This model requires you to accept the OpenRAIL++ license terms before download.
Description
PixArt-Σ XL — diffusion-transformer text-to-image model that uses a weak-to-strong training scheme to generate high-fidelity 1024px images (scalable toward 4K) with strong prompt alignment and efficient compute.
Tags
At a glance
| Kind | image generation model |
|---|---|
| Precision | float32 |
| Container size | 20.3 GiB |
| Weights inside | 20.3 GiB |
| Components | 3 |
| Licence | OpenRAIL++ |
| Origin | PixArt-alpha/PixArt-Sigma-XL-2-1024-MS @ e102b3591cc8 |
What is inside the container
A .nbx container packs every part of PixArt-Sigma-XL-2-1024-MS — 3 components with weights, graphs and runtime settings — so one file is all the runtime needs.
| Component | Blocks | Layers | Weights |
|---|---|---|---|
| transformer | 28 | 28 | 2.3 GiB |
| vae | 4 | — | 319 MiB |
| text_encoder | 24 | 24 | 17.7 GiB |
Default settings
The values the container ships with. They apply when you do not pass your own.
- Height (px)
- 1024
- Width (px)
- 1024
- Inference steps
- 20
- Guidance scale
- 4.5
Memory needed
The weights of PixArt-Sigma-XL-2-1024-MS take 20.3 GiB 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 20.3 GiB (21,828,335,504 bytes).
$ sha256sum PixArt-Sigma-XL-2-1024-MS.nbx
409590ec6d3b72f59944195a54649996f62c08d8071f0246467e86eaac91ee3d
Questions
What is PixArt-Sigma-XL-2-1024-MS?
A image generation model by PixArt-alpha, distributed as a float32 .nbx container and licensed OpenRAIL++.
How big is the download?
20.3 GiB, of which 20.3 GiB are weights.
How do I run it?
Install the runtime from the download page, then use the commands under “CLI Usage” below: neurobrix import PixArt-alpha/PixArt-Sigma-XL-2-1024-MS, then neurobrix serve --model PixArt-Sigma-XL-2-1024-MS.
CLI Usage
$ neurobrix import PixArt-alpha/PixArt-Sigma-XL-2-1024-MS
$ neurobrix serve --model PixArt-Sigma-XL-2-1024-MS
$ neurobrix run --prompt "your prompt here"
$ neurobrix stop