Evaluation of large language models for the generation of knowledge graphs

This repository contains:

• A technical report: PDF
• Code to download graphs from DBpedia in rdf and csv formats: query.py
• The required code to fine-tune an instance of Mistral on the downloaded graph: finetune.py.
• A script to run inference on the fine-tuned model: inference.py
• The code to train and evaluate tokenizers: tokenizer.py