BALM_paired: embeddings generated from a 1k subsample of the paired test set do not show grouping by V gene
lsantuari opened this issue · comments
Hello,
I cannot reproduce the grouping by V gene for the heavy and the light chain when using a subsample of 1000 sequences from your paired test set and generating the embeddings with the BALM_paired model.
Here is my notebook.
Update: The grouping with VH embeddings is clear when considering V gene groups instead of subgroups.
I am closing the issue.