log TFLOPs

[malay-nagda]

What does this PR do ?

Log TFLOPs per sec per GPU (for LLMs)

Collection: [common, nlp, Run CICD]

Changelog

• callbacks.append(FLOPsMeasurementCallback(self.cfg))

Usage

from pytorch_lightning import Trainer
from nemo.collections.common.metrics.perf_metrics import FLOPsMeasurementCallback

callbacks = []
flops_measurement_callback =  FLOPsMeasurementCallback(self.cfg)
callbacks.append(flops_measurement_callback)

trainer = Trainer(..., callbacks=callbacks)
...
trainer.fit(model)

GitHub Actions CI

The Jenkins CI system has been replaced by GitHub Actions self-hosted runners.

The GitHub Actions CI will run automatically when the "Run CICD" label is added to the PR.
To re-run CI remove and add the label again.
To run CI on an untrusted fork, a NeMo user with write access must first click "Approve and run".

Before your PR is "Ready for review"

Pre checks:

• Make sure you read and followed Contributor guidelines
• Did you write any new necessary tests?
• Did you add or update any necessary documentation?
• Does the PR affect components that are optional to install? (Ex: Numba, Pynini, Apex etc)
  • Reviewer: Does the PR have correct import guards for all optional libraries?

PR Type:

• New Feature
• Bugfix
• Documentation

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

• Related to Util for measuring MFU? #9673

[pablo-garay]

Was functionality tested? All is working well - as per the testing process & results?

[TJ-Solergibert]

Hi @malay-nagda , This callback just computes the TFLOPs on training end. Do you guys have in the roadmap logging other metrics such tokens per second during training instead of just on training end? I could work on that feature!