Add ESM2 Finetuning Benchmark Configuration #964
Merged
Conversation
This file contains hidden or 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
dorotat-nv
reviewed
Jul 4, 2025
dorotat-nv
reviewed
Jul 4, 2025
dorotat-nv
reviewed
Jul 4, 2025
dorotat-nv
reviewed
Jul 4, 2025
nvmvle
force-pushed
the
mvle/onboarding-esm2-finetuning
branch
2 times, most recently
from
9ca500f to
3ae0439
Compare
nvmvle
force-pushed
the
mvle/onboarding-esm2-finetuning
branch
from
a94c575 to
6a7c952
Compare
nvmvle
changed the title
Collaborator
Author
|
/ok to test |
@nvmvle, there was an error processing your request: See the following link for more information: https://docs.gha-runners.nvidia.com/cpr/e/1/ |
nvmvle
requested review from
cspades,
farhadrgh,
jomitchellnv,
jstjohn,
jwilber,
pstjohn,
sichu2023,
skothenhill-nv and
trvachov
as code owners
Collaborator
Author
|
/ok to test 6a7c952 |
pstjohn
reviewed
Jul 22, 2025
pstjohn
reviewed
Jul 22, 2025
jstjohn
reviewed
Jul 22, 2025
jstjohn
reviewed
Jul 22, 2025
sub-packages/bionemo-esm2/src/bionemo/esm2/scripts/finetune_esm2.py
Outdated
Show resolved
Hide resolved
jstjohn
reviewed
Jul 22, 2025
jwilber
reviewed
Jul 22, 2025
nvmvle
force-pushed
the
mvle/onboarding-esm2-finetuning
branch
from
e5ce3b9 to
0a8ee44
Compare
jwilber
approved these changes
Jul 24, 2025
…benchmarks to enhance monitoring capabilities during training.
…sure proper event file existence and validity, improving error reporting during training monitoring.
… enable or disable checkpoint creation via a new parameter. Update assertions to verify checkpoint behavior based on this parameter, enhancing test coverage and flexibility.
…ontrol. Increase time limit to 14400 seconds, adjust GPU and node settings, and introduce early stopping functionality via a new argument. Modify related test cases to validate early stopping behavior and ensure proper argument parsing.
…0 for extended training duration.
…nce benchmarks. Remove unused parameters, streamline script arguments, and ensure consistency in training settings across both configurations. Introduce new parameters for better control over training processes.
…ttings for improved performance. Adjust training parameters for consistency and enhance experiment naming convention to include tensor parallelism and pipeline parallelism details.
…pt. Update YAML files to include new argument for tflops calculation, and modify training logic to support this feature. Enhance tests to validate argument parsing for the new callback.
…micolon from early stop argument
…estoration argument and updating YAML files to include this parameter. Adjust time limit for performance benchmarks and modify test cases to ensure proper argument parsing for checkpoint restoration.
nvmvle
force-pushed
the
mvle/onboarding-esm2-finetuning
branch
from
18dd98a to
c05f863
Compare
…ecute based on checkpoint availability. Ensure proper validation of prediction outputs for classification and regression tasks. Update path handling for restored checkpoints.
yzhang123
reviewed
Jul 25, 2025
| workspace: /workspace/bionemo2 | ||
| data_base_path: /data/FLIP | ||
| restore_from_checkpoint_path: /data/esm2_650M_nemo2 | ||
| nodes: [4] |
Collaborator
There was a problem hiding this comment.
why do you need so many nodes?
yzhang123
reviewed
Jul 25, 2025
yzhang123
reviewed
Jul 25, 2025
yzhang123
reviewed
Jul 25, 2025
Collaborator
|
/ok to test 4a32bc1 |
trvachov
approved these changes
Jul 29, 2025
Codecov Report❌ Patch coverage is
Additional details and impacted files@@ Coverage Diff @@
## main #964 +/- ##
==========================================
- Coverage 83.32% 83.30% -0.03%
==========================================
Files 148 148
Lines 9746 9758 +12
==========================================
+ Hits 8121 8129 +8
- Misses 1625 1629 +4
|
Collaborator
|
Bypassing rules to merge to unblock My -- we just released so low risk of screwing up any timelines. |
Collaborator
|
/ok to test ef89eb5 |
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
This PR adds comprehensive benchmark configurations for ESM2 finetuning to support performance testing and validation. The changes introduce two new benchmark configurations (partial-conv and perf) along with enhanced finetuning capabilities including checkpointing control, TensorBoard logging, and TFLOPS measurement callbacks.
Key enhancements include:
--disable-checkpointingoption for faster benchmark runsType of changes
CI Pipeline Configuration
Configure CI behavior by applying the relevant labels:
Note
By default, the notebooks validation tests are skipped unless explicitly enabled.
Authorizing CI Runs
We use copy-pr-bot to manage authorization of CI
runs on NVIDIA's compute resources.
automatically be copied to a pull-request/ prefixed branch in the source repository (e.g. pull-request/123)
/ok to testcomment on the pull request to trigger CI. This will need to be done for each new commit.Pre-submit Checklist
Signed-off-by: My Le mvle@nvidia.com