This Retriever Reader is an agent which first retrieves from a database and then reads the dialogue + knowledge from the database to answer.

NOTE: this model only works for evaluation; it assumes all training is already done.

## Basic Examples¶

Evaluate a pre-trained model on Open SQuAD. This model uses a TF-IDF retriever and the DrQA model as the document reader.

python projects/drqa/eval_opensquad.py


Argument

Description

--retriever-model-file

--reader-model-file

--num-retrieved

How many passages to retrieve

Default: 5.

--split-paragraphs

Whether to split the retrieved passages into paragraphs

Default: True.