:2610.21204v1 Announce Type: new Abstract: When a speaker says ``some students passed'', listeners infer that not all of them did. Reasoning models now let users decide how long a model thinks be

ESG-Kor: A Korean Dataset for ESG-related Information Extraction and Practical Use Cases

arXiv:2610.21368v1 Announce Type: cross Abstract: With the expansion of pre-trained language model usage in recent years, the importance of datasets for performing tasks in specialized domains has sig

ETH: Data-Balanced Heterogeneous Pipeline Parallelism for Training Large Models

arXiv:2610.22035v1 Announce Type: cross Abstract: Current heterogeneity-aware training systems for large models balance workloads by assigning unequal numbers of layers across pipeline stages, a strat

EVA-Cap: Optimizing Audiovisual Video Captioning via Event-Centric Alignment

arXiv:2610.22186v1 Announce Type: cross Abstract: Recent advances in Omni Language Models (OLMs) provide a unified solution for audiovisual video captioning. Nevertheless, existing approaches predomin

safetyarxiv-cs-cl

Evo-DexBench: Benchmarking Generalization across Related Tasks for Dexterous Manipulation

model-releasesarxiv-cs-ro

arXiv:2610.21180v1 Announce Type: new Abstract: General-purpose dexterous manipulation requires policies to transfer knowledge from related tasks to new tasks with only a few target demonstrations. Ex

Failure Is the Mother of Success: Recovering Supervision with Grounding Hindsight

arXiv:2610.21018v1 Announce Type: new Abstract: When every rollout for a problem fails, GRPO receives neither a successful exemplar nor a nonzero group-relative advantage, so it learns nothing from ex

Fast-VLA: You Only Think Once for Robot Manipulation

model-releasesarxiv-cs-ro

arXiv:2610.21389v1 Announce Type: new Abstract: Vision-language-action (VLA) policies commonly place a multi-billion-parameter model inside the feedback loop, repeatedly recomputing task context whene

FastVLA: Task-Conditioned and Temporally Adaptive Visual Token Compression for Efficient Vision-Language-Action Models

arXiv:2610.22146v1 Announce Type: new Abstract: Vision-Language-Action (VLA) models present a promising paradigm for general-purpose robot manipulation, but closed-loop deployment is often limited by

researcharxiv-cs-ro

Fault Detection and Bit-Error Rate Diagnostics in Optical Wireless Links Using Edge-AI

arXiv:2610.21402v1 Announce Type: new Abstract: Optical Wireless Communication (OWC) systems provide high-bandwidth, spatially secure, and RF-interference-free digital transmission. However, direct fr

local-aiarxiv-cs-lg

Federated Art Chase: Privacy Leakage and Data Poisoning in Federated Fine-Tuning

arXiv:2610.21547v1 Announce Type: new Abstract: Federated learning (FL) is increasingly used to adapt a pretrained, centrally fine-tuned model to data that remain on user devices, beyond training a mo

FGSS-BO: Feedback-Guided Hypothesis Management for LLM-Based Bayesian Optimization

model-releasesarxiv-cs-cl

arXiv:2610.21246v1 Announce Type: new Abstract: Large language model (LLM)-guided Bayesian optimization (BO) uses past observations to guide search, but proposed directions and their revision can rema

FocusOPD: Structure-Guided On-Policy Distillation for Diffusion Models

model-releasesarxiv-cs-cv

arXiv:2610.21250v1 Announce Type: new Abstract: Multi-teacher on-policy distillation enables rectified-flow models with diffusion-transformer architectures to combine task-specific capabilities within

Forecast lead time and the timing of crew deployment for weather-driven power outages using AI weather models

arXiv:2610.20918v1 Announce Type: cross Abstract: Utilities request mutual aid crews to restore power after severe storms, and requesting early secures more crews but relies on less accurate forecasts

From Long Context to Selective Memory:A Survey of Memory-Centric Compression for Large Language Models

model-releasesarxiv-cs-cl

arXiv:2610.21165v1 Announce Type: new Abstract: Long-context capability has become essential for large language models (LLMs) to handle complex tasks, yet continuously expanding the context window doe

Tutorials

Full-Duplex Interaction with Speech, Expressions and Gestures

arXiv:2610.21452v1 Announce Type: cross Abstract: Human face-to-face conversation requires coordinating speech, gestures and expressions across both speaking and listening. We present ELLSA 2, an end-

tutorialsarxiv-cs-ai

Global-in-time Spectral Duhamel-Picard Method for Post-Exposure Bake Simulation and Physics-Informed Neural Operators for EUV Lithography

arXiv:2610.20655v1 Announce Type: cross Abstract: We present a global-in-time spectral Duhamel-Picard method with polynomial expansions for post-exposure bake (PEB) simulation. The coupled acid-quench

Applications

Graph Dirichlet Energy for Fitting-Free Bivariate Causal Inference

arXiv:2610.21892v1 Announce Type: new Abstract: We propose the Graph Dirichlet Energy Causal Inference (GDECI) criterion for distinguishing cause from effect in bivariate settings. For a candidate dir

applicationsarxiv-cs-lg

Agents

GraphACM: Graph-based Action Consistency Measurement for Uncertainty Quantification in LLM Agents

arXiv:2610.22031v1 Announce Type: new Abstract: Behavioral consistency across repeated executions provides an informative signal for assessing the reliability of large language model (LLM) agents. How

agentsarxiv-cs-ai

GraphBEAT: Graph-based Bucketized Event Aggregation Over Time for Early Risk Stratification of Gastrointestinal Cancers

arXiv:2610.22014v1 Announce Type: cross Abstract: Longitudinal electronic health records (EHRs) contain complementary diagnostic and laboratory information about patient health. However, irregularity,

Grid-Bench: Can LLMs Predict How Discrete Worlds Change?

model-releasesarxiv-cs-cl

arXiv:2610.22312v1 Announce Type: new Abstract: Spatial instructions describe how to move objects, transform regions, or modify their surroundings. When several operations are applied in sequence, eac

Grounding Generative Verification in 3D Content

model-releasesarxiv-cs-cv

arXiv:2610.22360v1 Announce Type: new Abstract: Recent advances in multimodal large language models (MLLMs) and 3D generation have expanded 3D understanding and synthesis. However, beyond visual fidel