Skip to content
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

[Feature] FP8 communication in ShardFormer #5899

Merged
merged 3 commits into from
Jul 18, 2024

Conversation

GuangyaoZhang
Copy link
Contributor

@GuangyaoZhang GuangyaoZhang commented Jul 10, 2024

📌 Checklist before creating the PR

  • I have created an issue for this PR for traceability
  • The title follows the standard format: [doc/gemini/tensor/...]: A concise description
  • I have added relevant tags if possible for us to better distinguish different PRs
  • I have installed pre-commit: pip install pre-commit && pre-commit install

🚨 Issue number

resolve #5916

📝 What does this PR do?

Accuracy on GPT2 finetune example

TP/SP Acc(FP16) Acc(FP8)
TP only 0.689 0.728
Split gather 0.728 0.729

It is wierd the accuracy is improved. Possibly the introduced noise by FP8 serves as a good Regularization item.

💥 Checklist before requesting a review

  • I have linked my PR to an issue (instruction)
  • My issue clearly describes the problem/feature/proposal, with diagrams/charts/table/code if possible
  • I have performed a self-review of my code
  • I have added thorough tests.
  • I have added docstrings for all the functions/methods I implemented

⭐️ Do you enjoy contributing to Colossal-AI?

  • 🌝 Yes, I do.
  • 🌚 No, I don't.

Tell us more if you don't enjoy contributing to Colossal-AI.

@GuangyaoZhang GuangyaoZhang changed the title [Draft] All to All sequence parallelism FP8 communication [Draft] FP8 communication in ShardFormer Jul 12, 2024
@GuangyaoZhang GuangyaoZhang changed the base branch from main to feature/fp8_comm July 16, 2024 07:12
@GuangyaoZhang GuangyaoZhang force-pushed the SP_fp8 branch 2 times, most recently from 537a3e8 to 8f7b9bb Compare July 17, 2024 05:45
@GuangyaoZhang GuangyaoZhang marked this pull request as ready for review July 17, 2024 05:46
@GuangyaoZhang GuangyaoZhang requested a review from a team as a code owner July 17, 2024 05:46
@GuangyaoZhang GuangyaoZhang force-pushed the SP_fp8 branch 3 times, most recently from 027fe98 to db09a5a Compare July 17, 2024 06:22
@GuangyaoZhang GuangyaoZhang changed the title [Draft] FP8 communication in ShardFormer FP8 communication in ShardFormer Jul 17, 2024
@GuangyaoZhang GuangyaoZhang changed the title FP8 communication in ShardFormer [Feature] FP8 communication in ShardFormer Jul 17, 2024
@GuangyaoZhang GuangyaoZhang requested a review from BurkeHulk July 17, 2024 07:31
@GuangyaoZhang GuangyaoZhang merged commit d0bdb51 into hpcaitech:feature/fp8_comm Jul 18, 2024
3 checks passed
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