issues
search
facebookresearch
/
ToMe
A method to increase the speed and lower the memory footprint of existing vision transformers.
Other
916
stars
67
forks
source link
issues
Newest
Newest
Most commented
Recently updated
Oldest
Least commented
Least recently updated
Question regarding scatter_reduce
#43
AshStuff
closed
3 weeks ago
0
About using ToMe in ImageEncoderViT within segment anything
#42
yangzijia
opened
1 month ago
2
Add CUDA extensions
#41
vinnamkim
closed
1 month ago
0
AttributeError: 'ToMeVisionTransformer' object has no attribute 'blocks'
#40
Guoli-w
opened
2 months ago
3
End-to-end inference doesn't be accelerated.
#39
rwshihhh
opened
2 months ago
3
About inference acceleration for multi-modal large language models
#38
JoaquinChou
closed
4 months ago
3
Video Classification off-the-shelf model
#37
JoakimHaurum
opened
9 months ago
0
How to implement ToMe to image encoders in CLIP model?
#36
Bostoncake
closed
1 month ago
1
Training Speedup for ToMe with Video
#35
dfan
closed
10 months ago
3
How to restore it to its original shape
#34
sweetwaterrain
closed
1 month ago
1
Apply Flash-attn for MAE-based models.
#33
JerryFlymi
opened
11 months ago
3
Has anyone been able to successfully convert the timm+tome model to onnx?
#32
Roman54228
closed
11 months ago
2
How do you understand the phrase "This can change the outcome of softmax attention: if we merge two tokens with the same key, that key has less effect in the softmax term" in Part Tracking Token Size ?
#31
Nanuion
closed
12 months ago
5
Used in the vanilla Transformer
#30
yuzhenmao
closed
9 months ago
6
Can ToMe be applied together with RoPE?
#29
cjdjr
closed
1 month ago
1
hyper-parm "r" in ViT-g/14
#28
rese1f
closed
1 year ago
4
Does ToMe work for Restormer?
#27
tttt-0814
closed
1 year ago
2
I keep getting this issue, how do I fix it?
#26
303Aki303
closed
1 year ago
1
(Question) LayerNorm for queries and keys
#25
sanagno
closed
1 year ago
2
Does ToMe work for focal modulation networks?
#24
subneed
closed
1 year ago
1
Does ToMe work for focal modulation networks?
#23
subneed
opened
1 year ago
2
How exactly tokens are reduced while there is no change in your model dimensions before and after tome.patch
#22
amirrezarajabi
closed
1 year ago
4
[WIP] Patch for CLIP
#21
FrancescoSaverioZuppichini
opened
1 year ago
9
pretrained model for ViT-L on Kinetics-400
#20
ChenMnZ
closed
1 year ago
1
eucl distance function in Tab. 1b
#19
ChenMnZ
closed
1 year ago
2
Clarify MAE ViT-L numbers in the README
#18
dbolya
closed
1 year ago
3
The difference between 85.96 and 85.66 original MAE accuracy
#17
LiuDongyang6
closed
1 year ago
2
Is ToMe sensitive to the image resolution?
#16
numb3r3
closed
1 year ago
1
Does not support swin-transformer?
#15
YoloEliwa
closed
1 year ago
1
Is there any way to generate heatmap with ToMe
#14
Nightwatch-Fox11
opened
1 year ago
0
Training with merging
#13
WayneCV
closed
1 year ago
4
AttributeError: 'ToMeVisionTransformer' object has no attribute 'norm'
#12
vignagajan
closed
1 year ago
1
Some doubts about Accuracy
#11
HelloWXJ1024
closed
1 year ago
3
How can I use ToMe under pytorch 1.10.1
#10
Nightwatch-Fox11
closed
1 year ago
2
About keeping token order
#9
Chenfeng1271
closed
1 year ago
2
ToMeBlock cannot be used with torch.utils.checkpoint
#8
powermano
closed
1 year ago
4
Add Stable Diffusion Patch + Notebook Example
#7
lalalune
opened
1 year ago
4
Initial pass at ToMe-ing Stable Diffusion unets
#6
blindcrone
closed
1 year ago
1
AttributeError: 'ToMeBlock' object has no attribute 'drop_path'
#5
kos94ok
closed
1 year ago
7
Stable Diffusion
#4
lalalune
closed
1 year ago
16
Candidate fix for a missing distillation token
#3
dbolya
closed
1 year ago
0
Adding support for HuggingFace vision Transformers
#2
NielsRogge
opened
1 year ago
3
`ToMeTransformer` object has no attribute 'dist_token'
#1
Raghvender1205
closed
1 year ago
12