ORTModule memory improvement #18924
Merged
ORTModule memory improvement #18924
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
pengwa
added
the
training
…pengwa/mem_improvement
…pengwa/mem_improvement
snnn
previously approved these changes
Jan 1, 2024
askhade
reviewed
Jan 9, 2024
orttraining/orttraining/python/training/ortmodule/_inference_manager.py
Outdated
Show resolved
Hide resolved
…pengwa/mem_improvement
askhade
approved these changes
Jan 15, 2024
snnn
approved these changes
Jan 15, 2024
justinchuby
reviewed
Jan 16, 2024
Comment on lines
+44
to
+49
| Args: | ||
| exported_model (ModelProto): The exported model. | ||
| named_params (Optional[Dict[str, torch.nn.parameter.Parameter]]): The full parameter map. | ||
|
|
||
| Returns: | ||
| tuple[bool, ModelProto]: A tuple of bool and ModelProto. The bool indicates whether the model is modified. |
There was a problem hiding this comment.
For future reference: no need to include type information in docstrings as they are in the function signature already.
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Outdated suggestions cannot be applied.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.
Dependency
#19007
ORTModule memory efficient gradient management
Previously I have tried to solve the coarsed-grained gradient accumulation/update problem in ORTModule with #8979, while that resolution somehow is not fully validated with DDP or there is user hooks on the gradient accumulation on torch parameter.
This PR is addressing the problem in the similar approach as PR 8979, e.g. trigger gradient accumulation once ORT computed the grad, but instead of use a AccumulateGrad op, this time with a ONNX operator PythonOp, internally it will call param.backward(grad), which will help handle all related hooks correctly.
Design
Check the details from
https://microsoftapc-my.sharepoint.com/:p:/g/personal/pengwa_microsoft_com/EaaBq4EzsFhOmsDEXCG7Ba4Bb9bwd0O2sFV_JXJ4jBLYLA?e=7Sz2g8&nav=eyJzSWQiOjI3MSwiY0lkIjozMjE4NzI1NDIzfQ
Convergence Validation:
differences are on mostly 0.000x, sometimes 0.00x, which may comes from the different order gradient apply happens before or after this change (on deepspeed zero stage 2)
TODO
Consolidate the logic with Stage3's similar logic.