dl20-doc-hgf-wp.yaml

---
corpus: msmarco-doc
corpus_path: collections/msmarco/msmarco-doc/

index_path: indexes/lucene-index.msmarco-doc-hgf-wp/
collection_class: JsonCollection
generator_class: DefaultLuceneDocumentGenerator
index_threads: 7
index_options: -storePositions -storeDocvectors -storeRaw -analyzeWithHuggingFaceTokenizer bert-base-uncased
index_stats:
  documents: 3213835
  documents (non-empty): 3213835
  total terms: 5326253111

metrics:
  - metric: AP@100
    command: target/appassembler/bin/trec_eval
    params: -c -M 100 -m map # Note, this is different DL 2020 passage ranking!
    separator: "\t"
    parse_index: 2
    metric_precision: 4
    can_combine: false
  - metric: nDCG@10
    command: target/appassembler/bin/trec_eval
    params: -c -m ndcg_cut.10
    separator: "\t"
    parse_index: 2
    metric_precision: 4
    can_combine: false
  - metric: R@100
    command: target/appassembler/bin/trec_eval
    params: -c -m recall.100
    separator: "\t"
    parse_index: 2
    metric_precision: 4
    can_combine: false
  - metric: R@1000
    command: target/appassembler/bin/trec_eval
    params: -c -m recall.1000
    separator: "\t"
    parse_index: 2
    metric_precision: 4
    can_combine: false

topic_reader: TsvInt
topics:
  - name: "[DL20 (Doc)](https://trec.nist.gov/data/deep2020.html)"
    id: dl20
    path: topics.dl20.txt
    qrel: qrels.dl20-doc.txt

models:
  - name: bm25-default
    display: BM25 (default)
    params: -bm25 -analyzeWithHuggingFaceTokenizer  bert-base-uncased
    results:
      AP@100:
        - 0.3258
      nDCG@10:
        - 0.5046
      R@100:
        - 0.5483
      R@1000:
        - 0.7436