lucidrains/vit-pytorch issues and pull requests

#341 - remove duplicated qkv computation in na_vit_nested_tensor_3d.py

Pull Request - State: closed - Opened by JacobLinCool 6 months ago - 2 comments

#340 - How to train？

Issue - State: open - Opened by LiJichen0114 7 months ago - 2 comments

#339 - Add option to set frame padding for 3D CCT

Pull Request - State: closed - Opened by kalekundert 7 months ago - 2 comments

#337 - How can I use a 1D Vision Transformer?

Issue - State: open - Opened by SmallPotato705 7 months ago - 1 comment

#336 - Allow to use classification token in CvT?

Issue - State: open - Opened by Yash-10 8 months ago

#335 - Why not always project out from Attention block?

Issue - State: open - Opened by fiskrt 8 months ago - 1 comment

#334 - Modified the forward function of Attention so that it can operate on …

Pull Request - State: open - Opened by EdgeObserver 9 months ago

#333 - try again

Pull Request - State: closed - Opened by lucidrains 9 months ago

#332 - train navit nest 3d error when backward

Issue - State: closed - Opened by HaloTrouvaille 9 months ago - 5 comments

#330 - RegionViT - Local token embedding

Issue - State: closed - Opened by minhquoc0712 10 months ago - 1 comment

#329 - The Total params: and Params size (MB) of the model printed by summary are different from the bit_base model in timm library. Theoretically, the same settings should be the same. What is the reason?

Issue - State: open - Opened by lucker26 10 months ago - 1 comment

#328 - MS-COCO training from Imagenet pretrained checkpoint

Issue - State: open - Opened by prateekiiest 10 months ago - 1 comment

#327 - Add ViViT variant with factorized self-attention

Pull Request - State: closed - Opened by roydenwa 11 months ago - 1 comment

#326 - update dep

Pull Request - State: closed - Opened by lucidrains 11 months ago

#325 - Nested navit

Pull Request - State: closed - Opened by lucidrains 11 months ago

#324 - Update distill.py to include device agnostic code for `distill_mlp` head and `distillation_token`

Pull Request - State: open - Opened by vivekh2000 about 1 year ago

#323 - Update simple_flash_attn_vit_3d.py

Pull Request - State: open - Opened by zhulinchng about 1 year ago

#322 - Multi-GPU training of NaViT model

Issue - State: closed - Opened by b5y about 1 year ago - 1 comment

#321 - Weight Initialization

Issue - State: open - Opened by simonaay about 1 year ago

#320 - SimpleViT misleading summary

Issue - State: open - Opened by asusdisciple about 1 year ago

#319 - anyone knows why _freeze_stages() starts from block[0]?

Issue - State: open - Opened by abc5z7 about 1 year ago

#318 - Update distill.py

Pull Request - State: open - Opened by vivekh2000 about 1 year ago

#317 - Update vit.py to include LayerNorm in the MLP head which is missing

Pull Request - State: open - Opened by vivekh2000 about 1 year ago - 1 comment

#315 - Choice for reduced order model / latent space

Issue - State: open - Opened by ramdhan1989 about 1 year ago

#314 - [MaxViT] Block/Grid Attention question

Issue - State: open - Opened by sonderlau about 1 year ago

#313 - Layer Norm modification

Pull Request - State: open - Opened by RyanKim17920 about 1 year ago - 2 comments

#312 - Swin UNet

Issue - State: open - Opened by sibi-venti about 1 year ago

#311 - Validation accuracy higher than training accuracy

Issue - State: closed - Opened by yoder460 about 1 year ago - 1 comment

#310 - Patch Embedding Design Choice?

Issue - State: open - Opened by tonyyunyang about 1 year ago

#309 - Why Remove PreNorm?

Issue - State: closed - Opened by tonyyunyang about 1 year ago

#308 - Request for Pre-trained Weights for Vit

Issue - State: open - Opened by ZSLsherly over 1 year ago

#307 - Whether to include pre-trained models

Issue - State: closed - Opened by KawaiiAsh over 1 year ago - 1 comment

#306 - Non-deterministic results based on group_max_seq_len in NaViT

Issue - State: closed - Opened by dempsey-ryan over 1 year ago - 3 comments

#304 - CrossViT does not handle other than three channel images

Issue - State: closed - Opened by Yash-10 over 1 year ago - 2 comments

#303 - Fix #302 Rendering Readme for project description on PyPi publish

Pull Request - State: closed - Opened by soumya1729 over 1 year ago - 3 comments

#302 - PyPi page markdown render

Issue - State: closed - Opened by soumya1729 over 1 year ago - 1 comment

#301 - display random images when previewing in cats_and_dogs.ipynb

Pull Request - State: open - Opened by berinaniesh over 1 year ago

#300 - Cuda memory for 3D VIT

Issue - State: open - Opened by JesseZZZZZ over 1 year ago - 2 comments

#298 - A question with ViT 3d

Issue - State: closed - Opened by JesseZZZZZ over 1 year ago

#297 - Add implementation of LongVit

Issue - State: closed - Opened by jpfeil over 1 year ago - 4 comments

#296 - Problems regarding training 3D Vision transformer : model does not converge

Issue - State: open - Opened by Uljibuh over 1 year ago - 1 comment

#295 - Multi-target Regression Question

Issue - State: open - Opened by stethemJ over 1 year ago

#294 - can we use CvT model for segmentation?

Issue - State: open - Opened by HawkingRadiation42 over 1 year ago

#293 - Masking attention with batches

Issue - State: open - Opened by ashrafflh over 1 year ago

#292 - Question regarding 1d fft use

Issue - State: closed - Opened by chengengliu over 1 year ago - 1 comment

#291 - Trouble loading ViT - Dino structure for channels>3?

Issue - State: open - Opened by AgentM-GEG over 1 year ago

#290 - First attempt

Pull Request - State: open - Opened by jefferson-bercaw over 1 year ago

#289 - Questions about distill_loss

Issue - State: open - Opened by haoren55555 over 1 year ago - 3 comments

#288 - how to train

Issue - State: open - Opened by lingxitong over 1 year ago - 2 comments

#287 - Layernorm in Cross attention

Issue - State: closed - Opened by turtleman99 almost 2 years ago - 4 comments

#286 - CvT with 1 channel input data

Issue - State: closed - Opened by tranlg99 almost 2 years ago - 2 comments

#285 - Fix typo in vit_1d LayerNorm

Pull Request - State: closed - Opened by l0wgear almost 2 years ago - 1 comment

#284 - add xcit

Pull Request - State: closed - Opened by lucidrains almost 2 years ago

#283 - Update README.md

Pull Request - State: closed - Opened by EIFY almost 2 years ago - 1 comment

#282 - Not correctly understanding the Multi Head Attention part of the ViT implementation...

Issue - State: closed - Opened by JavierUrenaPhDProjects almost 2 years ago - 3 comments

#281 - Potential regression with PT 2.0 and CUDA 12.2/CuDNN 8.9.4

Issue - State: closed - Opened by roywei almost 2 years ago - 1 comment

#280 - Using vision transformers for different image resolutions

Issue - State: open - Opened by Oussamab21 almost 2 years ago - 1 comment

#279 - vit_pytorch -> cross_vit.py(mistake)

Issue - State: closed - Opened by RufusRubin almost 2 years ago - 1 comment

#278 - Saving and loading model seems to be regressing to lower performance

Issue - State: closed - Opened by aperiamegh almost 2 years ago - 1 comment

#277 - structural 3D ViT

Issue - State: closed - Opened by aperiamegh almost 2 years ago - 4 comments

#276 - This ViT implementation as generative network

Issue - State: open - Opened by MrCorsair3 almost 2 years ago - 1 comment

#275 - TVM compilation failed on SimpleViT

Issue - State: open - Opened by yangxin0926 almost 2 years ago

#274 - Update vit.py

Pull Request - State: closed - Opened by LuYuchenOrRobert almost 2 years ago - 4 comments

#273 - begin work on NaViT (wip)

Pull Request - State: closed - Opened by lucidrains about 2 years ago - 7 comments

#272 - Support SimpleViT as encoder in MAE

Pull Request - State: closed - Opened by roydenwa about 2 years ago - 3 comments

#271 - MAE Training

Issue - State: open - Opened by mw9385 about 2 years ago

#270 - Is it possible to use the "Accessing Attention" of the vit-pytorch on the timm models?

Issue - State: open - Opened by Shima-shoki about 2 years ago

#269 - Dimension issues in Masked Patch Prediction

Issue - State: closed - Opened by KananVyas about 2 years ago - 4 comments

#267 - When running python train_cifar10.py, RuntimeError: An attempt has been made to start a new process before the current process has finished its bootstrapping phase.

Issue - State: closed - Opened by laserljy about 2 years ago

#266 - ViVit pos encoding

Issue - State: closed - Opened by eyalmazuz about 2 years ago - 2 comments

#265 - add ViTResiDual

Pull Request - State: closed - Opened by Hazqeel09 about 2 years ago - 7 comments

#264 - Number of patches in height and width dimension should be in a single…

Pull Request - State: closed - Opened by Vishu26 about 2 years ago - 2 comments

#263 - Integrate Aim - an open-source experiment tracker

Issue - State: closed - Opened by tatyusha over 2 years ago - 1 comment

#262 - PyTorch 2.0 support

Issue - State: open - Opened by kxzxvbk over 2 years ago - 2 comments

#261 - Multi-head attention part on ViT

Issue - State: closed - Opened by andreYoo over 2 years ago

#260 - Add Masked Position Prediction

Pull Request - State: closed - Opened by Vishu26 over 2 years ago - 1 comment

#259 - ViT for regression task such as Real Estate Price Prediction or Stock Exchange Datasets, any regression dataset.

Issue - State: closed - Opened by saifhassan over 2 years ago - 7 comments

#258 - Small Typo in CCT description

Issue - State: closed - Opened by DSARichard over 2 years ago - 1 comment

#257 - Why the accuracy rate to 100% using its examples' dataset

Issue - State: closed - Opened by zx-fxs over 2 years ago - 2 comments

#256 - Using SimpleVit to estimate odometry

Issue - State: open - Opened by Deadrosas over 2 years ago

#255 - Apply Tanh activation function to ViT - MLP Head

Issue - State: open - Opened by joeycouse over 2 years ago - 1 comment

#254 - How to use torchvision.models.feature_extraction.create_feature_extractor() with vit_pytorch?

Issue - State: closed - Opened by ArturasDruteika over 2 years ago

#253 - MAE bug!

Issue - State: closed - Opened by hotco87 over 2 years ago - 2 comments

#252 - [Feature Request] ViTDet

Issue - State: open - Opened by austinmw over 2 years ago - 7 comments
Labels: enhancement

#251 - Training a VIT from pre-trained patches embeddings

Issue - State: closed - Opened by AdrianBZG over 2 years ago - 1 comment

#249 - add an interpolate_embeddings helper function

Issue - State: open - Opened by DanTaranis over 2 years ago

#248 - The problem of reprinting vivit

Issue - State: closed - Opened by kuangxiaoye over 2 years ago - 1 comment

#247 - 自动驾驶更新笔记 Autopilot Updating Notes

Issue - State: closed - Opened by nwaysir over 2 years ago

#246 - How to use mask in ViT

Issue - State: closed - Opened by Ma-Zijing over 2 years ago - 3 comments

#244 - LayerNorm for Vit

Issue - State: closed - Opened by zjhdxh almost 3 years ago

#243 - Neighbourhood Attention Implementation

Issue - State: closed - Opened by RisabBiswas almost 3 years ago - 5 comments

#242 - Update mae.py

Pull Request - State: closed - Opened by Vishu26 almost 3 years ago - 4 comments

#241 - MAE `decoder_tokens` computation

Issue - State: closed - Opened by Vishu26 almost 3 years ago - 1 comment

#240 - How to retrain ViT

Issue - State: open - Opened by anusmitabose almost 3 years ago - 2 comments

#239 - Support for loading pretrained weights into networks:

Issue - State: open - Opened by PrithivirajDamodaran almost 3 years ago - 2 comments

#238 - Tensors must have same number of dimensions : got 5 and 3

Issue - State: closed - Opened by satwiksunnam19 almost 3 years ago - 8 comments

#237 - Quesiton about attention's qkv matrix

Issue - State: open - Opened by JearSYY almost 3 years ago - 2 comments

#236 - simplify `to_patch_embedding` using Conv2d

Issue - State: closed - Opened by avihu111 almost 3 years ago - 4 comments

#235 - Issues loading RegionVIT pre-trained checkpoints

Issue - State: open - Opened by PrithivirajDamodaran almost 3 years ago - 1 comment

#234 - Loading weights of custom ViT models

Issue - State: closed - Opened by PrithivirajDamodaran almost 3 years ago - 1 comment

GitHub / lucidrains/vit-pytorch issues and pull requests