Search: “llm”

How Much Can Language Models Gain from Test-Time Computation?

arXiv:2610.01110v1 Announce Type: new Abstract: How much can test-time computation improve a language model, and at what cost? Test-time scaling is widely proposed as a substitute for larger models, b

Human

Content type

All Blog X Post Paper YouTube Reddit GitHub

Clear filters

In Dialogue with Intelligence: Toward Insightful Co-Augmentation

DGX agent

arXiv:2505.22767v4 Announce Type: replace-cross Abstract: Dialogue with a large language model can lead a person to insight: a sudden change in how they understand a problem. This perspective asks how

researcharxiv-cs-ai

2 Oct 2026

InterEvolve: Test-Time Evolution of Reward Programs for Humanoid Loco-Manipulation

arXiv:2610.02196v1 Announce Type: cross Abstract: We study test-time evolution for humanoid loco-manipulation: solving tasks that a controller was never trained for by repurposing its existing skills,

agentsarxiv-cs-cv

ITC-MoE: Importance-guided Token-aware Compression for MoE Diffusion Language Models

arXiv:2610.01296v1 Announce Type: new Abstract: Mixture-of-Experts (MoE) Diffusion Language Models (DLMs) offer flexible parallel decoding and increased model capacity, but their large number of exper

Knowing When to Yield: Grounded Arbitration of User Corrections in Text-Based Embodied Agents

arXiv:2610.00282v1 Announce Type: new Abstract: How should an embodied agent respond when a person's correction may be wrong? We formulate grounded correction arbitration as a choice among accepting,

Learning to Predict Distributions over Weight Updates for Test-Time Adaptation

arXiv:2610.01934v1 Announce Type: new Abstract: Hypernetworks have recently shown success in dynamically adapting the parameters of Large Language Models (LLMs) at runtime based on signals such as tas

Legal Research Bench: Measuring End-to-End Reliability in Long-Horizon Legal Research Agents

arXiv:2610.00609v1 Announce Type: new Abstract: Legal research is a core and time-consuming legal workflow. Lawyers must identify controlling authority, verify that it remains valid, reconcile statute

LEGO-OPD: Factorized Teacher Composition for Multimodal On-Policy Distillation

arXiv:2610.00333v1 Announce Type: cross Abstract: Multimodal on-policy distillation (OPD) aims to improve visual grounding while preserving the strong reasoning capabilities of language models. Recent

Make Sparse Rewards Count: Density-Aware Reward Aggregation for Multi-Reward RL

arXiv:2610.00574v1 Announce Type: new Abstract: Multi-reward reinforcement learning trains large language models to satisfy multiple behavioral objectives simultaneously. Reward-wise normalization, as

safetyarxiv-cs-lg

MCRI: A Four-Dimensional Framework for Analyzing and Evaluating Agent Skills

arXiv:2610.01506v1 Announce Type: new Abstract: As agents evolve from single-tool systems into modular, composite architectures, skills are becoming an important mechanism for capability development a

MoRA: MoE Pruning via Router Bias Learning and Expert Approximation

arXiv:2610.00367v1 Announce Type: new Abstract: Mixture-of-Experts (MoE) models enable parameter scaling with limited per-token computation by activating only a small subset of experts for each token,

More agents don't mean higher performance. There is a coordination bottleneck to consider. Not to mention the unnecessary costs. So how many…

More agents don't mean higher performance. There is a coordination bottleneck to consider. Not to mention the unnecessary costs. So how many of a multi-agent team's actions actually help it finish the

model-releasesdair-ai--x

MorphoBranch: A Fine-Structure-Preserving Workbench for Morphometric Analysis of Branched Cellular Structures

model-releasesarxiv-cs-cv

arXiv:2610.00860v1 Announce Type: cross Abstract: Background and Objectives: Fluorescence-labeled cellular arbors provide readouts of neuronal and microglial morphology, but fine and weakly labeled pr

Pushing CPU Speech Synthesis to the Wall: Extreme Inference Tuning under Serverless Architecture and Billing

arXiv:2610.00063v1 Announce Type: cross Abstract: Instance-billed serverless platforms charge for CPU and memory over the lifetime of a warm instance, making idle inference state a direct serving cost

R-GroundBench: A Diagnostic Benchmark for R-Group Groundingin Markush Molecular Editing

arXiv:2610.00700v1 Announce Type: new Abstract: Recent advances in AI for scientific discovery enable molecular understandingand design, yet reasoning over incomplete chemical representations remainsu

Reconstruct, Practice, Go Real: Guided Self-Improvement for Embodied Agents

arXiv:2610.02204v1 Announce Type: cross Abstract: Building reliable robot capabilities across diverse tasks requires substantial human effort to develop and maintain skills, design rewards, and integr

Rethinking Probability-Based Reinforcement Learning From Posterior Concentration

arXiv:2610.01458v1 Announce Type: new Abstract: Verifier-free reinforcement learning with probability-based rewards offers a promising way to train LLMs on general reasoning tasks where external verif

Self-Evolving Coding Rules for AI Coding Agents

arXiv:2610.00650v1 Announce Type: cross Abstract: The performance of AI coding agents is highly dependent on their underlying coding rules. However, existing coding rules are typically hand-crafted an

Sentence Specificity Scores for Collaborative Technical Documentation: A Domain-Transfer Study

arXiv:2610.01046v1 Announce Type: cross Abstract: Collaboration depends on shared context, and technical documentation is one way that context persists across people and AI teammates. Specificity, the

SoK: Decentralized Agent Economic Infrastructure

arXiv:2610.01756v1 Announce Type: cross Abstract: Decentralized agent economies increasingly build a single task from protocols that were designed and secured separately. This creates a simple problem

Temporally-Resolved Token Attribution Reveals the Generation Dynamics of Diffusion Language Models

arXiv:2610.01177v1 Announce Type: cross Abstract: This work presents Diffusion Layer Integrated Gradients (DLIG), a token attribution method for diffusion language models (DLMs) that extends Integrate

TRACE: Trajectory Return Attribution and Contrastive Erasure for Multi-Turn Safety

arXiv:2610.01323v1 Announce Type: new Abstract: Safety-aligned large language models (LLMs) often refuse a harmful request but comply once the same goal is spread over several turns. Preference object

UniGuardian: A Unified Defense for Detecting Prompt Injection, Backdoor Attacks and Adversarial Attacks in Large Language Models

arXiv:2502.13141v2 Announce Type: replace-cross Abstract: Large Language Models (LLMs) are vulnerable to attacks like prompt injection, backdoor attacks, and adversarial attacks, which manipulate prom

What Makes Something Hard(er)? Explaining Question Difficulty in Natural Language

arXiv:2610.01627v1 Announce Type: cross Abstract: Difficulty is one of the most fundamental properties of a question: it determines whether the question can meaningfully discriminate between models of

Why Does Train-Validation Separation Emerge? Update-Pressure Density Dynamics in Pretrained Backbones

arXiv:2610.01425v1 Announce Type: new Abstract: Train-validation separation is the evolving difference between performance on observed training examples and a finite held-out validation set. We propos

A Quantitative Study of Sustained Focus in Large Language Models via Repetitive Deterministic Prediction Tasks

arXiv:2511.00763v3 Announce Type: replace Abstract: We investigate the performance of large language models (LLMs) on repetitive deterministic prediction tasks and study how the sequence accuracy rate

Tutorials

AdaGEPA: Adaptive Feedback Allocation for Reflective Prompt Optimization

arXiv:2609.39927v1 Announce Type: new Abstract: Prompt optimization improves the performance of language-model systems on downstream tasks by refining their prompts. Classical methods evaluate prompts

tutorialsarxiv-cs-cl

Also Small Models Can Reasonably Self-Evaluate Their Confidence

arXiv:2609.39478v1 Announce Type: new Abstract: This study systematically evaluates self-evaluation-based uncertainty quantification across different language models of varying sizes on question-answe

researcharxiv-cs-lg

Better Deck or Different Judge? Evaluating Agentic Harness Gains in Corporate and Investment Banking

arXiv:2609.39958v1 Announce Type: new Abstract: Corporate and investment banking teams use presentations to support credit decisions and advise clients on financing and transactions. Producing these d

Beyond Mode Collapse: Generating Diverse Synthetic Expert Conversations via Generative Flow Networks

arXiv:2609.38359v1 Announce Type: new Abstract: High quality synthetic data is central to post training LLMs for adaptive AI applications that represent the diverse expert strategies and decisions in

Built an SQLite-based memory system for Ollama that doesn't eat all your VRAM

Wanted to share a project I've been working on to give local Ollama models persistent document memory without needing heavy vector databases. It's called Hillock. The main problem I had with standard

model-releasesr-ollama

Co-Evolving Agents: Learning from Failures as Hard Negatives

arXiv:2511.22254v5 Announce Type: replace Abstract: Self-evolving agents improve their performance on long-horizon tasks by learning from their own interactions with an environment. A common approach

DataFlex: A Unified Framework for Data-Centric Dynamic Training of Large Language Models

model-releasesarxiv-cs-cl

arXiv:2603.26164v2 Announce Type: replace-cross Abstract: Data-centric training has emerged as a promising direction for improving large language models (LLMs) by optimizing not only model parameters

Local Ai

Decode-Latency Feedback Prefill: A Model-Free Controller and Its Generalization Limits

arXiv:2609.38386v1 Announce Type: new Abstract: Concurrent autoregressive inference creates a fundamental interference problem: prefilling a newly arrived long prompt can delay tokens for requests tha

local-aiarxiv-cs-ai

Disentangling Self-Distillation: Measuring and Modeling Acquisition and Retention

arXiv:2609.39494v1 Announce Type: new Abstract: Self-distillation with privileged context adapts a language model from demonstrations by letting the model, once conditioned on a reference response, te

DLRMv4: The Next-Generation Recommendation MLPerf Training Benchmark

Evolving MLPerf Training from feature-interaction baselines to production-scale sequential recommendation with HSTU and the 560 GB Yambda-5B embedding benchmark. The post DLRMv4: The Next-Generation R

model-releasesmlcommons

Do Self-Evolving Skills Generalize to Held-Out Tasks?

arXiv:2609.39148v1 Announce Type: new Abstract: AI agents can externalize what they learn from past tasks into reusable skills, such as procedures, checklists, code, or other executable artifacts, tha

Hardware

Efficient Expert-Parallel Communication on PCIe-Connected Consumer GPUs

arXiv:2609.40093v1 Announce Type: cross Abstract: Expert parallelism (EP) enables inference of large Mixture-of-Experts (MoE) models by placing their experts across multiple GPUs, but requires substan

hardwarearxiv-cs-lg

EngramBench: A Capability-Grounded Benchmark for Skill-Evolution Harnesses

arXiv:2609.39284v1 Announce Type: cross Abstract: While large language models have achieved remarkable success in isolated code generation, authentic software engineering requires sustained reasoning,

Evaluating Whether GPT-6 Astra Performs Unsanctioned Supply-Chain Attacks

arXiv:2609.38415v1 Announce Type: cross Abstract: This technical report presents an alignment evaluation developed and performed by the UK AI Security Institute for assessing whether advanced AI syste

EvoDuet: Bilevel Co-Evolution of Web Searching and Task Solving for Scientific Discovery

model-releasesarxiv-cs-cl

arXiv:2609.40340v1 Announce Type: new Abstract: Evolutionary search with large language models (LLMs) can stall when progress requires external knowledge the model lacks. Supplying relevant documents

Examining Variation in How Guided AI Tutors Resolve Student Impasses

arXiv:2609.38346v1 Announce Type: new Abstract: When a student is stuck, a tutor faces the assistance dilemma: help given too early can hinder productive struggle, while help withheld too long leaves

Excited to share that I have joined @MistralAI, where I will be building a team in Montreal and across Canada to push the frontier of safe, …

model-releasesarthur-mensch--x

Excited to share that I have joined @MistralAI, where I will be building a team in Montreal and across Canada to push the frontier of safe, capable, and open LLMs. Mistral has played, and will continu

Feature-Aware Token Attack for Compression-Triggered Stealthy Failures in Large Vision-Language Models

arXiv:2609.39134v1 Announce Type: new Abstract: Visual-token compression improves the efficiency of large vision-language models, but can expose failures that full-token evaluation misses. We study ad

researcharxiv-cs-cv

Fine-Tuning Diffusion Language Models with Context Selection and Target Weighting

arXiv:2609.38385v1 Announce Type: new Abstract: Supervised fine-tuning of discrete diffusion language models masks some response tokens and trains the model to recover their original values from the v

Finite-Horizon Fisher Memory in Two-Sided Power-Bounded Recurrent Systems

arXiv:2609.39800v1 Announce Type: new Abstract: We analyse allocation, admission and post-write retention in finite-horizon linear-Gaussian noisy recurrent memories. At every horizon, the directional

researcharxiv-cs-lg

From Given to Gathered Evidence: Agentic Learning for Longitudinal Medical Reasoning

model-releasesarxiv-cs-cv

arXiv:2609.39566v1 Announce Type: new Abstract: Foundation models can serve as clinical agents through tool-use harnesses. However, conventional medical benchmarks assess reasoning over preselected ev

From Image Interpretation to Clinical Reasoning: Upstream Physician-Context-Aware Multimodal Learning with Causal Reinforcement Learning

arXiv:2609.38924v1 Announce Type: new Abstract: Major adverse cardiovascular events (MACE) remain the leading cause of mortality worldwide. Opportunistic screening using routinely acquired clinical da

safetyarxiv-cs-cv