fusion_regression.beir-v1.0.0-hotpotqa.yaml Maven / Gradle / Ivy

Go to download
---
corpus: beir-v1.0.0-hotpotqa
corpus_path: collections/beir-v1.0.0/corpus/hotpotqa/

metrics:
  - metric: nDCG@10
    command: bin/trec_eval
    params: -c -m ndcg_cut.10
    separator: "	"
    parse_index: 2
    metric_precision: 4
    can_combine: false
  - metric: R@100
    command: bin/trec_eval
    params: -c -m recall.100
    separator: "	"
    parse_index: 2
    metric_precision: 4
    can_combine: false
  - metric: R@1000
    command: bin/trec_eval
    params: -c -m recall.1000
    separator: "	"
    parse_index: 2
    metric_precision: 4
    can_combine: false

topic_reader: TsvString
topics:
  - name: "BEIR (v1.0.0): HotpotQA"
    id: test
    path: topics.beir-v1.0.0-hotpotqa.test.tsv.gz
    qrel: qrels.beir-v1.0.0-hotpotqa.test.txt

# Run dependencies for fusion
runs:
  - name: flat-bm25
    dependency: beir-v1.0.0-hotpotqa.flat.yaml
    file: runs/run.inverted.beir-v1.0.0-hotpotqa.flat.test.bm25
  - name: bge-flat-onnx
    dependency: beir-v1.0.0-hotpotqa.bge-base-en-v1.5.parquet.flat.onnx.yaml
    file: runs/run.flat.beir-v1.0.0-hotpotqa.bge-base-en-v1.5.test.bge-flat-onnx

methods:
  - name: rrf
    k: 1000
    depth: 1000
    rrf_k: 60
    output: runs/runs.fuse.rrf.beir-v1.0.0-hotpotqa.flat.bm25.bge-base-en-v1.5.bge-flat-onnx.topics.beir-v1.0.0-hotpotqa.test.txt
    results:
      nDCG@10:
        - 0.7389
      R@100:
        - 0.8917
      R@1000:
        - 0.9473
  - name: average
    output: runs/runs.fuse.avg.beir-v1.0.0-hotpotqa.flat.bm25.bge-base-en-v1.5.bge-flat-onnx.topics.beir-v1.0.0-hotpotqa.test.txt
    results:
      nDCG@10:
        - 0.6497
      R@100:
        - 0.8184
      R@1000:
        - 0.8820
  - name: interpolation
    alpha: 0.5
    output: runs/runs.fuse.interp.beir-v1.0.0-hotpotqa.flat.bm25.bge-base-en-v1.5.bge-flat-onnx.topics.beir-v1.0.0-hotpotqa.test.txt
    results:      
      nDCG@10:
        - 0.6497
      R@100:
        - 0.8184
      R@1000:
        - 0.8820
  - name: normalize
    output: runs/runs.fuse.norm.beir-v1.0.0-hotpotqa.flat.bm25.bge-base-en-v1.5.bge-flat-onnx.topics.beir-v1.0.0-hotpotqa.test.txt
    results:      
      nDCG@10:
        - 0.7658
      R@100:
        - 0.8919
      R@1000:
        - 0.9477