5 papers
Hardening Agent Benchmarks with Adversarial Hacker-Fixer Loops
Ziqian Zhong, Ivgeni Segal, Ivan Bercovich +3
Agent benchmarks score submissions with outcome verifiers that are typically hand-written and brittle, leaving them open to reward hacking. We audit 1,968 tasks across five termina…
Terminal Wrench: A Dataset of 331 Reward-Hackable Environments and 3,632 Exploit Trajectories
Ivan Bercovich, Ivgeni Segal, Kexun Zhang +3
We release Terminal Wrench, a subset of 331 terminal-agent benchmark environments, copied from the popular open benchmarks that are demonstrably reward-hackable. The data set inclu…
Is Vibe Coding Safe? Benchmarking Vulnerability of Agent-Generated Code in Real-World Tasks
Songwen Zhao, Danqing Wang, Kexun Zhang +3
Vibe coding is a new software development paradigm in which human engineers prompt a large language model (LLM) agent to complete complex coding tasks with little supervision. Alth…
HardTests: Synthesizing High-Quality Test Cases for LLM Coding
Zhongmou He, Yee Man Choi, Kexun Zhang +6
Verifiers play a crucial role in large language model (LLM) reasoning, needed by post-training techniques such as reinforcement learning. However, reliable verifiers are hard to ge…
Revealing the Barriers of Language Agents in Planning
Jian Xie, Kexun Zhang, Jiangjie Chen +5
Autonomous planning has been an ongoing pursuit since the inception of artificial intelligence. Based on curated problem solvers, early planning agents could deliver precise soluti…