b25ff96768
wip - pre-rebase
3 years ago
e051dce354
Make all models FX traceable
3 years ago
0149ec30d7
wip - attempting to rebase
3 years ago
bc3d4eb403
wip -rebase
3 years ago
2ddef942b9
Better fix for #954 that doesn't break torchscript, pull torch._assert into timm namespace when it exists
3 years ago
4f0f9cb348
Fix #954 by bringing traceable _assert into timm to allow compat w/ PyTorch < 1.8
3 years ago
b745d30a3e
Fix formatting of last commit
3 years ago
3478f1d7f1
Traceability fix for vit models for some experiments
3 years ago
f658a72e72
Cleanup re-use of Dropout modules in Mlp modules after some twitter feedback :p
3 years ago
c02334d9fa
Add weights for regnetz_d and haloregnetz_c, update regnetz_c weights. Add commented PyTorch XLA code for halo attention
3 years ago
02daf2ab94
Add option to include relative pos embedding in the attention scaling as per references. See discussion #912
3 years ago
e2b8d44ff0
Halo, bottleneck attn, lambda layer additions and cleanup along w/ experimental model defs
...
* align interfaces of halo, bottleneck attn and lambda layer
* add qk_ratio to all of above, control q/k dim relative to output dim
* add experimental haloregnetz, and trionet (lambda + halo + bottle) models
3 years ago
007bc39323
Some halo and bottleneck attn code cleanup, add halonet50ts weights, use optimal crop ratios
3 years ago
b1c2e3eb92
Match rel_pos_indices attr rename in conv branch
3 years ago
b49630a138
Add relative pos embed option to LambdaLayer, fix last transpose/reshape.
3 years ago
b81e79aae9
Fix bottleneck attn transpose typo, hopefully these train better now..
3 years ago
515121cca1
Use reshape instead of view in std_conv, causing issues in recent PyTorch in channels_last
3 years ago
5bd04714e4
Cleanup weight init for byob/byoanet and related
3 years ago
8642401e88
Swap botnet 26/50 weights/models after realizing a mistake in arch def, now figuring out why they were so low...
3 years ago
5f12de4875
Add initial AttentionPool2d that's being trialed. Fix comment and still trying to improve reliability of sgd test.
3 years ago
492c0a4e20
Update HaloAttn comment
3 years ago
3b9032ea48
Use Tensor.unfold().unfold() for HaloAttn, fast like as_strided but more clarity
3 years ago
8449ba210c
Improve performance of HaloAttn, change default dim calc. Some cleanup / fixes for byoanet. Rename resnet26ts to tfs to distinguish (extra fc).
3 years ago
925e102982
Update attention / self-attn based models from a series of experiments:
...
* remove dud attention, involution + my swin attention adaptation don't seem worth keeping
* add or update several new 26/50 layer ResNe(X)t variants that were used in experiments
* remove models associated with dead-end or uninteresting experiment results
* weights coming soon...
3 years ago
01cb46a9a5
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.
3 years ago
8165cacd82
Realized LayerNorm2d won't work in all cases as is, fixed.
3 years ago
b9cfb64412
Support npz custom load for vision transformer hybrid models. Add posembed rescale for npz load.
3 years ago
8319e0c373
Add file docstring to std_conv.py
3 years ago
4d96165989
Merge branch 'master' into cleanup_xla_model_fixes
3 years ago
8880f696b6
Refactoring, cleanup, improved test coverage.
...
* Add eca_nfnet_l2 weights, 84.7 @ 384x384
* All 'non-std' (ie transformer / mlp) models have classifier / default_cfg test added
* Fix #694 reset_classifer / num_features / forward_features / num_classes=0 consistency for transformer / mlp models
* Add direct loading of npz to vision transformer (pure transformer so far, hybrid to come)
* Rename vit_deit* to deit_*
* Remove some deprecated vit hybrid model defs
* Clean up classifier flatten for conv classifiers and unusual cases (mobilenetv3/ghostnet)
* Remove explicit model fns for levit conv, just pass in arg
3 years ago
ba2ca4b464
One codepath for stdconv, switch layernorm to batchnorm so gain included. Tweak epsilon values for nfnet, resnetv2, vit hybrid.
3 years ago
b7a568f065
Fix torchscript issue in bat
3 years ago
8e4ac3549f
All ScaledStdConv and StdConv uses default to using F.layernorm so that they work with PyTorch XLA. eps value tweaking is a WIP.
3 years ago
bda8ab015a
Remove min channels for SelectiveKernel, divisor should cover cases well enough.
4 years ago
a27f4aec4a
Missed args for skresnext w/ refactoring.
4 years ago
307a935b79
Add non-local and BAT attention. Merge attn and self-attn factories into one. Add attention references to README. Add mlp 'mode' to ECA.
4 years ago
8bf63b6c6c
Able to use other attn layer in EfficientNet now. Create test ECA + GC B0 configs. Make ECA more configurable.
4 years ago
9611458e19
Throw in some FBNetV3 code I had lying around, some refactoring of SE reduction channel calcs for all EffNet archs.
4 years ago
f615474be3
Fix broken test, repvgg block doesn't have attn_last attr.
4 years ago
742c2d5247
Add Gather-Excite and Global Context attn modules. Refactor existing SE-like attn for consistency and refactor byob/byoanet for less redundancy.
4 years ago
9c78de8c02
Fix #661 , move hardswish out of default args for LeViT. Enable native torch support for hardswish, hardsigmoid, mish if present.
4 years ago
f45de37690
Merge branch 'master' into levit_visformer_rednet
4 years ago
d5af752117
Add preliminary gMLP and ResMLP impl to Mlp-Mixer
4 years ago
3bffc701f1
Merge branch 'master' into levit_visformer_rednet
4 years ago
ecc7552c5c
Add levit, levit_c, and visformer model defs. Largely untested and not finished cleanup.
4 years ago
165fb354b2
Add initial RedNet model / Involution layer impl for testing
4 years ago
c4f482a08b
EfficientNetV2 official impl w/ weights ported from TF. Cleanup/refactor of related EfficientNet classes and models.
4 years ago
715519a5ef
Rethink name of patch embed grid info
4 years ago
b2c305c2aa
Move Mlp and PatchEmbed modules into layers. Being used in lots of models now...
4 years ago
0721559511
Improved (hopefully) init for SA/SA-like layers used in ByoaNets
4 years ago
Alexander Soare