regression.msmarco-v1-doc-segmented.yaml Maven / Gradle / Ivy

Go to download
---
corpus: msmarco-doc-segmented
corpus_path: collections/msmarco/msmarco-doc-segmented/

index_path: indexes/lucene-inverted.msmarco-v1-doc-segmented/
collection_class: JsonCollection
generator_class: DefaultLuceneDocumentGenerator
index_threads: 16
index_options: -storePositions -storeDocvectors -storeRaw
index_stats:
  documents: 20545677
  documents (non-empty): 20545677
  total terms: 3200522554

metrics:
  - metric: AP@1000
    command: bin/trec_eval
    params: -c -m map
    separator: "\t"
    parse_index: 2
    metric_precision: 4
    can_combine: false
  - metric: RR@100
    command: bin/trec_eval
    params: -c -M 100 -m recip_rank
    separator: "\t"
    parse_index: 2
    metric_precision: 4
    can_combine: false
  - 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 Doc: Dev](https://github.com/microsoft/MSMARCO-Document-Ranking)"
    id: dev
    path: topics.msmarco-doc.dev.txt
    qrel: qrels.msmarco-doc.dev.txt

models:
  - name: bm25-default
    display: BM25 (default)
    params: -bm25 -hits 10000 -selectMaxPassage -selectMaxPassage.delimiter "#" -selectMaxPassage.hits 1000
    results:
      AP@1000:
        - 0.2690
      RR@100:
        - 0.2684
      R@100:
        - 0.7847
      R@1000:
        - 0.9178
# PRF regressions are no longer maintained for sparse judgments to reduce running times.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 -hits 10000 -selectMaxPassage -selectMaxPassage.delimiter "#" -selectMaxPassage.hits 1000
#    results:
#      AP@1000:
#        - 0.2420
#      RR@100:
#        - 0.2413
#      R@100:
#        - 0.7891
#      R@1000:
#        - 0.9351
#  - name: bm25-default+rocchio
#    display: +Rocchio
#    params: -bm25 -rocchio -hits 10000 -selectMaxPassage -selectMaxPassage.delimiter "#" -selectMaxPassage.hits 1000
#    results:
#      AP@1000:
#        - 0.2453
#      RR@100:
#        - 0.2447
#      R@100:
#        - 0.7915
#      R@1000:
#        - 0.9351
#  - name: bm25-default+rocchio-neg
#    display: +Rocchio*
#    params: -bm25 -rocchio -rocchio.useNegative -hits 10000 -selectMaxPassage -selectMaxPassage.delimiter "#" -selectMaxPassage.hits 1000
#    results:
#      AP@1000:
#        - 0.2464
#      RR@100:
#        - 0.2458
#      R@100:
#        - 0.7888
#      R@1000:
#        - 0.9339
#  - name: bm25-default+ax
#    display: +Ax
#    params: -bm25 -axiom -axiom.deterministic -rerankCutoff 20 -hits 10000 -selectMaxPassage -selectMaxPassage.delimiter "#" -selectMaxPassage.hits 1000
#    results:
#      AP@1000:
#        - 0.2208
#      RR@100:
#        - 0.2201
#      R@100:
#        - 0.7710
#      R@1000:
#        - 0.9264
#  - name: bm25-default+prf
#    display: +PRF
#    params: -bm25 -bm25prf -hits 10000 -selectMaxPassage -selectMaxPassage.delimiter "#" -selectMaxPassage.hits 1000
#    results:
#      AP@1000:
#        - 0.2325
#      RR@100:
#        - 0.2318
#      R@100:
#        - 0.7722
#      R@1000:
#        - 0.9185
  - name: bm25-tuned
    display: BM25 (tuned)
    params: -bm25 -bm25.k1 2.16 -bm25.b 0.61 -hits 10000 -selectMaxPassage -selectMaxPassage.delimiter "#" -selectMaxPassage.hits 1000
    results:
      AP@1000:
        - 0.2762
      RR@100:
        - 0.2756
      R@100:
        - 0.8013
      R@1000:
        - 0.9311
# 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-tuned+rm3
#    display: +RM3
#    params: -bm25 -bm25.k1 2.16 -bm25.b 0.61 -rm3 -hits 10000 -selectMaxPassage -selectMaxPassage.delimiter "#" -selectMaxPassage.hits 1000
#    results:
#      AP@1000:
#        - 0.2455
#      RR@100:
#        - 0.2448
#      R@100:
#        - 0.7978
#      R@1000:
#        - 0.9359
#  - name: bm25-tuned+rocchio
#    display: +Rocchio
#    params: -bm25 -bm25.k1 2.16 -bm25.b 0.61 -rocchio -hits 10000 -selectMaxPassage -selectMaxPassage.delimiter "#" -selectMaxPassage.hits 1000
#    results:
#      AP@1000:
#        - 0.2482
#      RR@100:
#        - 0.2475
#      R@100:
#        - 0.8013
#      R@1000:
#        - 0.9395
#  - name: bm25-tuned+rocchio-neg
#    display: +Rocchio*
#    params: -bm25 -bm25.k1 2.16 -bm25.b 0.61 -rocchio -rocchio.useNegative -hits 10000 -selectMaxPassage -selectMaxPassage.delimiter "#" -selectMaxPassage.hits 1000
#    results:
#      AP@1000:
#        - 0.2497
#      RR@100:
#        - 0.2491
#      R@100:
#        - 0.8042
#      R@1000:
#        - 0.9395
#  - name: bm25-tuned+ax
#    display: +Ax
#    params: -bm25 -bm25.k1 2.16 -bm25.b 0.61 -axiom -axiom.deterministic -rerankCutoff 20 -hits 10000 -selectMaxPassage -selectMaxPassage.delimiter "#" -selectMaxPassage.hits 1000
#    results:
#      AP@1000:
#        - 0.2330
#      RR@100:
#        - 0.2324
#      R@100:
#        - 0.7888
#      R@1000:
#        - 0.9353
#  - name: bm25-tuned+prf
#    display: +PRF
#    params: -bm25 -bm25.k1 2.16 -bm25.b 0.61 -bm25prf -hits 10000 -selectMaxPassage -selectMaxPassage.delimiter "#" -selectMaxPassage.hits 1000
#    results:
#      AP@1000:
#        - 0.2276
#      RR@100:
#        - 0.2269
#      R@100:
#        - 0.7687
#      R@1000:
#        - 0.9157