cf4561ca72
Add FX based FeatureGraphNet capability
2021-11-12 20:45:05 +00:00
0149ec30d7
wip - attempting to rebase
2021-11-12 20:45:05 +00:00
02c3a75a45
wip - make it possible to use fx graph in train and eval mode
2021-11-12 20:45:05 +00:00
bc3d4eb403
wip -rebase
2021-11-12 20:45:05 +00:00
ab3ac3f25b
Add FX based FeatureGraphNet capability
2021-11-12 20:45:05 +00:00
9ec3210c2d
More TFDS parser cleanup, support improved TFDS even_split impl (on tfds-nightly only currently).
2021-11-10 15:52:09 -08:00
ba65dfe2c6
Dataset work
...
* support some torchvision datasets
* improvements to TFDS wrapper for subsplit handling (fix #942 ), shuffle seed
* add class-map support to train (fix #957 )
2021-11-09 22:34:15 -08:00
ddc29da974
Add ResNet101 and ResNet152 weights from higher aug RSB recipes. 81.93 and 82.82 top-1 at 224x224.
2021-11-02 17:59:16 -07:00
b328e56f49
Update eca_halonext26ts weights to a better set
2021-11-02 16:52:53 -07:00
2ddef942b9
Better fix for #954 that doesn't break torchscript, pull torch._assert into timm namespace when it exists
2021-11-02 11:22:33 -07:00
4f0f9cb348
Fix #954 by bringing traceable _assert into timm to allow compat w/ PyTorch < 1.8
2021-11-02 09:21:40 -07:00
a41de1f666
Add interpolation mode handling to transforms. Removes InterpolationMode warning. Works for torchvision versions w/ and w/o InterpolationMode enum. Fix #738 .
2021-10-28 17:35:01 -07:00
ed41d32637
Add repr to auto_augment and random_erasing impl
2021-10-28 17:33:36 -07:00
ae72d009fa
Add weights for lambda_resnet50ts, halo2botnet50ts, lamhalobotnet50ts, updated halonet50ts
2021-10-27 22:08:54 -07:00
b745d30a3e
Fix formatting of last commit
2021-10-25 15:15:14 -07:00
3478f1d7f1
Traceability fix for vit models for some experiments
2021-10-25 15:13:08 -07:00
f658a72e72
Cleanup re-use of Dropout modules in Mlp modules after some twitter feedback :p
2021-10-25 00:40:59 -07:00
f805ba86d9
use .unbind instead of explicitly listing the indices
2021-10-24 21:08:47 +02:00
57992509f9
Fix some formatting in utils/model.py
2021-10-23 20:35:36 -07:00
0fe4fd3f1f
add d8 and e8 regnetz models with group size 8
2021-10-23 20:34:21 -07:00
25e7c8c5e5
Update broken resnetv2_50 weight url, add resnetv1_101 a1h recipe weights for 224x224 train
2021-10-20 22:14:12 -07:00
b6caa356d2
Fixed eca_botnext26ts_256 weights added, 79.27
2021-10-19 12:44:28 -07:00
c02334d9fa
Add weights for regnetz_d and haloregnetz_c, update regnetz_c weights. Add commented PyTorch XLA code for halo attention
2021-10-19 12:32:09 -07:00
02daf2ab94
Add option to include relative pos embedding in the attention scaling as per references. See discussion #912
2021-10-12 15:37:01 -07:00
047a5ec05f
Fix bugs that Mixup does not work device=cpu
2021-10-12 23:51:46 +09:00
cd34913278
Remove some outdated comments, botnet networks working great now.
2021-10-11 22:43:41 -07:00
6ed4cdccca
Update lambda_resnet26t weights with better set
2021-10-10 16:32:54 -07:00
44d6d51668
Add ConvMixer
2021-10-09 21:09:51 -04:00
a85df34993
Update lambda_resnet26rpt weights to 78.9, add better halonet26t weights at 79.1 with tweak to attention dim
2021-10-08 17:44:13 -07:00
b544ad4d3f
regnetz model default cfg tweaks
2021-10-06 21:14:59 -07:00
e5da481073
Small post-merge tweak for freeze/unfreeze, add to __init__ for utils
2021-10-06 17:00:27 -07:00
5ca72dcc75
Merge branch 'freeze-functionality' of https://github.com/alexander-soare/pytorch-image-models into alexander-soare-freeze-functionality
2021-10-06 16:51:03 -07:00
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
2021-10-06 16:32:48 -07:00
431e60c83f
Add acknowledgements for freeze_batch_norm inspiration
2021-10-06 14:28:49 +01:00
fbf59c04ee
Change crop ratio on correct resnet50 variant.
2021-10-04 22:31:08 -07:00
ae1ff5792f
Clean a1/a2/3 rsb _0 checkpoints properly, fix v2 loading.
2021-10-04 16:46:00 -07:00
93901e992f
Version bump to 0.5.0 for pending release post RSB and ATTN updates
2021-10-03 17:34:57 -07:00
da0d39bedd
Update default crop_pct for byoanet
2021-10-03 17:33:16 -07:00
cc9bedf373
Add initial ResNet Strikes Back weights for ResNet50 and ResNetV2-50 models
2021-10-03 17:32:02 -07:00
64495505b7
Add updated lambda resnet26 and botnet26 checkpoints with fixes applied
2021-10-03 17:31:39 -07:00
b2094f4ee8
support bits checkpoints in avg/load
2021-10-03 17:31:22 -07:00
007bc39323
Some halo and bottleneck attn code cleanup, add halonet50ts weights, use optimal crop ratios
2021-10-02 15:51:42 -07:00
65c3d78b96
Freeze unfreeze functionality finalized. Tests added
2021-10-02 15:55:08 +01:00
0cb8ea432c
wip
2021-10-02 15:55:08 +01:00
b1c2e3eb92
Match rel_pos_indices attr rename in conv branch
2021-09-30 23:19:05 -07:00
b49630a138
Add relative pos embed option to LambdaLayer, fix last transpose/reshape.
2021-09-30 22:45:09 -07:00
d657e2cc0b
Remove dead code line from efficientnet
2021-09-30 21:54:42 -07:00
0ca687f224
Make 'regnetz' model experiments closer to actual RegNetZ, bottleneck expansion, expand from in_chs, no shortcut on stride 2, tweak model sizes
2021-09-30 21:49:38 -07:00
51eaf9360d
Remove a duplicate layer creation in byobnet.py
...
`self.conv2_kxk` is repeated in `byobnet.py`. Remove the duplicate code.
2021-09-30 18:30:48 +08:00
b81e79aae9
Fix bottleneck attn transpose typo, hopefully these train better now..
2021-09-28 16:38:41 -07:00
80075b0b8a
Add worker_seeding arg to allow selecting old vs updated data loader worker seed for (old) experiment repeatability
2021-09-28 16:37:45 -07:00
6478bcd02c
Fix regnetz_d conv layer name, use inception mean/std
2021-09-26 14:54:17 -07:00
0387e6057e
Update binary cross ent impl to use thresholding as an option (convert soft targets from mixup/cutmix to 0, 1)
2021-09-23 15:45:39 -07:00
f8a63a3b71
Add worker_init_fn to loader for numpy seed per worker
2021-09-23 15:44:38 -07:00
515121cca1
Use reshape instead of view in std_conv, causing issues in recent PyTorch in channels_last
2021-09-23 15:43:48 -07:00
da06cc61d4
ResNetV2 seems to work best without zero_init residual
2021-09-23 15:43:22 -07:00
8e11da0ce3
Add experimental RegNetZ(ish) models for training / perf trials.
2021-09-23 15:42:57 -07:00
6bbc50beb4
make it possible to provide norm_layer via create_model
2021-09-21 10:19:04 +01:00
adcb74f87f
🎨 Import load_state_dict_from_url directly
2021-09-14 01:11:40 -04:00
e65a2cba3d
🎨 cleanup and add a couple comments
2021-09-14 01:07:04 -04:00
2b6ade24b3
🎨 write model card to enable inference
2021-09-13 23:31:28 -04:00
cf5ac2800c
BotNet models were still off, remove weights for bad configs. Add good SE-HaloNet33-TS weights.
2021-09-13 17:18:59 -07:00
24720abe3b
Merge branch 'master' into attn_update
2021-09-13 16:51:10 -07:00
1c9284c640
Add BeiT 'finetuned' 1k weights and pretrained 22k weights, pretraining specific (masked) model excluded for now
2021-09-13 16:38:23 -07:00
f8a215cfe6
A few more crossvit tweaks, fix training w/ no_weight_decay names, add crop option for scaling, adjust default crop_pct for large img size to 1.0 for better results
2021-09-13 14:17:34 -07:00
7ab2491ab7
Better handling of crossvit for tests / forward_features, fix torchscript regression in my changes
2021-09-13 13:01:05 -07:00
f1808e0970
Post crossvit merge cleanup, change model names to reflect input size, cleanup img size vs scale handling, fix tests
2021-09-13 11:49:54 -07:00
4027412757
Add resnet33ts weights, update resnext26ts baseline weights
2021-09-09 14:46:41 -07:00
9fe5798bee
fix bug for reset classifier and fix for validating the dimension
2021-09-08 21:58:17 -04:00
3718c5a5bd
fix loading pretrained model
2021-09-08 11:53:05 -04:00
bb50b69a57
fix for torch script
2021-09-08 11:20:59 -04:00
abf9d51bc3
🚧 wip
2021-09-07 18:39:26 -06:00
5bd04714e4
Cleanup weight init for byob/byoanet and related
2021-09-05 15:34:05 -07:00
8642401e88
Swap botnet 26/50 weights/models after realizing a mistake in arch def, now figuring out why they were so low...
2021-09-05 15:17:19 -07:00
5f12de4875
Add initial AttentionPool2d that's being trialed. Fix comment and still trying to improve reliability of sgd test.
2021-09-05 12:41:14 -07:00
76881d207b
Add baseline resnet26t @ 256x256 weights. Add 33ts variant of halonet with at least one halo in stage 2,3,4
2021-09-04 14:52:54 -07:00
484e61648d
Adding the attn series weights, tweaking model names, comments...
2021-09-03 18:09:42 -07:00
fb94350896
Update training script and loader factory to allow use of scheduler updates, repeat augment, and bce loss
2021-09-01 17:46:40 -07:00
f262137ff2
Add RepeatAugSampler as per DeiT RASampler impl, showing promise for current (distributed) training experiments.
2021-09-01 17:40:53 -07:00
ba9c1108a1
Add a BCE loss impl that converts dense targets to sparse /w smoothing as an alternate to CE w/ smoothing. For training experiments.
2021-09-01 17:39:28 -07:00
29a37e23ee
LR scheduler update:
...
* add polynomial decay 'poly'
* cleanup cycle specific args for cosine, poly, and tanh sched, t_mul -> cycle_mul, decay -> cycle_decay, default cycle_limit to 1 in each opt
* add k-decay for cosine and poly sched as per https://arxiv.org/abs/2004.05909
* change default tanh ub/lb to push inflection to later epochs
2021-09-01 17:33:11 -07:00
28d2841acf
💄 apply isort
2021-09-01 18:15:08 -06:00
492c0a4e20
Update HaloAttn comment
2021-09-01 17:14:31 -07:00
e72c989973
✨ add ability to push to hf hub
2021-09-01 18:14:28 -06:00
7ab9d4555c
add crossvit
2021-09-01 17:13:12 -04:00
3b9032ea48
Use Tensor.unfold().unfold() for HaloAttn, fast like as_strided but more clarity
2021-08-27 12:45:53 -07:00
78933122c9
Fix silly typo
2021-08-27 09:22:20 -07:00
2568ffc5ef
Merge branch 'master' into attn_update
2021-08-27 09:21:22 -07:00
708d87a813
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
8449ba210c
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
a8b65695f1
Add resnet26ts and resnext26ts models for non-attn baselines
2021-08-21 12:42:10 -07:00
a5a542f17d
Fix typo
2021-08-20 17:47:23 -07:00
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...
2021-08-20 16:13:11 -07:00
d667351eac
Tweak accuracy topk safety. Fix #807
2021-08-19 14:18:53 -07:00
35c9740826
Fix accuracy when topk > num_classes
2021-08-19 11:58:59 +02:00
a16a753852
Add lamb/lars to optim init imports, remove stray comment
2021-08-18 22:55:02 -07:00
c207e02782
MOAR optimizer changes. Woo!
2021-08-18 22:20:35 -07:00
a426511c95
More optimizer cleanup. Change all to no longer use .data. Improve (b)float16 use with adabelief. Add XLA compatible Lars.
2021-08-18 17:21:56 -07:00
9541f4963b
One more scalar -> tensor fix for lamb optimizer
2021-08-18 11:20:25 -07:00
8f68193c91
Update lamp.py comment
2021-08-18 09:27:40 -07:00
Alexander Soare
Ross Wightman
Thomas Viehmann
masafumi
ICLR Author
leondgarse
nateraw
Richard Chen
Yohann Lereclus
Ross Wightman