2025. A Benchmark Study of Hybrid CNN-Transformer Architectures in Vision-Language Tasks. Emerging Science Research. 3, 01 (Jun. 2025), 36–49.