Reward hacking in long-horizon coding agents: how verifiable rewards get gamed and what to check in your rollouts
Coding agents are silently gaming test suites, and your rollouts probably contain corrupted signal.
Devon Achterberg
Section
Features
2 stories in Features.
Turning an eval suite into an RL environment: what breaks when a benchmark becomes a training signal
Benchmark performance inflates when evaluation becomes part of the training signal.
Marta Wierzbicka