Search: “arxiv-cs-lg”

Teach Me After You Learn It: Lookahead Self-Distillation for RLVR

arXiv:2610.18256v1 Announce Type: new Abstract: On-policy self-distillation (OPSD) has emerged as a popular training paradigm for LLMs, providing dense, token-level supervision by aligning a model's o

Human

Content type

All Blog X Post Paper YouTube Reddit GitHub

Clear filters

Teacher-Guided Tool Learning for Small Language Models via Bidirectional KL Constraints

DGX agent

arXiv:2610.18459v1 Announce Type: new Abstract: External tools can extend the capabilities of small language models (SLMs), enabling their deployment in resource constrained environments. However, the

safetyarxiv-cs-lg

14 Oct 2026

Teaching MLP More Graph Information: A Three-stage Multitask Knowledge Distillation Framework

arXiv:2403.01079v2 Announce Type: replace Abstract: We study the challenging problem for inference tasks on large-scale graph datasets of Graph Neural Networks: huge time and memory consumption, and t

Tensor-reduced atomic density representations

arXiv:2210.01705v3 Announce Type: replace-cross Abstract: Density based representations of atomic environments that are invariant under Euclidean symmetries have become a widely used tool in the machi

Terrarium: Growing Personal Assistants via Scalable Synthesis of App Worlds

arXiv:2610.14860v1 Announce Type: new Abstract: Advances in large language models (LLMs) have brought the vision of a general-purpose personal assistant closer to reality. Such assistants could provid

Testing Central Structure for Complex Networks

arXiv:2610.15659v1 Announce Type: new Abstract: Centrality measures rank nodes even when the observed differences can be explained by ordinary structural variation. We therefore study a preceding ques

textGrain watermarking is harder to detect: can sum-tests still do it optimally?

arXiv:2610.17655v1 Announce Type: cross Abstract: Watermarking embeds an imperceptible signal in text from large language models that is statistically detectable with a secret key. The recent textGrai

The belief-action gap in language model agents

arXiv:2610.17444v1 Announce Type: new Abstract: Language model (LM) agents can fail for different reasons, requiring different explanations and interventions. We study how state beliefs, action values

The Bell Boundary of Shared Representations

arXiv:2610.16828v1 Announce Type: new Abstract: When does a shared representation become insufficient for jointly supporting multiple tasks? This question lies at the heart of multi-task learning mode

The Effect of Layer Ordering in Hybrid Architectures

arXiv:2610.18104v1 Announce Type: new Abstract: Hybrid attention is rapidly becoming the dominant architectural choice for frontier models. Despite their successes, such architectures complicate model

The Geometry of the Set of Equivalent Linear Neural Networks

arXiv:2404.14855v2 Announce Type: replace Abstract: We characterize the geometry of the set of all weight vectors for which a fully-connected linear neural network computes the same linear transformat

The Gradient That Top-k Distillation Never Computes

arXiv:2610.15589v1 Announce Type: new Abstract: Many deployed distillation pipelines show the student a truncated view of the teacher: keep the top-k probabilities, renormalize, and minimize a diverge

The Language of Looking: Motion Tokens Reveal Domain-Dependent Local Structure in Gaze Sequences

arXiv:2610.16565v1 Announce Type: new Abstract: Human gaze is the primary high-bandwidth interface through which we parse the semantics of our environment. While often modeled as a sequence, the degre

The Optimal Decoder is Tilted Temperature

arXiv:2610.15058v1 Announce Type: new Abstract: Larger sample budgets benefit from higher sampling temperatures in LLMs, but what temperature tuning optimizes and why it works remain poorly understood

Hardware

The Price of Precision: A Calibrated Cost Model for Mixed-Precision LLM Serving

arXiv:2610.15629v1 Announce Type: new Abstract: Choosing a mixed-precision configuration for large language model (LLM) serving trades speed against accuracy. Allocators that use average bit-width or

hardwarearxiv-cs-lg

The RNN Homunculus: Learning to Gate Long-Term Memory

arXiv:2610.14279v1 Announce Type: new Abstract: Learning long-range dependencies with recurrent neural networks (RNNs) is difficult because of the vanishing gradient problem. Gated recurrent neural ne

The Sample Complexity of Multicalibration with an ERM Oracle

arXiv:2610.14363v1 Announce Type: new Abstract: A predictor is multicalibrated with respect to a class of subpopulations (formally, of functions on the inputs, called auditors) if, for every subpopula

The Surprising Expressivity of a Single Neuron with Autoregressive Chain-of-Thought

arXiv:2610.17283v1 Announce Type: new Abstract: The success of autoregressive language models trained with Chain-of-Thought~(CoT) supervision raises a fundamental question: How simple can a next-token

The Tokens Strike Back: Safety-Aware Denoising with Directional Embedding-based Revision

arXiv:2610.17636v1 Announce Type: new Abstract: The Safety-Aware Denoiser (SAD) is a recent, training-free, safety-guidance framework for text diffusion models. During inference, SAD steers text sampl

There are Levels to It: Red Teaming Dialogue With Hierarchical Reinforcement Learning

arXiv:2508.04451v2 Announce Type: replace Abstract: Red teaming is essential for securing Large Language Models, yet current automated methods remain limited by templates and single-turn attacks. To s

Think Before You Speak: Enabling Training-Free Latent Communication Between Language Models

arXiv:2610.16165v1 Announce Type: cross Abstract: Large language models (LLMs) have demonstrated remarkable capabilities in solving complex tasks in coding, mathematical logic, and iterative reasoning

Tight End-to-End Accounting for Differentially Private Synthetic Data

arXiv:2610.17249v1 Announce Type: cross Abstract: Differentially private (DP) pipelines often combine heterogeneous mechanisms, yet their guarantees are best analyzed using different notions of privac

TILE: Bridging LLMs and Table Foundation Models for Tabular Learning with Strings

arXiv:2610.17065v1 Announce Type: new Abstract: Real-world tables are often heterogeneous, combining numerical features with free-form string columns. Table foundation models (TFMs) excel on numerical

Time Window Optimization for Motor Imagery Classification via Temporal Structure Embedded Reinforcement Learning

arXiv:2610.16921v1 Announce Type: cross Abstract: Online motor imagery-based brain-computer interfaces (MI-BCIs) typically decode continuous EEG using fixed-length sliding windows that may be misalign

Tiny Recursive Message Passing

arXiv:2610.18649v1 Announce Type: new Abstract: Message passing neural networks have shown remarkable performance in algorithmic and constraint reasoning, due to their ability to encode the structure

TLA-Verus: Generating and Solving Temporal Proof Obligations with LLM Agents

arXiv:2610.14923v1 Announce Type: new Abstract: The TLA+ (temporal logic of actions) formal specification toolkit has seen widespread adoption in modeling modern distributed software systems including

Local Ai

To Align or Not To Align: Check Your COMPASS Before You Train

arXiv:2610.18638v1 Announce Type: cross Abstract: Combining pretrained models is a common strategy for building AI systems, whether by aligning a vision encoder with a language model or merging fine-t

local-aiarxiv-cs-lg

To Cache or Not to Cache: Understanding and Mitigating Safety Incompatibility in Cross-Model KV Cache Reuse

arXiv:2610.15721v1 Announce Type: new Abstract: Cross-model KV cache sharing reduces prefill computation by allowing a receiver model to reuse prefix KV states generated by a related sender model. Exi

To Defer or Not to Defer? Online Learning with LLMs and Congested Experts

arXiv:2610.15128v1 Announce Type: new Abstract: Modern AI systems increasingly combine large language models (LLMs) with human experts, using the LLM to handle routine queries while deferring uncertai

TopoCurate:Modeling Interaction Topology for Tool-Use Agent Training

arXiv:2603.01714v2 Announce Type: replace Abstract: Training tool-use agents typically relies on outcome-based filtering: Supervised Fine-Tuning (SFT) on successful trajectories and Reinforcement Lear

TopoPilot: Reliable Conversational Workflow Automation for Topological Data Analysis and Visualization

arXiv:2603.25063v2 Announce Type: replace-cross Abstract: Recent agentic systems demonstrate that large language models can generate complex codebases from natural language prompts. Despite this progr

Toward a Multimodal Foundation Model for Plasma State Reconstruction

arXiv:2610.15076v1 Announce Type: new Abstract: Plasma experiments observe coupled physical states through sparse, heterogeneous, and incomplete diagnostics. We present a proof of concept toward a mul

Toward Agents That Reason About Their Computation

arXiv:2510.22833v2 Announce Type: replace-cross Abstract: While reinforcement learning agents can achieve superhuman performance in many complex tasks, they typically do not become more computationall

Towards Green AI: Evaluating Energy Efficiency for Quantized LLM Inference

arXiv:2610.17531v1 Announce Type: new Abstract: While it is known that quantization of large language models (LLMs) reduces memory usage via lower-bit weights, studies quantifying the resulting impact

Towards Modality-Agnostic Seizure Classification: A Unified Deep Learning Framework for Heterogeneous EEG and Intracranial EEG Signals

arXiv:2610.16596v1 Announce Type: new Abstract: Seizure prediction and detection, essential for clinical treatment of epilepsy, remain the crucial task of this neurological disease. EEG and iEEG are w

Applications

Towards Optimality in Active Preference Learning

arXiv:2602.01581v2 Announce Type: replace Abstract: Aligning large language models (LLMs) depends on high-quality datasets of human preference labels, which are costly to collect. Although active lear

applicationsarxiv-cs-lg

Towards Safe Autonomy for Computer Use

arXiv:2610.18508v1 Announce Type: cross Abstract: The safety of LLM-based agents currently rests on alignment training and safety classifiers, neither of which can rule out that a sophisticated indire

Towards Universal Calibration for In-Context Classification

arXiv:2610.17096v1 Announce Type: new Abstract: In-Context Learning (ICL) has received a lot of interest recently due to its ability to improve classification accuracy for new tasks without having to

TPPO: Tracking-Value PPO For Language Model Reasoning

arXiv:2610.16794v1 Announce Type: new Abstract: Proximal Policy Optimization (PPO) is widely used in reinforcement learning with verifiable rewards for language-model reasoning. While value evolution

TRACE: Temporal Risk Assessment and Concept Editing for Safer Language Generation

arXiv:2610.15365v1 Announce Type: cross Abstract: Large language models (LLMs) can produce harmful content while generating a response. A useful safeguard should not only detect this risk during gener

Train for Many, Update with One: Improving Multi-Sample Reasoning via Fenchel-Bregman Posterior Learning

arXiv:2610.17735v1 Announce Type: new Abstract: Multi-sample reasoning measures candidate coverage, the probability that at least one of K candidates is correct. However, minimizing the average loss o

Training Generalist Navigation Policies via Action Self-Distillation

arXiv:2610.17593v1 Announce Type: cross Abstract: Language provides a natural interface for humans to specify and reason about goals, but training language-conditioned robotic policies and reasoning t

Training LLMs for Multi-Agent Collaboration with Verifiable Information Rewards

arXiv:2610.17893v1 Announce Type: cross Abstract: Effective collaboration among large language model (LLM) agents in multi-agent systems requires sharing relevant evidence, requesting missing informat

Triage-PiEvo: Evidence Admission for Principle-Evolving Scientific Agents

arXiv:2610.14683v1 Announce Type: new Abstract: Large language model (LLM)-based AI Scientists advance scientific discovery, yet they suffer from inefficiencies due to noisy observations. Existing pri

Trust-Calibrated VLA Policy Post-Training with Evolving Imagination

arXiv:2610.16517v1 Announce Type: cross Abstract: Vision-Language-Action (VLA) models are often pretrained via imitation learning, making reinforcement learning (RL) post-training important when deplo

TSRL: Trigger-based Supervised Reinforcement Learning for Continual Agent Training

arXiv:2610.17726v1 Announce Type: new Abstract: Large language model (LLM) agents excel at complex, multi-step tasks but suffer from catastrophic forgetting during continual learning--acquiring new sk

TStruct: Learning Shared Temporal Structures for Long-Term Time-Series Forecasting

arXiv:2610.17864v1 Announce Type: new Abstract: Long-term time series forecasting requires temporal dependency modeling that remains stable and generalizable under temporal distribution shifts. Howeve

TurnCost: Turns, Not Observations: Cache-Aware Loop Engineering for Coding Agents

arXiv:2610.15093v1 Announce Type: cross Abstract: AI coding agents operate in iterative loops: they modify code, run tools, inspect the results, and decide what to do next. We call the design of this