Accès ouvert
2026
preprint
OpenAlex
Harsh Raj, David Lee, Anas Mahmoud, Renxiong Wang et autres
The increasing deployment of AI agents in long-horizon tasks yields massive execution logs. Diagnosing failures within these records is crucial for reliability, as it transforms outcome-level signals into actionable interventions. The sheer scale of the data renders human review impractical, driving the …
Accès ouvert
2026
preprint
OpenAlex
Harsh Raj, Vipul Gupta, Anas Mahmoud, Razvan-Gabriel Dumitru et autres
Existing evaluations often reduce agent failures to system-level outcomes, obscuring where the fault originated and which intervention would improve the agent system. This creates a repair-assignment problem: the same visible failure may call for model post-training, harness engineering, environment redesign, or benchmark …
Accès ouvert
2026
preprint
OpenAlex
MohammadHossein Rezaei, Anas Mahmoud, Zihao Wang, Utkarsh Tyagi et autres
Rubrics have emerged as an alternative to RLVR in open-ended domains where a single ground-truth final answer is not available. Existing rubric-based training methods rely on an LLM verifier that scores each rollout against rubrics. This introduces substantial training-time overhead, exposes optimization …
2026
dissertation
OpenAlex
Anas Mahmoud
3D scene understanding, enabled by camera and lidar sensors, is crucial for autonomous driving 3D tasks like object detection and segmentation. These sensors have complementary strengths, with lidar providing accurate depth data but returning sparse point clouds, while cameras deliver dense, color-rich …
Accès ouvert
2026
preprint
OpenAlex
Utkarsh Tyagi, Xingang Guo, MohammadHossein Rezaei, Daniel George et autres
Reinforcement learning with verifiable rewards has made post-training highly effective when correctness can be checked automatically. However, many important model behaviors require satisfying several qualitative criteria at once. Rubric-based rewards address this setting by grading prompt-specific criteria and aggregating them into a …
Accès ouvert
2026
preprint
OpenAlex
Anas Mahmoud, MohammadHossein Rezaei, Zihao Wang, Anisha Gunjal et autres
Reinforcement learning with verifiable rewards has enabled strong post-training gains in domains such as math and coding, though many open-ended settings rely on rubric-based rewards. We study reward hacking in rubric-based RL, where a policy is optimized against a training verifier but …