QuantStack/Wan2.1_T2V_14B_FusionX_VACE

Full fine-tune vs Wan-AI/Wan2.1-VACE-14B. Fine-tune: 90% of the weights changed, across 40 of 40 layers.

Tensors changed: Attention 640/768, Norms 240/288, MLP 160/192, other 4/71, Embeddings 10/12

1 derivatives checked

• QuantStack/Wan2.1_T2V_14B_FusionX_VACE-GGUF: Same weights, converted

On the Hub · Downloads on Model Pulse