metadata
tags:
- pyterrier
- pyterrier-artifact
- pyterrier-dr
task_categories:
- text-retrieval
viewer: false
language:
- en
pretty_name: Dense index of the BRIGHT Sustainable Living collection for PyTerrier
Dense index of the BRIGHT Sustainable Living collection for PyTerrier
Description
A dense index of the sustainable-living subset of the BRIGHT dataset build with PyTerrier's DR plugin.
Usage
You will need to install PyTerrier, and its DR engine:
%pip install pyterrier
%pip install pyterrier-dr
# Load the artifact
import pyterrier as pt
artifact = pt.Artifact.from_hf('pyterrier-tutorial/bright.sustainable.mini.dense')
# TODO: Show how you use the artifact
Reproduction
import pyterrier as pt
from pyterrier_dr import SBertBiEncoder, FlexIndex
dataset = pt.get_dataset('irds:bright/sustainable-living')
model_mini = SBertBiEncoder(
"sentence-transformers/all-MiniLM-L6-v2",
batch_size=32,
text_field="text"
)
dense_path_mini = "./bright-mini.flex"
if os.path.exists(dense_path_mini):
shutil.rmtree(dense_path_mini)
dense_index_mini = FlexIndex(dense_path_mini)
dense_indexer_mini = model_mini.doc_encoder() >> dense_index_mini.indexer(mode="overwrite")
dense_indexer_mini.index(dataset.get_corpus_iter())
Metadata
{
"type": "dense_index",
"format": "flex",
"vec_size": 384,
"doc_count": 60792
}