Skip to content

[Auto-Parallel] Support MoE on Auto-Parallel#657

Open
Xing-lil wants to merge 2 commits intoPaddlePaddle:developfrom
Xing-lil:auto_parallel_moe
Open

[Auto-Parallel] Support MoE on Auto-Parallel#657
Xing-lil wants to merge 2 commits intoPaddlePaddle:developfrom
Xing-lil:auto_parallel_moe

Conversation

@Xing-lil
Copy link
Copy Markdown

@Xing-lil Xing-lil commented Mar 20, 2026

Support MoE on Auto-Parallel

)
if self.down_proj.bias is not None:
self.down_proj.bias = dist.shard_tensor(
self.down_proj.bias, mesh, placements
Copy link
Copy Markdown

Choose a reason for hiding this comment

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

这里是否可能涉及还会其他层也做shard_tensor? 比如 Linear 层

Copy link
Copy Markdown
Author

Choose a reason for hiding this comment

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

对于MoE中为MLP,这里的shard_tensor是完备的,暂未支持MoE中为其他layer的情况。

@codecov-commenter
Copy link
Copy Markdown

Codecov Report

❌ Patch coverage is 15.83333% with 101 lines in your changes missing coverage. Please review.
⚠️ Please upload report for BASE (develop@0636f53). Learn more about missing BASE report.

Files with missing lines Patch % Lines
src/paddlefleet/transformer/moe/moe_layer.py 14.41% 93 Missing and 2 partials ⚠️
src/paddlefleet/transformer/mlp.py 25.00% 6 Missing ⚠️

❌ Your patch status has failed because the patch coverage (15.83%) is below the target coverage (90.00%). You can increase the patch coverage or adjust the target coverage.

Additional details and impacted files

Impacted file tree graph

@@            Coverage Diff             @@
##             develop     #657   +/-   ##
==========================================
  Coverage           ?   15.83%           
==========================================
  Files              ?        3           
  Lines              ?      120           
  Branches           ?       16           
==========================================
  Hits               ?       19           
  Misses             ?       99           
  Partials           ?        2           
Flag Coverage Δ
coverage_combine 15.83% <15.83%> (?)

Flags with carried forward coverage won't be shown. Click here to find out more.

Files with missing lines Coverage Δ
src/paddlefleet/fusions/fused_bias_swiglu.py 100.00% <100.00%> (ø)
src/paddlefleet/transformer/mlp.py 25.00% <25.00%> (ø)
src/paddlefleet/transformer/moe/moe_layer.py 14.41% <14.41%> (ø)
🚀 New features to boost your workflow:
  • ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

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.

3 participants