macavaney/d2q-msmarco-passage-scores-monot5

Doc2Query monoT5 Relevance Scores for msmarco-passage

This dataset provides the pre-computed query relevance scores for the msmarco-passage dataset, for use with Doc2Query--.

The generated queries come from macavaney/d2q-msmarco-passage and were scored with castorini/monot5-base-msmarco.

Getting started

This artefact is meant to be used with the pyterrier_doc2query pacakge. It can be installed as:

pip install git+https://github.com/terrierteam/pyterrier_doc2query

Depending on what you are using this aretefact for, you may also need the following additional packages:

pip install git+https://github.com/terrierteam/pyterrier_pisa # for indexing / retrieval
pip install git+https://github.com/terrierteam/pyterrier_t5 # for reproducing this aretefact

Using this artefact

The main use case is to use this aretefact in a Doc2Query−− indexing pipeline:

import pyterrier as pt ; pt.init()
from pyterrier_pisa import PisaIndex
from pyterrier_doc2query import QueryScoreStore, QueryFilter

store = QueryScoreStore.from_repo('https://huggingface.co/datasets/macavaney/d2q-msmarco-passage-scores-monot5')
index = PisaIndex('path/to/index')
pipeline = store.query_scorer(limit_k=40) >> QueryFilter(t=store.percentile(70)) >> index

dataset = pt.get_dataset('irds:msmarco-passage')
pipeline.index(dataset.get_corpus_iter())

You can also use the store directly as a dataset to look up or iterate over the data:

store.lookup('100')
# {'querygen': ..., 'querygen_store': ...}
for record in store:
   pass

Reproducing this aretefact

This aretefact can be reproduced using the following pipeline:

import pyterrier as pt ; pt.init()
from pyterrier_t5 import MonoT5ReRanker
from pyterrier_doc2query import Doc2QueryStore, QueryScoreStore, QueryScorer

doc2query_generator = Doc2QueryStore.from_repo('https://huggingface.co/datasets/macavaney/d2q-msmarco-passage').generator()
store = QueryScoreStore('path/to/store')
pipeline = doc2query_generator >> QueryScorer(MonoT5ReRanker()) >> store

dataset = pt.get_dataset('irds:msmarco-passage')
pipeline.index(dataset.get_corpus_iter())

Note that this process will take quite some time; it computes the relevance score for 80 generated queries for every document in the dataset.