From Correlation to Causation: Max-Pooling-Based Multi-Instance Learning Leads to More Robust Whole Slide Image Classification Paper • 2408.09449 • Published Dec 5, 2025 • 1
PhyRubric: Rubric-Grounded Diagnostic Evaluation of Physical Commonsense in Text-to-Video Generation Paper • 2611.13576 • Published 6 days ago • 13
SWE-Together: Evaluating Coding Agents in Interactive User Sessions Paper • 2606.29957 • Published Jun 29 • 15
From Sight to Foresight: Predictive Spatial Reasoning in Vision-Language Models Paper • 2610.04139 • Published Oct 2 • 10
TraceDance: An Automated System for Building Agent Behavior Benchmarks from Real-World Agent Deployment Traces Paper • 2609.33295 • Published Sep 27 • 62
Automating ML for Science: Can Frontier Agents Climb Scientific Hills in the Wild? Paper • 2610.28763 • Published 27 days ago • 19
T2J-BENCH: Benchmarking Coding Agents on Execution-Grounded PyTorch-to-JAX Repository Migration Paper • 2605.29054 • Published Oct 14 • 8
Self-Retrospection Distillation: Turning Post-hoc Experiences into Prior Foresight Paper • 2610.08077 • Published Oct 6 • 173
X-Tree: Tokenizing Reusable Experience for Efficient Agent Generalization Paper • 2609.32993 • Published Sep 26 • 54
TimeEvo: Failure-Driven Self-Evolution of a Time Series Agent Paper • 2609.27277 • Published Sep 23 • 28
TraceDance: An Automated System for Building Agent Behavior Benchmarks from Real-World Agent Deployment Traces Paper • 2609.33295 • Published Sep 27 • 62