E. Kelly Buchanan

Terminal-Bench: Benchmarking Agents on Hard, Realistic Tasks in Command Line Interfaces

Extracting task-relevant preserved dynamics from contrastive aligned neural recordings

Weaver: Shrinking the Generation-Verification Gap by Scaling Compute for Verification

Archon: An Architecture Search Framework for Inference-Time Techniques

An Architecture Search Framework for Inference-Time Techniques

On the interplay between learning and memory in deep state space models