8000 [Inference]fix deepseek_v3 with mtp in multi-gpu mode by freeliuzc · Pull Request #9894 · PaddlePaddle/PaddleNLP · GitHub
[go: up one dir, main page]
More Web Proxy on the site http://driver.im/
Skip to content

[Inference]fix deepseek_v3 with mtp in multi-gpu mode #9894

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.

Sign up for GitHub

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

Jump to bottom
Merged
merged 2 commits into from
Feb 19, 2025
Merged

[Inference]fix deepseek_v3 with mtp in multi-gpu mode #9894

merged 2 commits into from
Feb 19, 2025

Conversation

freeliuzc
Copy link
Collaborator

Before submitting

  • Lint code. If there are lint issues, please format the code first.
# Install and register `pre-commit` in the project folder
pip install pre-commit && pre-commit install

# Process previous code files separately
pre-commit run --file XXXX.py
  • Add test cases into tests folder. If there are codecov issues, please add tests cases first.

PR types

Bug fixes

PR changes

APIs

Description

  1. support multi-gpu mode in deepseekv3 with mtp.
  2. fix mtp rope bug

@paddle-bot Paddle-bot
Copy link
paddle-bot bot commented Feb 18, 2025

Thanks for your contribution!

yuanlehome
yuanlehome reviewed Feb 18, 2025
View reviewed changes
paddlenlp/experimental/transformers/proposers.py Outdated
@@ -108,6 +108,7 @@ def run(self, model_inputs: dict[str, paddle.Tensor], **kargs):
seq_lens_this_time,
seq_lens_encoder,
seq_lens_decoder,
model_inputs["max_dec_len"].cpu(),
Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

"max_length"

@codecov Codecov
Copy link
codecov bot commented Feb 18, 2025 •
edited
Loading

Codecov Report

Attention: Patch coverage is 0% with 8 lines in your changes missing coverage. Please review.

Project coverage is 51.41%. Comparing base (8e4ff07) to head (ace81fb).
Report is 332 commits behind head on develop.

Files with missing lines Patch % Lines
...enlp/experimental/transformers/generation_utils.py 0.00% 8 Missing ⚠️
Additional details and impacted files 
@@             Coverage Diff             @@
##           develop    #9894      +/-   ##
===========================================
- Coverage    51.41%   51.41%   -0.01%     
===========================================
  Files          745      745              
  Lines       118351   118356       +5     
===========================================
- Hits         60856    60854       -2     
- Misses       57495    57502       +7

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:
  • ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

@ZHUI ZHUI merged commit 1168406 into PaddlePaddle:develop Feb 19, 2025
10 of 12 checks passed
@freeliuzc freeliuzc deleted the support_mtp_fix_bug branch February 24, 2025 12:03
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Reviewers

@yuanlehome yuanlehome 50D4 yuanlehome 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.
3 participants
@freeliuzc @yuanlehome @ZHUI
0