All Downloads are FREE. Search and download functionalities are using the official Maven repository.

regression.msmarco-v2-passage.yaml Maven / Gradle / Ivy

---
corpus: msmarco-v2-passage
corpus_path: collections/msmarco/msmarco_v2_passage/

index_path: indexes/lucene-inverted.msmarco-v2-passage/
collection_class: MsMarcoV2PassageCollection
generator_class: DefaultLuceneDocumentGenerator
index_threads: 24
index_options: -storeRaw
index_stats:
  documents: 138364198
  documents (non-empty): 138364197
  total terms: 4673266800

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.txt
    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.txt
    qrel: qrels.msmarco-v2-passage.dev2.txt

models:
  - name: bm25-default
    display: BM25 (default)
    params: -bm25
    results:
      MAP@100:
        - 0.0709
        - 0.0794
      MRR@100:
        - 0.0719
        - 0.0802
      R@100:
        - 0.3397
        - 0.3459
      R@1000:
        - 0.5733
        - 0.5839
# PRF regressions are no longer maintained for sparse judgments to reduce running times.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: bm25-default+rm3
#    display: +RM3
#    params: -bm25 -rm3 -collection MsMarcoV2PassageCollection
#    results:
#      MAP@100:
#        - 0.0621
#        - 0.0651
#      MRR@100:
#        - 0.0630
#        - 0.0659
#      R@100:
#        - 0.3370
#        - 0.3435
#      R@1000:
#        - 0.5947
#        - 0.6062
#  - name: bm25-default+rocchio
#    display: +Rocchio
#    params: -bm25 -rocchio -collection MsMarcoV2PassageCollection
#    results:
#      MAP@100:
#        - 0.0622
#        - 0.0663
#      MRR@100:
#        - 0.0631
#        - 0.0671
#      R@100:
#        - 0.3413
#        - 0.3516
#      R@1000:
#        - 0.5968
#        - 0.6104




© 2015 - 2025 Weber Informatics LLC | Privacy Policy