regression.msmarco-v2-passage.unicoil-noexp-0shot.cached.yaml Maven / Gradle / Ivy

Go to download
---
corpus: msmarco-v2-passage-unicoil-noexp-0shot
corpus_path: collections/msmarco/msmarco_v2_passage_unicoil_noexp_0shot/

download_url: https://rgw.cs.uwaterloo.ca/JIMMYLIN-bucket0/data/msmarco_v2_passage_unicoil_noexp_0shot.tar
download_checksum: d9cc1ed3049746e68a2c91bf90e5212d
download_corpus: msmarco_v2_passage_unicoil_noexp_0shot

index_path: indexes/lucene-inverted.msmarco-v2-passage.unicoil-noexp-0shot/
collection_class: JsonVectorCollection
generator_class: DefaultLuceneDocumentGenerator
index_threads: 24
index_options: -impact -pretokenized -storeRaw
index_stats:
  documents: 138364198
  documents (non-empty): 138364198
  total terms: 411330032512

metrics:
  - metric: MAP@100
    command: bin/trec_eval
    params: -c -M 100 -m map
    separator: "\t"
    parse_index: 2
    metric_precision: 4
    can_combine: true
  - metric: MRR@100
    command: bin/trec_eval
    params: -c -M 100 -m recip_rank
    separator: "\t"
    parse_index: 2
    metric_precision: 4
    can_combine: true
  - metric: R@100
    command: bin/trec_eval
    params: -c -m recall.100
    separator: "\t"
    parse_index: 2
    metric_precision: 4
    can_combine: false
  - metric: R@1000
    command: bin/trec_eval
    params: -c -m recall.1000
    separator: "\t"
    parse_index: 2
    metric_precision: 4
    can_combine: false

topic_reader: TsvInt
topics:
  - name: "[MS MARCO V2 Passage: Dev](https://microsoft.github.io/msmarco/TREC-Deep-Learning.html)"
    id: dev
    path: topics.msmarco-v2-passage.dev.unicoil-noexp.0shot.tsv.gz
    qrel: qrels.msmarco-v2-passage.dev.txt
  - name: "[MS MARCO V2 Passage: Dev2](https://microsoft.github.io/msmarco/TREC-Deep-Learning.html)"
    id: dev2
    path: topics.msmarco-v2-passage.dev2.unicoil-noexp.0shot.tsv.gz
    qrel: qrels.msmarco-v2-passage.dev2.txt

models:
  - name: unicoil-noexp-0shot-cached_q
    display: uniCOIL (noexp) zero-shot
    params: -parallelism 16 -impact -pretokenized
    results:
      MAP@100:
        - 0.1333
        - 0.1374
      MRR@100:
        - 0.1342
        - 0.1385
      R@100:
        - 0.4976
        - 0.5217
      R@1000:
        - 0.7010
        - 0.7114
# PRF regressions are no longer maintained for sparse judgments to reduce running times.
# (commenting out instead of removing; in case these numbers are needed, just uncomment and rerun.)
#
#  - name: unicoil-noexp-0shot-cached_q+rm3
#    display: +RM3
#    params: -parallelism 16 -impact -pretokenized -rm3 -collection JsonVectorCollection
#    results:
#      MAP@100:
#        - 0.1115
#        - 0.1190
#      MRR@100:
#        - 0.1124
#        - 0.1197
#      R@100:
#        - 0.5006
#        - 0.5129
#      R@1000:
#        - 0.7258
#        - 0.7346
#  - name: unicoil-noexp-0shot-cached_q+rocchio
#    display: +Rocchio
#    params: -parallelism 16 -impact -pretokenized -rocchio -collection JsonVectorCollection
#    results:
#      MAP@100:
#        - 0.1163
#        - 0.1179
#      MRR@100:
#        - 0.1172
#        - 0.1186
#      R@100:
#        - 0.5082
#        - 0.5218
#      R@1000:
#        - 0.7299
#        - 0.7387