This model requires you to accept the Apache 2.0 license terms before download.
Description
Open text-to-video generation, MMDiT architecture, 11B parameters.
Tags
At a glance
| Kind | video generation model |
|---|---|
| Precision | bfloat16 |
| Container size | 42.4 GiB |
| Weights inside | 42.4 GiB |
| Components | 4 |
| Licence | Apache 2.0 |
What is inside the container
A .nbx container packs every part of Open-Sora-v2 — 4 components with weights, graphs and runtime settings — so one file is all the runtime needs.
| Component | Blocks | Layers | Weights |
|---|---|---|---|
| transformer | 38 | — | 22.1 GiB |
| vae | 4 | — | 470 MiB |
| text_encoder | 24 | 24 | 18.2 GiB |
| text_encoder_2 | 24 | — | 1.6 GiB |
Default settings
The values the container ships with. They apply when you do not pass your own.
- Frames
- 51
- Frames per second
- 24
- Inference steps
- 100
- Guidance scale
- 7
Memory needed
The weights of Open-Sora-v2 take 42.4 GiB in bfloat16. 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 42.4 GiB (45,564,545,454 bytes).
$ sha256sum Open-Sora-v2.nbx
ad9387eaee34d4d56f0b23c3efcf8da456f1766d982adf7cbb328fb085a347c1
Questions
What is Open-Sora-v2?
A video generation model by hpcai-tech, distributed as a bfloat16 .nbx container and licensed Apache 2.0.
How big is the download?
42.4 GiB, of which 42.4 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 hpcai-tech/Open-Sora-v2, then neurobrix serve --model Open-Sora-v2.
CLI Usage
$ neurobrix import hpcai-tech/Open-Sora-v2
$ neurobrix serve --model Open-Sora-v2
$ neurobrix run --prompt "your prompt here"
$ neurobrix stop