add zero optimizer parallel #593
Merged
+971
−2
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
SamitHuang
reviewed
Jul 13, 2024
| print(f"rank_id: {rank_id}, group_size: {group_size}") | ||
| ms.reset_auto_parallel_context() | ||
| ms.set_auto_parallel_context( | ||
| parallel_mode=ms.ParallelMode.DATA_PARALLEL, |
There was a problem hiding this comment.
How is the difference between it ParallelMode.SEMI_AUTO_PARALLEL?
There was a problem hiding this comment.
Implemented a new version of zero optimizer parallel in DATA_PARALLEL refer to DeepSpeed. Not use the MindSpore automatic parallel process.
CaitinZhao
force-pushed
the
master
branch
8 times, most recently
from
0e46851 to
c0a8757
Compare
CaitinZhao
force-pushed
the
master
branch
2 times, most recently
from
2755198 to
dc2d467
Compare
CaitinZhao
force-pushed
the
master
branch
2 times, most recently
from
e4978fe to
69cce29
Compare
SamitHuang
approved these changes
Sep 3, 2024
zhtmike
reviewed
Sep 3, 2024
mindone/trainers/train_step.py
Outdated
| @@ -104,6 +115,10 @@ def construct(self, *inputs): | |||
|
|
|||
| # 1. compute gradients (of the up-scaled loss w.r.t. the model weights) | |||
| grads = self.grad(self.network, weights)(*inputs, scaling_sens_filled) | |||
|
|
|||
| # Gradient communication | |||
| grads = self.zero_helper.cal_gradients(grads) | |||
There was a problem hiding this comment.
self.zero_helper can be None, need to to add a if condition
CaitinZhao
force-pushed
the
master
branch
2 times, most recently
from
60c7aaa to
a19b806
Compare
added 2 commits
September 10, 2024 10:46
CaitinZhao
force-pushed
the
master
branch
2 times, most recently
from
d9b877e to
bb21538
Compare
vigo999
approved these changes
Sep 11, 2024
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
What does this PR do?
参考deepspeed的zero，在MindSpore的数据并行模式下实现优化器并行算法
Fixes # (issue)
Adds # (feature)
Before submitting
What's New. Here are thedocumentation guidelines
Who can review?
Anyone in the community is free to review the PR once the tests have passed. Feel free to tag
members/contributors who may be interested in your PR.
@xxx