kekmodel/MPL-pytorch Issues
Typo
Updated 1Teacher Gradients
Updated 13MPL framework performance
UpdatedAccuracy with or without finetune
Updated 2Incorrect cross entropy?
UpdatedWhy do we expand labels?
UpdatedQuick Dumb Question
Closeddumb theoretical questions
Closed 2learning rate not match?
Closed 1Training set should shuffle?
Closed 2License
Closed 1CE(pseudo hard label, teacher_us)
Updated 1Models
Updated 1Encounter with error in DDP
Closed 4Errors
Updated 1CIFAR-10-4K Number of Steps
UpdatedQuestion about teacher loss
Closed 1About dot product approximation
Closed 1