This repository contains the resources, notebooks for fine-tuning and results of working with sota models (for the start of 2023) for relation extraction tasks in Russian:
-
The data provided by NEREL authors were converted to the docred format using notebook
-
The final version of the data for document level RE will be published separately (to get access, you can contact by mail)
- DREEAM
- DocUNET
Here you can download some fine-tuned models.
"Document level relation extraction in Russian" (presented at the AIST-2023, the preprint will be linked here once it is available)