AI.info
Yuki M. Asano
Explore Yuki M. Asano on AI.info.
- AVQ-Attention: Adaptive Vector-Quantized Attention
- 3D sans 3D Scans: Scalable Pre-training from Video-Generated Point Clouds
- Same Content, Different Answers: Cross-Modal Inconsistency in MLLMs
- Elastic ViTs from Pretrained Models without Retraining
- What Layers When: Learning to Skip Compute in LLMs with Residual Gates