GPT support for BF16 grad reductions #5920
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
|
|
timmoon10
force-pushed
the
dist-adam-bf16-grad-sync
branch
2 times, most recently
from
bd9176d to
ee8500b
Compare
timmoon10
force-pushed
the
dist-adam-bf16-grad-sync
branch
from
24ef1f6 to
17cc4ae
Compare
Signed-off-by: Tim Moon <[email protected]>
timmoon10
force-pushed
the
dist-adam-bf16-grad-sync
branch
from
5476cf1 to
59429ce
Compare
Signed-off-by: Tim Moon <[email protected]>
timmoon10
force-pushed
the
dist-adam-bf16-grad-sync
branch
from
44bcd5f to
68d7441
Compare
![github-advanced-security[bot]](https://avatars.githubusercontent.com/in/57789?s=60&v=4){kind=small}
Signed-off-by: Tim Moon <[email protected]>
timmoon10
force-pushed
the
dist-adam-bf16-grad-sync
branch
from
5b9b09d to
b80a1f4
Compare
Signed-off-by: Tim Moon <[email protected]>
|
This PR is stale because it has been open for 14 days with no activity. Remove stale label or comment or update or this will be closed in 7 days. |
mikolajblaz
reviewed
Mar 23, 2023
|
|
||
| # Compute norm of local gradients for explicit FP32 optimizer | ||
| if self._fp32_optim is not None: | ||
| _fp32_optim_grad_sync() |
There was a problem hiding this comment.
self._fp32_optim_grad_sync()?
Comment on lines
+75
to
+76
| if getattr(param, '_with_fp32_optimizer', False): | ||
| main_param = param.detach().clone().float() |
There was a problem hiding this comment.
param -> model_param?
titu1994
pushed a commit
to titu1994/NeMo
that referenced
this pull request
Mar 24, 2023
* Add support for BF16 grad reductions with distopt Signed-off-by: Tim Moon <[email protected]> * Fix style issues Signed-off-by: Tim Moon <[email protected]> * Fix style issues Signed-off-by: Tim Moon <[email protected]> * Update Apex commit Signed-off-by: Tim Moon <[email protected]> --------- Signed-off-by: Tim Moon <[email protected]>
This was referenced Mar 27, 2023
timmoon10
added a commit
to timmoon10/NeMo
that referenced
this pull request
Mar 31, 2023
* Add support for BF16 grad reductions with distopt Signed-off-by: Tim Moon <[email protected]> * Fix style issues Signed-off-by: Tim Moon <[email protected]> * Fix style issues Signed-off-by: Tim Moon <[email protected]> * Update Apex commit Signed-off-by: Tim Moon <[email protected]> --------- Signed-off-by: Tim Moon <[email protected]>
8 tasks
ericharper
pushed a commit
that referenced
this pull request
Apr 3, 2023
* GPT support for BF16 grad reductions (#5920) * Add support for BF16 grad reductions with distopt Signed-off-by: Tim Moon <[email protected]> * Fix style issues Signed-off-by: Tim Moon <[email protected]> * Fix style issues Signed-off-by: Tim Moon <[email protected]> * Update Apex commit Signed-off-by: Tim Moon <[email protected]> --------- Signed-off-by: Tim Moon <[email protected]> * Add custom functions to launch distopt communication in interleaved pipeline parallelism (#6183) Signed-off-by: Tim Moon <[email protected]> * Bugfix for BF16 grad reductions with distopt (#6340) * Debug distopt support for BF16 grad reductions Signed-off-by: Tim Moon <[email protected]> * Dump and load FP32 main params Signed-off-by: Tim Moon <[email protected]> * Style tweaks Signed-off-by: Tim Moon <[email protected]> --------- Signed-off-by: Tim Moon <[email protected]> Co-authored-by: Mikołaj Błaż <[email protected]> --------- Signed-off-by: Tim Moon <[email protected]> Co-authored-by: Mikołaj Błaż <[email protected]>
hsiehjackson
pushed a commit
to hsiehjackson/NeMo
that referenced
this pull request
Jun 2, 2023
* Add support for BF16 grad reductions with distopt Signed-off-by: Tim Moon <[email protected]> * Fix style issues Signed-off-by: Tim Moon <[email protected]> * Fix style issues Signed-off-by: Tim Moon <[email protected]> * Update Apex commit Signed-off-by: Tim Moon <[email protected]> --------- Signed-off-by: Tim Moon <[email protected]> Signed-off-by: hsiehjackson <[email protected]>
8 tasks
8 tasks
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 ?
Adds GPT support for BF16/FP16 gradient reductions, with embedding grad reductions in FP32.
Collection: NLP
Changelog
Usage
Set the optimizer to
distributed_fused_adamin the config file:NeMo/examples/nlp/language_modeling/conf/megatron_gpt_config.yaml
Line 193 in 65c277b
Configure the optimizer with
grad_sync_dtype: bf16.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