Search: “models”

Protecting Language Models Against Unauthorized Distillation through Trace Rewriting

arXiv:2602.15143v2 Announce Type: replace Abstract: Knowledge distillation is a widely adopted technique for transferring capabilities from LLMs to smaller, more efficient student models. However, una

researcharxiv-cs-ai

Temporal Contrastive Decoding: A Training-Free Method for Large Audio-Language Models

DGX agent

arXiv:2604.15383v1 Announce Type: cross Abstract: Large audio-language models (LALMs) generalize across speech, sound, and music, but unified decoders can exhibit a temporal smoothing bias: transient

safetyarxiv-cs-ai

20 Apr 2026

TRIDENT: Enhancing Large Language Model Safety with Tri-Dimensional Diversified Red-Teaming Data Synthesis

arXiv:2505.24672v2 Announce Type: replace Abstract: Large Language Models (LLMs) excel in various natural language processing tasks but remain vulnerable to generating harmful content or being exploit

VLegal-Bench: Cognitively Grounded Benchmark for Vietnamese Legal Reasoning of Large Language Models

model-releasesarxiv-cs-ai

arXiv:2512.14554v5 Announce Type: replace-cross Abstract: The rapid advancement of large language models (LLMs) has enabled new possibilities for applying artificial intelligence within the legal doma

Safety

When Search Goes Wrong: Red-Teaming Web-Augmented Large Language Models

arXiv:2510.09689v3 Announce Type: replace-cross Abstract: Large Language Models (LLMs) have been augmented with web search to overcome the limitations of the static knowledge boundary by accessing up-

safetyarxiv-cs-ai

Anthropic locked Claude Code to native apps in Jan 2026. Are we still comparing models or just ecosystems

Anthropic's Claude Code desktop app is strictly optimized for Anthropic's models , creating a 'walled garden' effect that restricts users to Claude exclusively. The redesigned Claude Code desktop app

model-releasesr-chatgpt

19 Apr 2026

Mistral, which once aimed for top open models, now leans on being an alternative to Chinese and US labs, says it's on track for $80M in monthly revenue by Dec. (Iain Martin/Forbes)

Iain Martin / Forbes: Mistral, which once aimed for top open models, now leans on being an alternative to Chinese and US labs, says it's on track for $80M in monthly revenue by Dec. — Paris-based Mist

model-releasestechmeme

18 Apr 2026

RIP Anthropic API fees. Someone just made Claude Code run 100% locally on a MacBook for $0/month. 122B parameter model. 65 tokens per second…

model-releasesclem-delangue--x

RIP Anthropic API fees. Someone just made Claude Code run 100% locally on a MacBook for $0/month. 122B parameter model. 65 tokens per second. Nothing touches the cloud. The trick everyone else missed:

18 Apr 2026

Atropos: Improving Cost-Benefit Trade-off of LLM-based Agents under Self-Consistency with Early Termination and Model Hotswap

arXiv:2604.15075v1 Announce Type: cross Abstract: Open-weight Small Language Models(SLMs) can provide faster local inference at lower financial cost, but may not achieve the same performance level as

local-aiarxiv-cs-lg

CI-CBM: Class-Incremental Concept Bottleneck Model for Interpretable Continual Learning

arXiv:2604.14519v1 Announce Type: cross Abstract: Catastrophic forgetting remains a fundamental challenge in continual learning, in which models often forget previous knowledge when fine-tuned on a ne

researcharxiv-cs-cv

Dissecting Failure Dynamics in Large Language Model Reasoning

arXiv:2604.14528v1 Announce Type: cross Abstract: Large Language Models (LLMs) achieve strong performance through extended inference-time deliberation, yet how their reasoning failures arise remains p

local-aiarxiv-cs-cl

DLink: Distilling Layer-wise and Dominant Knowledge from EEG Foundation Models

arXiv:2604.15016v1 Announce Type: new Abstract: EEG foundation models (FMs) achieve strong cross-subject and cross-task generalization but impose substantial computational and memory costs that hinder

Doubly Outlier-Robust Online Infinite Hidden Markov Model

arXiv:2604.14322v1 Announce Type: cross Abstract: We derive a robust update rule for the online infinite hidden Markov model (iHMM) for when the streaming data contains outliers and the model is missp

Figma stock closed down 6.84% on Friday after Anthropic launched Claude Design, a dedicated app powered by its latest model Claude Opus 4.7 (Jon Keegan/Sherwood News)

Jon Keegan / Sherwood News: Figma stock closed down 6.84% on Friday after Anthropic launched Claude Design, a dedicated app powered by its latest model Claude Opus 4.7 — Today Anthropic launched Claud

model-releasestechmeme

HARNESS: Lightweight Distilled Arabic Speech Foundation Models

arXiv:2604.14186v1 Announce Type: cross Abstract: Large self-supervised speech (SSL) models achieve strong downstream performance, but their size limits deployment in resource-constrained settings. We

applicationsarxiv-cs-cl

How to Fine-Tune a Reasoning Model? A Teacher-Student Cooperation Framework to Synthesize Student-Consistent SFT Data

arXiv:2604.14164v1 Announce Type: new Abstract: A widely adopted strategy for model enhancement is to use synthetic data generated by a stronger model for supervised fine-tuning (SFT). However, for em

tutorialsarxiv-cs-cl

Large Vision Model-Guided Masked Low-Rank Approximation for Ground-Roll Attenuation

arXiv:2604.00998v2 Announce Type: replace Abstract: Ground roll is a common type of coherent noise in seismic records, and its attenuation remains challenging due to its substantial overlap with usefu

local-aiarxiv-cs-cv

Safety

LeapAlign: Post-Training Flow Matching Models at Any Generation Step by Building Two-Step Trajectories

arXiv:2604.15311v1 Announce Type: new Abstract: This paper focuses on the alignment of flow matching models with human preferences. A promising way is fine-tuning by directly backpropagating reward gr

safetyarxiv-cs-cv

Agents

MCPThreatHive: Automated Threat Intelligence for Model Context Protocol Ecosystems

arXiv:2604.13849v1 Announce Type: cross Abstract: The rapid proliferation of Model Context Protocol (MCP)-based agentic systems has introduced a new category of security threats that existing framewor

agentsarxiv-cs-ai

Nova Forge SDK series part 2: Practical guide to fine-tune Nova models using data mixing capabilities

This hands-on guide walks through every step of fine-tuning an Amazon Nova model with the Amazon Nova Forge SDK, from data preparation to training with data mixing to evaluation, giving you a repeatab

tutorialsaws-ml-blog

SPaCe: Unlocking Sample-Efficient Large Language Models Training With Self-Pace Curriculum Learning

arXiv:2508.05015v2 Announce Type: replace Abstract: Large language models (LLMs) have shown strong reasoning capabilities when fine-tuned with reinforcement learning (RL). However, such methods requir

TennisTV: Do Multimodal Large Language Models Understand Tennis Rallies?

model-releasesarxiv-cs-cv

arXiv:2509.15602v5 Announce Type: replace Abstract: Multimodal large language models (MLLMs) excel at general video understanding but struggle with fast, high-frequency sports like tennis, where rally

The Mirror Design Pattern: Strict Data Geometry over Model Scale for Prompt Injection Detection

model-releasesarxiv-cs-ai

arXiv:2603.11875v2 Announce Type: replace-cross Abstract: Prompt injection defenses are often framed as semantic understanding problems and delegated to increasingly large neural detectors. For the fi

VisPCO: Visual Token Pruning Configuration Optimization via Budget-Aware Pareto-Frontier Learning for Vision-Language Models

arXiv:2604.15188v1 Announce Type: new Abstract: Visual token pruning methods effectively mitigate the quadratic computational growth caused by processing high-resolution images and video frames in vis

researcharxiv-cs-cv

Wow I can already say after just 5 hours using @AnthropicAI Opus 4.7 that this is the first model that 'gets' what I'm doing when I'm workin…

tutorialsjeremy-howard--x

Wow I can already say after just 5 hours using @AnthropicAI Opus 4.7 that this is the first model that 'gets' what I'm doing when I'm working. It feels aligned with me in a way no previous model did.

180 tok/s generation on a 4090 with qwen 3.6. if you're on a 4090 and not running this model yet you're leaving performance on the table. 3B…

model-releasesclem-delangue--x

180 tok/s generation on a 4090 with qwen 3.6. if you're on a 4090 and not running this model yet you're leaving performance on the table. 3B active params at that speed is insane for agentic coding. t

A KL Lens on Quantization: Fast, Forward-Only Sensitivity for Mixed-Precision SSM-Transformer Models

arXiv:2604.13440v1 Announce Type: new Abstract: Deploying Large Language Models (LLMs) on edge devices faces severe computational and memory constraints, limiting real-time processing and on-device in

local-aiarxiv-cs-lg

A1: A Fully Transparent Open-Source, Adaptive and Efficient Truncated Vision-Language-Action Model

arXiv:2604.05672v3 Announce Type: replace Abstract: Vision-Language-Action (VLA) models have emerged as a powerful paradigm for open-world robot manipulation, but their practical deployment is often c

researcharxiv-cs-ro

Abstract 3D Perception for Spatial Intelligence in Vision-Language Models

arXiv:2511.10946v3 Announce Type: replace Abstract: Vision-language models (VLMs) struggle with 3D-related tasks such as spatial cognition and physical understanding, which are crucial for real-world

applicationsarxiv-cs-cv

Adaptive Learning via Off-Model Training and Importance Sampling for Fully Non-Markovian Optimal Stochastic Control. Complete version

arXiv:2604.13147v1 Announce Type: cross Abstract: This paper studies continuous-time stochastic control problems whose controlled states are fully non-Markovian and depend on unknown model parameters.

Addressing Overthinking in Large Vision-Language Models via Gated Perception-Reasoning Optimization

arXiv:2601.04442v2 Announce Type: replace-cross Abstract: Large Vision-Language Models (LVLMs) have exhibited strong reasoning capabilities through chain-of-thought mechanisms that generate step-by-st

tutorialsarxiv-cs-cl

Before the First Token: Scale-Dependent Emergence of Hallucination Signals in Autoregressive Language Models

arXiv:2604.13068v1 Announce Type: new Abstract: When do large language models decide to hallucinate? Despite serious consequences in healthcare, law, and finance, few formal answers exist. Recent work

applicationsarxiv-cs-cl

Better and Worse with Scale: How Contextual Entrainment Diverges with Model Size

arXiv:2604.13275v1 Announce Type: new Abstract: Larger language models become simultaneously better and worse at handling contextual information -- better at ignoring false claims, worse at ignoring i

researcharxiv-cs-cl

Big Tech companies say a $90B data center buildout in Spain's Aragón, one of Europe's fastest-growing hubs, should be an EU model, as local residents push back (Clara Hernanz Lizarraga/Bloomberg)

Clara Hernanz Lizarraga / Bloomberg: Big Tech companies say a $90B data center buildout in Spain's Aragón, one of Europe's fastest-growing hubs, should be an EU model, as local residents push back — B

local-aitechmeme

Decoding the Delta: Unifying Remote Sensing Change Detection and Understanding with Multimodal Large Language Models

model-releasesarxiv-cs-cv

arXiv:2604.14044v1 Announce Type: new Abstract: While Multimodal Large Language Models (MLLMs) excel in general vision-language tasks, their application to remote sensing change understanding is hinde

Democratising Pathology Co-Pilots: An Open Pipeline and Dataset for Whole-Slide Vision-Language Modelling

arXiv:2512.17326v2 Announce Type: replace Abstract: Vision-language models (VLMs) have the potential to become co-pilots for pathologists. However, most VLMs either focus on small regions of interest

researcharxiv-cs-cv

Feed-Forward 3D Scene Modeling: A Problem-Driven Perspective

arXiv:2604.14025v1 Announce Type: new Abstract: Reconstructing 3D representations from 2D inputs is a fundamental task in computer vision and graphics, serving as a cornerstone for understanding and i

applicationsarxiv-cs-cv

Google’s Gemini 3.1 Flash TTS model offers unparalleled control over AI voices

model-releasessiliconangle

Google LLC’s DeepMind artificial intelligence unit today rolled out a new text-to-speech model called Gemini 3.1 Flash TTS. Unlike its earlier, robotic predecessors, it enables users to direct the voc

Safety

Heavy-Tailed Class-Conditional Priors for Long-Tailed Generative Modeling

arXiv:2509.02154v2 Announce Type: replace-cross Abstract: Variational Autoencoders (VAEs) with global priors trained under an imbalanced empirical class distribution can lead to underrepresentation of

safetyarxiv-cs-cv

LaoBench: A Large-Scale Multidimensional Lao Benchmark for Large Language Models

arXiv:2511.11334v3 Announce Type: replace Abstract: The rapid advancement of large language models (LLMs) has not been matched by their evaluation in low-resource languages, especially Southeast Asian

Mitigating Barren Plateaus in Quantum Denoising Diffusion Probabilistic Model

arXiv:2512.06695v2 Announce Type: replace Abstract: Quantum generative models exploit quantum superposition and entanglement to enhance learning efficiency for both classical and quantum data. Recentl

MOONSHOT : A Framework for Multi-Objective Pruning of Vision and Large Language Models

model-releasesarxiv-cs-lg

arXiv:2604.13287v1 Announce Type: new Abstract: Weight pruning is a common technique for compressing large neural networks. We focus on the challenging post-training one-shot setting, where a pre-trai

MulDimIF: A Multi-Dimensional Constraint Framework for Evaluating and Improving Instruction Following in Large Language Models

arXiv:2505.07591v2 Announce Type: replace Abstract: Instruction following refers to the ability of large language models (LLMs) to generate outputs that satisfy all specified constraints. Existing res

OpenAI launches GPT-Rosalind, an AI model for life sciences research, including drug discovery, as a research preview for customers such as Moderna and Amgen (Megan Morrone/Axios)

Megan Morrone / Axios: OpenAI launches GPT-Rosalind, an AI model for life sciences research, including drug discovery, as a research preview for customers such as Moderna and Amgen — OpenAI announced

model-releasestechmeme

Opus 4.7 is a model I’ve loved working with in Claude Code. It’s more agentic and instruction following but also incredibly smart and creati…

Opus 4.7 is a model I’ve loved working with in Claude Code. It’s more agentic and instruction following but also incredibly smart and creative. I think it takes a slight adjustment to get used to, but

model-releasesthariq--x

Reward Design for Physical Reasoning in Vision-Language Models

arXiv:2604.13993v1 Announce Type: cross Abstract: Physical reasoning over visual inputs demands tight integration of visual perception, domain knowledge, and multi-step symbolic inference. Yet even st

UniBlendNet: Unified Global, Multi-Scale, and Region-Adaptive Modeling for Ambient Lighting Normalization

model-releasesarxiv-cs-cv

arXiv:2604.13383v1 Announce Type: new Abstract: Ambient Lighting Normalization (ALN) aims to restore images degraded by complex, spatially varying illumination conditions. Existing methods, such as IF

VLMs Need Words: Vision Language Models Ignore Visual Detail In Favor of Semantic Anchors

arXiv:2604.02486v2 Announce Type: replace-cross Abstract: Vision-language models (VLMs) have achieved impressive performance across a wide range of multimodal tasks. However, they often fail on tasks

researcharxiv-cs-cl