[Distributed] Add dp_gradient_sync_after_accumulate

[AndSonder]

PR types

Others

PR changes

Others

Description

添加 dp_gradient_sync_after_accumulate 参数

相关PR：

• [Auto Parallel] Move reduce to opt stage Paddle#62157
• [AutoParallel] Change switch name to gradient_sync_after_accumulate Paddle#62441

[paddle-bot]

Thanks for your contribution!

[codecov]

Codecov Report

Attention: Patch coverage is 0% with 3 lines in your changes are missing coverage. Please review.

Project coverage is 56.51%. Comparing base (e3cb5d2) to head (1168be9).
Report is 10 commits behind head on develop.

Files	Patch %	Lines
paddlenlp/trainer/training_args.py	0.00%	3 Missing ⚠️

Additional details and impacted files

@@             Coverage Diff             @@
##           develop    #8045      +/-   ##
===========================================
- Coverage    56.54%   56.51%   -0.04%     
===========================================
  Files          592      592              
  Lines        91067    91128      +61     
===========================================
+ Hits         51493    51497       +4     
- Misses       39574    39631      +57     

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

[From00]

LGTM

[ZHUI]

LGTM