Ross Wightman cf5fec5047 Cleanup experimental vit weight init a bit

2021-03-20 09:44:24 -07:00

..

__init__.py

Cleanup experimental vit weight init a bit

2021-03-20 09:44:24 -07:00

activations_jit.py

…

activations_me.py

Merge pull request #282 from tigert1998/patch-1

2021-02-04 12:18:40 -08:00

activations.py

Fix inplace arg compat for GELU and PreLU via activation factory

2020-11-30 13:27:40 -08:00

adaptive_avgmax_pool.py

Add 'fast' global pool option, remove redundant SEModule from tresnet, normal one is now 'fast'

2020-09-02 09:11:48 -07:00

anti_aliasing.py

…

blur_pool.py

…

cbam.py

AdaptiveAvgPool2d -> mean((2,3)) for all SE/attn layers to avoid NaN with AMP + channels_last layout. See https://github.com/pytorch/pytorch/issues/43992

2020-09-01 16:05:32 -07:00

classifier.py

ResNetV2 pre-act and non-preact model, w/ BiT pretrained weights and support for ViT R50 model. Tweaks for in21k num_classes passing. More to do... tests failing.

2020-12-28 16:59:15 -08:00

cond_conv2d.py

Initial Vision Transformer impl w/ patch and hybrid variants. Refactor tuple helpers.

2020-10-13 13:33:44 -07:00

config.py

…

conv2d_same.py

…

conv_bn_act.py

Fix regression in models with 1001 class pretrained weights. Improve batchnorm arg and BatchNormAct layer handling in several models.

2021-02-09 16:22:52 -08:00

create_act.py

Fix inplace arg compat for GELU and PreLU via activation factory

2020-11-30 13:27:40 -08:00

create_attn.py

Fix regression in models with 1001 class pretrained weights. Improve batchnorm arg and BatchNormAct layer handling in several models.

2021-02-09 16:22:52 -08:00

create_conv2d.py

Use in_channels for depthwise groups, allows using out_channels=N * in_channels (does not impact existing models). Fix #354.

2021-02-09 16:22:52 -08:00

create_norm_act.py

Fix regression in models with 1001 class pretrained weights. Improve batchnorm arg and BatchNormAct layer handling in several models.

2021-02-09 16:22:52 -08:00

drop.py

Initial Vision Transformer impl w/ patch and hybrid variants. Refactor tuple helpers.

2020-10-13 13:33:44 -07:00

eca.py

AdaptiveAvgPool2d -> mean((2,3)) for all SE/attn layers to avoid NaN with AMP + channels_last layout. See https://github.com/pytorch/pytorch/issues/43992

2020-09-01 16:05:32 -07:00

evo_norm.py

Fix a silly bug in Sample version of EvoNorm missing x* part of swish, update EvoNormBatch to accumulated unbiased variance.

2020-08-13 18:25:01 -07:00

helpers.py

update collections.abc import

2021-02-10 23:54:35 +11:00

inplace_abn.py

…

linear.py

A few more changes for 0.3.2 maint release. Linear layer change for mobilenetv3 and inception_v3, support no bias for linear wrapper.

2020-11-30 16:19:52 -08:00

median_pool.py

Initial Vision Transformer impl w/ patch and hybrid variants. Refactor tuple helpers.

2020-10-13 13:33:44 -07:00

mixed_conv2d.py

Use in_channels for depthwise groups, allows using out_channels=N * in_channels (does not impact existing models). Fix #354.

2021-02-09 16:22:52 -08:00

norm_act.py

Fix regression in models with 1001 class pretrained weights. Improve batchnorm arg and BatchNormAct layer handling in several models.

2021-02-09 16:22:52 -08:00

padding.py

…

pool2d_same.py

Initial Vision Transformer impl w/ patch and hybrid variants. Refactor tuple helpers.

2020-10-13 13:33:44 -07:00

se.py

Initial Normalizer-Free Reg/ResNet impl. A bit of related layer refactoring.

2021-01-27 22:06:57 -08:00

selective_kernel.py

AdaptiveAvgPool2d -> mean((2,3)) for all SE/attn layers to avoid NaN with AMP + channels_last layout. See https://github.com/pytorch/pytorch/issues/43992

2020-09-01 16:05:32 -07:00

separable_conv.py

Fix regression in models with 1001 class pretrained weights. Improve batchnorm arg and BatchNormAct layer handling in several models.

2021-02-09 16:22:52 -08:00

space_to_depth.py

…

split_attn.py

…

split_batchnorm.py

…

std_conv.py

Update notes in ScaledStdConv impl

2021-02-18 12:44:08 -08:00

test_time_pool.py

Add new weights for ecaresnet26t/50t/269d models. Remove distinction between 't' and 'tn' (tiered models), tn is now t. Add test time img size spec to default cfg.

2021-02-06 16:30:02 -08:00

weight_init.py

Cleanup experimental vit weight init a bit

2021-03-20 09:44:24 -07:00