AI
ReguSim: Evaluating LLM Agent Rule Grounding in Financial Compliance
Researchers have developed ReguSim, a controlled environment for testing the ability of large langua...
AI
Optimal Skill Selection for LLM Agents with Provable Bicriteria Guarantees
Large language model (LLM) agents acquire task-specific capabilities by loading reusable skill docum...
AI
ExPhy: A Benchmark for Explicit Physical Property Learning in Multi-Object Trajectory Forecasting
Researchers have created a new benchmark called ExPhy to help evaluate AI models' ability to learn a...
AI
Manifold Drift in Flow Preference Optimization: A Root Cause of Reward Hacking
Researchers have identified a problem in preference optimization for generative models, known as man...
AI
Contrastive Mixed Prompt Learning for Incomplete Multimodal Sentiment Analysis with Unseen Modality Combination
Researchers have proposed a new model called CMPL (Contrastive Mixed Prompt Lear...
AI
A three-dimensional typology of agency for advanced AI systems
Researchers have proposed a three-dimensional typology for understanding the con...
AI
On the Applicability of Safety Nets: A Safety-By-Design Solution for Certifying Neural Networks
Researchers have developed a 'safety net' system to ensure the reliability of AI...
AI
What You Can't See Is What You Learn: Restricted Evidence Visibility Favors Compositional Generalization in Shared-Genome Language-Model Societies
Researchers have found that restricting what a shared language model can see in ...