WhiteAiZ/Illustrious-xl-v1.0

Full fine-tune vs OnomaAIResearch/Illustrious-xl-early-release-v0. Fine-tune: 94% of the weights changed, across 11 of 12 layers.

Tensors changed: Embeddings 453/589, other 505/662, Attention 700/716, Norms 268/546

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

On the Hub · Downloads on Model Pulse