capreolus.searcher.special

Module Contents

Classes

MsmarcoPsgSearcherMixin

MsmarcoPsg

Skip the searching on training set by converting the official training triplet into a “fake” runfile.

MsmarcoPsgBm25

Skip the searching on training set by converting the official training triplet into a “fake” runfile.

StaticTctColBertDev

Skip the searching on training set by converting the official training triplet into a “fake” runfile.

Attributes

logger

SUPPORTED_TRIPLE_FILE

capreolus.searcher.special.logger[source]
capreolus.searcher.special.SUPPORTED_TRIPLE_FILE = ['small', 'large.v1', 'large.v2'][source]
class capreolus.searcher.special.MsmarcoPsgSearcherMixin[source]
static convert_to_trec_runs(msmarco_top1k_fn, style='eval')[source]
static get_fn_from_url(url)[source]
get_url(self)[source]
download_and_prepare_train_set(self, tmp_dir)[source]
class capreolus.searcher.special.MsmarcoPsg(config=None, provide=None, share_dependency_objects=False, build=True)[source]

Bases: capreolus.searcher.Searcher, MsmarcoPsgSearcherMixin

Skip the searching on training set by converting the official training triplet into a “fake” runfile. Use the offical runfile for the development and the test set.

module_name = msmarcopsg[source]
dependencies[source]
config_spec[source]
class capreolus.searcher.special.MsmarcoPsgBm25(config=None, provide=None, share_dependency_objects=False, build=True)[source]

Bases: capreolus.searcher.anserini.BM25, MsmarcoPsgSearcherMixin

Skip the searching on training set by converting the official training triplet into a “fake” runfile. Conduct configurable BM25 search on the development and the test set.

module_name = msmarcopsgbm25[source]
dependencies[source]
config_spec[source]
class capreolus.searcher.special.StaticTctColBertDev(config=None, provide=None, share_dependency_objects=False, build=True)[source]

Bases: capreolus.searcher.Searcher, MsmarcoPsgSearcherMixin

Skip the searching on training set by converting the official training triplet into a “fake” runfile. Use the runfile pre-prepared using TCT-ColBERT (https://cs.uwaterloo.ca/~jimmylin/publications/Lin_etal_2021_RepL4NLP.pdf)

module_name = static_tct_colbert[source]
dependencies[source]
config_spec[source]