Training l3 model with new audio video dataset

[YNawal]

Hello everyone,

I'm trying to training L3 Net with a new dataset. I have a folder with 200K videos. the train.py script in the l3embeddig folder has no help messages about the format of data in th trai_data_dir/valid_data_dir.
What I constated that the data generator in train.py reads f5 files.
Can anyon help hot to fed my videos data to the train script.
Thanks.

@auroracramer