…View arXiv page View PDF GitHub 2 Add to collection Community Automatically hardening benchmarks and training environments with the hacker–fixer loop. Paper: https://arxiv.org/abs/2606.08960 Code: https://github…
…To demonstrate its utility, we analyze different judge biases from the perspectives of discoverability and exploitability, and explore an agent-based system for automatically detecting reward hacking onset from training logs . The…
…This is an automated message from the Librarian Bot . I found the following papers similar to this paper. The following papers were recommended by the Semantic Scholar API Reward Hacking in the…
…This is an automated message from the Librarian Bot . I found the following papers similar to this paper. The following papers were recommended by the Semantic Scholar API SpecBench: Measuring Reward Hacking…
…First, we formalize Surrogate Objective Hacking : a differentiable softmax router exposed to the PPO surrogate receives a direct gradient toward advantage heads that are numerically favorable for the current update, even when…
…Observability-Driven Automatic Evolution of Coding-Agent Harnesses (2026) Reward Hacking Benchmark: Measuring Exploits in LLM Agents with Tool Use (2026) Continual Harness: Online Adaptation for Self-Improving Foundation Agents (2026) RoboPhD…
…reasoning over search regularization, thereby avoiding reward hacking. Extensive experiments demonstrate that SAAS substantially reduces over-search, while maintaining accuracy. This is an automated message from the Librarian Bot . I found the…
…i'd love to see an ablation on gating strength across data scales to confirm it's the main driver rather than a stability hack. the arXivLens breakdown helped me parse the…
…View arXiv page View PDF Add to collection Community Trust-Region Behavior Blending for On-Policy Distillation https://x.com/AMyashka/status/2061431653797425288 This is an automated message from the Librarian Bot…
…This rubric reward is applied only to responses with correct final answers (positive-only strategy), distinguishing the reasoning quality among correct responses and preventing reward hacking . Experiments on three reasoning LLMs (4B…