arXiv ·
BanglaShield: A Script-Robust Safety Classifier and Benchmark for Bangla Native, Romanized, and Code-Mixed Prompts
arXiv 的公开论文，完整内容可在来源页面查看。
首页 / 浏览
全部内容
浏览全部已收录的公开内容，包括尚未归入主分类的条目。
10330 条公开内容
arXiv ·
Beyond Clean Accuracy: The Practical Challenges and Benefits of Pixel-Based Language Models
arXiv 的公开论文，完整内容可在来源页面查看。
arXiv ·
What Matters for NVFP4 Training? A Comparative Study of Low-Precision Pretraining Recipes
arXiv 的公开论文，完整内容可在来源页面查看。
arXiv ·
FlashMap: Flow Map Speculative Decoding
arXiv 的公开论文，完整内容可在来源页面查看。
arXiv ·
Ask Less, Ask Better: Allocating User Attention in Proactive LLM Agents with Interruption Budgets
arXiv 的公开论文，完整内容可在来源页面查看。
arXiv ·
Toward a Universal Local Speech Feature Extractor Through Distillation
arXiv 的公开论文，完整内容可在来源页面查看。
arXiv ·
When Opposite Premises Translate Identically: A Source-Conditioned Diagnostic for Multilingual Commonsense Items
arXiv 的公开论文，完整内容可在来源页面查看。
arXiv ·
Which Long Documents Matter? A Controlled Study of Data Selection for Context Extension
arXiv 的公开论文，完整内容可在来源页面查看。
arXiv ·
GardenBench: A lightweight, daily evaluation of LLM capabilities
arXiv 的公开论文，完整内容可在来源页面查看。
arXiv ·
Doing Less with More: A First-Principles Exploration of the Suitability of Agentic Computing over Alternative Architectural Choices
arXiv 的公开论文，完整内容可在来源页面查看。
arXiv ·
A Controlled LLM Workflow for Evidence-Grounded Diagnosis of Risk Score Differences
arXiv 的公开论文，完整内容可在来源页面查看。
arXiv ·
What Social Media Teaches Us About Accumulative AI Risks
arXiv 的公开论文，完整内容可在来源页面查看。
arXiv ·
Beyond Explicit Intent: Discovering and Shaping Artistic Intent through Human-AI Co-Creation
arXiv 的公开论文，完整内容可在来源页面查看。
arXiv ·
Unified Multiparadigm Fusion Intelligence for Rare Disease Diagnosis
arXiv 的公开论文，完整内容可在来源页面查看。
arXiv ·
LogicVal:A Logic-Formalized Benchmark for Constraint Satisfaction in Text-to-Image Generation
arXiv 的公开论文，完整内容可在来源页面查看。
arXiv ·
World of SWECraft: Agent Collaboration at Scale via Simulated Software Worlds
arXiv 的公开论文，完整内容可在来源页面查看。
arXiv ·
Persistent Memory and Iterative Improvement in an LLM Agent for Chinese-Character Drawing
arXiv 的公开论文，完整内容可在来源页面查看。
arXiv ·
Cache2LoRA: Understanding and Improving Context Internalization
arXiv 的公开论文，完整内容可在来源页面查看。
arXiv ·
Joint, Decoupled, or Agent: When Each Multimodal Alignment Paradigm Wins
arXiv 的公开论文，完整内容可在来源页面查看。
arXiv ·
A Conditional No-Hair Theorem for Transformers: How Language Models Forget Their Prompt
arXiv 的公开论文，完整内容可在来源页面查看。
arXiv ·
TorchTPU - A Hybrid Eager/Compiled Architecture for Native PyTorch Execution on TPU
arXiv 的公开论文，完整内容可在来源页面查看。
arXiv ·
Self-Improving Video-to-Action Models
arXiv 的公开论文，完整内容可在来源页面查看。
arXiv ·
Agents' Last Doubt: Knowing When to Act and When to Step Back in Tool-Using Agents
arXiv 的公开论文，完整内容可在来源页面查看。
arXiv ·
Distillation, Self-Training, and the Fragility of Learned Chemical Reasoning
arXiv 的公开论文，完整内容可在来源页面查看。
arXiv ·
Training-Free Prompt Selection for Efficient Reinforcement Learning With Partial Rollouts
arXiv 的公开论文，完整内容可在来源页面查看。
arXiv ·
Probing Ambiguity: Adversarial Intent Resolution for Robust Text-to-SQL
arXiv 的公开论文，完整内容可在来源页面查看。
arXiv ·
Cortex: A Fixed-Point Theory of Governed Coding Agents - Technical Report
arXiv 的公开论文，完整内容可在来源页面查看。
arXiv ·
AM-Bench: A Unified Taxonomy and Evaluation Suite for Agentic Misalignment
arXiv 的公开论文，完整内容可在来源页面查看。
arXiv ·
Can Compact Open-Weight Large Language Models Recover Mobility Events from Incomplete Trajectories? A Controlled Evaluation Using Air Trips
arXiv 的公开论文，完整内容可在来源页面查看。
arXiv ·
From SEC Filings to Financial Signals: Evaluating LLMs on Stock Market Reactions and Analyst Forecasts
arXiv 的公开论文，完整内容可在来源页面查看。