This repository contains codes and access to dataset used for our accepted LLM-KBC workshop paper titled Do Instruction-tuned Large Language Models Help with Relation Extraction?

The REBEL dataset we used for fine-tuning Dolly-v2-3b can be found here: https://zenodo.org/record/6139236#.YhJdiJPMJhH

In the sample_data folder, you can find the 100 samples we used for manual evaluation.

To convert a relation extraction dataset to an instruction-tuning dataset, run the following command.

python data_transform.py

For fine-tuning Dolly with LoRA, run the following code in a command line with providing your own transformed data path in the script.

python fine-tuning.py

Once the adapter is trained, you can run inference using the following command.

python inference.py

If you do have further questions about running the code, feel free to raise an issue or drop me an email at: x DOT li3 AT uva DOT nl .