Picture for Kai Han

Kai Han

A Survey on 3D Human Avatar Modeling -- From Reconstruction to Generation

Jun 06, 2024
Viaarxiv icon

GeminiFusion: Efficient Pixel-wise Multimodal Fusion for Vision Transformer

Add code
Jun 04, 2024
Viaarxiv icon

SA-GS: Semantic-Aware Gaussian Splatting for Large Scene Reconstruction with Geometry Constrain

Add code
May 28, 2024
Viaarxiv icon

SciFIBench: Benchmarking Large Multimodal Models for Scientific Figure Interpretation

Add code
May 14, 2024
Viaarxiv icon

EMS-SD: Efficient Multi-sample Speculative Decoding for Accelerating Large Language Models

Add code
May 13, 2024
Viaarxiv icon

Memory-Space Visual Prompting for Efficient Vision-Language Fine-Tuning

Add code
May 09, 2024
Viaarxiv icon

Kangaroo: Lossless Self-Speculative Decoding via Double Early Exiting

Add code
Apr 29, 2024
Viaarxiv icon

GhostNetV3: Exploring the Training Strategies for Compact Models

Add code
Apr 17, 2024
Viaarxiv icon

SPTNet: An Efficient Alternative Framework for Generalized Category Discovery with Spatial Prompt Tuning

Add code
Mar 20, 2024
Figure 1 for SPTNet: An Efficient Alternative Framework for Generalized Category Discovery with Spatial Prompt Tuning
Figure 2 for SPTNet: An Efficient Alternative Framework for Generalized Category Discovery with Spatial Prompt Tuning
Figure 3 for SPTNet: An Efficient Alternative Framework for Generalized Category Discovery with Spatial Prompt Tuning
Figure 4 for SPTNet: An Efficient Alternative Framework for Generalized Category Discovery with Spatial Prompt Tuning
Viaarxiv icon

DenseMamba: State Space Models with Dense Hidden Connection for Efficient Large Language Models

Add code
Mar 05, 2024
Figure 1 for DenseMamba: State Space Models with Dense Hidden Connection for Efficient Large Language Models
Figure 2 for DenseMamba: State Space Models with Dense Hidden Connection for Efficient Large Language Models
Figure 3 for DenseMamba: State Space Models with Dense Hidden Connection for Efficient Large Language Models
Figure 4 for DenseMamba: State Space Models with Dense Hidden Connection for Efficient Large Language Models
Viaarxiv icon