Training on larger datasets

Hi, congratulations on the great work, this approach looks very promising!
Are there any plans to release checkpoints trained on larger datasets?

The issue names no files, tests, or training entry points. First clarify whether releasing larger-dataset checkpoints is planned and which datasets, training configuration, and artifacts are in scope. Done would require an agreed release plan and published checkpoints, rather than a narrowly defined code change.

Written by the indexing model from the issue text.