Skip to content

Conversation

@jlamypoirier
Copy link
Collaborator

@jlamypoirier jlamypoirier commented Sep 18, 2025

✨ Description

Version to 0.3.0 and associated cleanup:

  • Explicitly prevent loading checkpoint from earlier version, with appropriate error message.
  • Rename training_dtype -> compute_dtype
  • Remove unused flat config format
  • Remove most backward compatibility code, as we no longer support previous versions anyway
  • Remove legacy dataset configuration. Add back a minimalistic version in test_match_megatron for testing purposes.
  • Remove concatenated memmap dataset.
  • Postpone or remove remaining todos for v0.3

@jlamypoirier jlamypoirier marked this pull request as ready for review September 18, 2025 21:09
Base automatically changed from block_interface_convert to main September 18, 2025 21:17
Copy link
Collaborator

@tscholak tscholak left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

As discussed, LGTM!

@jlamypoirier jlamypoirier merged commit ca3b000 into main Sep 18, 2025
2 checks passed
@jlamypoirier jlamypoirier deleted the v0.3.0 branch September 18, 2025 21:43
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.

3 participants