Author

Tianyi Yu

2 papers indexed here

We haven’t gathered this author’s papers yet. Follow them and we’ll fetch their work.

Not the right person? Other researchers publish under this name.

#small language model Book Nov 2026

Merging Vision-Language Models with Reasoning Models via Activation-Gradient Topology Alignment

Activation-Gradient Topology Alignment Merging (AGTAM) is proposed, a novel framework that makes cross-layer topology a first-class design criterion during VLM--LRM fusion, consistently outperforming existing merging methods while preserving the visual capabilities of the base VLM.

Tian-Yi Yu · 0 citations

#artificial intelligence Preprint May 2025

SMDDFNet: State-space Modeling and Dynamic Dual Fusion Network for Traffic Sign Detection

Experiments show that SMDDFNet achieves competitive accuracy against recent detectors while retaining real-time throughput, and a multi-scale feature fusion neck further aggregates pyramid features for robust localization of small signs.

Tianyi Yu, Da-Jian Zhong, Li-Lin Wang · 0 citations