ERROR lora diffusion_model.double_blocks.16.txt_attn.proj.weight "addmm_cuda" not implemented for 'Float8_e4m3fn'

[hong2610]

This option only appears when <weight_dtype> is selected in - <weight_dtype>. It does not appear when <fp8_e5m2> is selected.
cuda11.8升级到12.4时出现这个异常，但是仍然出图。unet加载器flux-dev-fp8.safetensors枝剪类型选择<weight_dtype>有错误信息，选其他两个<fp8_e5m2>没有错误信息。
pytorch version: 2.5.0+cu124
xformers version: 0.0.28.post2

model weight dtype torch.float8_e4m3fn, manual cast: torch.bfloat16 model_type FLUX Requested to load Flux Loading 1 new model loaded partially 10745.827979049682 10745.7216796875 46 ERROR lora diffusion_model.double_blocks.0.img_attn.qkv.weight "addmm_cuda" not implemented for 'Float8_e4m3fn' ................ ERROR lora diffusion_model.double_blocks.16.txt_attn.proj.weight "addmm_cuda" not implemented for 'Float8_e4m3fn' ERROR lora diffusion_model.double_blocks.17.img_attn.proj.weight "addmm_cuda" not implemented for 'Float8_e4m3fn' ERROR lora diffusion_model.double_blocks.17.txt_attn.proj.weight "addmm_cuda" not implemented for 'Float8_e4m3fn' ERROR lora diffusion_model.double_blocks.18.img_attn.proj.weight "addmm_cuda" not implemented for 'Float8_e4m3fn' ERROR lora diffusion_model.double_blocks.18.txt_attn.proj.weight "addmm_cuda" not implemented for 'Float8_e4m3fn'