Commits

Branch selector

 main

User selector

All users

Datepicker

All time

Commit History

Commits on May 4, 2022

• add simple vit, from https://arxiv.org/abs/2205.01580

  lucidrains

  committed
  b3e90a2

Commits on May 3, 2022

• add EsViT, by popular request, an alternative to Dino that is compatible with efficient ViTs with accounting for regional self-supervised loss

  lucidrains

  committed
  4ef72fc

Commits on Apr 7, 2022

• fix bibtex typo (lucidrains#212)

  vztu

  authored
  c2aab05

Commits on Apr 6, 2022

• fix mbconv residual block

  lucidrains

  committed
  81661e3

• fix maxvit - need feedforwards after attention

  lucidrains

  committed
  13f8e12

• link to maxvit in readme

  lucidrains

  committed
  2d4089c

• maxvit intent to build (lucidrains#211)

  Show description for c7bb5fc
  lucidrains

  authored
  c7bb5fc

• sponsor button

  lucidrains

  authored
  946b19b

Commits on Mar 31, 2022

• let windowed tokens exchange information across heads a la talking heads prior to pointwise attention in sep-vit

  lucidrains

  committed
  d93cd84

• cleanup sepvit

  lucidrains

  committed
  5d4c798

• intent to build (lucidrains#210)

  Show description for d65a742
  lucidrains

  authored
  d65a742

• do not layernorm on last transformer block for scalable vit, as there is already one in mlp head

  lucidrains

  committed
  8c54e01

• complete learnable memory ViT, for efficient fine-tuning and potentially plays into continual learning

  lucidrains

  committed
  df656fe

Commits on Mar 30, 2022

• correct need for post-attention dropout

  lucidrains

  committed
  4e6a42a

Commits on Mar 28, 2022

• link to tensorflow2 translation by @taki0112

  lucidrains

  committed
  6d7298d

Commits on Mar 26, 2022

• CCT allow for rectangular images

  lucidrains

  committed
  9cd56ff

Commits on Mar 23, 2022

• add proposed parallel vit from facebook ai for exploration purposes

  lucidrains

  committed
  2aae406

• fix window size of none for scalable vit for rectangular images

  lucidrains

  committed
  c2b2db2

• some better defaults for scalable vit

  lucidrains

  committed
  719048d

• add scalable vit, from bytedance AI

  lucidrains

  committed
  d27721a

Commits on Mar 22, 2022

• update to einops 0.4, which is torchscript jit friendly

  lucidrains

  committed
  cb22cbb

Commits on Mar 2, 2022

• add patch merger

  lucidrains

  committed
  6db20de

Commits on Jan 31, 2022

• allow for rectangular images for efficient adapter

  lucidrains

  committed
  1bae5d3

Commits on Jan 29, 2022

• return None from extractor if no attention layers

  lucidrains

  committed
  25b3842

Commits on Jan 25, 2022

• epsilon should be inside square root

  lucidrains

  committed
  64a07f5

Commits on Jan 23, 2022

• fix block repeats in readme example for Nest

  lucidrains

  committed
  126d204

Commits on Jan 22, 2022

• fix feature maps in Nest, thanks to @MarkYangjiayi

  lucidrains

  committed
  c1528ac

Commits on Jan 6, 2022

• decoder positional embedding needs to be reapplied https://twitter.com/giffmana/status/1479195631587631104

  lucidrains

  committed
  1cc0f18

Commits on Jan 3, 2022

• 0.26.2

  lucidrains

  committed
  28eaba6

• build @jrounds suggestion

  lucidrains

  committed
  0082301

Commits on Dec 31, 2021

• 0.26.1

  lucidrains

  authored
  91ed738

• Merge pull request lucidrains#186 from chinhsuanwu/mobilevit

  Show description for 1b58daa
  lucidrains

  authored
  1b58daa

Commits on Dec 29, 2021

• Update MobileViT

  

  chinhsuanwu

  committed
  f2414b2

• readme

  lucidrains

  committed
  891b92e

Commits on Dec 28, 2021

• add ViT for small datasets https://arxiv.org/abs/2112.13492

  lucidrains

  committed
  70ba532

Pagination

PreviousNext