This is a basic proof of concept for fine-tuning a large language model (LLM) to generate TextFSM templates based on raw text and the expected output.
The ntc-templates repository provides a collection of TextFSM templates, along with unit tests that include raw data and expected outputs. These resources serve as the foundation for this fine-tuning process.
Low-Rank Adaptation (LoRA) is used during fine-tuning to reduce memory consumption.
The code is in textfsmllm.ipynb