∞ Message

Documentation

• Class name: LLMChatMessages
• Category: SALT/Language Toolkit/Messages
• Output node: False

The LLMChatMessagesAdv node is designed to prepare and structure chat messages for interaction between a system and a user. It encapsulates the process of converting raw text inputs into a formatted list of chat messages, facilitating a structured dialogue flow.

Input types

Required

• prompt
  • unknown
  • Comfy dtype: STRING
  • Python dtype: unknown
• role
  • unknown
  • Comfy dtype: COMBO[STRING]
  • Python dtype: unknown

Output types

• llm_message
  • Comfy dtype: LIST
  • A list of structured chat messages, representing the dialogue between the system and the user. This output facilitates further processing or interaction within a chat-based application.
  • Python dtype: List[ChatMessage]

Usage tips

• Infra type: CPU
• Common nodes: unknown

Source code

class LLMChatMessages:
    def __init__(self):
        pass

    @classmethod
    def INPUT_TYPES(cls):
        return {
            "required": {
                "prompt": ("STRING", {"multiline": True, "dynamicPrompts": False, "default": "prompt"}),
                "role": (["SYSTEM", "USER"],),
            },
        }

    RETURN_TYPES = ("LIST", )
    RETURN_NAMES = ("llm_message", )

    FUNCTION = "prepare_messages"
    CATEGORY = f"{MENU_NAME}/{SUB_MENU_NAME}/Messages"

    def prepare_messages(self, prompt, role):
        messages = [
                ChatMessage(role=MessageRole.SYSTEM if role == "SYSTEM" else MessageRole.USER, content=repr(prompt) ),
        ]
        return (messages,)