gigant/whisper-medium-romanian
Full fine-tune vs openai/whisper-medium. Fine-tune: 100% of the weights changed, across 24 of 24 layers.
Tensors changed: Embeddings 2/3, Norms 244/244, Attention 504/504, MLP 192/192, other 4/4
Full fine-tune vs openai/whisper-medium. Fine-tune: 100% of the weights changed, across 24 of 24 layers.
Tensors changed: Embeddings 2/3, Norms 244/244, Attention 504/504, MLP 192/192, other 4/4