Metadata-Version: 2.4
Name: resselt
Version: 1.3.2
Summary: Image Super-Resolution PyTorch architectures loader
License-File: LICENSE.md
Requires-Python: >=3.10
Requires-Dist: einops>=0.8.0
Requires-Dist: numpy>=1
Requires-Dist: safetensors<0.6
Requires-Dist: torch>=2.6.0
Provides-Extra: cu118
Requires-Dist: torch>=2.6.0; extra == 'cu118'
Provides-Extra: cu124
Requires-Dist: torch>=2.6.0; extra == 'cu124'
Provides-Extra: cu126
Requires-Dist: torch>=2.6.0; extra == 'cu126'
Provides-Extra: cu128
Requires-Dist: torch>=2.7.0; extra == 'cu128'
Description-Content-Type: text/markdown

# resselt

## Usage

```py
import torch
from resselt import load_from_file, load_from_state_dict

# Load model from file
model = load_from_file("spanplus_2x.pth")

# or from state dict
state_dict = torch.load("spanplus_2x.pth")
model = load_from_state_dict(state_dict)

# and after you can get model metadata: ModelMetadata(name, in_channels, out_channels, upscale)
print(model.parameters_info)
```

## Credits

* Based on [chaiNNer-org/spandrel](https://github.com/chaiNNer-org/spandrel)
