TY  - RPRT
TI  - Evaluating the Scalability and Adversarial Generalization of GRPO-Trained NLI Models
AU  - Pablo Miralles-González
AU  - Javier Huertas-Tato
AU  - Alejandro Martín
AU  - David Camacho
PY  - 2026
UR  - https://arxiv.org/abs/2504.18376
ID  - 2504.18376
ER  -