ValueError: Classification metrics can't handle a mix of multilabel-indicator and multiclass targets

Hey everyone, I am new to Huggingface and I’m trying to Fine tune roberta-base on go_emotions dataset for Multi Label classification.
My final train dataset is a dictionary where the labels are one hot encoded.

I am using the Trainer class of Huggingface to train my model and when I run the trainer class an error pops up saying

“Value error: Classification metrics can’t handle a mix of multilabel-indicator and multiclass targets”

I have tried decoding it by changing the shapes in the metrics function but nothing works.

Can someone please help me out with this?