chore(deps): update dependency timm to v1 #1267
Open
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
This PR contains the following updates:
==0.4.12
->==1.0.3
Release Notes
huggingface/pytorch-image-models (timm)
v1.0.3
Compare Source
May 14, 2024
normalize=
flag for transorms, return non-normalized torch.Tensor with original dytpe (forchug
)May 11, 2024
Searching for Better ViT Baselines (For the GPU Poor)
weights and vit variants released. Exploring model shapes between Tiny and Base.timm
models. See example in https://github.com/huggingface/pytorch-image-models/discussions/1232#discussioncomment-9320949forward_intermediates()
API refined and added to more models including some ConvNets that have other extraction methods.features_only=True
feature extraction. Remaining 34 architectures can be supported but based on priority requests.April 11, 2024
features_only=True
support for ViT models with flat hidden states or non-std module layouts (so far covering'vit_*', 'twins_*', 'deit*', 'beit*', 'mvitv2*', 'eva*', 'samvit_*', 'flexivit*'
)forward_intermediates()
API that can be used with a feature wrapping module or direclty.v0.9.16
Compare Source
Feb 19, 2024
Jan 8, 2024
Datasets & transform refactoring
--dataset hfids:org/dataset
)datasets
and webdataset wrapper streaming from HF hub with recenttimm
ImageNet uploads to https://huggingface.co/timm--input-size 1 224 224
or--in-chans 1
, sets PIL image conversion appropriately in dataset--val-split ''
) in train script--bce-sum
(sum over class dim) and--bce-pos-weight
(positive weighting) args for training as they're common BCE loss tweaks I was often hard codingv0.9.12
Compare Source
Nov 23, 2023
v0.9.11
Compare Source
Nov 20, 2023
model_args
config entry.model_args
will be passed as kwargs through to models on creation.vision_transformer.py
typing and doc cleanup by Laureηtv0.9.10
Compare Source
Nov 4
Nov 3, 2023
quickgelu
ViT variants for OpenAI, DFN, MetaCLIP weights that use it (less efficient)convnext_xxlarge
v0.9.9
Compare Source
Nov 3, 2023
quickgelu
ViT variants for OpenAI, DFN, MetaCLIP weights that use it (less efficient)convnext_xxlarge
v0.9.8
Compare Source
Oct 20, 2023
vision_transformer.py
.v0.9.7
Compare Source
Small bug fix & extra model from v0.9.6
Sep 1, 2023
v0.9.6
Compare Source
Aug 28, 2023
vision_transformer.py
,vision_transformer_hybrid.py
,deit.py
, andeva.py
w/o breaking backward compat.dynamic_img_size=True
to args at model creation time to allow changing the grid size (interpolate abs and/or ROPE pos embed each forward pass).dynamic_img_pad=True
to allow image sizes that aren't divisible by patch size (pad bottom right to patch size each forward pass).img_size
(interpolate pretrained embed weights once) on creation still works.patch_size
(resize pretrained patch_embed weights once) on creation still works.python validate.py /imagenet --model vit_base_patch16_224 --amp --amp-dtype bfloat16 --img-size 255 --crop-pct 1.0 --model-kwargs dynamic_img_size=True dyamic_img_pad=True
Aug 25, 2023
--reparam
arg tobenchmark.py
,onnx_export.py
, andvalidate.py
to trigger layer reparameterization / fusion for models with any one ofreparameterize()
,switch_to_deploy()
orfuse()
Aug 11, 2023
python validate.py /imagenet --model swin_base_patch4_window7_224.ms_in22k_ft_in1k --amp --amp-dtype bfloat16 --input-size 3 256 320 --model-kwargs window_size=8,10 img_size=256,320
v0.9.5
Compare Source
Minor updates and bug fixes. New ResNeXT w/ highest ImageNet eval I'm aware of in the ResNe(X)t family (
seresnextaa201d_32x8d.sw_in12k_ft_in1k_384
)Aug 3, 2023
selecsls*
model naming regressionJuly 27, 2023
seresnextaa201d_32x8d.sw_in12k_ft_in1k_384
weights (and.sw_in12k
pretrain) with 87.3% top-1 on ImageNet-1k, best ImageNet ResNet family model I'm aware of.v0.9.2
Compare Source
v0.9.1
Compare Source
The first non pre-release since Oct 2022 with a long list of changes from 0.6.x releases...
May 12, 2023
May 11, 2023
timm
0.9 released, transition from 0.8.xdev releasesMay 10, 2023
timm
get_intermediate_layers
function on vit/deit models for grabbing hidden states (inspired by DINO impl). This is WIP and may change significantly... feedback welcome.pretrained=True
and no weights exist (instead of continuing with random initialization)bnb
prefix, iebnbadam8bit
timm
out of pre-release stateApril 27, 2023
timm
models uploaded to HF Hub and almost all updated to support multi-weight pretrained configsApril 21, 2023
--grad-accum-steps
), thanks Taeksang Kim--head-init-scale
and--head-init-bias
to train.py to scale classiifer head and set fixed bias for fine-tuneinplace_abn
) use, replaced use in tresnet with standard BatchNorm (modified weights accordingly).April 12, 2023
drop_rate
(classifier dropout),proj_drop_rate
(block mlp / out projections),pos_drop_rate
(position embedding drop),attn_drop_rate
(attention dropout). Also add patch dropout (FLIP) to vit and eva models.April 5, 2023
timm
trained weights added with recipe based tags to differentiateresnetaa50d.sw_in12k_ft_in1k
- 81.7 @ 224, 82.6 @ 288resnetaa101d.sw_in12k_ft_in1k
- 83.5 @ 224, 84.1 @ 288seresnextaa101d_32x8d.sw_in12k_ft_in1k
- 86.0 @ 224, 86.5 @ 288seresnextaa101d_32x8d.sw_in12k_ft_in1k_288
- 86.5 @ 288, 86.7 @ 320March 31, 2023
March 22, 2023
regnet.py
,rexnet.py
,byobnet.py
,resnetv2.py
,swin_transformer.py
,swin_transformer_v2.py
,swin_transformer_v2_cr.py
swinv2_cr_*
, and NHWC for all others) and spatial embedding outputs.timm
weights:rexnetr_200.sw_in12k_ft_in1k
- 82.6 @ 224, 83.2 @ 288rexnetr_300.sw_in12k_ft_in1k
- 84.0 @ 224, 84.5 @ 288regnety_120.sw_in12k_ft_in1k
- 85.0 @ 224, 85.4 @ 288regnety_160.lion_in12k_ft_in1k
- 85.6 @ 224, 86.0 @ 288regnety_160.sw_in12k_ft_in1k
- 85.6 @ 224, 86.0 @ 288 (compare to SWAG PT + 1k FT this is same BUT much lower res, blows SEER FT away)Feb 26, 2023
convnext_xxlarge
default LayerNorm eps to 1e-5 (for CLIP weights, improved stability)Feb 20, 2023
convnext_large_mlp.clip_laion2b_ft_320
andconvnext_lage_mlp.clip_laion2b_ft_soup_320
CLIP image tower weights for features & fine-tuneFeb 16, 2023
safetensor
checkpoint support addedvit_*
,vit_relpos*
,coatnet
/maxxvit
(to start)features_only=True
Feb 7, 2023
convnext_base.clip_laion2b_augreg_ft_in1k
- 86.2% @ 256x256convnext_base.clip_laiona_augreg_ft_in1k_384
- 86.5% @ 384x384convnext_large_mlp.clip_laion2b_augreg_ft_in1k
- 87.3% @ 256x256convnext_large_mlp.clip_laion2b_augreg_ft_in1k_384
- 87.9% @ 384x384features_only=True
. Adapted from https://github.com/dingmyu/davit by Fredo.features_only=True
.features_only=True
support to newconv
variants, weight remap required./results
totimm/data/_info
.timm
inference.py
to use, try:python inference.py /folder/to/images --model convnext_small.in12k --label-type detail --topk 5
Jan 20, 2023
Add two convnext 12k -> 1k fine-tunes at 384x384
convnext_tiny.in12k_ft_in1k_384
- 85.1 @ 384convnext_small.in12k_ft_in1k_384
- 86.2 @ 384Push all MaxxViT weights to HF hub, and add new ImageNet-12k -> 1k fine-tunes for
rw
base MaxViT and CoAtNet 1/2 modelsJan 11, 2023
.in12k
tags)convnext_nano.in12k_ft_in1k
- 82.3 @ 224, 82.9 @ 288 (previously released)convnext_tiny.in12k_ft_in1k
- 84.2 @ 224, 84.5 @ 288convnext_small.in12k_ft_in1k
- 85.2 @ 224, 85.3 @ 288Jan 6, 2023
--model-kwargs
and--opt-kwargs
to scripts to pass through rare args directly to model classes from cmd linetrain.py /imagenet --model resnet50 --amp --model-kwargs output_stride=16 act_layer=silu
train.py /imagenet --model vit_base_patch16_clip_224 --img-size 240 --amp --model-kwargs img_size=240 patch_size=12
Jan 5, 2023
convnext.py
Dec 23, 2022 🎄☃
efficientnet_b5.in12k_ft_in1k
- 85.9 @ 448x448vit_medium_patch16_gap_384.in12k_ft_in1k
- 85.5 @ 384x384vit_medium_patch16_gap_256.in12k_ft_in1k
- 84.5 @ 256x256convnext_nano.in12k_ft_in1k
- 82.9 @ 288x288Dec 8, 2022
vision_transformer.py
, MAE style ViT-L/14 MIM pretrain w/ EVA-CLIP targets, FT on ImageNet-1k (w/ ImageNet-22k intermediate for some)Dec 6, 2022
beit.py
.Dec 5, 2022
0.8.0dev0
) of multi-weight support (model_arch.pretrained_tag
). Install withpip install --pre timm
--torchcompile
argumentOct 15, 2022
--amp-impl apex
, bfloat16 supportedf via--amp-dtype bfloat16
v0.9.0
Compare Source
First non pre-release in a loooong while, changelog from 0.6.x below...
May 11, 2023
timm
0.9 released, transition from 0.8.xdev releasesMay 10, 2023
timm
get_intermediate_layers
function on vit/deit models for grabbing hidden states (inspired by DINO impl). This is WIP and may change significantly... feedback welcome.pretrained=True
and no weights exist (instead of continuing with random initialization)bnb
prefix, iebnbadam8bit
timm
out of pre-release stateApril 27, 2023
timm
models uploaded to HF Hub and almost all updated to support multi-weight pretrained configsApril 21, 2023
--grad-accum-steps
), thanks Taeksang Kim--head-init-scale
and--head-init-bias
to train.py to scale classiifer head and set fixed bias for fine-tuneinplace_abn
) use, replaced use in tresnet with standard BatchNorm (modified weights accordingly).April 12, 2023
drop_rate
(classifier dropout),proj_drop_rate
(block mlp / out projections),pos_drop_rate
(position embedding drop),attn_drop_rate
(attention dropout). Also add patch dropout (FLIP) to vit and eva models.April 5, 2023
timm
trained weights added with recipe based tags to differentiateresnetaa50d.sw_in12k_ft_in1k
- 81.7 @ 224, 82.6 @ 288resnetaa101d.sw_in12k_ft_in1k
- 83.5 @ 224, 84.1 @ 288seresnextaa101d_32x8d.sw_in12k_ft_in1k
- 86.0 @ 224, 86.5 @ 288seresnextaa101d_32x8d.sw_in12k_ft_in1k_288
- 86.5 @ 288, 86.7 @ 320March 31, 2023
March 22, 2023
regnet.py
,rexnet.py
,byobnet.py
,resnetv2.py
,swin_transformer.py
,swin_transformer_v2.py
,swin_transformer_v2_cr.py
swinv2_cr_*
, and NHWC for all others) and spatial embedding outputs.timm
weights:rexnetr_200.sw_in12k_ft_in1k
- 82.6 @ 224, 83.2 @ 288rexnetr_300.sw_in12k_ft_in1k
- 84.0 @ 224, 84.5 @ 288regnety_120.sw_in12k_ft_in1k
-Configuration
📅 Schedule: Branch creation - At any time (no schedule defined), Automerge - At any time (no schedule defined).
🚦 Automerge: Disabled by config. Please merge this manually once you are satisfied.
♻ Rebasing: Whenever PR becomes conflicted, or you tick the rebase/retry checkbox.
🔕 Ignore: Close this PR and you won't be reminded about this update again.
This PR has been generated by Mend Renovate. View repository job log here.