Ross Wightman 3b9032ea48 Use Tensor.unfold().unfold() for HaloAttn, fast like as_strided but more clarity

2021-08-27 12:45:53 -07:00

..

layers

Use Tensor.unfold().unfold() for HaloAttn, fast like as_strided but more clarity

2021-08-27 12:45:53 -07:00

pruned

…

__init__.py

wip xcit

2021-07-11 09:39:38 +01:00

byoanet.py

Improve performance of HaloAttn, change default dim calc. Some cleanup / fixes for byoanet. Rename resnet26ts to tfs to distinguish (extra fc).

2021-08-26 21:56:44 -07:00

byobnet.py

Improve performance of HaloAttn, change default dim calc. Some cleanup / fixes for byoanet. Rename resnet26ts to tfs to distinguish (extra fc).

2021-08-26 21:56:44 -07:00

cait.py

Refactoring, cleanup, improved test coverage.

2021-06-12 16:40:02 -07:00

coat.py

Refactoring, cleanup, improved test coverage.

2021-06-12 16:40:02 -07:00

convit.py

Merge branch 'master' into cleanup_xla_model_fixes

2021-06-12 23:19:25 -07:00

cspnet.py

…

densenet.py

…

dla.py

Refactoring, cleanup, improved test coverage.

2021-06-12 16:40:02 -07:00

dpn.py

Refactoring, cleanup, improved test coverage.

2021-06-12 16:40:02 -07:00

efficientnet_blocks.py

Bring EfficientNet SE layer in line with others, pull se_ratio outside of blocks. Allows swapping w/ other attn layers.

2021-05-29 23:41:38 -07:00

efficientnet_builder.py

Able to use other attn layer in EfficientNet now. Create test ECA + GC B0 configs. Make ECA more configurable.

2021-05-30 12:47:02 -07:00

efficientnet.py

Add gc_efficientnetv2_rw_t weights (global context instead of SE attn). Add TF XL weights even though the fine-tuned ones don't validate that well. Change default arg for GlobalContext to use scal (mul) mode.

2021-08-07 16:45:29 -07:00

factory.py

…

features.py

…

ghostnet.py

Fix a few issues loading pretrained vit/bit npz weights w/ num_classes=0 __init__ arg. Missed a few other small classifier handling detail on Mlp, GhostNet, Levit. Should fix #713

2021-06-22 23:16:05 -07:00

gluon_resnet.py

…

gluon_xception.py

…

hardcorenas.py

Bring EfficientNet SE layer in line with others, pull se_ratio outside of blocks. Allows swapping w/ other attn layers.

2021-05-29 23:41:38 -07:00

helpers.py

Refactoring, cleanup, improved test coverage.

2021-06-12 16:40:02 -07:00

hrnet.py

…

hub.py

…

inception_resnet_v2.py

…

inception_v3.py

…

inception_v4.py

…

levit.py

Fix a few issues loading pretrained vit/bit npz weights w/ num_classes=0 __init__ arg. Missed a few other small classifier handling detail on Mlp, GhostNet, Levit. Should fix #713

2021-06-22 23:16:05 -07:00

mlp_mixer.py

Add gMLP-S weights, 79.6 top-1

2021-06-23 10:40:30 -07:00

mobilenetv3.py

Refactoring, cleanup, improved test coverage.

2021-06-12 16:40:02 -07:00

nasnet.py

…

nest.py

Use bicubic interpolation in resize_pos_embed()

2021-07-12 10:38:31 -07:00

nfnet.py

Refactoring, cleanup, improved test coverage.

2021-06-12 16:40:02 -07:00

pit.py

Refactoring, cleanup, improved test coverage.

2021-06-12 16:40:02 -07:00

pnasnet.py

…

registry.py

Refactoring, cleanup, improved test coverage.

2021-06-12 16:40:02 -07:00

regnet.py

Add Gather-Excite and Global Context attn modules. Refactor existing SE-like attn for consistency and refactor byob/byoanet for less redundancy.

2021-05-27 18:03:29 -07:00

res2net.py

…

resnest.py

Add non-local and BAT attention. Merge attn and self-attn factories into one. Add attention references to README. Add mlp 'mode' to ECA.

2021-05-31 13:18:11 -07:00

resnet.py

Refactoring, cleanup, improved test coverage.

2021-06-12 16:40:02 -07:00

resnetv2.py

Fix features for resnetv2_50t

2021-06-28 15:56:24 -07:00

rexnet.py

Post merge cleanup

2021-06-07 14:38:30 -07:00

selecsls.py

…

senet.py

…

sknet.py

Remove min channels for SelectiveKernel, divisor should cover cases well enough.

2021-05-31 15:38:56 -07:00

swin_transformer.py

Refactoring, cleanup, improved test coverage.

2021-06-12 16:40:02 -07:00

tnt.py

…

tresnet.py

Add Gather-Excite and Global Context attn modules. Refactor existing SE-like attn for consistency and refactor byob/byoanet for less redundancy.

2021-05-27 18:03:29 -07:00

twins.py

Refactoring, cleanup, improved test coverage.

2021-06-12 16:40:02 -07:00

vgg.py

…

visformer.py

Fix a few issues loading pretrained vit/bit npz weights w/ num_classes=0 __init__ arg. Missed a few other small classifier handling detail on Mlp, GhostNet, Levit. Should fix #713

2021-06-22 23:16:05 -07:00

vision_transformer_hybrid.py

AugReg release

2021-06-20 17:46:06 -07:00

vision_transformer.py

Fix ViT SAM weight compat as weights at URL changed to not use repr layer. Fix #825. Tweak optim test.

2021-08-27 09:20:13 -07:00

vovnet.py

…

xception_aligned.py

…

xception.py

…

xcit.py

Allow act_layer switch for xcit, fix in_chans for some variants

2021-07-12 13:27:29 -07:00