This is AIO file (Diffusion Model+TextEncoder+VAE) for convenience. Enjoy PonyV7!Original Resources are below:<ul><li><a target="_blank" rel="ugc" href="https://civarchive.com/models/1901521/pony-v7-base">https://civarchive.com/models/1901521/pony-v7-base</a></li><li><a target="_blank" rel="ugc" href="https://huggingface.co/purplesmartai/pony-v7-base">https://huggingface.co/purplesmartai/pony-v7-base</a></li></ul>Notice:2025/1/3Scaled fp8 Diffusion Model + fp16 TextEncoder + fp16 VAE is now available, have fun!2025/1/2I finally found the origin of the problem. In fp8 quantization of PonyV7, we need "scaled fp8". Otherwise, it'll break the critical part of the model weight.2025/12/31Currently, fp16 AIO has no saturation problem. And, I'll try to recreate correct fp8 AIO.2025/12/31I found image saturation problem in fp8 AIO, so I'll upload fixed version for that.

PonyV7_fp8_fp16_All_In_One_Safetensors

Scaled fp8 Diffusion Model + fp16 TextEncoder + fp16 VAEScale Algorithm: AbsMaxExclude Patterns: bias, t_embedder, init_x_linear, final_linear, positional_encoding, register_tokens, cond_seq_linearThank you Anthropic. Claude gave me the conversion script what I wanted ;-)

Files with SHA256: ece931ae2fb0a1a8b60eb70fad32820fc1a7daebadef4a54d950d490ba58aab8