Question answering based on documents with citations

Beginners
1

I am looking for an NLP model, that can do the following tasks:

  • I provide it with a question.
  • I provide it with some list of documents (say, 10), that somehow relate to that question.
  • Based on the question and the documents, the neural network returns an answer with knowledge from the documents and for each important aspect of the answer it cites document A, B, C etc.
  • At the bottom, we have a bibliography of the documents for all cited ones. (But that can be done manually).

Is there some model that can do this task?

2

I found this repo: GitHub - princeton-nlp/ALCE: [EMNLP 2023] Enabling Large Language Models to Generate Text with Citations. Paper: https://arxiv.org/abs/2305.14627
which may relate to the task. But, I also don’t know how to use it to fine-tune existing LLM, so that the LLM have new skill to cite from given sources.