aicps / llm_4_plc

LLM4PLC: Harnessing Large Language Models for Verifiable Programming of PLCs in Industrial Control Systems

LLM4PLC: Harnessing Large Language Models for Verifiable Model-Based Design in Industrial Control Systems

Please refer to the project website for the complete description and result showcase.

This research project aims to develop a novel approach to Model-Based Design (MBD) for Industrial Control Systems (ICS) by leveraging iterative prompting of Large Language Models (LLMs) to generate code for Programmable Logic Controllers (PLCs). This repository contains published code, prompts, and descriptions of the LLM4PLC project.

Architecture

To provide a high-level overview of the architecture of the LLM4PLC project, the following diagram is provided.

Pre-Trained LoRAs

The pretrained loras are available here. To use these, please refer to the instructions in text-generation-webui.

Contributors: Mohamad Fakih - [email protected]

Rahul Dharmaji - [email protected]

Yasamin Moghaddas - [email protected]

llm_4_plc's Issues

Provided examples result in a compilation error.

Good afternoon,

first, very nice work. I have tried to compile the provided files in "LLM_Test" using the bash scripts you provided, however it results in a compilation error "error: unknown syntax error." . Is it also possible to provide the training dataset you used for training?

Thanks,