Search: “arxiv-cs-lg”

Complementing reinforcement learning with SFT through logit averaging in the post training of LLMs

arXiv:2605.20555v1 Announce Type: new Abstract: We introduce a novel method that averages the logits of a frozen reference policy (e.g., SFT) and a trainable policy, and incorporate the method into Gr

Content type

All Blog X Post Paper YouTube Reddit GitHub

Clear filters

Agents

Compositional Transduction with Latent Analogies for Offline Goal-Conditioned Reinforcement Learning

DGX agent

arXiv:2605.20609v1 Announce Type: new Abstract: Compositional generalization is essential for reaching unseen goals under novel contextual variations in offline goal-conditioned reinforcement learning

agentsarxiv-cs-lg

21 May 2026

Computational-Statistical Trade-off in Kernel Two-Sample Testing with Random Fourier Features

arXiv:2407.08976v2 Announce Type: replace-cross Abstract: Recent years have seen a surge in methods for two-sample testing, among which the Maximum Mean Discrepancy (MMD) test has emerged as an effect

Compute Only Once: UG-Separation for Efficient Large Recommendation Models

arXiv:2602.10455v2 Announce Type: replace-cross Abstract: Driven by scaling laws, recommender systems increasingly rely on larger-scale models to capture complex feature interactions and user behavior

Concentration of General Stochastic Approximation Under Heavy-Tailed Markovian Noise

arXiv:2605.20999v1 Announce Type: cross Abstract: We establish maximal concentration bounds for the iterates generated by stochastic approximation algorithms with general step sizes, where the noise h

Conditional Equivalence of DPO and RLHF: Implicit Assumption, Failure Modes, and Provable Alignment

arXiv:2605.20834v1 Announce Type: cross Abstract: Direct Preference Optimization (DPO) has emerged as a popular alternative to Reinforcement Learning from Human Feedback (RLHF), offering theoretical e

Conditioning Gaussian Processes on Almost Anything

arXiv:2605.21041v1 Announce Type: cross Abstract: Gaussian processes (GPs) offer a principled probabilistic model over functions, but exact inference is restricted to the linear-Gaussian regime. We es

Local Ai

Conformal Selective Acting: Anytime-Valid Risk Control for RLVR-Trained LLMs

arXiv:2605.20270v1 Announce Type: new Abstract: A local specialist LLM, fine-tuned with reinforcement learning from verifiable rewards (RLVR) on operator-local data, is installed in a regulated organi

local-aiarxiv-cs-lg

Consistent Geometric Deep Learning via Hilbert Bundles and Cellular Sheaves

arXiv:2605.06395v2 Announce Type: replace Abstract: Modern deep learning architectures increasingly contend with sophisticated signals that are natively infinite-dimensional, such as time series, prob

Consistently Informative Soft-Label Temperature for Knowledge Distillation

arXiv:2605.20357v1 Announce Type: new Abstract: Knowledge distillation (KD) transfers knowledge from a high-capacity teacher to a compact student by matching their predictive distributions, with tempe

Contradiction Graphs Determine VC Dimension

arXiv:2605.20434v1 Announce Type: cross Abstract: We study the contradiction graphs associated with binary concept classes. For a class H subseteq {0,1}^X, the order-m contradiction graph G_m(H) has a

Control and optimization for Neural Partial Differential Equations in Supervised Learning

arXiv:2506.20764v2 Announce Type: replace-cross Abstract: Although there is a substantial body of literature on control and optimization problems for parabolic and hyperbolic systems, the specific pro

Control, Optimal Transport and Neural Differential Equations in Supervised Learning

arXiv:2503.15105v4 Announce Type: replace-cross Abstract: We study the fundamental computational problem of approximating optimal transport (OT) equations using neural differential equations (Neural O

Corrected Integrated Laplace Approximation for Bayesian Inference in Latent Gaussian Models

arXiv:2605.20345v1 Announce Type: cross Abstract: Latent Gaussian models (LGMs) are a popular class of Bayesian hierarchical models that include Gaussian processes, as well as certain spatial models a

Correcting Stochastic Update Bias in Preconditioned Language Model Optimizers

arXiv:2605.20756v1 Announce Type: new Abstract: Preconditioned optimizers are central to language model training, but their stochastic update rules are usually treated as direct approximations to popu

CRAFT: Conflict-Resolved Aggregation for Federated Training

arXiv:2605.21317v1 Announce Type: new Abstract: The aggregation of conflicting client updates remains a fundamental bottleneck in federated learning (FL) under heterogeneous data distributions. Naive

CRANE: Correcting Errors in Raw Nanopore Signals Using Hidden Markov Models

arXiv:2603.20420v2 Announce Type: replace-cross Abstract: Nanopore sequencing can read substantially longer sequences of nucleic acid molecules, called reads, than other sequencing methods, which has

CT-OT Flow: Estimating Continuous-Time Dynamics from Discrete Temporal Snapshots

arXiv:2505.17354v3 Announce Type: replace Abstract: In many real-world settings--e.g., single-cell RNA sequencing, mobility sensing, and environmental monitoring--data are observed only as temporally

Cumulative Meta-Learning from Active Learning Queries for Robustness to Spurious Correlations

arXiv:2605.20771v1 Announce Type: new Abstract: Spurious correlations in real-world datasets cause machine learning models to rely on irrelevant patterns, undermining reliability, generalization, and

Data-Efficient Neural Operator Training via Physics-Based Active Learning

arXiv:2605.21348v1 Announce Type: new Abstract: Solving partial differential equations with neural operators significantly reduces computational costs but remains bottlenecked by high training data re

DC-LA: Difference-of-Convex Langevin Algorithm

arXiv:2601.22932v2 Announce Type: replace Abstract: We study a sampling problem whose target distribution is pi propto exp(-f-r) where the data fidelity term f is Lipschitz smooth while the regularize

Decision-Path Patterns as Tree Reliability Signals: Path-based Adaptive Weighting for Random Forest Classification

arXiv:2605.20716v1 Announce Type: new Abstract: Random forests aggregate tree votes by simple majority, treating all trees as equally informative. We observe that the topological pattern along each tr

Decomposing MXFP4 quantization error for LLM reinforcement learning: reducible bias, recoverable deadzone, and an irreducible floor

arXiv:2605.20402v1 Announce Type: new Abstract: MXFP4 arithmetic can dramatically accelerate reinforcement learning (RL) post-training of large language models (LLMs), yet the quantization error intro

DeCoR: Design and Control Co-Optimization for Urban Streets Using Reinforcement Learning

arXiv:2605.21311v1 Announce Type: new Abstract: Modern vision systems can detect, track, and forecast urban actors at scale, yet translating perception outputs to urban design remains limited. We intr

Decoupling Communication from Policy: Robust MARL under Bandwidth Constraints

arXiv:2605.21085v1 Announce Type: cross Abstract: Communication enables coordination in multi-agent reinforcement learning (MARL), but many real-world applications, e.g., search-and-rescue with drone

Deep Learning Surrogates for Emulating Stochastic Climate Tipping Dynamics

arXiv:2605.20580v1 Announce Type: new Abstract: This work explores a dynamics-informed Temporal Fusion Transformer (TFT) as a data-driven surrogate for computationally intensive Earth system simulatio

Deep Neural Networks as Discrete Dynamical Systems: Implications for Physics-Informed Learning

arXiv:2601.00473v3 Announce Type: replace Abstract: We revisit the analogy between feed-forward deep neural networks (DNNs) and discrete dynamical systems derived from neural integral equations and th

Design for Manufacturing: A Manufacturability Knowledge-Integrated Reinforcement Learning Framework for Free-Form Pipe Routing in Aeroengines

arXiv:2605.20644v1 Announce Type: new Abstract: Design for manufacturing plays a critical role in advanced aeroengine development, where complex components necessitate careful consideration of manufac

Diagnosing Overhead in Dispatch Operations: Cross-architecture Observatory

arXiv:2605.20982v1 Announce Type: cross Abstract: AlltoAll dispatch is the dominant bottleneck of MoE expert parallelism, and the interconnect community has responded with four families of mitigations

Local Ai

Diffusion Models Memorize in Training -- and Generalize in Inference

arXiv:2603.13419v2 Announce Type: replace Abstract: Diffusion models generalize well in practice. However, an optimal diffusion model fully memorizes the training data and therefore fails to generaliz

local-aiarxiv-cs-lg

DISC: Decoupling Instruction from State-Conditioned Control via Policy Generation

arXiv:2605.20856v1 Announce Type: cross Abstract: Language-conditioned manipulation policies typically process instructions and observations through shared network parameters. This task-state entangle

Disentangling Bias by Modeling Intra- and Inter-modal Causal Attention for Multimodal Sentiment Analysis

arXiv:2508.04999v2 Announce Type: replace Abstract: Multimodal sentiment analysis (MSA) aims to understand human emotions by integrating information from multiple modalities, such as text, audio, and

Distributed Direct Preference Optimization

arXiv:2605.20696v1 Announce Type: new Abstract: Preference-based reinforcement learning (RL) is a key paradigm for aligning policies with human judgments, yet its theoretical behavior in distributed s

Divide and Contrast: Learning Robust Temporal Features without Augmentation

arXiv:2605.21241v1 Announce Type: new Abstract: Self-supervised learning for time-series representation aims to reduce reliance on labeled data while maintaining strong downstream performance, yet man

Divide et Calibra: Multiclass Local Calibration via Vector Quantization

arXiv:2605.21060v1 Announce Type: new Abstract: Accurate and well-calibrated Machine Learning (ML) models are mandatory in high-stakes settings, yet effective multiclass calibration remains challengin

Domain-Adaptable Reinforcement Learning for Code Generation with Dense Rewards

arXiv:2605.21180v1 Announce Type: new Abstract: Large language models show strong potential for automated code generation, but lack guarantees for correctness, quality, safety, and domain-specific con

Dynamic Shapley Computation

arXiv:2605.20620v1 Announce Type: new Abstract: Shapley-based data valuation provides a principled way to quantify the contribution of training data, but its high computational cost makes it impractic

Dynamic TMoE: A Drift-Aware Dynamic Mixture of Experts Framework for Non-Stationary Time Series Forecasting

arXiv:2605.20678v1 Announce Type: new Abstract: Non-stationary time series forecasting is challenged by evolving distribution shifts that static models struggle to capture. While Mixture-of-Experts (M

E-PCN: Jet Tagging with Explainable Particle Chebyshev Networks Using Kinematic Features

arXiv:2512.07420v2 Announce Type: cross Abstract: The identification and classification of collimated particle sprays, or jets, are essential for interpreting data from high-energy collider experiment

ECUAS{n}: A family of metrics for principled evaluation of uncertainty-augmented systems

arXiv:2605.20490v1 Announce Type: cross Abstract: In high-stakes automated decision-making, access to predictive uncertainty is essential for enabling users -- human or downstream systems -- to accept

Effective Model Pruning: Measure The Redundancy of Model Components

arXiv:2509.25606v3 Announce Type: replace Abstract: This article initiates the study of a basic question about model pruning. Given a vector s of importance scores assigned to model components, how ma

Efficient Banzhaf-Based Data Valuation for k-Nearest Neighbors Classification

arXiv:2605.21033v1 Announce Type: new Abstract: Data valuation, the task of quantifying the contribution of individual data points to model performance, has emerged as a fundamental challenge in machi

Efficient Learning of Deep State Space Models via Importance Smoothing

arXiv:2605.21108v1 Announce Type: new Abstract: Latent state space systems are ubiquitous in statistical modelling, arising naturally when a time series is observed through a noisy measurement functio

Tutorials

Efficient numeracy in language models through single-token number embeddings

arXiv:2510.06824v2 Announce Type: replace Abstract: To drive progress in science and engineering, large language models (LLMs) must be able to process large amounts of numerical data and solve long ca

tutorialsarxiv-cs-lg

Enhanced Reinforcement Learning-based Process Synthesis via Quantum Computing

arXiv:2605.21213v1 Announce Type: cross Abstract: In this work, we present quantum reinforcement learning (RL) as a solution strategy for process synthesis problems. Building on our prior work, we dev

Ensemble RL through Classifier Models: Enhancing Risk-Return Trade-offs in Trading Strategies

arXiv:2502.17518v2 Announce Type: replace Abstract: This paper presents a comprehensive study on the use of ensemble Reinforcement Learning (RL) models in financial trading strategies, leveraging clas

Epistemic Uncertainty Quantification for Pre-trained VLMs via Riemannian Flow Matching

arXiv:2601.21662v2 Announce Type: replace Abstract: Vision-Language Models (VLMs) are typically deterministic in nature and lack intrinsic mechanisms to quantify epistemic uncertainty, which reflects

Equilibrium Reasoners: Learning Attractors Enables Scalable Reasoning

arXiv:2605.21488v1 Announce Type: new Abstract: Scaling test-time compute by iteratively updating a latent state has emerged as a powerful paradigm for reasoning. Yet the internal mechanisms that enab