# Spandrel
[](https://pypi.org/project/spandrel/)
[](https://github.com/chaiNNer-org/spandrel/releases)
[](https://pypi.org/project/spandrel/#files)
[](https://pypi.org/project/spandrel/#files:~:text=Requires%3A%20Python%20%3C3.12%2C%20%3E%3D3.8)
[](https://chainner.app/spandrel/)
[](https://github.com/chaiNNer-org/spandrel/actions)
[](https://github.com/chaiNNer-org/spandrel/blob/main/LICENSE)
[](https://github.com/chaiNNer-org/spandrel/graphs/contributors)
Spandrel is a library for loading and running pre-trained PyTorch models. It automatically detects the model architecture and hyperparameters from model files, and provides a unified interface for running models.
After seeing many projects extract out [chaiNNer](https://github.com/chaiNNer-org/chaiNNer)'s model support into their own projects, I decided to create this PyPi package for the architecture support and model loading functionality. I'm also hoping that by having a central package anyone can use, the community will be encouraged [to help add support for more models](CONTRIBUTING.md).
This package does not yet have easy inference code, but porting that code is planned as well.
## Installation
Spandrel is available through pip:
```shell
pip install spandrel
```
## Basic Usage
While Spandrel supports different kinds of models, this is how you would run a super resolution model (e.g. ESRGAN, SwinIR, HAT, etc.):
```python
from spandrel import ImageModelDescriptor, ModelLoader
import torch
# load a model from disk
model = ModelLoader().load_from_file(r"path/to/model.pth")
# make sure it's an image to image model
assert isinstance(model, ImageModelDescriptor)
# send it to the GPU and put it in inference mode
model.cuda().eval()
# use the model
def process(image: torch.Tensor) -> torch.Tensor:
with torch.no_grad():
return model(image)
```
Note that `model` is a [`ModelDescriptor`](https://chainner.app/spandrel/#ModelDescriptor) object, which is a wrapper around the actual PyTorch model. This wrapper provides a unified interface for running models, and also contains metadata about the model. See [`ImageModelDescriptor`](https://chainner.app/spandrel/spandrel.ImageModelDescriptor.html) for more details about the metadata contained and how to call the model.
> **_NOTE: `ImageModelDescriptor` will NOT convert an image to a tensor for you. It is purely making the forward passes of these models more convenient to use, since the actual forward passes are not always as simple as image in/image out._**
### More architectures
If you are working on a non-commercial open-source project or a private project, you should use `spandrel` and `spandrel_extra_arches` to get everything spandrel has to offer. The `spandrel` package only contains architectures with [permissive and public domain licenses](https://en.wikipedia.org/wiki/Permissive_software_license) (MIT, Apache 2.0, public domain), so it is fit for every use case. Architectures with restrictive licenses (e.g. non-commercial) are implemented in the `spandrel_extra_arches` package.
```python
import spandrel
import spandrel_extra_arches
# add extra architectures before `ModelLoader` is used
spandrel_extra_arches.install()
# load a model from disk
model = spandrel.ModelLoader().load_from_file(r"path/to/model.pth")
... # use model
```
## Supported File Types
Spandrel mainly supports loading `.pth` files for all supported architectures. This is what you will typically find from official repos and community trained models. However, Spandrel also supports loading TorchScript traced models (`.pt`), certain types of `.ckpt` files, and `.safetensors` files for any supported architecture saved in one of these formats.
## Model Architecture Support
> **_NOTE: By its very nature, Spandrel will never be able to support every model architecture. The goal is just to support as many as is realistically possible._**
Spandrel currently supports a limited amount of network architectures. If the architecture you need is not supported, feel free to [request it](https://github.com/chaiNNer-org/spandrel/issues) or try [adding it](CONTRIBUTING.md).
#### Single Image Super Resolution
- [ESRGAN](https://github.com/xinntao/ESRGAN) (RRDBNet)
- This includes regular [ESRGAN](https://github.com/xinntao/ESRGAN), [ESRGAN+](https://github.com/ncarraz/ESRGANplus), "new-arch ESRGAN" ([RealSR](https://github.com/jixiaozhong/RealSR), [BSRGAN](https://github.com/cszn/BSRGAN)), and [Real-ESRGAN](https://github.com/xinntao/Real-ESRGAN)
- Models: [Community ESRGAN](https://openmodeldb.info) | [ESRGAN+](https://drive.google.com/drive/folders/1lNky9afqEP-qdxrAwDFPJ1g0ui4x7Sin) | [BSRGAN](https://github.com/cszn/BSRGAN/tree/main/model_zoo) | [RealSR](https://github.com/jixiaozhong/RealSR#pre-trained-models) | [Real-ESRGAN](https://github.com/xinntao/Real-ESRGAN/blob/master/docs/model_zoo.md)
- [Real-ESRGAN Compact](https://github.com/xinntao/Real-ESRGAN) (SRVGGNet) | [Models](https://github.com/xinntao/Real-ESRGAN/blob/master/docs/model_zoo.md)
- [Swift-SRGAN](https://github.com/Koushik0901/Swift-SRGAN) | [Models](https://github.com/Koushik0901/Swift-SRGAN/releases/tag/v0.1)
- [SwinIR](https://github.com/JingyunLiang/SwinIR) | [Models](https://github.com/JingyunLiang/SwinIR/releases/tag/v0.0)
- [Swin2SR](https://github.com/mv-lab/swin2sr) | [Models](https://github.com/mv-lab/swin2sr/releases/tag/v0.0.1)
- [HAT](https://github.com/XPixelGroup/HAT) | [Models](https://drive.google.com/drive/folders/1HpmReFfoUqUbnAOQ7rvOeNU3uf_m69w0)
- [Omni-SR](https://github.com/Francis0625/Omni-SR) | [Models](https://github.com/Francis0625/Omni-SR#preparation)
- [SRFormer](https://github.com/HVision-NKU/SRFormer) (+) | [Models](https://github.com/HVision-NKU/SRFormer#pretrain-models)
- [DAT](https://github.com/zhengchen1999/DAT) | [Models](https://github.com/zhengchen1999/DAT#testing)
- [FeMaSR](https://github.com/chaofengc/FeMaSR) (+) | [Models](https://github.com/chaofengc/FeMaSR/releases/tag/v0.1-pretrain_models)
- [GRL](https://github.com/ofsoundof/GRL-Image-Restoration) | [Models](https://github.com/ofsoundof/GRL-Image-Restoration/releases/tag/v1.0.0)
- [DITN](https://github.com/yongliuy/DITN) | [Models](https://drive.google.com/drive/folders/1XpHW27H5j2S4IH8t4lccgrgHkIjqrS-X)
- [MM-RealSR](https://github.com/TencentARC/MM-RealSR) | [Models](https://github.com/TencentARC/MM-RealSR/releases/tag/v1.0.0)
- [SPAN](https://github.com/hongyuanyu/SPAN) | [Models](https://drive.google.com/file/d/1iYUA2TzKuxI0vzmA-UXr_nB43XgPOXUg/view?usp=sharing)
- [Real-CUGAN](https://github.com/bilibili/ailab/tree/main/Real-CUGAN) | [Models](https://drive.google.com/drive/folders/1jAJyBf2qKe2povySwsGXsVMnzVyQzqDD), [Pro Models](https://drive.google.com/drive/folders/1hfT4WwnNUaS43ErrgXk0J1R5Ik8s5NVo)
- [CRAFT](https://github.com/AVC2-UESTC/CRAFT-SR) | [Models](https://drive.google.com/file/d/13wAmc93BPeBUBQ24zUZOuUpdBFG2aAY5/view?usp=sharing)
- [SAFMN](https://github.com/sunny2109/SAFMN) | [Models](https://drive.google.com/drive/folders/12O_xgwfgc76DsYbiClYnl6ErCDrsi_S9?usp=share_link), [JPEG model](https://github.com/sunny2109/SAFMN/releases/tag/v0.1.1)
- [RGT](https://github.com/zhengchen1999/RGT) | [RGT Models](https://drive.google.com/drive/folders/1zxrr31Kp2D_N9a-OUAPaJEn_yTaSXTfZ?usp=drive_link), [RGT-S Models](https://drive.google.com/drive/folders/1j46WHs1Gvyif1SsZXKy1Y1IrQH0gfIQ1?usp=drive_link)
- [DCTLSA](https://github.com/zengkun301/DCTLSA) | [Models](https://github.com/zengkun301/DCTLSA/tree/main/pretrained)
- [ATD](https://github.com/LabShuHangGU/Adaptive-Token-Dictionary) | [Models](https://drive.google.com/drive/folders/1D3BvTS1xBcaU1mp50k3pBzUWb7qjRvmB?usp=sharing)
- [AdaCode](https://github.com/kechunl/AdaCode) | [Models](https://github.com/kechunl/AdaCode/releases/tag/v0-pretrain_models)
- [DRCT](https://github.com/ming053l/DRCT)
- [PLKSR](https://github.com/dslisleedh/PLKSR) and [RealPLKSR](https://github.com/muslll/neosr/blob/master/neosr/archs/realplksr_arch.py) | [Models](https://drive.google.com/drive/u/1/folders/1lIkZ00y9cRQpLU9qmCIB2XtS-2ZoqKq8)
- [SeemoRe](https://github.com/eduardzamfir/seemoredetails) | [Models](https://drive.google.com/drive/folders/15jtvcS4jL_6QqEwaRodEN8FBrqVPrO2u?usp=share_link)
- [MoSR](https://github.com/umzi2/MoSR) | [Models](https://drive.google.com/drive/u/0/folders/1HPy7M4Zzq8oxhdsQ2cnfqy73klmQWp_r)
#### Face Restoration
- [GFPGAN](https://github.com/TencentARC/GFPGAN) | [1.2](https://github.com/TencentARC/GFPGAN/releases/download/v1.3.0/GFPGANv1.2.pth), [1.3](https://github.com/TencentARC/GFPGAN/releases/download/v1.3.0/GFPGANv1.3.pth), [1.4](https://github.com/TencentARC/GFPGAN/releases/download/v1.3.4/GFPGANv1.4.pth)
- [RestoreFormer](https://github.com/wzhouxiff/RestoreFormer) | [Model](https://github.com/TencentARC/GFPGAN/releases/download/v1.3.4/RestoreFormer.pth)
- [CodeFormer](https://github.com/sczhou/CodeFormer) (+) | [Model](https://github.com/sczhou/CodeFormer/releases/download/v0.1.0/codeformer.pth)
#### Inpainting
- [LaMa](https://github.com/advimman/lama) | [Model](https://github.com/Sanster/models/releases/download/add_big_lama/big-lama.pt)
- [MAT](https://github.com/fenglinglwb/MAT) (+) | [Model](https://github.com/Sanster/models/releases/download/add_mat/Places_512_FullData_G.pth)
#### Denoising
- [SCUNet](https://github.com/cszn/SCUNet) | [GAN Model](https://github.com/cszn/KAIR/releases/download/v1.0/scunet_color_real_gan.pth) | [PSNR Model](https://github.com/cszn/KAIR/releases/download/v1.0/scunet_color_real_psnr.pth)
- [Uformer](https://github.com/ZhendongWang6/Uformer) | [Denoise SIDD Model](https://mailustceducn-my.sharepoint.com/:u:/g/personal/zhendongwang_mail_ustc_edu_cn/Ea7hMP82A0xFlOKPlQnBJy0B9gVP-1MJL75mR4QKBMGc2w?e=iOz0zz) | [Deblur GoPro Model](https://mailustceducn-my.sharepoint.com/:u:/g/personal/zhendongwang_mail_ustc_edu_cn/EfCPoTSEKJRAshoE6EAC_3YB7oNkbLUX6AUgWSCwoJe0oA?e=jai90x)
- [KBNet](https://github.com/zhangyi-3/KBNet) | [Models](https://mycuhk-my.sharepoint.com/:f:/g/personal/1155135732_link_cuhk_edu_hk/EofsV3eVcAxNlrW72JXqzRUBhkM1Mzw50pJ3BHlAyMYnVw?e=MeMB5H)
- [NAFNet](https://github.com/megvii-research/NAFNet) | [Models](https://github.com/megvii-research/NAFNet#results-and-pre-trained-models)
- [Restormer](https://github.com/swz30/Restormer) (+) | [Models](https://github.com/swz30/Restormer/releases/tag/v1.0)
- [FFTformer](https://github.com/kkkls/FFTformer) | [Models](https://github.com/kkkls/FFTformer/releases/tag/pretrain_model)
- [M3SNet](https://github.com/Tombs98/M3SNet) (+) | [Models](https://drive.google.com/drive/folders/1y4BEX7LagtXVO98ZItSbJJl7WWM3gnbD)
- [MPRNet](https://github.com/swz30/MPRNet) (+) | [Deblurring](https://drive.google.com/file/d/1QwQUVbk6YVOJViCsOKYNykCsdJSVGRtb/view?usp=sharing), [Deraining](https://drive.google.com/file/d/1O3WEJbcat7eTY6doXWeorAbQ1l_WmMnM/view?usp=sharing), [Denoising](https://drive.google.com/file/d/1LODPt9kYmxwU98g96UrRA0_Eh5HYcsRw/view?usp=sharing)
- [MIRNet2](https://github.com/swz30/MIRNetv2) (+) | [Models](https://github.com/swz30/MIRNetv2/releases/tag/v1.0.0) (SR not supported)
- [DnCNN, FDnCNN](https://github.com/cszn/DPIR) | [Models](https://github.com/cszn/KAIR/releases/tag/v1.0)
- [DRUNet](https://github.com/cszn/DPIR) | [Models](https://github.com/cszn/KAIR/releases/tag/v1.0)
- [IPT](https://github.com/huawei-noah/Pretrained-IPT) | [Models](https://drive.google.com/drive/folders/1MVSdUX0YBExauG0fFz4ANiWTrq9xZEj7?usp=sharing)
#### DeJPEG
- [FBCNN](https://github.com/jiaxi-jiang/FBCNN) | [Models](https://github.com/jiaxi-jiang/FBCNN/releases/tag/v1.0)
#### Colorization
- [DDColor](https://github.com/piddnad/DDColor) (+) | [Models](https://github.com/piddnad/DDColor/blob/master/MODEL_ZOO.md)
#### Dehazing
- [MixDehazeNet](https://github.com/AmeryXiong/MixDehazeNet) | [Models](https://drive.google.com/drive/folders/1ep6W4H3vNxshYjq71Tb3MzxrXGgaiM6C?usp=drive_link)
#### Low-light Enhancement
- [RetinexFormer](https://github.com/caiyuanhao1998/Retinexformer) | [Models](https://drive.google.com/drive/folders/1ynK5hfQachzc8y96ZumhkPPDXzHJwaQV?usp=drive_link)
- [HVI-CIDNet](https://github.com/Fediory/HVI-CIDNet) | [Models](https://github.com/Fediory/HVI-CIDNet/#weights-and-results-)
(All architectures marked with a `+` are only part of `spandrel_extra_arches`.)
## Security
Use `.safetensors` files for guaranteed security.
As you may know, loading `.pth` files [poses a security risk](https://github.com/pytorch/pytorch/issues/52596) due to python's `pickle` module being inherently unsafe and vulnerable to arbitrary code execution (ACE). To mitigate this, Spandrel only allows deserializing certain types of data. This helps to improve security, but it still doesn't fully solve the issue of ACE.
## Used By
Here are some cool projects that use Spandrel:
- [AUTOMATIC1111's SD WebUI](https://github.com/AUTOMATIC1111/stable-diffusion-webui)
- [ComfyUI](https://github.com/comfyanonymous/ComfyUI)
- [InvokeAI](https://github.com/invoke-ai/InvokeAI)
- [chaiNNer](https://github.com/chaiNNer-org/chaiNNer)
- [imaginAIry](https://github.com/brycedrennan/imaginAIry)
- [dgenerate](https://github.com/Teriks/dgenerate)
## License
This repo is bounded by the MIT license. However, the code of implemented architectures (everything inside an `__arch/` directory) is bound by their original respective licenses (which are included in their respective `__arch/` directories).
Raw data
{
"_id": null,
"home_page": null,
"name": "spandrel",
"maintainer": null,
"docs_url": null,
"requires_python": ">=3.8",
"maintainer_email": null,
"keywords": "spandrel, pytorch architecture, pytorch arch, model arch, model architecture",
"author": "chaiNNer team",
"author_email": null,
"download_url": "https://files.pythonhosted.org/packages/bb/7e/2f3a41c69651cf90b636857147568eb1806f56af5b8015ac7638a930abe1/spandrel-0.4.0.tar.gz",
"platform": null,
"description": "# Spandrel\n\n[](https://pypi.org/project/spandrel/)\n[](https://github.com/chaiNNer-org/spandrel/releases)\n[](https://pypi.org/project/spandrel/#files)\n[](https://pypi.org/project/spandrel/#files:~:text=Requires%3A%20Python%20%3C3.12%2C%20%3E%3D3.8)\n[](https://chainner.app/spandrel/)\n\n[](https://github.com/chaiNNer-org/spandrel/actions)\n[](https://github.com/chaiNNer-org/spandrel/blob/main/LICENSE)\n[](https://github.com/chaiNNer-org/spandrel/graphs/contributors)\n\nSpandrel is a library for loading and running pre-trained PyTorch models. It automatically detects the model architecture and hyperparameters from model files, and provides a unified interface for running models.\n\nAfter seeing many projects extract out [chaiNNer](https://github.com/chaiNNer-org/chaiNNer)'s model support into their own projects, I decided to create this PyPi package for the architecture support and model loading functionality. I'm also hoping that by having a central package anyone can use, the community will be encouraged [to help add support for more models](CONTRIBUTING.md).\n\nThis package does not yet have easy inference code, but porting that code is planned as well.\n\n## Installation\n\nSpandrel is available through pip:\n\n```shell\npip install spandrel\n```\n\n## Basic Usage\n\nWhile Spandrel supports different kinds of models, this is how you would run a super resolution model (e.g. ESRGAN, SwinIR, HAT, etc.):\n\n```python\nfrom spandrel import ImageModelDescriptor, ModelLoader\nimport torch\n\n# load a model from disk\nmodel = ModelLoader().load_from_file(r\"path/to/model.pth\")\n\n# make sure it's an image to image model\nassert isinstance(model, ImageModelDescriptor)\n\n# send it to the GPU and put it in inference mode\nmodel.cuda().eval()\n\n# use the model\ndef process(image: torch.Tensor) -> torch.Tensor:\n with torch.no_grad():\n return model(image)\n```\n\nNote that `model` is a [`ModelDescriptor`](https://chainner.app/spandrel/#ModelDescriptor) object, which is a wrapper around the actual PyTorch model. This wrapper provides a unified interface for running models, and also contains metadata about the model. See [`ImageModelDescriptor`](https://chainner.app/spandrel/spandrel.ImageModelDescriptor.html) for more details about the metadata contained and how to call the model.\n\n> **_NOTE: `ImageModelDescriptor` will NOT convert an image to a tensor for you. It is purely making the forward passes of these models more convenient to use, since the actual forward passes are not always as simple as image in/image out._**\n\n### More architectures\n\nIf you are working on a non-commercial open-source project or a private project, you should use `spandrel` and `spandrel_extra_arches` to get everything spandrel has to offer. The `spandrel` package only contains architectures with [permissive and public domain licenses](https://en.wikipedia.org/wiki/Permissive_software_license) (MIT, Apache 2.0, public domain), so it is fit for every use case. Architectures with restrictive licenses (e.g. non-commercial) are implemented in the `spandrel_extra_arches` package.\n\n```python\nimport spandrel\nimport spandrel_extra_arches\n\n# add extra architectures before `ModelLoader` is used\nspandrel_extra_arches.install()\n\n# load a model from disk\nmodel = spandrel.ModelLoader().load_from_file(r\"path/to/model.pth\")\n\n... # use model\n```\n\n## Supported File Types\n\nSpandrel mainly supports loading `.pth` files for all supported architectures. This is what you will typically find from official repos and community trained models. However, Spandrel also supports loading TorchScript traced models (`.pt`), certain types of `.ckpt` files, and `.safetensors` files for any supported architecture saved in one of these formats.\n\n## Model Architecture Support\n\n> **_NOTE: By its very nature, Spandrel will never be able to support every model architecture. The goal is just to support as many as is realistically possible._**\n\nSpandrel currently supports a limited amount of network architectures. If the architecture you need is not supported, feel free to [request it](https://github.com/chaiNNer-org/spandrel/issues) or try [adding it](CONTRIBUTING.md).\n\n#### Single Image Super Resolution\n\n- [ESRGAN](https://github.com/xinntao/ESRGAN) (RRDBNet)\n - This includes regular [ESRGAN](https://github.com/xinntao/ESRGAN), [ESRGAN+](https://github.com/ncarraz/ESRGANplus), \"new-arch ESRGAN\" ([RealSR](https://github.com/jixiaozhong/RealSR), [BSRGAN](https://github.com/cszn/BSRGAN)), and [Real-ESRGAN](https://github.com/xinntao/Real-ESRGAN)\n - Models: [Community ESRGAN](https://openmodeldb.info) | [ESRGAN+](https://drive.google.com/drive/folders/1lNky9afqEP-qdxrAwDFPJ1g0ui4x7Sin) | [BSRGAN](https://github.com/cszn/BSRGAN/tree/main/model_zoo) | [RealSR](https://github.com/jixiaozhong/RealSR#pre-trained-models) | [Real-ESRGAN](https://github.com/xinntao/Real-ESRGAN/blob/master/docs/model_zoo.md)\n- [Real-ESRGAN Compact](https://github.com/xinntao/Real-ESRGAN) (SRVGGNet) | [Models](https://github.com/xinntao/Real-ESRGAN/blob/master/docs/model_zoo.md)\n- [Swift-SRGAN](https://github.com/Koushik0901/Swift-SRGAN) | [Models](https://github.com/Koushik0901/Swift-SRGAN/releases/tag/v0.1)\n- [SwinIR](https://github.com/JingyunLiang/SwinIR) | [Models](https://github.com/JingyunLiang/SwinIR/releases/tag/v0.0)\n- [Swin2SR](https://github.com/mv-lab/swin2sr) | [Models](https://github.com/mv-lab/swin2sr/releases/tag/v0.0.1)\n- [HAT](https://github.com/XPixelGroup/HAT) | [Models](https://drive.google.com/drive/folders/1HpmReFfoUqUbnAOQ7rvOeNU3uf_m69w0)\n- [Omni-SR](https://github.com/Francis0625/Omni-SR) | [Models](https://github.com/Francis0625/Omni-SR#preparation)\n- [SRFormer](https://github.com/HVision-NKU/SRFormer) (+) | [Models](https://github.com/HVision-NKU/SRFormer#pretrain-models)\n- [DAT](https://github.com/zhengchen1999/DAT) | [Models](https://github.com/zhengchen1999/DAT#testing)\n- [FeMaSR](https://github.com/chaofengc/FeMaSR) (+) | [Models](https://github.com/chaofengc/FeMaSR/releases/tag/v0.1-pretrain_models)\n- [GRL](https://github.com/ofsoundof/GRL-Image-Restoration) | [Models](https://github.com/ofsoundof/GRL-Image-Restoration/releases/tag/v1.0.0)\n- [DITN](https://github.com/yongliuy/DITN) | [Models](https://drive.google.com/drive/folders/1XpHW27H5j2S4IH8t4lccgrgHkIjqrS-X)\n- [MM-RealSR](https://github.com/TencentARC/MM-RealSR) | [Models](https://github.com/TencentARC/MM-RealSR/releases/tag/v1.0.0)\n- [SPAN](https://github.com/hongyuanyu/SPAN) | [Models](https://drive.google.com/file/d/1iYUA2TzKuxI0vzmA-UXr_nB43XgPOXUg/view?usp=sharing)\n- [Real-CUGAN](https://github.com/bilibili/ailab/tree/main/Real-CUGAN) | [Models](https://drive.google.com/drive/folders/1jAJyBf2qKe2povySwsGXsVMnzVyQzqDD), [Pro Models](https://drive.google.com/drive/folders/1hfT4WwnNUaS43ErrgXk0J1R5Ik8s5NVo)\n- [CRAFT](https://github.com/AVC2-UESTC/CRAFT-SR) | [Models](https://drive.google.com/file/d/13wAmc93BPeBUBQ24zUZOuUpdBFG2aAY5/view?usp=sharing)\n- [SAFMN](https://github.com/sunny2109/SAFMN) | [Models](https://drive.google.com/drive/folders/12O_xgwfgc76DsYbiClYnl6ErCDrsi_S9?usp=share_link), [JPEG model](https://github.com/sunny2109/SAFMN/releases/tag/v0.1.1)\n- [RGT](https://github.com/zhengchen1999/RGT) | [RGT Models](https://drive.google.com/drive/folders/1zxrr31Kp2D_N9a-OUAPaJEn_yTaSXTfZ?usp=drive_link), [RGT-S Models](https://drive.google.com/drive/folders/1j46WHs1Gvyif1SsZXKy1Y1IrQH0gfIQ1?usp=drive_link)\n- [DCTLSA](https://github.com/zengkun301/DCTLSA) | [Models](https://github.com/zengkun301/DCTLSA/tree/main/pretrained)\n- [ATD](https://github.com/LabShuHangGU/Adaptive-Token-Dictionary) | [Models](https://drive.google.com/drive/folders/1D3BvTS1xBcaU1mp50k3pBzUWb7qjRvmB?usp=sharing)\n- [AdaCode](https://github.com/kechunl/AdaCode) | [Models](https://github.com/kechunl/AdaCode/releases/tag/v0-pretrain_models)\n- [DRCT](https://github.com/ming053l/DRCT)\n- [PLKSR](https://github.com/dslisleedh/PLKSR) and [RealPLKSR](https://github.com/muslll/neosr/blob/master/neosr/archs/realplksr_arch.py) | [Models](https://drive.google.com/drive/u/1/folders/1lIkZ00y9cRQpLU9qmCIB2XtS-2ZoqKq8)\n- [SeemoRe](https://github.com/eduardzamfir/seemoredetails) | [Models](https://drive.google.com/drive/folders/15jtvcS4jL_6QqEwaRodEN8FBrqVPrO2u?usp=share_link)\n- [MoSR](https://github.com/umzi2/MoSR) | [Models](https://drive.google.com/drive/u/0/folders/1HPy7M4Zzq8oxhdsQ2cnfqy73klmQWp_r)\n\n#### Face Restoration\n\n- [GFPGAN](https://github.com/TencentARC/GFPGAN) | [1.2](https://github.com/TencentARC/GFPGAN/releases/download/v1.3.0/GFPGANv1.2.pth), [1.3](https://github.com/TencentARC/GFPGAN/releases/download/v1.3.0/GFPGANv1.3.pth), [1.4](https://github.com/TencentARC/GFPGAN/releases/download/v1.3.4/GFPGANv1.4.pth)\n- [RestoreFormer](https://github.com/wzhouxiff/RestoreFormer) | [Model](https://github.com/TencentARC/GFPGAN/releases/download/v1.3.4/RestoreFormer.pth)\n- [CodeFormer](https://github.com/sczhou/CodeFormer) (+) | [Model](https://github.com/sczhou/CodeFormer/releases/download/v0.1.0/codeformer.pth)\n\n#### Inpainting\n\n- [LaMa](https://github.com/advimman/lama) | [Model](https://github.com/Sanster/models/releases/download/add_big_lama/big-lama.pt)\n- [MAT](https://github.com/fenglinglwb/MAT) (+) | [Model](https://github.com/Sanster/models/releases/download/add_mat/Places_512_FullData_G.pth)\n\n#### Denoising\n\n- [SCUNet](https://github.com/cszn/SCUNet) | [GAN Model](https://github.com/cszn/KAIR/releases/download/v1.0/scunet_color_real_gan.pth) | [PSNR Model](https://github.com/cszn/KAIR/releases/download/v1.0/scunet_color_real_psnr.pth)\n- [Uformer](https://github.com/ZhendongWang6/Uformer) | [Denoise SIDD Model](https://mailustceducn-my.sharepoint.com/:u:/g/personal/zhendongwang_mail_ustc_edu_cn/Ea7hMP82A0xFlOKPlQnBJy0B9gVP-1MJL75mR4QKBMGc2w?e=iOz0zz) | [Deblur GoPro Model](https://mailustceducn-my.sharepoint.com/:u:/g/personal/zhendongwang_mail_ustc_edu_cn/EfCPoTSEKJRAshoE6EAC_3YB7oNkbLUX6AUgWSCwoJe0oA?e=jai90x)\n- [KBNet](https://github.com/zhangyi-3/KBNet) | [Models](https://mycuhk-my.sharepoint.com/:f:/g/personal/1155135732_link_cuhk_edu_hk/EofsV3eVcAxNlrW72JXqzRUBhkM1Mzw50pJ3BHlAyMYnVw?e=MeMB5H)\n- [NAFNet](https://github.com/megvii-research/NAFNet) | [Models](https://github.com/megvii-research/NAFNet#results-and-pre-trained-models)\n- [Restormer](https://github.com/swz30/Restormer) (+) | [Models](https://github.com/swz30/Restormer/releases/tag/v1.0)\n- [FFTformer](https://github.com/kkkls/FFTformer) | [Models](https://github.com/kkkls/FFTformer/releases/tag/pretrain_model)\n- [M3SNet](https://github.com/Tombs98/M3SNet) (+) | [Models](https://drive.google.com/drive/folders/1y4BEX7LagtXVO98ZItSbJJl7WWM3gnbD)\n- [MPRNet](https://github.com/swz30/MPRNet) (+) | [Deblurring](https://drive.google.com/file/d/1QwQUVbk6YVOJViCsOKYNykCsdJSVGRtb/view?usp=sharing), [Deraining](https://drive.google.com/file/d/1O3WEJbcat7eTY6doXWeorAbQ1l_WmMnM/view?usp=sharing), [Denoising](https://drive.google.com/file/d/1LODPt9kYmxwU98g96UrRA0_Eh5HYcsRw/view?usp=sharing)\n- [MIRNet2](https://github.com/swz30/MIRNetv2) (+) | [Models](https://github.com/swz30/MIRNetv2/releases/tag/v1.0.0) (SR not supported)\n- [DnCNN, FDnCNN](https://github.com/cszn/DPIR) | [Models](https://github.com/cszn/KAIR/releases/tag/v1.0)\n- [DRUNet](https://github.com/cszn/DPIR) | [Models](https://github.com/cszn/KAIR/releases/tag/v1.0)\n- [IPT](https://github.com/huawei-noah/Pretrained-IPT) | [Models](https://drive.google.com/drive/folders/1MVSdUX0YBExauG0fFz4ANiWTrq9xZEj7?usp=sharing)\n\n#### DeJPEG\n\n- [FBCNN](https://github.com/jiaxi-jiang/FBCNN) | [Models](https://github.com/jiaxi-jiang/FBCNN/releases/tag/v1.0)\n\n#### Colorization\n\n- [DDColor](https://github.com/piddnad/DDColor) (+) | [Models](https://github.com/piddnad/DDColor/blob/master/MODEL_ZOO.md)\n\n#### Dehazing\n\n- [MixDehazeNet](https://github.com/AmeryXiong/MixDehazeNet) | [Models](https://drive.google.com/drive/folders/1ep6W4H3vNxshYjq71Tb3MzxrXGgaiM6C?usp=drive_link)\n\n#### Low-light Enhancement\n\n- [RetinexFormer](https://github.com/caiyuanhao1998/Retinexformer) | [Models](https://drive.google.com/drive/folders/1ynK5hfQachzc8y96ZumhkPPDXzHJwaQV?usp=drive_link)\n- [HVI-CIDNet](https://github.com/Fediory/HVI-CIDNet) | [Models](https://github.com/Fediory/HVI-CIDNet/#weights-and-results-)\n\n(All architectures marked with a `+` are only part of `spandrel_extra_arches`.)\n\n## Security\n\nUse `.safetensors` files for guaranteed security.\n\nAs you may know, loading `.pth` files [poses a security risk](https://github.com/pytorch/pytorch/issues/52596) due to python's `pickle` module being inherently unsafe and vulnerable to arbitrary code execution (ACE). To mitigate this, Spandrel only allows deserializing certain types of data. This helps to improve security, but it still doesn't fully solve the issue of ACE.\n\n## Used By\n\nHere are some cool projects that use Spandrel:\n\n- [AUTOMATIC1111's SD WebUI](https://github.com/AUTOMATIC1111/stable-diffusion-webui)\n- [ComfyUI](https://github.com/comfyanonymous/ComfyUI)\n- [InvokeAI](https://github.com/invoke-ai/InvokeAI)\n- [chaiNNer](https://github.com/chaiNNer-org/chaiNNer)\n- [imaginAIry](https://github.com/brycedrennan/imaginAIry)\n- [dgenerate](https://github.com/Teriks/dgenerate)\n\n## License\n\nThis repo is bounded by the MIT license. However, the code of implemented architectures (everything inside an `__arch/` directory) is bound by their original respective licenses (which are included in their respective `__arch/` directories).\n",
"bugtrack_url": null,
"license": "MIT",
"summary": "Give your project support for a variety of PyTorch model architectures, including auto-detecting model architecture from just .pth files. spandrel gives you arch support.",
"version": "0.4.0",
"project_urls": null,
"split_keywords": [
"spandrel",
" pytorch architecture",
" pytorch arch",
" model arch",
" model architecture"
],
"urls": [
{
"comment_text": "",
"digests": {
"blake2b_256": "5bade27e971a75deaaac7817e022174f88fae4db203216ac59c1ace99f3e56eb",
"md5": "8e00f2b167ce5fa4591e6046e0c1ce89",
"sha256": "819f3f15fe944fe589f43ba33150458a00d0ed0c2e8ef1eef0198615320127b1"
},
"downloads": -1,
"filename": "spandrel-0.4.0-py3-none-any.whl",
"has_sig": false,
"md5_digest": "8e00f2b167ce5fa4591e6046e0c1ce89",
"packagetype": "bdist_wheel",
"python_version": "py3",
"requires_python": ">=3.8",
"size": 297453,
"upload_time": "2024-09-16T01:36:33",
"upload_time_iso_8601": "2024-09-16T01:36:33.993902Z",
"url": "https://files.pythonhosted.org/packages/5b/ad/e27e971a75deaaac7817e022174f88fae4db203216ac59c1ace99f3e56eb/spandrel-0.4.0-py3-none-any.whl",
"yanked": false,
"yanked_reason": null
},
{
"comment_text": "",
"digests": {
"blake2b_256": "bb7e2f3a41c69651cf90b636857147568eb1806f56af5b8015ac7638a930abe1",
"md5": "7cb9fbd0fcf367e331f395ef460604ab",
"sha256": "f45526893f923a12ef37542c44e444b1208976593bc7ccdfa54fd04c7c3e80ca"
},
"downloads": -1,
"filename": "spandrel-0.4.0.tar.gz",
"has_sig": false,
"md5_digest": "7cb9fbd0fcf367e331f395ef460604ab",
"packagetype": "sdist",
"python_version": "source",
"requires_python": ">=3.8",
"size": 228294,
"upload_time": "2024-09-16T01:36:35",
"upload_time_iso_8601": "2024-09-16T01:36:35.784716Z",
"url": "https://files.pythonhosted.org/packages/bb/7e/2f3a41c69651cf90b636857147568eb1806f56af5b8015ac7638a930abe1/spandrel-0.4.0.tar.gz",
"yanked": false,
"yanked_reason": null
}
],
"upload_time": "2024-09-16 01:36:35",
"github": false,
"gitlab": false,
"bitbucket": false,
"codeberg": false,
"lcname": "spandrel"
}
JSON