Distributed optimizer reduces GPT embedding grads in FP32 #8792
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
Signed-off-by: Tim Moon <tmoon@nvidia.com>
|
jenkins |
Signed-off-by: Tim Moon <tmoon@nvidia.com>
|
jenkins |
|
jenkins |
|
jenkins |
timmoon10
changed the title
timmoon10
changed the title
erhoo82
approved these changes
Apr 5, 2024
This was referenced Apr 9, 2024
anmolgupt
pushed a commit
to anmolgupt/NeMo
that referenced
this pull request
Apr 11, 2024
* Make sure embedding grads are reduced in FP32 Signed-off-by: Tim Moon <tmoon@nvidia.com> * Access correct attr to get position embeddings Signed-off-by: Tim Moon <tmoon@nvidia.com> --------- Signed-off-by: Tim Moon <tmoon@nvidia.com>
8 tasks
8 tasks
suiyoubi
pushed a commit
that referenced
this pull request
May 2, 2024
* Make sure embedding grads are reduced in FP32 Signed-off-by: Tim Moon <tmoon@nvidia.com> * Access correct attr to get position embeddings Signed-off-by: Tim Moon <tmoon@nvidia.com> --------- Signed-off-by: Tim Moon <tmoon@nvidia.com> Signed-off-by: Ao Tang <aot@nvidia.com>
rohitrango
pushed a commit
to rohitrango/NeMo
that referenced
this pull request
Jun 25, 2024
* Make sure embedding grads are reduced in FP32 Signed-off-by: Tim Moon <tmoon@nvidia.com> * Access correct attr to get position embeddings Signed-off-by: Tim Moon <tmoon@nvidia.com> --------- Signed-off-by: Tim Moon <tmoon@nvidia.com>
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Outdated suggestions cannot be applied.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 ?
When training Megatron-core GPT with the distributed optimizer, the embedding gradients were reduced in the grad sync dtype (usually BF16). However, we want to reduce in FP32 to improve convergence.
Collection: NLP
Changelog
Usage
Run GPT, e.g. with the config at https://github.com/NVIDIA/NeMo/blob/main/examples/nlp/language_modeling/conf/megatron_gpt_config.yaml.
Enable mcore with
model.mcore_gpt=Trueand the distributed optimizer withmodel.optim.name=distributed_fused_adam.Jenkins CI
To run Jenkins, a NeMo User with write access must comment
jenkinson the PR.Before your PR is "Ready for review"
Pre checks:
PR Type:
If you haven't finished some of the above items you can still open "Draft" PR.
Who can review?
Anyone in the community is free to review the PR once the checks have passed.
Contributor guidelines contains specific people who can review PRs to various areas.
Additional Information