This is an implementation of the DocLLM paper for Llama models. Based on the paper "DocLLM: A layout-aware generative language model for multimodal document understanding".
Most of the code in this repository is published under MIT license. However, the script "src/external_scripts/document_tokenization/document_tokenization_pymupdf.py" is published GNU Affero General Public License due to it using PyMuPDF. If another license for PyMuPDF is acquired, the script may also be used under that license.