Yntec/AnythingV4

Full fine-tune vs Yntec/AnythingV3. Fine-tune: 92% of the weights changed, across 12 of 12 layers.

Tensors changed: Embeddings 7/7, Norms 178/282, MLP 48/48, Attention 256/272, other 394/522

1 derivatives checked

On the Hub · Downloads on Model Pulse