Bai L, Xiong Z, Lin H, Xu G, Xie X, Guo R, et al. Frozen Language Models Are Gradient Coherence Rectifiers in Vision Transformers. AAAI [Internet]. 2025 Apr. 11 [cited 2026 May 12];39(2):1817-25. Available from: https://ojs.aaai.org/index.php/AAAI/article/view/32176