Delete README.md

README.md

@@ -1,88 +0,0 @@
----
-license: other
-license_name: research-use
-license_link: LICENSE
-library_name: pytorch
-pipeline_tag: image-to-image
-tags:
-  - super-resolution
-  - image-super-resolution
-  - image-to-image
-  - pytorch
-  - onnx
-datasets:
-  - eugenesiow/Div2k
-model-index:
-  - name: srnet-x4
-    results:
-      - task:
-          type: image-to-image
-          name: Image super-resolution (x4)
-        dataset:
-          name: DIV2K validation (100 images, bicubic degradation)
-          type: eugenesiow/Div2k
-        metrics:
-          - type: psnr
-            name: PSNR (Y channel, shave 4)
-            value: 30.28
----
-
-# SRNet x4 — small, fast single-image super-resolution
-
-A compact residual CNN (~1.25 M parameters) that upscales images 4x. It predicts a residual on top of a bicubic upscale, so it starts from — and improves on — plain bicubic interpolation. Trained in 30 minutes on 2x T4 GPUs.
-
-## Results
-
-Evaluated on the 100 DIV2K validation images, full images, bicubic-x4 degradation, Y-channel PSNR with a 4 px border shave, 8-bit outputs:
-
-| Method | PSNR (dB) |
-|---|---|
-| Bicubic | 28.23 |
-| **SRNet x4** | **30.28** (+2.05) |
-
-Metrics are from this repo's own training notebook (antialiased bicubic downscale, not the official MATLAB `imresize`), so treat them as comparable between methods here but only approximately comparable to published tables.
-
-## Usage
-
-```python
-import numpy as np, torch
-from PIL import Image
-from huggingface_hub import hf_hub_download
-import importlib.util
-
-spec = importlib.util.spec_from_file_location("srnet", hf_hub_download("your-username/srnet-x4", "srnet.py"))
-srnet = importlib.util.module_from_spec(spec); spec.loader.exec_module(srnet)
-
-model = srnet.SRNet.from_pretrained("your-username/srnet-x4").eval()
-img = np.asarray(Image.open("photo.png").convert("RGB"), np.float32) / 255
-x = torch.from_numpy(img).permute(2, 0, 1)[None]
-with torch.inference_mode():
-    y = model(x).clamp(0, 1)[0].permute(1, 2, 0).numpy()
-Image.fromarray((y * 255).round().astype("uint8")).save("photo_x4.png")
-```
-
-Large images: process in tiles with a margin of at least `2*nb + 6 = 38` input pixels to get output identical to whole-image inference (see the demo Space's `app.py`).
-
-**ONNX:** `srnet_x4.onnx` takes `input` (float32 NCHW RGB in [0, 1], dynamic batch/height/width) and returns `output` (clip to [0, 1]).
-
-## Model details
-
-- Architecture: 3x3 conv head -> 16 residual blocks (conv-ReLU-conv, 64 channels) -> 3x3 conv with long skip -> conv to 48 channels + PixelShuffle(4) -> added to a bicubic upscale of the input. The final conv is zero-initialised, so training starts at exact bicubic.
-- Compute: all convolutions run at input resolution; about 2.5 MFLOPs per input pixel.
-
-## Training
-
-- Data: DF2K (DIV2K + Flickr2K) HR training images; DIV2K validation images held out.
-- Degradation: antialiased bicubic x4 downscale, quantised to 8 bit. No noise, blur or compression.
-- 16,157 steps, batch 32, 64x64 LR patches (256x256 HR), random flips/rotations.
-- AdamW, peak LR 4e-4 with cosine decay, L1 loss, then MSE for the final 8% of the schedule; EMA (0.999) weights evaluated alongside raw weights and the better one kept; fp16 mixed precision.
-
-## Limitations
-
-- Trained only on clean bicubic-downscaled images. On real-world inputs with noise, blur or JPEG artefacts it will not denoise or deblock and may amplify artefacts.
-- Optimised for PSNR with L1/MSE losses, so fine texture (fur, feathers, foliage) comes out smoother than the original. It does not hallucinate detail the way GAN-based upscalers do.
-- x4 only.
-
-## License / data
-
-Set the license to suit you. The training data has its own terms (DIV2K is provided for research use; Flickr2K images carry their original Flickr licenses), which may restrict commercial use of the weights — check them before choosing a license and update the YAML header above.