Skip to content

stdxl optimizations #34

Merged
yeonsily merged 5 commits into
dev_ae_stdxl_ftfrom
stdxl_hpugraph
Feb 14, 2024
Merged

stdxl optimizations #34
yeonsily merged 5 commits into
dev_ae_stdxl_ftfrom
stdxl_hpugraph

Conversation

@libinta
Copy link
Copy Markdown
Collaborator

@libinta libinta commented Feb 13, 2024 •
edited
Loading

Following changes:

  1. Added hpu graph for training, and disable it due to OOM.
  2. changed resolution to 1024,
  3. changed dataloader_num_process to 8
  4. Changed the noise scheduler to DDPM for accuracy.

What does this PR do?

Fixes # (issue)

Before submitting

  • This PR fixes a typo or improves the docs (you can dismiss the other checks if that's the case).
  • Did you make sure to update the documentation with your changes?
  • Did you write any new necessary tests?

@libinta libinta changed the title [UCS-2658] Added hpu graph for training, changed resolution to 1024, … stdxl optimizations Feb 14, 2024
@yeonsily yeonsily merged commit 98a63ec into dev_ae_stdxl_ft Feb 14, 2024
astachowiczhabana added a commit that referenced this pull request Nov 28, 2024
@astachowiczhabana @chaojun-zhang
[SW_208086] implement fused sdpa for wav2vec2 (#18) (#34)
077167f 
* [SW_208086] [PT][Gaudi2][8x][Wave2Vec2-AC-HF][Torch.compile]][Perf] Perf drop -11%

 improve perf with FusedSDPA

* [SW-208086] Fix formatting issues + add logger

---------

Co-authored-by: Chaojun Zhang <chzhang@habana.ai>
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Reviewers

@yeonsily yeonsily yeonsily approved these changes

Assignees

No one assigned

Labels

None yet

Projects

None yet

Milestone

No milestone

Development

Successfully merging this pull request may close these issues.

2 participants

@libinta @yeonsily