ESM2 Finetuning refactor #574
Open
+2,052
−876
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: Farhad Ramezanghorbani <[email protected]>
Signed-off-by: Farhad Ramezanghorbani <[email protected]>
Signed-off-by: Farhad Ramezanghorbani <[email protected]>
|
Ported over the changes from #546 |
Signed-off-by: Farhad Ramezanghorbani <[email protected]>
Signed-off-by: Farhad Ramezanghorbani <[email protected]>
Signed-off-by: Farhad Ramezanghorbani <[email protected]>
Signed-off-by: Farhad Ramezanghorbani <[email protected]>
Signed-off-by: Farhad Ramezanghorbani <[email protected]>
Signed-off-by: Farhad Ramezanghorbani <[email protected]>
Signed-off-by: Farhad Ramezanghorbani <[email protected]>
farhadrgh
requested review from
jstjohn,
malcolmgreaves,
skothenhill-nv,
ohadmo,
pstjohn,
trvachov and
dorotat-nv
as code owners
Signed-off-by: Farhad Ramezanghorbani <[email protected]>
Signed-off-by: Farhad Ramezanghorbani <[email protected]>
Signed-off-by: Farhad Ramezanghorbani <[email protected]>
|
/build-ci |
sichu2023
reviewed
Jan 13, 2025
| @@ -180,12 +180,12 @@ def get_loss_reduction_class(self) -> Type[RegressorLossReduction]: | |||
| return RegressorLossReduction | |||
|
|
|||
|
|
|||
| class InMemorySingleValueDataset(Dataset): | |||
| class InMemorySingleValueDataset(InMemoryCSVDataset): | |||
There was a problem hiding this comment.
Should we move this into dataset.py or anywhere under data?
sichu2023
reviewed
Jan 13, 2025
| @@ -205,12 +190,12 @@ def get_loss_reduction_class(self) -> Type[ClassifierLossReduction]: | |||
| return ClassifierLossReduction | |||
|
|
|||
|
|
|||
| class InMemoryPerTokenValueDataset(Dataset): | |||
| class InMemoryPerTokenValueDataset(InMemoryCSVDataset): | |||
There was a problem hiding this comment.
Similarly should we move this under dataset.py? Like esm2/data/finetune/dataset.py or similar.
sichu2023
reviewed
Jan 13, 2025
Comment on lines
+611
to
+618
| dataset_class_options: Dict[str, Type[InMemoryCSVDataset]] = SUPPORTED_DATASETS | ||
|
|
||
| def dataset_class_type(desc: str) -> Type[InMemoryCSVDataset]: | ||
| try: | ||
| return dataset_class_options[desc] | ||
| except KeyError: | ||
| raise argparse.ArgumentTypeError( | ||
| f"Do not recognize key {desc}, valid options are: {dataset_class_options.keys()}" |
There was a problem hiding this comment.
@pstjohn had a similar approach by inheriting from both str and enum to streamline argument parsing.
https://github.com/NVIDIA/bionemo-framework/blob/main/sub-packages/bionemo-llm/src/bionemo/llm/model/biobert/transformer_specs.py#L47
https://github.com/NVIDIA/bionemo-framework/blob/main/sub-packages/bionemo-esm2/src/bionemo/esm2/scripts/train_esm2.py#L586
sichu2023
reviewed
Jan 13, 2025
| config_class: Type[BioBertConfig] = ESM2FineTuneSeqConfig, | ||
| metric_tracker: Callback | None = None, | ||
| overlap_grad_reduce: bool = True, | ||
| overlap_param_gather: bool = False, # TODO waiting for a NeMo fix |
There was a problem hiding this comment.
sichu2023
reviewed
Jan 13, 2025
| dataset_class=args.dataset_class, | ||
| config_class=args.config_class, | ||
| overlap_grad_reduce=not args.no_overlap_grad_reduce, | ||
| overlap_param_gather=args.overlap_param_gather, |
There was a problem hiding this comment.
sichu2023
reviewed
Jan 13, 2025
Comment on lines
+574
to
+578
| parser.add_argument( | ||
| "--overlap-param-gather", | ||
| action="store_true", | ||
| default=False, | ||
| ) # TODO waiting for a NeMo fix |
There was a problem hiding this comment.
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.
Description
Type of changes
CI Pipeline Configuration
Configure CI behavior by checking relevant boxes below. This will automatically apply labels.
Note
By default, the notebooks validation tests are skipped unless explicitly enabled.
Usage
Pre-submit Checklist