LeapLabTHU/DAT
Repository of Vision Transformer with Deformable Attention (CVPR2022) and DAT++: Spatially Dynamic Vision Transformerwith Deformable Attention
PythonApache-2.0
Issues
- 8
How to visualize the learned offsets?
#11 opened by zzwei1 - 0
- 0
- 5
- 1
处理文本数据或一维序列数据
#32 opened by zhengchunge - 3
how to visualize Figure 5,Figure 6, Figure7
#15 opened by chenxinhe1 - 1
- 1
- 0
Deformable Attention Journal Paper not referenced
#21 opened by kumartr - 1
About Low Accuracy
#20 opened by z1zzzz1 - 1
Controlling the number of keys per query
#26 opened by tdchua - 1
How to run the model
#30 opened by SITUSITU - 1
About the range of offset
#31 opened by Chris-city - 1
- 1
- 1
您好,DAT可以当作即插即用的模块吗
#29 opened by zi1zi - 6
- 2
RuntimeError: Trying to create tensor with negative dimension -96: [-96, 1, 9, 9]
#4 opened by nullxjx - 1
- 1
训练时计算维度出错
#22 opened by JusticeLin - 3
- 3
- 2
Changing model input size from 384 -> 1024
#19 opened by tdchua - 1
- 2
The computational cost of deformabel attention
#16 opened by linjing7 - 1
what does -1 mean in strides and groups variable
#14 opened by Hua-YS - 1
训练问题
#8 opened by xiaojin-sfe - 1
固定输入图片尺寸
#7 opened by houjie8888 - 4
Pretrain Model
#6 opened by xxtars - 1
- 1
Error: Trying to create tensor with negative dimension -96: [-96, 1, 9, 9]
#10 opened by ChidanandKumarKS - 1
displacement problem
#3 opened by ZhuangLii - 1
Different depthwise convolution kernel sizes?
#5 opened by xskxzr - 1
build_loader function missing
#1 opened by arezaz