search

DushyantaDhyani / kdtf

Knowledge Distillation using Tensorflow
MIT License
142 stars 47 forks source link

Will the teacher model be trained when training student model #1

Open mmpang opened 5 years ago

mmpang commented 5 years ago

When init from checkpoint, the variables in teacher model are still trainable, right? So they will be trained again in the process training the student model, I guess.

DushyantaDhyani commented 5 years ago

No they won't be retrained. The performance reported from these two checks https://github.com/DushyantaDhyani/kdtf/blob/master/main.py#L72 and https://github.com/DushyantaDhyani/kdtf/blob/master/main.py#L83 are exactly to determine that