Skip to content

[TorchAcc] update accelerate API and add llama3-70B #1400

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
merged 1 commit into from
Jul 16, 2024

Conversation

baoleai
Copy link
Collaborator

@baoleai baoleai commented Jul 16, 2024

PR type

  • Bug Fix
  • New Feature
  • Document Updates
  • More Models or Datasets Support

PR information

  1. Update torchacc accelerate API;
  2. fix accuracy calculation;
  3. add llama3-70b examples.

Experiment results

swift deepspeed zeros3: 2.029 samples/s
torchacc fsdp: 2.641 samples/s

@baoleai baoleai requested a review from tastelikefeet July 16, 2024 02:44
@baoleai baoleai merged commit 243e5e3 into modelscope:main Jul 16, 2024
2 checks passed
@baoleai baoleai deleted the features/llama3_70b branch July 16, 2024 03:34
hjh0119 pushed a commit to hjh0119/swift that referenced this pull request Jul 22, 2024
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

2 participants