arguana.terrier

Description

Terrier index for Arguana

Usage

# Load the artifact
import pyterrier as pt
index = pt.Artifact.from_hf('pyterrier/arguana.terrier')
index.bm25()

Benchmarks

name	nDCG@10	R@1000
bm25	0.4914	0.9922
dph	0.4319	0.99

Reproduction

import pyterrier as pt
from tqdm import tqdm
import ir_datasets
dataset = ir_datasets.load('beir/arguana')
meta_docno_len = dataset.metadata()['docs']['fields']['doc_id']['max_len']
indexer = pt.IterDictIndexer("./arguana.terrier", meta={'docno': meta_docno_len, 'text': 4096})
docs = ({'docno': d.doc_id, 'text': d.default_text()} for d in tqdm(dataset.docs))
indexer.index(docs)

Metadata

{
  "type": "sparse_index",
  "format": "terrier",
  "package_hint": "python-terrier"
}