Skip to content

feat: modernize JAX/TPU distribution with Keras 3 on Kaggle v5e-8 (Fixes #275)#403

Open
Paramveersingh-S wants to merge 4 commits into
google-gemma:mainfrom
Paramveersingh-S:feat/tpu-v5e-parallelism
Open

feat: modernize JAX/TPU distribution with Keras 3 on Kaggle v5e-8 (Fixes #275)#403
Paramveersingh-S wants to merge 4 commits into
google-gemma:mainfrom
Paramveersingh-S:feat/tpu-v5e-parallelism

Conversation

@Paramveersingh-S

Copy link
Copy Markdown

Description

This PR addresses Issue #275 by providing a modernized data parallelism guide that relies on the Keras 3 backend rather than the legacy Flax abstractions.

Key Changes

  • Created tutorials/Gemma_3_Keras_TPU_Parallelism.ipynb.
  • Demonstrates how to instantiate a keras.distribution.DataParallel mesh specifically for the Kaggle TPU v5e-8 environment.
  • Added the notebook to tutorials/README.md under the Inference Capabilities table.

Fixes #275.

@review-notebook-app

Copy link
Copy Markdown

Check out this pull request on  ReviewNB

See visual diffs & provide feedback on Jupyter Notebooks.


Powered by ReviewNB

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Projects

None yet

Development

Successfully merging this pull request may close these issues.

[Feature Request]: Modernize JAX/TPU Parallelism with Gemma 3 (Kaggle v5e-8 Support)

1 participant