| # HVAC dataset | |
| This dataset was created to benchmark our approach for multi-modal retrieval (images and tables specifically) on pdf documents. It consists of two components: | |
| 1) A set of pdf documents | |
| 2) A csv file which contains questions and the groundtruth document names as well as page numbers |