Uh oh!
There was an error while loading. Please reload this page.
[docs] Models - #12248
Conversation
HuggingFaceDocBuilderDev
commented
Aug 27, 2025
The docs for this PR live here. All of your documentation changes will be reflected on that endpoint. The docs are available until 30 days after the last update. |
sayakpaul
left a comment
There was a problem hiding this comment.
Thanks! Left some comments, LMK if they are unclear.
| | `"cuda"` | places model or pipeline on CUDA device | | ||
| | `"balanced"` | evenly distributes model or pipeline on all GPUs | | ||
| | `"auto"` | distribute model from fastest device first to slowest | | ||
| | `"cuda"` | places pipeline on CUDA device | |
There was a problem hiding this comment.
"cuda" is just an example. If someone wants to do it for any other supported accelerator, I believe they pass it by their name 👀
| |`"cuda"`| places pipeline on CUDA device | | |
| |`"cuda"`| places pipeline on CUDA (or supported accelerator) device | |
Uh oh!
There was an error while loading. Please reload this page.
Uh oh!
There was an error while loading. Please reload this page.
Uh oh!
There was an error while loading. Please reload this page.
Uh oh!
There was an error while loading. Please reload this page.
Uh oh!
There was an error while loading. Please reload this page.
Splits off the
Modelssection fromLoad schedulers and modelsand creates a dedicated section for models to include device placement, torch dtype,AutoModelAPI, and saving as shards.