An efficient memory-enhanced encoder-decoder language model for long document summarization: combining text segmentation and cross-memory attention.
Check human evaluation results.
This project is released under the CC-BY-SA 3.0 IT license (see LICENSE).