README.md

Spider-NQ: Context Encoder

This is the context encoder of the model fine-tuned on Natural Questions (and initialized from Spider) discussed in our paper Learning to Retrieve Passages without Supervision.

Usage

We used weight sharing for the query encoder and passage encoder, so the same model should be applied for both.

Note! We format the passages similar to DPR, i.e. the title and the text are separated by a [SEP] token, but token type ids are all 0-s.

An example usage:

from transformers import AutoTokenizer, DPRContextEncoder

tokenizer = AutoTokenizer.from_pretrained("NAACL2022/spider-nq-ctx-encoder")
model = DPRContextEncoder.from_pretrained("NAACL2022/spider-nq-ctx-encoder")

title = "Sauron"
context = "Sauron is the title character and main antagonist of J. R. R. Tolkien's \"The Lord of the Rings\"."

input_dict = tokenizer(title, context, return_tensors="pt")
del input_dict["token_type_ids"]

outputs = model(**input_dict)