Repackage of PiD v1.5 from NVIDIA to safetensors files and int8 conversion done with Silveroxides/convert_to_quant, targeting only patch_blocks attn_(qkv/proj_y) / mlp (but .w2) / adaLN_modulation layers.
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support