Entry for LM-KBC: Knowledge Base Construction from Pre-trained Language Models (3rd Edition)

This repository hosts the implementation for the paper "Navigating Nulls, Numbers and Numerous Entities: Robust Knowledge Base Construction from Large Language Models".

This repository is a fork of the official LM-KBC challenge at ISWC 2024 repository (https://github.com/lm-kbc/dataset2024).

This repository contains:

• Our model pipeline - see models/dual_llama_3_chat
• Various prompts used in our experiments - see prompt_templates/
• Config file for with our final pipeline settings
• Code for generating additional contextual information - see data/Generating_contextual_information.ipynb

To run our implementation:

Config file: configs/custom-llama-3-8b-instruct.yaml

python baseline.py -c configs/custom-llama-3-8b-instruct.yaml -i data/val.jsonl
python evaluate.py -g data/val.jsonl -p output/configs/custom-llama-3-8b-instruct.jsonl