Load Checkpoint

Documentation

• Class name: CheckpointLoaderSimple
• Category: loaders
• Output node: False

The CheckpointLoaderSimple node is designed for loading model checkpoints without requiring a configuration name. It simplifies the checkpoint loading process by automatically guessing the appropriate configuration based on the checkpoint name, making it easier to use for quick model setups or testing.

Input types

Required

• ckpt_name
  • Specifies the name of the checkpoint to load. This parameter is crucial for identifying which checkpoint file to retrieve and use for model initialization.
  • Comfy dtype: COMBO[STRING]
  • Python dtype: str

Output types

• model
  • Comfy dtype: MODEL
  • Returns the main model loaded from the specified checkpoint.
  • Python dtype: torch.nn.Module
• clip
  • Comfy dtype: CLIP
  • Returns the CLIP model associated with the loaded checkpoint, if available.
  • Python dtype: torch.nn.Module
• vae
  • Comfy dtype: VAE
  • Returns the VAE model associated with the loaded checkpoint, if available.
  • Python dtype: torch.nn.Module

Usage tips

• Infra type: GPU
• Common nodes:
  • CLIPTextEncode
  • VAEDecode
  • LoraLoader
  • CLIPSetLastLayer
  • Reroute
  • CR Apply LoRA Stack
  • VAEEncode
  • KSampler
  • UltimateSDUpscale

Source code

class CheckpointLoaderSimple:
    @classmethod
    def INPUT_TYPES(s):
        return {"required": { "ckpt_name": (folder_paths.get_filename_list("checkpoints"), ),
                             }}
    RETURN_TYPES = ("MODEL", "CLIP", "VAE")
    FUNCTION = "load_checkpoint"

    CATEGORY = "loaders"

    def load_checkpoint(self, ckpt_name):
        ckpt_path = folder_paths.get_full_path("checkpoints", ckpt_name)
        out = comfy.sd.load_checkpoint_guess_config(ckpt_path, output_vae=True, output_clip=True, embedding_directory=folder_paths.get_folder_paths("embeddings"))
        return out[:3]