Limits of Difficulty Scaling: Hard Samples Yield Diminishing Returns in GRPO-Tuned SLMs
arXiv:2604.06298v1 Announce Type: new Abstract: Recent alignment work on Large Language Models (LLMs) suggests preference optimization can improve reasoning by shifting probability mass toward better solutions. We test this claim in a resource-constrained setting by applying GRPO with LoRA to...
Distributional Open-Ended Evaluation of LLM Cultural Value Alignment Based on Value Codebook
arXiv:2604.06210v1 Announce Type: new Abstract: As LLMs are globally deployed, aligning their cultural value orientations is critical for safety and user engagement. However, existing benchmarks face the Construct-Composition-Context ($C^3$) challenge: relying on discriminative, multiple-choice formats that probe value knowledge rather...
OpenAI releases a new safety blueprint to address the rise in child sexual exploitation
OpenAI's new Child Safety Blueprint aims to tackle the alarming rise in child sexual exploitation linked to advancements in AI.
The Depth Ceiling: On the Limits of Large Language Models in Discovering Latent Planning
arXiv:2604.06427v1 Announce Type: new Abstract: The viability of chain-of-thought (CoT) monitoring hinges on models being unable to reason effectively in their latent representations. Yet little is known about the limits of such latent reasoning in LLMs. We test these limits...
ValueGround: Evaluating Culture-Conditioned Visual Value Grounding in MLLMs
arXiv:2604.06484v1 Announce Type: new Abstract: Cultural values are expressed not only through language but also through visual scenes and everyday social practices. Yet existing evaluations of cultural values in language models are almost entirely text-only, making it unclear whether models...
Illocutionary Explanation Planning for Source-Faithful Explanations in Retrieval-Augmented Language Models
arXiv:2604.06211v1 Announce Type: new Abstract: Natural language explanations produced by large language models (LLMs) are often persuasive, but not necessarily scrutable: users cannot easily verify whether the claims in an explanation are supported by evidence. In XAI, this motivates a...
Learning to Interrupt in Language-based Multi-agent Communication
arXiv:2604.06452v1 Announce Type: new Abstract: Multi-agent systems using large language models (LLMs) have demonstrated impressive capabilities across various domains. However, current agent communication suffers from verbose output that overload context and increase computational costs. Although existing approaches focus on compressing...
Tubi is the first streamer to launch a native app within ChatGPT
Tubi becomes the first streaming service to offer an app integration within ChatGPT, the AI chatbot that millions of users turn to for answers.
Discrete Flow Matching Policy Optimization
arXiv:2604.06491v1 Announce Type: new Abstract: We introduce Discrete flow Matching policy Optimization (DoMinO), a unified framework for Reinforcement Learning (RL) fine-tuning Discrete Flow Matching (DFM) models under a broad class of policy gradient methods. Our key idea is to view...
AE-ViT: Stable Long-Horizon Parametric Partial Differential Equations Modeling
arXiv:2604.06475v1 Announce Type: new Abstract: Deep Learning Reduced Order Models (ROMs) are becoming increasingly popular as surrogate models for parametric partial differential equations (PDEs) due to their ability to handle high-dimensional data, approximate highly nonlinear mappings, and utilize GPUs. Existing...
DiffuMask: Diffusion Language Model for Token-level Prompt Pruning
arXiv:2604.06627v1 Announce Type: new Abstract: In-Context Learning and Chain-of-Thought prompting improve reasoning in large language models (LLMs). These typically come at the cost of longer, more expensive prompts that may contain redundant information. Prompt compression based on pruning offers a...
Does a Global Perspective Help Prune Sparse MoEs Elegantly?
arXiv:2604.06542v1 Announce Type: new Abstract: Empirical scaling laws for language models have encouraged the development of ever-larger LLMs, despite their growing computational and memory costs. Sparse Mixture-of-Experts (MoEs) offer a promising alternative by activating only a subset of experts per...
FMI@SU ToxHabits: Evaluating LLMs Performance on Toxic Habit Extraction in Spanish Clinical Texts
arXiv:2604.06403v1 Announce Type: new Abstract: The paper presents an approach for the recognition of toxic habits named entities in Spanish clinical texts. The approach was developed for the ToxHabits Shared Task. Our team participated in subtask 1, which aims to...
The Illusion of Superposition? A Principled Analysis of Latent Thinking in Language Models
arXiv:2604.06374v1 Announce Type: new Abstract: Latent reasoning via continuous chain-of-thoughts (Latent CoT) has emerged as a promising alternative to discrete CoT reasoning. Operating in continuous space increases expressivity and has been hypothesized to enable superposition: the ability to maintain multiple...
To beat Altman in court, Musk offers to give all damages to OpenAI nonprofit
Musk won’t seek a “single dollar” in OpenAI suit after asking to pocket up to $134 billion.
Consistency-Guided Decoding with Proof-Driven Disambiguation for Three-Way Logical Question Answering
arXiv:2604.06196v1 Announce Type: new Abstract: Three-way logical question answering (QA) assigns $True/False/Unknown$ to a hypothesis $H$ given a premise set $S$. While modern large language models (LLMs) can be accurate on isolated examples, we identify two recurring failure modes in...
TalkLoRA: Communication-Aware Mixture of Low-Rank Adaptation for Large Language Models
arXiv:2604.06291v1 Announce Type: new Abstract: Low-Rank Adaptation (LoRA) enables parameter-efficient fine-tuning of Large Language Models (LLMs), and recent Mixture-of-Experts (MoE) extensions further enhance flexibility by dynamically combining multiple LoRA experts. However, existing MoE-augmented LoRA methods assume that experts operate independently,...
How our digital devices are putting our right to privacy at risk
Law professor Andrew Guthrie Ferguson chats with Ars about his new book,Your Data Will Be Used Against You.
SHAPE: Stage-aware Hierarchical Advantage via Potential Estimation for LLM Reasoning
arXiv:2604.06636v1 Announce Type: new Abstract: Process supervision has emerged as a promising approach for enhancing LLM reasoning, yet existing methods fail to distinguish meaningful progress from mere verbosity, leading to limited reasoning capabilities and unresolved token inefficiency. To address this,...
Improving Robustness In Sparse Autoencoders via Masked Regularization
arXiv:2604.06495v1 Announce Type: new Abstract: Sparse autoencoders (SAEs) are widely used in mechanistic interpretability to project LLM activations onto sparse latent spaces. However, sparsity alone is an imperfect proxy for interpretability, and current training objectives often result in brittle latent...
Inference-Time Code Selection via Symbolic Equivalence Partitioning
arXiv:2604.06485v1 Announce Type: new Abstract: "Best-of-N" selection is a popular inference-time scaling method for code generation using Large Language Models (LLMs). However, to reliably identify correct solutions, existing methods often depend on expensive or stochastic external verifiers. In this paper,...
Distributed Interpretability and Control for Large Language Models
arXiv:2604.06483v1 Announce Type: new Abstract: Large language models that require multiple GPU cards to host are usually the most capable models. It is necessary to understand and steer these models, but the current technologies do not support the interpretability and...
RAGEN-2: Reasoning Collapse in Agentic RL
arXiv:2604.06268v1 Announce Type: new Abstract: RL training of multi-turn LLM agents is inherently unstable, and reasoning quality directly determines task performance. Entropy is widely used to track reasoning stability. However, entropy only measures diversity within the same input, and cannot...
FLeX: Fourier-based Low-rank EXpansion for multilingual transfer
arXiv:2604.06253v1 Announce Type: new Abstract: Cross-lingual code generation is critical in enterprise environments where multiple programming languages coexist. However, fine-tuning large language models (LLMs) individually for each language is computationally prohibitive. This paper investigates whether parameter-efficient fine-tuning methods and optimizer...
Scientific Knowledge-driven Decoding Constraints Improving the Reliability of LLMs
arXiv:2604.06603v1 Announce Type: new Abstract: Large language models (LLMs) have shown strong knowledge reserves and task-solving capabilities, but still face the challenge of severe hallucination, hindering their practical application. Though scientific theories and rules can efficiently direct the behaviors of...
The Illusion of Stochasticity in LLMs
arXiv:2604.06543v1 Announce Type: new Abstract: In this work, we demonstrate that reliable stochastic sampling is a fundamental yet unfulfilled requirement for Large Language Models (LLMs) operating as agents. Agentic systems are frequently required to sample from distributions, often inferred from...
State-of-the-Art Arabic Language Modeling with Sparse MoE Fine-Tuning and Chain-of-Thought Distillation
arXiv:2604.06421v1 Announce Type: new Abstract: This paper introduces Arabic-DeepSeek-R1, an application-driven open-source Arabic LLM that leverages a sparse MoE backbone to address the digital equity gap for under-represented languages, and establishes a new SOTA across the entire Open Arabic LLM...
Attention Flows: Tracing LLM Conceptual Engagement via Story Summaries
arXiv:2604.06416v1 Announce Type: new Abstract: Although LLM context lengths have grown, there is evidence that their ability to integrate information across long-form texts has not kept pace. We evaluate one such understanding task: generating summaries of novels. When human authors...
ART: Attention Replacement Technique to Improve Factuality in LLMs
arXiv:2604.06393v1 Announce Type: new Abstract: Hallucination in large language models (LLMs) continues to be a significant issue, particularly in tasks like question answering, where models often generate plausible yet incorrect or irrelevant information. Although various methods have been proposed to...
TelcoAgent-Bench: A Multilingual Benchmark for Telecom AI Agents
arXiv:2604.06209v1 Announce Type: new Abstract: The integration of large language model (LLM) agents into telecom networks introduces new challenges, related to intent recognition, tool execution, and resolution generation, while taking into consideration different operational constraints. In this paper, we introduce...