Skip to content

Conversation

copybara-service[bot]
Copy link

@copybara-service copybara-service bot commented Aug 7, 2025

This CL introduces an end-to-end Flax implementation of the DLRM-HSTU model, along with individual unit tests for the various components.

  • Converted all necessary PyTorch modules to Flax/Linen
  • Full training loop test for end to end model
  • Added unit tests for individual components (e.g. ActionEncoder, ContentEncoder, STU modules)
  • Ensured model runs correctly on TPUs.
  • Verified shape-correctness of all modules and parameters.

Reverts changelist 793734230

… model, along with individual unit tests for the various components.

- Converted all necessary PyTorch modules to Flax/Linen
- Full training loop test for end to end model
- Added unit tests for individual components (e.g. `ActionEncoder`, `ContentEncoder`, `STU` modules)
- Ensured model runs correctly on TPUs.
- Verified shape-correctness of all modules and parameters.

Reverts changelist 793734230

PiperOrigin-RevId: 791875864
@copybara-service copybara-service bot changed the title Replace jax.sharding.use_mesh with jax.set_mesh. jax.set_mesh can act as a global setter or a context manager. This CL introduces an end-to-end Flax implementation of the DLRM-HSTU model, along with individual unit tests for the various components. Aug 25, 2025
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

1 participant