1.6 KiB
1.6 KiB
Documentation
- Class name: Automatic CFG - Fastest
- Category: model_patches/automatic_cfg/presets
- Output node: False
这个节点为模型应用了一个高速配置补丁,优化模型以获得更快的性能,同时不影响生成内容的质量。它使用了一种先进的配置来简化模型的操作,使其非常适合需要快速调整的场景。
Input types
Required
- model
- 需要应用高速配置补丁的模型。它是经过优化以提高性能的核心组件。
- Comfy dtype: MODEL
- Python dtype: torch.nn.Module
Output types
- model
- 经过高速配置优化的模型,具有更好的性能表现。
- Comfy dtype: MODEL
- Python dtype: torch.nn.Module
Usage tips
- Infra type:
CPU - Common nodes: unknown
Source code
class simpleDynamicCFGHighSpeed:
@classmethod
def INPUT_TYPES(s):
return {"required": {
"model": ("MODEL",),
}}
RETURN_TYPES = ("MODEL",)
FUNCTION = "patch"
CATEGORY = "model_patches/automatic_cfg/presets"
def patch(self, model):
advcfg = advancedDynamicCFG()
m = advcfg.patch(model=model, automatic_cfg = "hard",
skip_uncond = True, uncond_sigma_start = 7.5, uncond_sigma_end = 1,
latent_intensity_rescale = False, latent_intensity_rescale_cfg = 7.6,
latent_intensity_rescale_sigma_start = 15, latent_intensity_rescale_sigma_end = 7.5,
latent_intensity_rescale_method = "hard")[0]
return (m, )