Skip to content

Latest commit

 

History

History
130 lines (94 loc) · 8.53 KB

regressions-msmarco-v2-passage-unicoil-noexp-0shot.md

File metadata and controls

130 lines (94 loc) · 8.53 KB
Raw

Anserini Regressions: MS MARCO (V2) Passage Ranking

Model: uniCOIL (without any expansions) zero-shot

This page describes regression experiments for passage ranking on the MS MARCO V2 Passage Corpus using the dev queries, which is integrated into Anserini's regression testing framework. Here, we cover experiments with the uniCOIL model trained on the MS MARCO V1 passage ranking test collection, applied in a zero-shot manner, without any expansions.

The uniCOIL model is described in the following paper:

Jimmy Lin and Xueguang Ma. A Few Brief Notes on DeepImpact, COIL, and a Conceptual Framework for Information Retrieval Techniques. arXiv:2106.14807.

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.

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 msmarco-v2-passage-unicoil-noexp-0shot

We make available a version of the corpus that has already been processed with uniCOIL, i.e., we have performed model inference on every document and stored the output sparse vectors. Thus, no neural inference is involved.

From any machine, the following command will download the corpus and perform the complete regression, end to end:

python src/main/python/run_regression.py --download --index --verify --search --regression msmarco-v2-passage-unicoil-noexp-0shot

The run_regression.py script automates the following steps, but if you want to perform each step manually, simply copy/paste from the commands below and you'll obtain the same regression results.

Corpus Download

Download, unpack, and prepare the corpus:

# Download
wget https://rgw.cs.uwaterloo.ca/JIMMYLIN-bucket0/data/msmarco_v2_passage_unicoil_noexp_0shot.tar -P collections/

# Unpack
tar -xvf collections/msmarco_v2_passage_unicoil_noexp_0shot.tar -C collections/

# Rename (indexer is expecting corpus under a slightly different name)
mv collections/msmarco_v2_passage_unicoil_noexp_0shot collections/msmarco-v2-passage-unicoil-noexp-0shot

To confirm, msmarco_v2_passage_unicoil_noexp_0shot.tar is 24 GB and has an MD5 checksum of d9cc1ed3049746e68a2c91bf90e5212d. With the corpus downloaded, the following command will perform the remaining steps below:

python src/main/python/run_regression.py --index --verify --search --regression msmarco-v2-passage-unicoil-noexp-0shot \
  --corpus-path collections/msmarco-v2-passage-unicoil-noexp-0shot

Indexing

Sample indexing command:

target/appassembler/bin/IndexCollection \
  -collection JsonVectorCollection \
  -input /path/to/msmarco-v2-passage-unicoil-noexp-0shot \
  -index indexes/lucene-index.msmarco-v2-passage-unicoil-noexp-0shot/ \
  -generator DefaultLuceneDocumentGenerator \
  -threads 24 -impact -pretokenized -storeRaw \
  >& logs/log.msmarco-v2-passage-unicoil-noexp-0shot &

The path /path/to/msmarco-v2-passage-unicoil-noexp-0shot/ should point to the corpus downloaded above.

The important indexing options to note here are -impact -pretokenized: the first tells Anserini not to encode BM25 doclengths into Lucene's norms (which is the default) and the second option says not to apply any additional tokenization on the uniCOIL tokens. Upon completion, we should have an index with 138,364,198 documents.

For additional details, see explanation of common indexing options.

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:

target/appassembler/bin/SearchCollection \
  -index indexes/lucene-index.msmarco-v2-passage-unicoil-noexp-0shot/ \
  -topics tools/topics-and-qrels/topics.msmarco-v2-passage.dev.unicoil-noexp.0shot.tsv.gz \
  -topicreader TsvInt \
  -output runs/run.msmarco-v2-passage-unicoil-noexp-0shot.unicoil-noexp-0shot.topics.msmarco-v2-passage.dev.unicoil-noexp.0shot.txt \
  -parallelism 16 -impact -pretokenized &
target/appassembler/bin/SearchCollection \
  -index indexes/lucene-index.msmarco-v2-passage-unicoil-noexp-0shot/ \
  -topics tools/topics-and-qrels/topics.msmarco-v2-passage.dev2.unicoil-noexp.0shot.tsv.gz \
  -topicreader TsvInt \
  -output runs/run.msmarco-v2-passage-unicoil-noexp-0shot.unicoil-noexp-0shot.topics.msmarco-v2-passage.dev2.unicoil-noexp.0shot.txt \
  -parallelism 16 -impact -pretokenized &

Evaluation can be performed using trec_eval:

tools/eval/trec_eval.9.0.4/trec_eval -c -m recall.100 tools/topics-and-qrels/qrels.msmarco-v2-passage.dev.txt runs/run.msmarco-v2-passage-unicoil-noexp-0shot.unicoil-noexp-0shot.topics.msmarco-v2-passage.dev.unicoil-noexp.0shot.txt
tools/eval/trec_eval.9.0.4/trec_eval -c -m recall.1000 tools/topics-and-qrels/qrels.msmarco-v2-passage.dev.txt runs/run.msmarco-v2-passage-unicoil-noexp-0shot.unicoil-noexp-0shot.topics.msmarco-v2-passage.dev.unicoil-noexp.0shot.txt
tools/eval/trec_eval.9.0.4/trec_eval -c -M 100 -m map -c -M 100 -m recip_rank tools/topics-and-qrels/qrels.msmarco-v2-passage.dev.txt runs/run.msmarco-v2-passage-unicoil-noexp-0shot.unicoil-noexp-0shot.topics.msmarco-v2-passage.dev.unicoil-noexp.0shot.txt
tools/eval/trec_eval.9.0.4/trec_eval -c -m recall.100 tools/topics-and-qrels/qrels.msmarco-v2-passage.dev2.txt runs/run.msmarco-v2-passage-unicoil-noexp-0shot.unicoil-noexp-0shot.topics.msmarco-v2-passage.dev2.unicoil-noexp.0shot.txt
tools/eval/trec_eval.9.0.4/trec_eval -c -m recall.1000 tools/topics-and-qrels/qrels.msmarco-v2-passage.dev2.txt runs/run.msmarco-v2-passage-unicoil-noexp-0shot.unicoil-noexp-0shot.topics.msmarco-v2-passage.dev2.unicoil-noexp.0shot.txt
tools/eval/trec_eval.9.0.4/trec_eval -c -M 100 -m map -c -M 100 -m recip_rank tools/topics-and-qrels/qrels.msmarco-v2-passage.dev2.txt runs/run.msmarco-v2-passage-unicoil-noexp-0shot.unicoil-noexp-0shot.topics.msmarco-v2-passage.dev2.unicoil-noexp.0shot.txt

Effectiveness

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

MAP@100 uniCOIL (noexp) zero-shot
MS MARCO V2 Passage: Dev 0.1333
MS MARCO V2 Passage: Dev2 0.1374
MRR@100 uniCOIL (noexp) zero-shot
MS MARCO V2 Passage: Dev 0.1342
MS MARCO V2 Passage: Dev2 0.1385
R@100 uniCOIL (noexp) zero-shot
MS MARCO V2 Passage: Dev 0.4976
MS MARCO V2 Passage: Dev2 0.5217
R@1000 uniCOIL (noexp) zero-shot
MS MARCO V2 Passage: Dev 0.7010
MS MARCO V2 Passage: Dev2 0.7114

Reproduction Log*

To add to this reproduction log, modify this template and run bin/build.sh to rebuild the documentation.