[CPU] FusedAdam and CPU training support #3991
Merged
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
delock
requested review from
jeffra,
tjruwase,
samyam,
mrwyattii,
RezaYazdaniAminabadi,
cmikeh2,
awan-10 and
arashb
as code owners
tjruwase
reviewed
Jul 19, 2023
tjruwase
reviewed
Jul 19, 2023
tjruwase
reviewed
Jul 19, 2023
|
@tjruwase note we also have a PR in DeepSpeedExamples to demostrate CIFAR10 training with CPU. The PR also fix test broken for pytorch 2.0 because of API change. |
|
Also added CPUAdamBuilder for CPU accelerator, so HelloDeepSpeed which use cpu offload can run. |
tjruwase
approved these changes
Jul 25, 2023
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Outdated suggestions cannot be applied.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.
This PR implements FusedAdam and also enables CPU training with ZeRO stage 0-3. Note FP16 data type are not supported for CPU training. BF16 is the preferred datatype.
For ZeRO stage 0, this PR depends on #3842.
Whats in this PR:
This PR had been verified on CIFAR10 in DeepSpeedExamples, A PR had been submitted to support BF16 data type for CIFAR10 deepspeedai/DeepSpeedExamples#651 . Note for Zero0/1, there is an accuracy bug in DeepSpeed supporting BF16 training (#3979), for Zero2/3, accuracy is correct.