This model requires you to accept the Apache 2.0 license terms before download.
Description
HAT-S x4 - XPixel Group Hybrid Attention Transformer (Small) super-resolution model that activates more input pixels through combined channel and window attention for 4x image upscaling. Works at any input size. Apache 2.0.
Tags
At a glance
| Kind | image upscaler |
|---|---|
| Precision | float32 |
| Container size | 34.8 MiB |
| Weights inside | 38.4 MiB |
| Components | 1 |
| Licence | Apache 2.0 |
| Origin | jaideepsingh/upscale_models @ unrecorded |
What is inside the container
A .nbx container packs every part of HAT-HAT-S_SRx4 — its single component with weights, graphs and runtime settings — so one file is all the runtime needs.
| Component | Blocks | Layers | Weights |
|---|---|---|---|
| model | 0 | — | 38.4 MiB |
Memory needed
The weights of HAT-HAT-S_SRx4 take 38.4 MiB in float32. That is the floor for the memory the runtime has to place, spread over the devices it finds; activations come on top and grow with the size and length you ask for. A measured peak per device class is not published for this model yet.
Verify what you downloaded
Every download carries its SHA-256, and the file endpoint sends it in the X-NBX-SHA256 header. The container is 34.8 MiB (36,498,074 bytes).
$ sha256sum HAT-HAT-S_SRx4.nbx
8e960f78fc800643f03d59d29181717523ff4634e1ef3809b950b29fb5f7870f
Questions
What is HAT-HAT-S_SRx4?
A image upscaler by XPixelGroup, distributed as a float32 .nbx container and licensed Apache 2.0.
How big is the download?
34.8 MiB, of which 38.4 MiB are weights.
How do I run it?
Install the runtime from the download page, then use the commands under “CLI Usage” below: neurobrix import XPixelGroup/HAT-HAT-S_SRx4, then neurobrix serve --model HAT-HAT-S_SRx4.
CLI Usage
$ neurobrix import XPixelGroup/HAT-HAT-S_SRx4
$ neurobrix serve --model HAT-HAT-S_SRx4
$ neurobrix run --prompt "your prompt here"
$ neurobrix stop