feat: added mps support for inference #81
Conversation
|
Hi @Skylerwiernik! Thank you for your pull request and welcome to our community. Action RequiredIn order to merge any pull request (code, docs, etc.), we require contributors to sign our Contributor License Agreement, and we don't seem to have one on file for you. ProcessIn order for us to review and merge your suggested changes, please sign at https://code.facebook.com/cla. If you are contributing on behalf of someone else (eg your employer), the individual CLA may not be sufficient and your employer may need to sign the corporate CLA. Once the CLA is signed, our tooling will perform checks and validations. Afterwards, the pull request will be tagged with If you have received this in error or have any questions, please contact us at [email protected]. Thanks! |
meta-cla
bot
added
the
CLA Signed
|
Thank you for signing our Contributor License Agreement. We can now accept your code for this (and any) Meta Open Source project. Thanks! |
|
I haven't reviewed yet, but I would like to comment that this would be a nice feature to have if we can review it carefully and make sure it doesn't break existing code. It's a small change that would allow people running the code on more hardware platforms. CC @russellhowes |
mmuckley
left a comment
mmuckley
left a comment
There was a problem hiding this comment.
After reading through this, I think it is mostly good, except there is one function whose default signature is changed. Can you update it so that it has something like device="cuda:0" in the function signature?
notebooks/vjepa2_demo.py
Outdated
|
|
||
|
|
||
| def forward_vjepa_video(model_hf, model_pt, hf_transform, pt_transform): | ||
| def forward_vjepa_video(model_hf, model_pt, hf_transform, pt_transform, device): |
There was a problem hiding this comment.
This changes the default function signature. Can you modify it so that the default device is cuda, as was the case with the original function signature?
mmuckley
left a comment
mmuckley
left a comment
There was a problem hiding this comment.
Hi @Skylerwiernik, this looks pretty good but I noticed one case where cuda still isn't the default. I also think the notebook diff looks too big - is there any way to just change the lines relevant to mps?
| if torch.backends.mps.is_available(): | ||
| device = "mps" | ||
| else: | ||
| device = "cuda:0" |
There was a problem hiding this comment.
Default should be cuda here, not "mps".
| "cells": [ | ||
| { | ||
| "cell_type": "markdown", | ||
| "metadata": {}, |
There was a problem hiding this comment.
The noteobook diff is too big - any way to just change the mps lines?
| def forward_vjepa_video(model_hf, model_pt, hf_transform, pt_transform, device="cuda"): | ||
| # Run a sample inference with VJEPA | ||
| with torch.inference_mode(): | ||
| # Read and pre-process the image | ||
| video = get_video() # T x H x W x C | ||
| video = torch.from_numpy(video).permute(0, 3, 1, 2) # T x C x H x W | ||
| x_pt = pt_transform(video).cuda().unsqueeze(0) | ||
| x_hf = hf_transform(video, return_tensors="pt")["pixel_values_videos"].to("cuda") | ||
| x_pt = pt_transform(video).to(device).unsqueeze(0) | ||
| x_hf = hf_transform(video, return_tensors="pt")["pixel_values_videos"].to(device) |
2 participants
Fixes #80