regression.dl22-passage.splade-pp-ed.cached.yaml Maven / Gradle / Ivy

Go to download
---
corpus: msmarco-v2-passage-splade-pp-ed
corpus_path: collections/msmarco/msmarco_v2_passage_splade_pp_ed/

download_url: https://rgw.cs.uwaterloo.ca/pyserini/data/msmarco_v2_passage_splade_pp_ed.tar
download_checksum: 2cdb2adc259b8fa6caf666b20ebdc0e8
download_corpus: msmarco_v2_passage_splade_pp_ed

index_path: indexes/lucene-inverted.msmarco-v2-passage.splade-pp-ed/
collection_class: JsonVectorCollection
generator_class: DefaultLuceneDocumentGenerator
index_threads: 24
index_options: -impact -pretokenized -storeRaw
index_stats:
  documents: 138364198
  documents (non-empty): 138364198
  total terms: 861058871141

metrics:
  - metric: MAP@100
    command: bin/trec_eval
    params: -c -M 100 -m map -l 2
    separator: "\t"
    parse_index: 2
    metric_precision: 4
    can_combine: false
  - metric: MRR@100
    command: bin/trec_eval
    params: -c -M 100 -m recip_rank -l 2
    separator: "\t"
    parse_index: 2
    metric_precision: 4
    can_combine: false
  - metric: nDCG@10
    command: bin/trec_eval
    params: -c -m ndcg_cut.10
    separator: "\t"
    parse_index: 2
    metric_precision: 4
    can_combine: false
  - metric: R@100
    command: bin/trec_eval
    params: -c -m recall.100 -l 2
    separator: "\t"
    parse_index: 2
    metric_precision: 4
    can_combine: false
  - metric: R@1000
    command: bin/trec_eval
    params: -c -m recall.1000 -l 2
    separator: "\t"
    parse_index: 2
    metric_precision: 4
    can_combine: false

topic_reader: TsvInt
topics:
  - name: "[DL22 (Passage)](https://microsoft.github.io/msmarco/TREC-Deep-Learning)"
    id: dl22
    path: topics.dl22.splade-pp-ed.tsv.gz
    qrel: qrels.dl22-passage.txt

models:
  - name: splade-pp-ed-cached
    display: SPLADE++ CoCondenser-EnsembleDistil
    params: -parallelism 16 -impact -pretokenized
    results:
      MAP@100:
        - 0.1846
      MRR@100:
        - 0.7005
      nDCG@10:
        - 0.5705
      R@100:
        - 0.3742
      R@1000:
        - 0.6629
  - name: splade-pp-ed-cached+rm3
    display: +RM3
    params: -parallelism 16 -impact -pretokenized -rm3 -collection JsonVectorCollection
    results:
      MAP@100:
        - 0.1930
      MRR@100:
        - 0.7161
      nDCG@10:
        - 0.5711
      R@100:
        - 0.3561
      R@1000:
        - 0.6367
  - name: splade-pp-ed-cached+rocchio
    display: +Rocchio
    params: -parallelism 16 -impact -pretokenized -rocchio -collection JsonVectorCollection
    results:
      MAP@100:
        - 0.1968
      MRR@100:
        - 0.6938
      nDCG@10:
        - 0.5897
      R@100:
        - 0.3822
      R@1000:
        - 0.6778