reazon-research/japanese-wav2vec2-large-rs35kh

Full fine-tune vs reazon-research/japanese-wav2vec2-large. Fine-tune: 99% of the weights changed, across 24 of 24 layers.

Tensors changed: Norms 100/102, Attention 192/192, MLP 96/96, Embeddings 4/4, other 2/9

1 other repos hold byte-for-byte identical weight files.

On the Hub · Downloads on Model Pulse