Skip to content

bright-economics.bge-large-en-v1.5.flat.onnx.md

Latest commit

 

History

History
83 lines (59 loc) · 4.98 KB

bright-economics.bge-large-en-v1.5.flat.onnx.md

File metadata and controls

83 lines (59 loc) · 4.98 KB

Anserini Regressions: BRIGHT — Economics

Model: BGE-large-en-v1.5 with flat indexes (using ONNX for on-the-fly query encoding)

This page documents regression experiments, integrated into Anserini's regression testing framework, for BRIGHT — Economics using BGE-large-en-v1.5. The model itself can be download here. See the following paper for more details:

Shitao Xiao, Zheng Liu, Peitian Zhang, and Niklas Muennighoff. C-Pack: Packaged Resources To Advance General Chinese Embedding. arXiv:2309.07597, 2023.

In these experiments, we are using ONNX to perform query encoding on the fly.

The exact configurations for these regressions are stored in this YAML file. Note that this page is automatically generated from this template as part of Anserini's regression pipeline, so do not modify this page directly; modify the template instead and build Anserini to rebuild the documentation.

From one of our Waterloo servers (e.g., orca), the following command will perform the complete regression, end to end:

python src/main/python/run_regression.py --index --verify --search --regression bright-economics.bge-large-en-v1.5.flat.onnx

All the BRIGHT corpora, encoded by the BGE-large-en-v1.5 model, are available for download:

wget https://huggingface.co/datasets/castorini/collections-bright/resolve/main/bright-bge-large-en-v1.5.tar -P collections/
tar xvf collections/bright-bge-large-en-v1.5.tar -C collections/

The tarball is 13 GB and has MD5 checksum 0ce2634d34d3d467cd1afd74f2f63c7b. After download and unpacking the corpora, the run_regression.py command above should work without any issue.

Indexing

Typical indexing command, building flat indexes:

bin/run.sh io.anserini.index.IndexFlatDenseVectors \
  -threads 16 \
  -collection JsonDenseVectorCollection \
  -input /path/to/bright-economics.bge-large-en-v1.5 \
  -generator DenseVectorDocumentGenerator \
  -index indexes/lucene-flat.bright-economics.bge-large-en-v1.5/ \
  >& logs/log.bright-economics.bge-large-en-v1.5 &

The path /path/to/bright-economics.bge-large-en-v1.5/ should point to the corpus downloaded above.

Retrieval

Topics and qrels are stored here, which is linked to the Anserini repo as a submodule.

After indexing has completed, you should be able to perform retrieval as follows:

bin/run.sh io.anserini.search.SearchFlatDenseVectors \
  -index indexes/lucene-flat.bright-economics.bge-large-en-v1.5/ \
  -topics tools/topics-and-qrels/topics.bright-economics.tsv.gz \
  -topicReader TsvString \
  -output runs/run.bright-economics.bge-large-en-v1.5.bge-flat-onnx.topics.bright-economics.txt \
  -encoder BgeLargeEn15 -hits 1000 -removeQuery -threads 16 &

Evaluation can be performed using trec_eval:

bin/trec_eval -c -m ndcg_cut.10 tools/topics-and-qrels/qrels.bright-economics.txt runs/run.bright-economics.bge-large-en-v1.5.bge-flat-onnx.topics.bright-economics.txt
bin/trec_eval -c -m recall.100 tools/topics-and-qrels/qrels.bright-economics.txt runs/run.bright-economics.bge-large-en-v1.5.bge-flat-onnx.topics.bright-economics.txt
bin/trec_eval -c -m recall.1000 tools/topics-and-qrels/qrels.bright-economics.txt runs/run.bright-economics.bge-large-en-v1.5.bge-flat-onnx.topics.bright-economics.txt

Effectiveness

With the above commands, you should be able to reproduce the following results:

nDCG@10 BGE-large-en-v1.5
BRIGHT: Economics 0.1662
R@100 BGE-large-en-v1.5
BRIGHT: Economics 0.4875
R@1000 BGE-large-en-v1.5
BRIGHT: Economics 0.7745

With ONNX query encoding on non-quantized flat indexes, observed results may differ slightly (typically, lower), but scores should generally be within 0.001 of the results reported above (with some outliers).