issues
search
raoyongming
/
DynamicViT
[NeurIPS 2021] [T-PAMI] DynamicViT: Efficient Vision Transformers with Dynamic Token Sparsification
https://dynamicvit.ivg-research.xyz/
MIT License
576
stars
72
forks
source link
issues
Newest
Newest
Most commented
Recently updated
Oldest
Least commented
Least recently updated
Question about the keep/drop token
#48
King4819
opened
6 months ago
0
About BP problem mentioned in the introduction
#47
Cooperx521
opened
7 months ago
2
About BP problem
#46
Cooperx521
closed
7 months ago
0
Token keep ratio setting
#45
King4819
opened
7 months ago
0
RuntimeError: Expected all tensors to be on the same device, but found at least two devices, cuda:0 and cuda:1! (when checking argument for argument index in method wrapper_scatter__value)
#44
King4819
closed
8 months ago
2
Gradient accumulation code implement
#43
King4819
opened
10 months ago
0
DeiT lr question
#42
King4819
closed
11 months ago
1
Reproduce the result on single gpu
#41
King4819
closed
11 months ago
6
Flops after discarding tokens
#40
King4819
closed
1 year ago
1
ViT on CIFAR-100
#39
King4819
closed
1 year ago
7
Dynamic token on training speedup
#38
ZK-Zhou
closed
1 year ago
1
About some innovative ideas
#37
vegetableclean
opened
1 year ago
0
．
#36
vegetableclean
closed
1 year ago
0
Regarding connection elimination of the dropped patches.
#35
Alihjt
closed
1 year ago
3
MMdetection version
#34
Rkyzzy
opened
1 year ago
0
A few question about Deit Model used in Dynamic Vit
#33
secretu
closed
1 year ago
1
The problem of giving up a fixed number of tokens during the inference stage.
#32
MooresS
closed
1 year ago
1
Reproducing throughput results
#31
ivanke1
closed
1 year ago
1
Any hints on Batch=1 inference?
#30
lixinghe1999
closed
1 year ago
2
Code for Table 1 in the paper
#29
xXuHaiyang
closed
1 year ago
0
请教一下关于class “AdaSwinTransformerBlock"中 forward function 中x1，x2的问题
#28
LucasZhan
closed
1 year ago
3
Fail to reproduce accuracy of DynamicViT-B/0.7: lower accuracy than reported
#27
ShiFengyuan1999
closed
1 year ago
8
Temperature in Gumbel Softmax
#26
kaikai23
closed
1 year ago
1
[Questions] Dyswin feature output shape
#25
zafirshi
closed
2 years ago
2
Is the subscript 'i' of Z_global in equation 4 a mis-type?
#24
LucasZhan
closed
2 years ago
2
Implementation details are so largely different from the paper description
#23
ming1993li
closed
2 years ago
4
Some questions about your code
#22
leoozy
closed
2 years ago
2
Have you tested your latency on GPU?
#21
leoozy
closed
2 years ago
1
update
#20
liuzuyan
closed
2 years ago
0
Flops tools
#19
waynelrs
closed
2 years ago
2
DynamicVIT training stored checkpoint
#18
SwapnilDreams100
closed
2 years ago
2
Can't reproduce the accuracy of pre-trained models
#17
xiyiyia
closed
2 years ago
2
论文细节请教
#16
wangning7149
closed
2 years ago
1
关于训练和测试阶段裁剪token的策略
#15
1171000410
closed
2 years ago
6
GFLOPs and Throughput
#14
AmeenAli
closed
2 years ago
2
[ LV-ViT-S Pretrained Model ]
#13
IemProg
closed
2 years ago
1
About distill
#12
hegc
closed
2 years ago
2
Attention mask computation during training
#11
mtchiu2
closed
2 years ago
2
关于block中的残差通道
#10
bestfleer
closed
2 years ago
1
FLOPs
#9
Cydia2018
closed
2 years ago
2
test-pr
#8
dirtycomputer
closed
2 years ago
1
pretrained model download
#7
dirtycomputer
closed
2 years ago
2
Loss is nan when training my own dataset
#6
InfinityBox
closed
2 years ago
2
Structural downsampling and static token sparsification
#5
Yeez-lee
opened
3 years ago
3
Pretrain LV-ViT-S and LV-ViT-M
#4
kristenkn
closed
3 years ago
1
关于Attention Masking创新点的疑问
#3
xmu-xiaoma666
closed
3 years ago
3
update visualization example
#2
wl-zhao
closed
3 years ago
0
visualization
#1
wl-zhao
closed
3 years ago
0