star092304/vi-sign-language-videomae-base
Full fine-tune vs MCG-NJU/videomae-base-finetuned-kinetics. Fine-tune: 100% of the weights changed, across 12 of 12 layers.
Tensors changed: Norms 50/50, Embeddings 2/2, Attention 84/84, other 48/48
Full fine-tune vs MCG-NJU/videomae-base-finetuned-kinetics. Fine-tune: 100% of the weights changed, across 12 of 12 layers.
Tensors changed: Norms 50/50, Embeddings 2/2, Attention 84/84, other 48/48