What does this PR do ?

Patch release r2.5.0 with changes related to nv-one-logger which enable NeMo using OneLogger for job performance instrumentation.

Original PR !13437

Collection: [Note which collection this PR will affect]
Will keep updating

• llm
• speechlm2
• vlm
• core
• lightning

Changelog

• Add new CallbackGroup and Callback ABC in NeMo/nemo/lightning/pytorch/callbacks/callback_group.py and base_callback.py used by various collections.
• collections/llm
  • Add try finally logic for trainer in collections/llm/api.py to ensure the on_app_end() call is always called.
  • Add __init_subclass__ method to ensure the callbacks are hooked in collections/llm/fn/mixin.py.
  • Fix minor fail safe check in collections/llm/modelopt/speculative/model_transform.py.
  • Add data mocks for collections/llm/gpt/data/mock.py, collections/llm/t5/data/mock.py.
• collections/speechlm2
  • minor fix of regex in optim_setup.py.
• collections/vlm
  • add on_dataloader_init_start/end() instrumentation hook call in preloaded.py.
• collections/core
  • add one_model_init_start/end() on_dataloader_init_start/end(), on_load_checkpoint_start/end() instrumentation hook call in modelPT.py.
  • fix minor fail safe code value of optimizer_name in modelPT.py.
  • add finally to make sure with exception in _run_hydra still instrument on_app_end().
  • fix circular dependency issue in optim/lr_scheduler.py
  • fix circular dependency issue in optim/optimizers.py
• collections/lightning
  • add callback related classes.
  • import hook of OneLogger module for lightning boot up in collections/lightning/io/mixin.py
  • fix bug in model_checkpoint.py which result in pipeline failure in collections/lightning/pytorch/callbacks.
• tests
  • related unit tests are added.

Usage

Not changed.

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The Jenkins CI system has been replaced by GitHub Actions self-hosted runners.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 # (issue)