Disty0/Z-Image-Turbo-SDNQ-int8
Full fine-tune (fewer layers) vs Tongyi-MAI/Z-Image-Turbo. Tongyi-MAI/Z-Image-Turbo matches part of Disty0/Z-Image-Turbo-SDNQ-int8 (521 of its 790 tensors; the rest of Disty0/Z-Image-Turbo-SDNQ-int8 has no counterpart). On the shared part: Fine-tune: 100% of the weights changed, across 30 of 30 layers.
Tensors changed: other 31/70, Embeddings 0/9, Norms 0/204, Attention 136/136, MLP 102/102