YANC_LMStudio

A custom node for a LMStudio integration into ComfyUI.

Features

• Prompts can be sent to LMStudio via HTTP API endpoint.
• Handles 400 errors during API requests, attempting alternative approaches if needed.
• Supports system messages for structured prompting.
• Allows configuration of model identifiers and parameters like temperature and max_tokens.
• Includes optional model unloading in LMStudio and ComfyUI after usage.

Installation

To install the required dependencies browse to your custom_nodes folder inside of ComfyUI, open a command line and clone this repository:

git clone https://github.com/ALatentPlace/YANC_LMStudio.git

I will try to get it also to the ComfyUI Manager as soon as possible.

Usage

The node will put itself in the folders of my other node collection (YANC). You will then find it in the subfolder "LMStudio".

Configuration Options

• prompt: The input prompt to be sent to the AI model inside of LMStudio.
• model_identifier: Identifier for the specific model used. Can be found in LMStudio.
• system_message: A multi-line system prompt that sets instructions to the LLM.
• seed: Random seed (0 by default). Set it to fixed to prevent generations.
• ip: Hostname or IP address of the LMStudio service.
• port: Port number to connect to.
• temperature: Controls randomness in output, between 0.1 and 1.0.
• max_tokens: Maximum length of the generated response.

Model Handling

• unload_llm: Boolean flag to unload the LLM after use (default: False).
• unload_comfy_models: Boolean flag to unload COMFY models before sending a request to LMStudio (default: False).

Error Handling

• Status code 400 errors are caught and handled with alternative approaches when possible. This is to prevent prediction errors in LMStudio, which are usually caused by the system message.
• Automatic removal of the <think> block of Deepseek.