CLIPSave

Documentation

• Class name: CLIPSave
• Category: advanced/model_merging
• Output node: True

The CLIPSave node is designed for persisting CLIP model states or outputs along with optional metadata to a specified file. It encapsulates the functionality to serialize and save the model's state, facilitating the sharing or storage of model configurations and their associated data.

Input types

Required

• clip
  • The CLIP model instance to be saved. This parameter is crucial as it represents the model whose state or output is being persisted.
  • Comfy dtype: CLIP
  • Python dtype: supported_models_base.ClipTarget
• filename_prefix
  • A prefix for the filename under which the CLIP model and its data will be saved. This allows for organized storage and easy retrieval of saved models.
  • Comfy dtype: STRING
  • Python dtype: str

Output types

The node doesn't have output types

Usage tips

• Infra type: CPU
• Common nodes: unknown

Source code

class CLIPSave:
    def __init__(self):
        self.output_dir = folder_paths.get_output_directory()

    @classmethod
    def INPUT_TYPES(s):
        return {"required": { "clip": ("CLIP",),
                              "filename_prefix": ("STRING", {"default": "clip/ComfyUI"}),},
                "hidden": {"prompt": "PROMPT", "extra_pnginfo": "EXTRA_PNGINFO"},}
    RETURN_TYPES = ()
    FUNCTION = "save"
    OUTPUT_NODE = True

    CATEGORY = "advanced/model_merging"

    def save(self, clip, filename_prefix, prompt=None, extra_pnginfo=None):
        prompt_info = ""
        if prompt is not None:
            prompt_info = json.dumps(prompt)

        metadata = {}
        if not args.disable_metadata:
            metadata["prompt"] = prompt_info
            if extra_pnginfo is not None:
                for x in extra_pnginfo:
                    metadata[x] = json.dumps(extra_pnginfo[x])

        comfy.model_management.load_models_gpu([clip.load_model()], force_patch_weights=True)
        clip_sd = clip.get_sd()

        for prefix in ["clip_l.", "clip_g.", ""]:
            k = list(filter(lambda a: a.startswith(prefix), clip_sd.keys()))
            current_clip_sd = {}
            for x in k:
                current_clip_sd[x] = clip_sd.pop(x)
            if len(current_clip_sd) == 0:
                continue

            p = prefix[:-1]
            replace_prefix = {}
            filename_prefix_ = filename_prefix
            if len(p) > 0:
                filename_prefix_ = "{}_{}".format(filename_prefix_, p)
                replace_prefix[prefix] = ""
            replace_prefix["transformer."] = ""

            full_output_folder, filename, counter, subfolder, filename_prefix_ = folder_paths.get_save_image_path(filename_prefix_, self.output_dir)

            output_checkpoint = f"{filename}_{counter:05}_.safetensors"
            output_checkpoint = os.path.join(full_output_folder, output_checkpoint)

            current_clip_sd = comfy.utils.state_dict_prefix_replace(current_clip_sd, replace_prefix)

            comfy.utils.save_torch_file(current_clip_sd, output_checkpoint, metadata=metadata)
        return {}