Commit Graph

7 Commits (4fcd7ee0728a5fd96312e608cd8e32424baec56c)

Author SHA1 Message Date
Ma Zerun f9a2b04cee
[Feature] Add DeiT backbone and checkpoints. (#576)
* Support DeiT backbone.

* Use hook to automatically resize pos embed

* Update ViT training setting

* Add deit configs and update docs

* Fix vit arch assertion

* Remove useless init function

* Add unit tests.

* Fix resize_pos_embed for DeiT

* Improve according to comments.
2021-12-15 22:44:57 +08:00
Zhiliang Peng 18f6bb0b10
[Feature] Implement the conformer backbone. (#494)
* implement the conformer

* format code style

* format code style

* reuse the TransformerEncoderLayer in the vision_transformer.py

* Modify variable name

* delete unused params

* Remove warning info in Conformer head since it already exists in
Conformer.

* Rename some variables

* Add unit tests

* Use `getattr` instead of `get_submodule`.

* Remove some useless layers

* Refactor conformer and add configs

* Update configs and add metafile.

* Fix unit tests

* Update README

Co-authored-by: mzr1996 <mzr1996@163.com>
2021-12-07 14:00:17 +08:00
Ma Zerun f9eb9b409b
[Docs] Add Copyright information. (#413) 2021-08-17 19:52:42 +08:00
Ma Zerun 65410b05ad
Fix Mobilenetv3 structure and add pretrained model (#291)
* Refactor Mobilenetv3 structure and add ConvClsHead.

* Change model's name from 'MobileNetv3' to 'MobileNetV3'

* Modify configs for MobileNetV3 on CIFAR10.

And add MobileNetV3 configs for imagenet

* Fix activate setting bugs in MobileNetV3.

And remove bias in SELayer.

* Modify unittest

* Remove useless config and file.

* Fix mobilenetv3-large arch setting

* Add dropout option in ConvClsHead

* Fix MobilenetV3 structure according to torchvision version.

1. Remove with_expand_conv option in InvertedResidual, it should be decided by channels.

2. Revert activation function, should before SE layer.

* Format code.

* Rename MobilenetV3 arch "big" to "large".

* Add mobilenetv3_small torchvision training recipe

* Modify default `out_indices` of MobilenetV3, now it will change
according to `arch` if not specified.

* Add MobilenetV3 large config.

* Add mobilenetv3 README

* Modify InvertedResidual unit test.

* Refactor ConvClsHead to StackedLinearClsHead, and add unit tests.

* Add unit test for `simple_test` of `StackedLinearClsHead`.

* Fix typo

Co-authored-by: Yidi Shao <ydshao@smail.nju.edu.cn>
2021-06-27 23:19:36 +08:00
whcao affb39fe07
[Feature]Add Vit (#214)
* add imagenet bs 4096

* add vit_base_patch16_224_finetune

* add vit_base_patch16_224_pretrain

* add vit_base_patch16_384_finetune

* add vit_base_patch16_384_finetune

* add vit_b_p16_224_finetune_imagenet

* add vit_b_p16_224_pretrain_imagenet

* add vit_b_p16_384_finetune_imagenet

* add vit

* add vit

* add vit head

* vit unitest

* keep up with ClsHead

* test vit

* add flag to determiine whether to calculate acc during training

* Changes related to mmcv1.3.0

* change checkpoint saving interval to 10

* add label smooth

* default_runtime.py recovery

* docformatter

* docformatter

* delete 2 lines of comments

* delete configs/_base_/schedules/imagenet_bs4096.py

* add configs/_base_/schedules/imagenet_bs2048_AdamW.py

* rename imagenet_bs4096.py to imagenet_bs2048_AdamW.py

* add helpers.py

* test vit hybrid backbone

* fix HybridEmbed

* use to_2tuple instead
2021-04-16 19:22:41 +08:00
LXXXXR 07bb15e5fd
[Feature] Add heads and config for multilabel task (#145)
* resolve conflicts
add heads and config for multilabel tasks

* minor change

* remove evaluating mAP in head

* add baseline config

* add configs

* reserve only one config

* minor change

* fix minor bug

* minor change

* minor change

* add unittests and fix docstrings
2021-01-25 18:10:14 +08:00
yanglei fbc68bf266 Add classifiers, heads, necks and losses 2020-07-07 19:32:06 +08:00