Feb 21, 2025

• SigLIP 2 ViT image encoders added (https://huggingface.co/collections/timm/siglip-2-67b8e72ba08b09dd97aecaf9)
  • Variable resolution / aspect NaFlex versions are a WIP
• Add 'SO150M2' ViT weights trained with SBB recipes, great results, better for ImageNet than previous attempt w/ less training.
  • vit_so150m2_patch16_reg1_gap_448.sbb_e200_in12k_ft_in1k - 88.1% top-1
  • vit_so150m2_patch16_reg1_gap_384.sbb_e200_in12k_ft_in1k - 87.9% top-1
  • vit_so150m2_patch16_reg1_gap_256.sbb_e200_in12k_ft_in1k - 87.3% top-1
  • vit_so150m2_patch16_reg4_gap_256.sbb_e200_in12k
• Updated InternViT-300M '2.5' weights
• Release 1.0.15

Feb 1, 2025

• FYI PyTorch 2.6 & Python 3.13 are tested and working w/ current main and released version of timm

Jan 27, 2025

• Add Kron Optimizer (PSGD w/ Kronecker-factored preconditioner)
  • Code from https://github.com/evanatyourservice/kron_torch
  • See also https://sites.google.com/site/lixilinx/home/psgd

What's Changed

• Fix metavar for --input-size by @JosuaRieder in #2417
• Add arguments to the respective argument groups by @JosuaRieder in #2416
• Add missing training flag to convert_sync_batchnorm by @collinmccarthy in #2423
• Fix num_classes update in reset_classifier and RDNet forward head call by @brianhou0208 in #2421
• timm: add all to init by @adamjstewart in #2399
• Fiddling with Kron (PSGD) optimizer by @rwightman in #2427
• Try to force numpy<2.0 for torch 1.13 tests, update newest tested torch to 2.5.1 by @rwightman in #2429
• Kron flatten improvements + stochastic weight decay by @rwightman in #2431
• PSGD: unify RNG by @ClashLuke in #2433
• Add vit so150m2 weights by @rwightman in #2439
• adapt_input_conv: add type hints by @adamjstewart in #2441
• SigLIP 2 by @rwightman in #2440
• timm.models: explicitly export attributes by @adamjstewart in #2442

New Contributors

• @collinmccarthy made their first contribution in #2423
• @ClashLuke made their first contribution in #2433

Full Changelog: v1.0.14...v1.0.15

Jan 19, 2025

• Fix loading of LeViT safetensor weights, remove conversion code which should have been deactivated
• Add 'SO150M' ViT weights trained with SBB recipes, decent results, but not optimal shape for ImageNet-12k/1k pretrain/ft
  • vit_so150m_patch16_reg4_gap_256.sbb_e250_in12k_ft_in1k - 86.7% top-1
  • vit_so150m_patch16_reg4_gap_384.sbb_e250_in12k_ft_in1k - 87.4% top-1
  • vit_so150m_patch16_reg4_gap_256.sbb_e250_in12k
• Misc typing, typo, etc. cleanup
• 1.0.14 release to get above LeViT fix out

What's Changed

• Fix nn.Module type hints by @adamjstewart in #2400
• Add missing paper title by @JosuaRieder in #2405
• fix 'timm recipe scripts' link by @JosuaRieder in #2404
• fix typo in EfficientNet docs by @JosuaRieder in #2403
• disable abbreviating csv inference output with ellipses by @JosuaRieder in #2402
• fix incorrect LaTeX formulas by @JosuaRieder in #2406
• VGG ConvMlp: fix layer defaults/types by @adamjstewart in #2409
• Implement --no-console-results in inference.py by @JosuaRieder in #2408
• LeViT safetensors load is broken by conversion code that wasn't deactivated by @rwightman in #2412
• A few more weights by @rwightman in #2413
• Fix typos by @JosuaRieder in #2415

New Contributors

• @adamjstewart made their first contribution in #2400

Full Changelog: v1.0.13...v1.0.14

Jan 9, 2025

• Add support to train and validate in pure bfloat16 or float16
• wandb project name arg added by https://github.com/caojiaolong, use arg.experiment for name
• Fix old issue w/ checkpoint saving not working on filesystem w/o hard-link support (e.g. FUSE fs mounts)
• 1.0.13 release

Jan 6, 2025

• Add torch.utils.checkpoint.checkpoint() wrapper in timm.models that defaults use_reentrant=False, unless TIMM_REENTRANT_CKPT=1 is set in env.

Dec 31, 2024

• convnext_nano 384x384 ImageNet-12k pretrain & fine-tune. https://huggingface.co/models?search=convnext_nano%20r384
• Add AIM-v2 encoders from https://github.com/apple/ml-aim, see on Hub: https://huggingface.co/models?search=timm%20aimv2
• Add PaliGemma2 encoders from https://github.com/google-research/big_vision to existing PaliGemma, see on Hub: https://huggingface.co/models?search=timm%20pali2
• Add missing L/14 DFN2B 39B CLIP ViT, vit_large_patch14_clip_224.dfn2b_s39b
• Fix existing RmsNorm layer & fn to match standard formulation, use PT 2.5 impl when possible. Move old impl to SimpleNorm layer, it's LN w/o centering or bias. There were only two timm models using it, and they have been updated.
• Allow override of cache_dir arg for model creation
• Pass through trust_remote_code for HF datasets wrapper
• inception_next_atto model added by creator
• Adan optimizer caution, and Lamb decoupled weighgt decay options
• Some feature_info metadata fixed by https://github.com/brianhou0208
• All OpenCLIP and JAX (CLIP, SigLIP, Pali, etc) model weights that used load time remapping were given their own HF Hub instances so that they work with hf-hub: based loading, and thus will work with new Transformers TimmWrapperModel

What's Changed

• Punch cache_dir through model factory / builder / pretrain helpers by @rwightman in #2356
• Yuweihao inception next atto merge by @rwightman in #2360
• Dataset trust remote tweaks by @rwightman in #2361
• Add --dataset-trust-remote-code to the train.py and validate.py scripts by @grodino in #2328
• Fix feature_info.reduction by @brianhou0208 in #2369
• Add caution to Adan. Add decouple decay option to LAMB. by @rwightman in #2357
• Switching to timm specific weight instances for open_clip image encoders by @rwightman in #2376
• Fix broken image link in Quickstart doc by @ariG23498 in #2381
• Supporting aimv2 encoders by @rwightman in #2379
• fix: minor typos in markdowns by @ruidazeng in #2382
• Add 384x384 in12k pretrain and finetune for convnext_nano by @rwightman in #2384
• Fixed unfused attn2d scale by @laclouis5 in #2387
• Fix MQA V2 by @laclouis5 in #2388
• Wrap torch checkpoint() fn to default use_reentrant flag to False and allow env var override by @rwightman in #2394
• Add half-precision (bfloat16, float16) support to train & validate scripts by @rwightman in #2397
• Merging wandb project name chages w/ addition by @rwightman in #2398

New Contributors

• @brianhou0208 made their first contribution in #2369
• @ariG23498 made their first contribution in #2381
• @ruidazeng made their first contribution in #2382
• @laclouis5 made their first contribution in #2387

Full Changelog: v1.0.12...v1.0.13

Nov 28, 2024

• More optimizers
  • Add MARS optimizer (https://arxiv.org/abs/2411.10438, https://github.com/AGI-Arena/MARS)
  • Add LaProp optimizer (https://arxiv.org/abs/2002.04839, https://github.com/Z-T-WANG/LaProp-Optimizer)
  • Add masking from 'Cautious Optimizers' (https://arxiv.org/abs/2411.16085, https://github.com/kyleliang919/C-Optim) to Adafactor, Adafactor Big Vision, AdamW (legacy), Adopt, Lamb, LaProp, Lion, NadamW, RMSPropTF, SGDW
  • Cleanup some docstrings and type annotations re optimizers and factory
• Add MobileNet-V4 Conv Medium models pretrained on in12k and fine-tuned in1k @ 384x384
  • https://huggingface.co/timm/mobilenetv4_conv_medium.e250_r384_in12k_ft_in1k
  • https://huggingface.co/timm/mobilenetv4_conv_medium.e250_r384_in12k
  • https://huggingface.co/timm/mobilenetv4_conv_medium.e180_ad_r384_in12k
  • https://huggingface.co/timm/mobilenetv4_conv_medium.e180_r384_in12k
• Add small cs3darknet, quite good for the speed
  • https://huggingface.co/timm/cs3darknet_focus_s.ra4_e3600_r256_in1k

Nov 12, 2024

• Optimizer factory refactor
  • New factory works by registering optimizers using an OptimInfo dataclass w/ some key traits
  • Add list_optimizers, get_optimizer_class, get_optimizer_info to reworked create_optimizer_v2 fn to explore optimizers, get info or class
  • deprecate optim.optim_factory, move fns to optim/_optim_factory.py and optim/_param_groups.py and encourage import via timm.optim
• Add Adopt (https://github.com/iShohei220/adopt) optimizer
• Add 'Big Vision' variant of Adafactor (https://github.com/google-research/big_vision/blob/main/big_vision/optax.py) optimizer
• Fix original Adafactor to pick better factorization dims for convolutions
• Tweak LAMB optimizer with some improvements in torch.where functionality since original, refactor clipping a bit
• dynamic img size support in vit, deit, eva improved to support resize from non-square patch grids, thanks https://github.com/wojtke

Oct 31, 2024

Add a set of new very well trained ResNet & ResNet-V2 18/34 (basic block) weights. See https://huggingface.co/blog/rwightman/resnet-trick-or-treat

Oct 19, 2024

• Cleanup torch amp usage to avoid cuda specific calls, merge support for Ascend (NPU) devices from MengqingCao that should work now in PyTorch 2.5 w/ new device extension autoloading feature. Tested Intel Arc (XPU) in Pytorch 2.5 too and it (mostly) worked.

What's Changed

• mambaout.py: fixed bug by @NightMachinery in #2305
• Cleanup some amp related behaviour to better support different (non-cuda) devices by @rwightman in #2308
• Add NPU backend support for val and inference by @MengqingCao in #2109
• Update some clip pretrained weights to point to new hub locations by @rwightman in #2311
• ResNet vs MNV4 v1/v2 18 & 34 weights by @rwightman in #2316
• Replace deprecated positional argument with --data-dir by @JosuaRieder in #2322
• Fix typo in train.py: bathes > batches by @JosuaRieder in #2321
• Fix positional embedding resampling for non-square inputs in ViT by @wojtke in #2317
• Add trust_remote_code argument to ReaderHfds by @grodino in #2326
• Extend train epoch schedule by warmup_epochs if warmup_prefix enabled by @rwightman in #2325
• Extend existing unit tests using Cover-Agent by @mrT23 in #2331
• An impl of adafactor as per big vision (scaling vit) changes by @rwightman in #2320
• Add py.typed file as recommended by PEP 561 by @antoinebrl in #2252
• Add CODE_OF_CONDUCT.md and CITATION.cff files by @AlinaImtiaz018 in #2333
• Add some 384x384 small model weights by @rwightman in #2334
• In dist training, update loss running avg every step, sync on log by @rwightman in #2340
• Improve WandB logging by @sinahmr in #2341
• A few weights to merge Friday by @rwightman in #2343
• Update timm torchvision resnet weight urls to the updated urls in torchvision by @JohannesTheo in #2346
• More optimizer updates, add MARS, LaProp, add Adopt fix and more by @rwightman in #2347
• Cautious optimizer impl plus some typing cleanup. by @rwightman in #2349
• Add cautious mars, improve test reliability by skipping grad diff for… by @rwightman in #2351
• See if we can avoid some model / layer pickle issues with the aa attr in ConvNormAct by @rwightman in #2353

New Contributors

• @MengqingCao made their first contribution in #2109
• @JosuaRieder made their first contribution in #2322
• @wojtke made their first contribution in #2317
• @grodino made their first contribution in #2326
• @AlinaImtiaz018 made their first contribution in #2333
• @sinahmr made their first contribution in #2341
• @JohannesTheo made their first contribution in #2346

Full Changelog: v1.0.11...v1.0.12

Quick turnaround from 1.0.10 to fix an error impacting 3rd party packages that still import through a deprecated path that isn't tested.

Oct 16, 2024

• Fix error on importing from deprecated path timm.models.registry, increased priority of existing deprecation warnings to be visible
• Port weights of InternViT-300M (https://huggingface.co/OpenGVLab/InternViT-300M-448px) to timm as vit_intern300m_patch14_448

Oct 14, 2024

• Pre-activation (ResNetV2) version of 18/18d/34/34d ResNet model defs added by request (weights pending)
• Release 1.0.10

Oct 11, 2024

• MambaOut (https://github.com/yuweihao/MambaOut) model & weights added. A cheeky take on SSM vision models w/o the SSM (essentially ConvNeXt w/ gating). A mix of original weights + custom variations & weights.

• SigLIP SO400M ViT fine-tunes on ImageNet-1k @ 378x378, added 378x378 option for existing SigLIP 384x384 models
  • vit_so400m_patch14_siglip_378.webli_ft_in1k - 89.42 top-1
  • vit_so400m_patch14_siglip_gap_378.webli_ft_in1k - 89.03
• SigLIP SO400M ViT encoder from recent multi-lingual (i18n) variant, patch16 @ 256x256 (https://huggingface.co/timm/ViT-SO400M-16-SigLIP-i18n-256). OpenCLIP update pending.
• Add two ConvNeXt 'Zepto' models & weights (one w/ overlapped stem and one w/ patch stem). Uses RMSNorm, smaller than previous 'Atto', 2.2M params.
  • convnext_zepto_rms_ols.ra4_e3600_r224_in1k - 73.20 top-1 @ 224
  • convnext_zepto_rms.ra4_e3600_r224_in1k - 72.81 @ 224

Sept 2024

• Add a suite of tiny test models for improved unit tests and niche low-resource applications (https://huggingface.co/blog/rwightman/timm-tiny-test)
• Add MobileNetV4-Conv-Small (0.5x) model (https://huggingface.co/posts/rwightman/793053396198664)
  • mobilenetv4_conv_small_050.e3000_r224_in1k - 65.81 top-1 @ 256, 64.76 @ 224
• Add MobileNetV3-Large variants trained with MNV4 Small recipe
  • mobilenetv3_large_150d.ra4_e3600_r256_in1k - 81.81 @ 320, 80.94 @ 256
  • mobilenetv3_large_100.ra4_e3600_r224_in1k - 77.16 @ 256, 76.31 @ 224

Oct 14, 2024

• Pre-activation (ResNetV2) version of 18/18d/34/34d ResNet model defs added by request (weights pending)
• Release 1.0.10

Oct 11, 2024

• MambaOut (https://github.com/yuweihao/MambaOut) model & weights added. A cheeky take on SSM vision models w/o the SSM (essentially ConvNeXt w/ gating). A mix of original weights + custom variations & weights.

• SigLIP SO400M ViT fine-tunes on ImageNet-1k @ 378x378, added 378x378 option for existing SigLIP 384x384 models
  • vit_so400m_patch14_siglip_378.webli_ft_in1k - 89.42 top-1
  • vit_so400m_patch14_siglip_gap_378.webli_ft_in1k - 89.03
• SigLIP SO400M ViT encoder from recent multi-lingual (i18n) variant, patch16 @ 256x256 (https://huggingface.co/timm/ViT-SO400M-16-SigLIP-i18n-256). OpenCLIP update pending.
• Add two ConvNeXt 'Zepto' models & weights (one w/ overlapped stem and one w/ patch stem). Uses RMSNorm, smaller than previous 'Atto', 2.2M params.
  • convnext_zepto_rms_ols.ra4_e3600_r224_in1k - 73.20 top-1 @ 224
  • convnext_zepto_rms.ra4_e3600_r224_in1k - 72.81 @ 224

Sept 2024

• Add a suite of tiny test models for improved unit tests and niche low-resource applications (https://huggingface.co/blog/rwightman/timm-tiny-test)
• Add MobileNetV4-Conv-Small (0.5x) model (https://huggingface.co/posts/rwightman/793053396198664)
  • mobilenetv4_conv_small_050.e3000_r224_in1k - 65.81 top-1 @ 256, 64.76 @ 224
• Add MobileNetV3-Large variants trained with MNV4 Small recipe
  • mobilenetv3_large_150d.ra4_e3600_r256_in1k - 81.81 @ 320, 80.94 @ 256
  • mobilenetv3_large_100.ra4_e3600_r224_in1k - 77.16 @ 256, 76.31 @ 224

Aug 21, 2024

• Updated SBB ViT models trained on ImageNet-12k and fine-tuned on ImageNet-1k, challenging quite a number of much larger, slower models

• MobileNet-V1 1.25, EfficientNet-B1, & ResNet50-D weights w/ MNV4 baseline challenge recipe

• Add SAM2 (HieraDet) backbone arch & weight loading support

• Add Hiera Small weights trained w/ abswin pos embed on in12k & fine-tuned on 1k

Aug 8, 2024

• Add RDNet ('DenseNets Reloaded', https://arxiv.org/abs/2403.19588), thanks Donghyun Kim

July 28, 2024

• Add mobilenet_edgetpu_v2_m weights w/ ra4 mnv4-small based recipe. 80.1% top-1 @ 224 and 80.7 @ 256.
• Release 1.0.8

July 26, 2024

• More MobileNet-v4 weights, ImageNet-12k pretrain w/ fine-tunes, and anti-aliased ConvLarge models

• Impressive MobileNet-V1 and EfficientNet-B0 baseline challenges (https://huggingface.co/blog/rwightman/mobilenet-baselines)

• Prototype of set_input_size() added to vit and swin v1/v2 models to allow changing image size, patch size, window size after model creation.
• Improved support in swin for different size handling, in addition to set_input_size, always_partition and strict_img_size args have been added to __init__ to allow more flexible input size constraints
• Fix out of order indices info for intermediate 'Getter' feature wrapper, check out or range indices for same.
• Add several tiny < .5M param models for testing that are actually trained on ImageNet-1k

• Fix vit reg token init, thanks Promisery
• Other misc fixes

June 24, 2024

• 3 more MobileNetV4 hyrid weights with different MQA weight init scheme

• florence2 weight loading in DaViT model

June 12, 2024

• MobileNetV4 models and initial set of timm trained weights added:

• Apple MobileCLIP (https://arxiv.org/pdf/2311.17049, FastViT and ViT-B) image tower model support & weights added (part of OpenCLIP support).
• ViTamin (https://arxiv.org/abs/2404.02132) CLIP image tower model & weights added (part of OpenCLIP support).
• OpenAI CLIP Modified ResNet image tower modelling & weight support (via ByobNet). Refactor AttentionPool2d.
• Refactoring & improvements, especially related to classifier_reset and num_features vs head_hidden_size for forward_features() vs pre_logits

May 14, 2024

• Support loading PaliGemma jax weights into SigLIP ViT models with average pooling.
• Add Hiera models from Meta (https://github.com/facebookresearch/hiera).
• Add normalize= flag for transorms, return non-normalized torch.Tensor with original dytpe (for chug)
• Version 1.0.3 release

May 11, 2024

• Searching for Better ViT Baselines (For the GPU Poor) weights and vit variants released. Exploring model shapes between Tiny and Base.

• AttentionExtract helper added to extract attention maps from timm models. See example in #1232 (comment)
• forward_intermediates() API refined and added to more models including some ConvNets that have other extraction methods.
• 1017 of 1047 model architectures support features_only=True feature extraction. Remaining 34 architectures can be supported but based on priority requests.
• Remove torch.jit.script annotated functions including old JIT activations. Conflict with dynamo and dynamo does a much better job when used.

April 11, 2024

• Prepping for a long overdue 1.0 release, things have been stable for a while now.
• Significant feature that's been missing for a while, 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*')
• Above feature support achieved through a new forward_intermediates() API that can be used with a feature wrapping module or direclty.

model = timm.create_model('vit_base_patch16_224')
final_feat, intermediates = model.forward_intermediates(input) 
output = model.forward_head(final_feat)  # pooling + classifier head

print(final_feat.shape)
torch.Size([2, 197, 768])

for f in intermediates:
    print(f.shape)
torch.Size([2, 768, 14, 14])
torch.Size([2, 768, 14, 14])
torch.Size([2, 768, 14, 14])
torch.Size([2, 768, 14, 14])
torch.Size([2, 768, 14, 14])
torch.Size([2, 768, 14, 14])
torch.Size([2, 768, 14, 14])
torch.Size([2, 768, 14, 14])
torch.Size([2, 768, 14, 14])
torch.Size([2, 768, 14, 14])
torch.Size([2, 768, 14, 14])
torch.Size([2, 768, 14, 14])

print(output.shape)
torch.Size([2, 1000])

model = timm.create_model('eva02_base_patch16_clip_224', pretrained=True, img_size=512, features_only=True, out_indices=(-3, -2,))
output = model(torch.randn(2, 3, 512, 512))

for o in output:    
    print(o.shape)   
torch.Size([2, 768, 32, 32])
torch.Size([2, 768, 32, 32])

• TinyCLIP vision tower weights added, thx Thien Tran