pyserini/2cr/msmarco-v1-doc.yaml

conditions:
  - name: bm25-doc-tuned
    display: BM25 doc (k1=4.46, b=0.82)
    display-html: BM25 doc (<i>k<sub><small>1</small></sub></i>=4.46, <i>b</i>=0.82)
    command: python -m pyserini.search.lucene --threads 16 --batch-size 128 --index msmarco-v1-doc-slim --topics $topics --output $output --bm25
    topics:
      - topic_key: msmarco-doc-dev
        eval_key: msmarco-doc-dev
        scores:
          - MRR@10: 0.2767
            R@1K: 0.9357
      - topic_key: dl19-doc
        eval_key: dl19-doc
        scores:
          - MAP: 0.2336
            nDCG@10: 0.5233
            R@1K: 0.6757
      - topic_key: dl20
        eval_key: dl20-doc
        scores:
          - MAP: 0.3581
            nDCG@10: 0.5061
            R@1K: 0.7776
  - name: bm25-doc-default
    display: BM25 doc (k1=0.9, b=0.4)
    display-html: BM25 doc (<i>k<sub><small>1</small></sub></i>=0.9, <i>b</i>=0.4)
    display-row: "[<a href=\"#\" data-mdb-toggle=\"tooltip\" title=\"Ma et al. (SIGIR 2021) Document Expansions and Learned Sparse Lexical Representations for MS MARCO V1 and V2.\">1</a>] &mdash; (1a)"
    command: python -m pyserini.search.lucene --threads 16 --batch-size 128 --index msmarco-v1-doc-slim --topics $topics --output $output --bm25 --k1 0.9 --b 0.4
    topics:
      - topic_key: msmarco-doc-dev
        eval_key: msmarco-doc-dev
        scores:
          - MRR@10: 0.2299
            R@1K: 0.8856
      - topic_key: dl19-doc
        eval_key: dl19-doc
        scores:
          - MAP: 0.2434
            nDCG@10: 0.5176
            R@1K: 0.6966
      - topic_key: dl20
        eval_key: dl20-doc
        scores:
          - MAP: 0.3793
            nDCG@10: 0.5286
            R@1K: 0.8085
  - name: bm25-doc-segmented-tuned
    display: BM25 doc segmented (k1=2.16, b=0.61)
    display-html: BM25 doc segmented (<i>k<sub><small>1</small></sub></i>=2.16, <i>b</i>=0.61)
    command: python -m pyserini.search.lucene --threads 16 --batch-size 128 --index msmarco-v1-doc-segmented-slim --topics $topics --output $output --bm25 --hits 10000 --max-passage-hits 1000 --max-passage
    topics:
      - topic_key: msmarco-doc-dev
        eval_key: msmarco-doc-dev
        scores:
          - MRR@10: 0.2756
            R@1K: 0.9311
      - topic_key: dl19-doc
        eval_key: dl19-doc
        scores:
          - MAP: 0.2398
            nDCG@10: 0.5389
            R@1K: 0.6565
      - topic_key: dl20
        eval_key: dl20-doc
        scores:
          - MAP: 0.3458
            nDCG@10: 0.5213
            R@1K: 0.7725
  - name: bm25-doc-segmented-default
    display: BM25 doc segmented (k1=0.9, b=0.4)
    display-html: BM25 doc segmented (<i>k<sub><small>1</small></sub></i>=0.9, <i>b</i>=0.4)
    display-row: "[<a href=\"#\" data-mdb-toggle=\"tooltip\" title=\"Ma et al. (SIGIR 2021) Document Expansions and Learned Sparse Lexical Representations for MS MARCO V1 and V2.\">1</a>] &mdash; (1b)"
    command: python -m pyserini.search.lucene --threads 16 --batch-size 128 --index msmarco-v1-doc-segmented-slim --topics $topics --output $output --bm25 --k1 0.9 --b 0.4 --hits 10000 --max-passage-hits 1000 --max-passage
    topics:
      - topic_key: msmarco-doc-dev
        eval_key: msmarco-doc-dev
        scores:
          - MRR@10: 0.2684
            R@1K: 0.9178
      - topic_key: dl19-doc
        eval_key: dl19-doc
        scores:
          - MAP: 0.2449
            nDCG@10: 0.5302
            R@1K: 0.6871
      - topic_key: dl20
        eval_key: dl20-doc
        scores:
          - MAP: 0.3586
            nDCG@10: 0.5281
            R@1K: 0.7755
  - name: bm25-rm3-doc-tuned
    display: BM25+RM3 doc (k1=4.46, b=0.82)
    display-html: BM25+RM3 doc (<i>k<sub><small>1</small></sub></i>=4.46, <i>b</i>=0.82)
    command: python -m pyserini.search.lucene --threads 16 --batch-size 128 --index msmarco-v1-doc-full --topics $topics --output $output --bm25 --rm3
    topics:
      - topic_key: msmarco-doc-dev
        eval_key: msmarco-doc-dev
        scores:
          - MRR@10: 0.2227
            R@1K: 0.9303
      - topic_key: dl19-doc
        eval_key: dl19-doc
        scores:
          - MAP: 0.2638
            nDCG@10: 0.5526
            R@1K: 0.7188
      - topic_key: dl20
        eval_key: dl20-doc
        scores:
          - MAP: 0.3610
            nDCG@10: 0.5195
            R@1K: 0.8180
  - name: bm25-rm3-doc-default
    display: BM25+RM3 doc (k1=0.9, b=0.4)
    display-html: BM25+RM3 doc (<i>k<sub><small>1</small></sub></i>=0.9, <i>b</i>=0.4)
    display-row: "[<a href=\"#\" data-mdb-toggle=\"tooltip\" title=\"Ma et al. (SIGIR 2021) Document Expansions and Learned Sparse Lexical Representations for MS MARCO V1 and V2.\">1</a>] &mdash; (1c)"
    command: python -m pyserini.search.lucene --threads 16 --batch-size 128 --index msmarco-v1-doc-full --topics $topics --output $output --bm25 --rm3 --k1 0.9 --b 0.4
    topics:
      - topic_key: msmarco-doc-dev
        eval_key: msmarco-doc-dev
        scores:
          - MRR@10: 0.1618
            R@1K: 0.8783
      - topic_key: dl19-doc
        eval_key: dl19-doc
        scores:
          - MAP: 0.2773
            nDCG@10: 0.5174
            R@1K: 0.7507
      - topic_key: dl20
        eval_key: dl20-doc
        scores:
          - MAP: 0.4015
            nDCG@10: 0.5254
            R@1K: 0.8259
  - name: bm25-rm3-doc-segmented-tuned
    display: BM25+RM3 doc segmented (k1=2.16, b=0.61)
    display-html: BM25+RM3 doc segmented (<i>k<sub><small>1</small></sub></i>=2.16, <i>b</i>=0.61)
    command: python -m pyserini.search.lucene --threads 16 --batch-size 128 --index msmarco-v1-doc-segmented-full --topics $topics --output $output --bm25 --rm3 --hits 10000 --max-passage-hits 1000 --max-passage
    topics:
      - topic_key: msmarco-doc-dev
        eval_key: msmarco-doc-dev
        scores:
          - MRR@10: 0.2448
            R@1K: 0.9359
      - topic_key: dl19-doc
        eval_key: dl19-doc
        scores:
          - MAP: 0.2655
            nDCG@10: 0.5392
            R@1K: 0.7037
      - topic_key: dl20
        eval_key: dl20-doc
        scores:
          - MAP: 0.3471
            nDCG@10: 0.5030
            R@1K: 0.8056
  - name: bm25-rm3-doc-segmented-default
    display: BM25+RM3 doc segmented (k1=0.9, b=0.4)
    display-html: BM25+RM3 doc segmented (<i>k<sub><small>1</small></sub></i>=0.9, <i>b</i>=0.4)
    display-row: "[<a href=\"#\" data-mdb-toggle=\"tooltip\" title=\"Ma et al. (SIGIR 2021) Document Expansions and Learned Sparse Lexical Representations for MS MARCO V1 and V2.\">1</a>] &mdash; (1d)"
    command: python -m pyserini.search.lucene --threads 16 --batch-size 128 --index msmarco-v1-doc-segmented-full --topics $topics --output $output --bm25 --rm3 --k1 0.9 --b 0.4 --hits 10000 --max-passage-hits 1000 --max-passage
    topics:
      - topic_key: msmarco-doc-dev
        eval_key: msmarco-doc-dev
        scores:
          - MRR@10: 0.2413
            R@1K: 0.9351
      - topic_key: dl19-doc
        eval_key: dl19-doc
        scores:
          - MAP: 0.2892
            nDCG@10: 0.5684
            R@1K: 0.7368
      - topic_key: dl20
        eval_key: dl20-doc
        scores:
          - MAP: 0.3792
            nDCG@10: 0.5202
            R@1K: 0.8023
  - name: bm25-rocchio-doc-tuned
    display: BM25+Rocchio doc (k1=4.46, b=0.82)
    display-html: BM25+Rocchio doc (<i>k<sub><small>1</small></sub></i>=4.46, <i>b</i>=0.82)
    command: python -m pyserini.search.lucene --threads 16 --batch-size 128 --index msmarco-v1-doc-full --topics $topics --output $output --bm25 --rocchio
    topics:
      - topic_key: msmarco-doc-dev
        eval_key: msmarco-doc-dev
        scores:
          - MRR@10: 0.2242
            R@1K: 0.9314
      - topic_key: dl19-doc
        eval_key: dl19-doc
        scores:
          - MAP: 0.2657
            nDCG@10: 0.5584
            R@1K: 0.7299
      - topic_key: dl20
        eval_key: dl20-doc
        scores:
          - MAP: 0.3628
            nDCG@10: 0.5199
            R@1K: 0.8217
  - name: bm25-rocchio-doc-default
    display: BM25+Rocchio doc (k1=0.9, b=0.4)
    display-html: BM25+Rocchio doc (<i>k<sub><small>1</small></sub></i>=0.9, <i>b</i>=0.4)
    command: python -m pyserini.search.lucene --threads 16 --batch-size 128 --index msmarco-v1-doc-full --topics $topics --output $output --bm25 --rocchio --k1 0.9 --b 0.4
    topics:
      - topic_key: msmarco-doc-dev
        eval_key: msmarco-doc-dev
        scores:
          - MRR@10: 0.1624
            R@1K: 0.8789
      - topic_key: dl19-doc
        eval_key: dl19-doc
        scores:
          - MAP: 0.2811
            nDCG@10: 0.5256
            R@1K: 0.7546
      - topic_key: dl20
        eval_key: dl20-doc
        scores:
          - MAP: 0.4089
            nDCG@10: 0.5192
            R@1K: 0.8273
  - name: bm25-rocchio-doc-segmented-tuned
    display: BM25+Rocchio doc segmented (k1=2.16, b=0.61)
    display-html: BM25+Rocchio doc segmented (<i>k<sub><small>1</small></sub></i>=2.16, <i>b</i>=0.61)
    command: python -m pyserini.search.lucene --threads 16 --batch-size 128 --index msmarco-v1-doc-segmented-full --topics $topics --output $output --bm25 --rocchio --hits 10000 --max-passage-hits 1000 --max-passage
    topics:
      - topic_key: msmarco-doc-dev
        eval_key: msmarco-doc-dev
        scores:
          - MRR@10: 0.2475
            R@1K: 0.9395
      - topic_key: dl19-doc
        eval_key: dl19-doc
        scores:
          - MAP: 0.2672
            nDCG@10: 0.5421
            R@1K: 0.7115
      - topic_key: dl20
        eval_key: dl20-doc
        scores:
          - MAP: 0.3521
            nDCG@10: 0.4997
            R@1K: 0.8042
  - name: bm25-rocchio-doc-segmented-default
    display: BM25+Rocchio doc segmented (k1=0.9, b=0.4)
    display-html: BM25+Rocchio doc segmented (<i>k<sub><small>1</small></sub></i>=0.9, <i>b</i>=0.4)
    command: python -m pyserini.search.lucene --threads 16 --batch-size 128 --index msmarco-v1-doc-segmented-full --topics $topics --output $output --bm25 --rocchio --k1 0.9 --b 0.4 --hits 10000 --max-passage-hits 1000 --max-passage
    topics:
      - topic_key: msmarco-doc-dev
        eval_key: msmarco-doc-dev
        scores:
          - MRR@10: 0.2447
            R@1K: 0.9351
      - topic_key: dl19-doc
        eval_key: dl19-doc
        scores:
          - MAP: 0.2889
            nDCG@10: 0.5570
            R@1K: 0.7423
      - topic_key: dl20
        eval_key: dl20-doc
        scores:
          - MAP: 0.3830
            nDCG@10: 0.5226
            R@1K: 0.8102
  - name: bm25-d2q-t5-doc-tuned
    display: BM25 w/ doc2query-T5 doc (k1=4.68, b=0.87)
    display-html: BM25 w/ doc2query-T5 doc (<i>k<sub><small>1</small></sub></i>=4.68, <i>b</i>=0.87)
    command: python -m pyserini.search.lucene --threads 16 --batch-size 128 --index msmarco-v1-doc-d2q-t5 --topics $topics --output $output --bm25
    topics:
      - topic_key: msmarco-doc-dev
        eval_key: msmarco-doc-dev
        scores:
          - MRR@10: 0.3269
            R@1K: 0.9553
      - topic_key: dl19-doc
        eval_key: dl19-doc
        scores:
          - MAP: 0.2620
            nDCG@10: 0.5972
            R@1K: 0.6867
      - topic_key: dl20
        eval_key: dl20-doc
        scores:
          - MAP: 0.4099
            nDCG@10: 0.5852
            R@1K: 0.8105
  - name: bm25-d2q-t5-doc-default
    display: BM25 w/ doc2query-T5 doc (k1=0.9, b=0.4)
    display-html: BM25 w/ doc2query-T5 doc (<i>k<sub><small>1</small></sub></i>=0.9, <i>b</i>=0.4)
    display-row: "[<a href=\"#\" data-mdb-toggle=\"tooltip\" title=\"Ma et al. (SIGIR 2021) Document Expansions and Learned Sparse Lexical Representations for MS MARCO V1 and V2.\">1</a>] &mdash; (2a)"
    command: python -m pyserini.search.lucene --threads 16 --batch-size 128 --index msmarco-v1-doc-d2q-t5 --topics $topics --output $output --bm25 --k1 0.9 --b 0.4
    topics:
      - topic_key: msmarco-doc-dev
        eval_key: msmarco-doc-dev
        scores:
          - MRR@10: 0.2880
            R@1K: 0.9259
      - topic_key: dl19-doc
        eval_key: dl19-doc
        scores:
          - MAP: 0.2700
            nDCG@10: 0.5968
            R@1K: 0.7190
      - topic_key: dl20
        eval_key: dl20-doc
        scores:
          - MAP: 0.4230
            nDCG@10: 0.5885
            R@1K: 0.8403
  - name: bm25-d2q-t5-doc-segmented-tuned
    display: BM25 w/ doc2query-T5 doc segmented (k1=2.56, b=0.59)
    display-html: BM25 w/ doc2query-T5 doc segmented (<i>k<sub><small>1</small></sub></i>=2.56, <i>b</i>=0.59)
    command: python -m pyserini.search.lucene --threads 16 --batch-size 128 --index msmarco-v1-doc-segmented-d2q-t5 --topics $topics --output $output --bm25 --hits 10000 --max-passage-hits 1000 --max-passage
    topics:
      - topic_key: msmarco-doc-dev
        eval_key: msmarco-doc-dev
        scores:
          - MRR@10: 0.3209
            R@1K: 0.9530
      - topic_key: dl19-doc
        eval_key: dl19-doc
        scores:
          - MAP: 0.2658
            nDCG@10: 0.6273
            R@1K: 0.6707
      - topic_key: dl20
        eval_key: dl20-doc
        scores:
          - MAP: 0.4047
            nDCG@10: 0.5943
            R@1K: 0.7968
  - name: bm25-d2q-t5-doc-segmented-default
    display: BM25 w/ doc2query-T5 doc segmented (k1=0.9, b=0.4)
    display-html: BM25 w/ doc2query-T5 doc segmented (<i>k<sub><small>1</small></sub></i>=0.9, <i>b</i>=0.4)
    display-row: "[<a href=\"#\" data-mdb-toggle=\"tooltip\" title=\"Ma et al. (SIGIR 2021) Document Expansions and Learned Sparse Lexical Representations for MS MARCO V1 and V2.\">1</a>] &mdash; (2b)"
    command: python -m pyserini.search.lucene --threads 16 --batch-size 128 --index msmarco-v1-doc-segmented-d2q-t5 --topics $topics --output $output --bm25 --k1 0.9 --b 0.4 --hits 10000 --max-passage-hits 1000 --max-passage
    topics:
      - topic_key: msmarco-doc-dev
        eval_key: msmarco-doc-dev
        scores:
          - MRR@10: 0.3179
            R@1K: 0.9490
      - topic_key: dl19-doc
        eval_key: dl19-doc
        scores:
          - MAP: 0.2798
            nDCG@10: 0.6119
            R@1K: 0.7165
      - topic_key: dl20
        eval_key: dl20-doc
        scores:
          - MAP: 0.4150
            nDCG@10: 0.5957
            R@1K: 0.8046
  - name: bm25-rm3-d2q-t5-doc-tuned
    display: BM25+RM3 w/ doc2query-T5 doc (k1=4.68, b=0.87)
    display-html: BM25+RM3 w/ doc2query-T5 doc (<i>k<sub><small>1</small></sub></i>=4.68, <i>b</i>=0.87)
    command: python -m pyserini.search.lucene --threads 16 --batch-size 128 --index msmarco-v1-doc-d2q-t5-docvectors --topics $topics --output $output --bm25 --rm3
    topics:
      - topic_key: msmarco-doc-dev
        eval_key: msmarco-doc-dev
        scores:
          - MRR@10: 0.2623
            R@1K: 0.9522
      - topic_key: dl19-doc
        eval_key: dl19-doc
        scores:
          - MAP: 0.2813
            nDCG@10: 0.6091
            R@1K: 0.7184
      - topic_key: dl20
        eval_key: dl20-doc
        scores:
          - MAP: 0.4100
            nDCG@10: 0.5745
            R@1K: 0.8238
  - name: bm25-rm3-d2q-t5-doc-default
    display: BM25+RM3 w/ doc2query-T5 doc (k1=0.9, b=0.4)
    display-html: BM25+RM3 w/ doc2query-T5 doc (<i>k<sub><small>1</small></sub></i>=0.9, <i>b</i>=0.4)
    display-row: "[<a href=\"#\" data-mdb-toggle=\"tooltip\" title=\"Ma et al. (SIGIR 2021) Document Expansions and Learned Sparse Lexical Representations for MS MARCO V1 and V2.\">1</a>] &mdash; (2c)"
    command: python -m pyserini.search.lucene --threads 16 --batch-size 128 --index msmarco-v1-doc-d2q-t5-docvectors --topics $topics --output $output --bm25 --rm3 --k1 0.9 --b 0.4
    topics:
      - topic_key: msmarco-doc-dev
        eval_key: msmarco-doc-dev
        scores:
          - MRR@10: 0.1834
            R@1K: 0.9126
      - topic_key: dl19-doc
        eval_key: dl19-doc
        scores:
          - MAP: 0.3045
            nDCG@10: 0.5904
            R@1K: 0.7737
      - topic_key: dl20
        eval_key: dl20-doc
        scores:
          - MAP: 0.4230
            nDCG@10: 0.5427
            R@1K: 0.8631
  - name: bm25-rm3-d2q-t5-doc-segmented-tuned
    display: BM25+RM3 w/ doc2query-T5 doc segmented (k1=2.56, b=0.59)
    display-html: BM25+RM3 w/ doc2query-T5 doc segmented (<i>k<sub><small>1</small></sub></i>=2.56, <i>b</i>=0.59)
    command: python -m pyserini.search.lucene --threads 16 --batch-size 128 --index msmarco-v1-doc-segmented-d2q-t5-docvectors --topics $topics --output $output --bm25 --rm3 --hits 10000 --max-passage-hits 1000 --max-passage
    topics:
      - topic_key: msmarco-doc-dev
        eval_key: msmarco-doc-dev
        scores:
          - MRR@10: 0.2973
            R@1K: 0.9563
      - topic_key: dl19-doc
        eval_key: dl19-doc
        scores:
          - MAP: 0.2892
            nDCG@10: 0.6247
            R@1K: 0.7069
      - topic_key: dl20
        eval_key: dl20-doc
        scores:
          - MAP: 0.4016
            nDCG@10: 0.5711
            R@1K: 0.8156
  - name: bm25-rm3-d2q-t5-doc-segmented-default
    display: BM25+RM3 w/ doc2query-T5 doc segmented (k1=0.9, b=0.4)
    display-html: BM25+RM3 w/ doc2query-T5 doc segmented (<i>k<sub><small>1</small></sub></i>=0.9, <i>b</i>=0.4)
    display-row: "[<a href=\"#\" data-mdb-toggle=\"tooltip\" title=\"Ma et al. (SIGIR 2021) Document Expansions and Learned Sparse Lexical Representations for MS MARCO V1 and V2.\">1</a>] &mdash; (2d)"
    command: python -m pyserini.search.lucene --threads 16 --batch-size 128 --index msmarco-v1-doc-segmented-d2q-t5-docvectors --topics $topics --output $output --bm25 --rm3 --k1 0.9 --b 0.4 --hits 10000 --max-passage-hits 1000 --max-passage
    topics:
      - topic_key: msmarco-doc-dev
        eval_key: msmarco-doc-dev
        scores:
          - MRR@10: 0.2803
            R@1K: 0.9551
      - topic_key: dl19-doc
        eval_key: dl19-doc
        scores:
          - MAP: 0.3030
            nDCG@10: 0.6290
            R@1K: 0.7483
      - topic_key: dl20
        eval_key: dl20-doc
        scores:
          - MAP: 0.4271
            nDCG@10: 0.5851
            R@1K: 0.8266
  - name: unicoil-noexp-otf
    display: "uniCOIL (noexp): otf"
    display-html: "uniCOIL (noexp): on-the-fly query inference"
    command: python -m pyserini.search.lucene --threads 16 --batch-size 128 --index msmarco-v1-doc-segmented-unicoil-noexp --topics $topics --encoder castorini/unicoil-noexp-msmarco-passage --output $output --impact --hits 10000 --max-passage-hits 1000 --max-passage
    topics:
      - topic_key: msmarco-doc-dev
        eval_key: msmarco-doc-dev
        scores:
          - MRR@10: 0.3410
            R@1K: 0.9420
      - topic_key: dl19-doc
        eval_key: dl19-doc
        scores:
          - MAP: 0.2661
            nDCG@10: 0.6347
            R@1K: 0.6385
      - topic_key: dl20
        eval_key: dl20-doc
        scores:
          - MAP: 0.3698
            nDCG@10: 0.5906
            R@1K: 0.7621
  - name: unicoil-noexp
    display: "uniCOIL (noexp): pre-encoded"
    display-html: "uniCOIL (noexp): pre-encoded queries"
    display-row: "[<a href=\"#\" data-mdb-toggle=\"tooltip\" title=\"Ma et al. (SIGIR 2021) Document Expansions and Learned Sparse Lexical Representations for MS MARCO V1 and V2.\">1</a>] &mdash; (3a)"
    command: python -m pyserini.search.lucene --threads 16 --batch-size 128 --index msmarco-v1-doc-segmented-unicoil-noexp --topics $topics --output $output --impact --hits 10000 --max-passage-hits 1000 --max-passage
    topics:
      - topic_key: msmarco-doc-dev-unicoil-noexp
        eval_key: msmarco-doc-dev
        scores:
          - MRR@10: 0.3409
            R@1K: 0.9420
      - topic_key: dl19-doc-unicoil-noexp
        eval_key: dl19-doc
        scores:
          - MAP: 0.2665
            nDCG@10: 0.6349
            R@1K: 0.6391
      - topic_key: dl20-unicoil-noexp
        eval_key: dl20-doc
        scores:
          - MAP: 0.3698
            nDCG@10: 0.5893
            R@1K: 0.7623
  - name: unicoil-otf
    display: "uniCOIL (w/ doc2query-T5): otf"
    display-html: "uniCOIL (w/ doc2query-T5): on-the-fly query inference"
    command: python -m pyserini.search.lucene --threads 16 --batch-size 128 --index msmarco-v1-doc-segmented-unicoil --topics $topics --encoder castorini/unicoil-msmarco-passage --output $output --impact --hits 10000 --max-passage-hits 1000 --max-passage
    topics:
      - topic_key: msmarco-doc-dev
        eval_key: msmarco-doc-dev
        scores:
          - MRR@10: 0.3532
            R@1K: 0.9546
      - topic_key: dl19-doc
        eval_key: dl19-doc
        scores:
          - MAP: 0.2789
            nDCG@10: 0.6396
            R@1K: 0.6654
      - topic_key: dl20
        eval_key: dl20-doc
        scores:
          - MAP: 0.3881
            nDCG@10: 0.6030
            R@1K: 0.7866
  - name: unicoil
    display: "uniCOIL (w/ doc2query-T5): pre-encoded"
    display-html: "uniCOIL (w/ doc2query-T5): pre-encoded queries"
    display-row: "[<a href=\"#\" data-mdb-toggle=\"tooltip\" title=\"Ma et al. (SIGIR 2021) Document Expansions and Learned Sparse Lexical Representations for MS MARCO V1 and V2.\">1</a>] &mdash; (3b)"
    command: python -m pyserini.search.lucene --threads 16 --batch-size 128 --index msmarco-v1-doc-segmented-unicoil --topics $topics --output $output --impact --hits 10000 --max-passage-hits 1000 --max-passage
    topics:
      - topic_key: msmarco-doc-dev-unicoil
        eval_key: msmarco-doc-dev
        scores:
          - MRR@10: 0.3531
            R@1K: 0.9546
      - topic_key: dl19-doc-unicoil
        eval_key: dl19-doc
        scores:
          - MAP: 0.2789
            nDCG@10: 0.6396
            R@1K: 0.6652
      - topic_key: dl20-unicoil
        eval_key: dl20-doc
        scores:
          - MAP: 0.3882
            nDCG@10: 0.6033
            R@1K: 0.7869