Why the train file smaller than test file?

facebookresearch / EmpatheticDialogues

Dialogue model that produces empathetic responses when trained on the EmpatheticDialogues dataset.

Other

444 stars 63 forks source link

Why the train file smaller than test file? #50

Open huangfu170 opened 1 year ago

huangfu170 commented 1 year ago

I have read the ED article, the train/valid/test conversations is 19533 / 2770 / 2547. But the train file is 16.9M. the valid, test file is about 36M. I want to know what make the contradiction between the conversations number and the file size. Any response will help me a lot.