∞ YAML Composer

Documentation

• Class name: LLMYamlComposer
• Category: SALT/Language Toolkit/Tools/YAML
• Output node: False

The LLMYamlComposer node is designed to generate YAML documents from given text inputs, utilizing specified classifiers and any additional directions provided. It abstracts the complexity of YAML composition, leveraging language models to ensure the output is correctly formatted and comprehensive.

Input types

Required

• llm_model
  • Specifies the language model to use for generating the YAML document. It plays a crucial role in interpreting the input text and classifiers to produce a well-structured YAML output.
  • Comfy dtype: LLM_MODEL
  • Python dtype: Dict[str, Any]
• text_input
  • The primary text input from which the YAML document will be generated. This input forms the basis of the content that will be structured into YAML format.
  • Comfy dtype: STRING
  • Python dtype: str
• classifier_list
  • A comma-separated list of classifiers that guide the language model in structuring the input text into a YAML document. These classifiers help in categorizing and organizing the data more effectively.
  • Comfy dtype: STRING
  • Python dtype: str

Optional

• extra_directions
  • Additional instructions to further guide the language model in generating the YAML document. This allows for more tailored and specific YAML outputs.
  • Comfy dtype: STRING
  • Python dtype: str

Output types

• yaml_output
  • Comfy dtype: STRING
  • The generated YAML document, structured and formatted based on the input text, classifiers, and any additional directions provided.
  • Python dtype: str

Usage tips

• Infra type: CPU
• Common nodes: unknown

Source code

class LLMYamlComposer:
    @classmethod
    def INPUT_TYPES(cls):
        return {
            "required": {
                "llm_model": ("LLM_MODEL",),
                "text_input": ("STRING", {"multiline": True, "dynamicPrompts": False, "placeholder": "Data..."}),
                "classifier_list": ("STRING", {"multiline": False, "dynamicPrompts": False}),
            },
            "optional": {
                "extra_directions": ("STRING", {"multiline": True, "dynamicPrompts": False, "placeholder": "Extra directions for the LLM to follow..."}),
            }
        }

    RETURN_TYPES = ("STRING",)
    RETURN_NAMES = ("yaml_output",)

    FUNCTION = "compose_yaml"
    CATEGORY = f"{MENU_NAME}/{SUB_MENU_NAME}/Tools/YAML"

    def compose_yaml(self, llm_model, text_input, classifier_list, extra_directions=""):
        classifier_list = [item.strip() for item in classifier_list.split(",") if item.strip()]
        prompt = (
            f"{text_input}\n\n###\n\n"
            "Given the above text, create a valid YAML document utilizing *all* of the data; "
            f"using the following classifiers: {classifier_list}.\n\n"
            f"{extra_directions}\n\n"
            "Please ensure the YAML output is properly formatted, and does not omit any data."
        )

        response = llm_model['llm'].complete(prompt)

        return (response.text,)