training_extensions icon indicating copy to clipboard operation
training_extensions copied to clipboard
verified publisher icon openvinotoolkit

Enforcing the loss type to be either CrossEntropy or IBLoss in classification task

Open JihwanEom opened this issue 2 years ago • 0 comments

Describe the bug image https://github.com/openvinotoolkit/training_extensions/blob/develop/otx/algorithms/classification/adapters/mmcls/configurer.py#L632-L647

Current loss patching logic is dynamically and forcefully changing the loss type to either CrossEntropy or IBLoss. This code prevents the loss type from being updated to LabelSmoothingLoss or FocalLoss, which are supported in the MM family.

Environment: Latest develop branch, commit hash: fb1ecb6ed328e0aae6839bb9760ba0949ddf068b

JihwanEom avatar May 16 '23 05:05 JihwanEom