Enable model and data sharding #96
Draft
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
- create partition manager object - make MAP compatible - migrate to Orbax checkpointing - refactor predictive
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Outdated suggestions cannot be applied.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.
To scale to very large model, we need to enable model parallelization and FSDP in Fortuna. We can do so by exploiting JAX sharding functionalities. In this PR, we plan to make training and fine-tuning methods in Fortuna working with shardings. The user will determine how many GPUs to allocate for each type of parallelization (data, FSD and model parallelization) and some simple model partitioning rules. Fortuna will do the rest.
Pull request type
Please check the type of change your PR introduces: