Terry Yue Zhuo

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

Cyber-Zero: Training Cybersecurity Agents without Runtime

HackWorld: Evaluating Computer-Use Agents on Exploiting Web Application Vulnerabilities

Training Language Model Agents to Find Vulnerabilities with CTF-Dojo

A Mixture of Linear Corrections Generates Secure Code

BigCodeArena: Unveiling More Reliable Human Preferences in Code Generation via Execution

BigCodeBench: Benchmarking Code Generation with Diverse Function Calls and Complex Instructions

Unmasking the Version-Switching Capabilities of Code Generation Models