From ad97377e463e70ee8b2f501ac7c41134af53e976 Mon Sep 17 00:00:00 2001 From: Jimmy Lin Date: Sat, 15 Jun 2024 07:42:28 -0400 Subject: [PATCH] Refactor regression documentation to fix consistency issues (#2524) --- ...regressions-beir-v1.0.0-arguana.splade-pp-ed.cached.md | 6 +++--- .../regressions-beir-v1.0.0-arguana.splade-pp-ed.onnx.md | 4 ++-- ...egressions-beir-v1.0.0-arguana.unicoil-noexp.cached.md | 4 +++- .../regressions-beir-v1.0.0-bioasq.splade-pp-ed.cached.md | 6 +++--- .../regressions-beir-v1.0.0-bioasq.splade-pp-ed.onnx.md | 4 ++-- ...regressions-beir-v1.0.0-bioasq.unicoil-noexp.cached.md | 4 +++- ...sions-beir-v1.0.0-climate-fever.splade-pp-ed.cached.md | 6 +++--- ...essions-beir-v1.0.0-climate-fever.splade-pp-ed.onnx.md | 4 ++-- ...ions-beir-v1.0.0-climate-fever.unicoil-noexp.cached.md | 4 +++- ...beir-v1.0.0-cqadupstack-android.splade-pp-ed.cached.md | 6 +++--- ...s-beir-v1.0.0-cqadupstack-android.splade-pp-ed.onnx.md | 4 ++-- ...eir-v1.0.0-cqadupstack-android.unicoil-noexp.cached.md | 4 +++- ...beir-v1.0.0-cqadupstack-english.splade-pp-ed.cached.md | 6 +++--- ...s-beir-v1.0.0-cqadupstack-english.splade-pp-ed.onnx.md | 4 ++-- ...eir-v1.0.0-cqadupstack-english.unicoil-noexp.cached.md | 4 +++- ...-beir-v1.0.0-cqadupstack-gaming.splade-pp-ed.cached.md | 6 +++--- ...ns-beir-v1.0.0-cqadupstack-gaming.splade-pp-ed.onnx.md | 4 ++-- ...beir-v1.0.0-cqadupstack-gaming.unicoil-noexp.cached.md | 4 +++- ...ons-beir-v1.0.0-cqadupstack-gis.splade-pp-ed.cached.md | 6 +++--- ...sions-beir-v1.0.0-cqadupstack-gis.splade-pp-ed.onnx.md | 4 ++-- ...ns-beir-v1.0.0-cqadupstack-gis.unicoil-noexp.cached.md | 4 +++- ...-v1.0.0-cqadupstack-mathematica.splade-pp-ed.cached.md | 6 +++--- ...ir-v1.0.0-cqadupstack-mathematica.splade-pp-ed.onnx.md | 4 ++-- ...v1.0.0-cqadupstack-mathematica.unicoil-noexp.cached.md | 4 +++- ...beir-v1.0.0-cqadupstack-physics.splade-pp-ed.cached.md | 6 +++--- ...s-beir-v1.0.0-cqadupstack-physics.splade-pp-ed.onnx.md | 4 ++-- ...eir-v1.0.0-cqadupstack-physics.unicoil-noexp.cached.md | 4 +++- ...-v1.0.0-cqadupstack-programmers.splade-pp-ed.cached.md | 6 +++--- ...ir-v1.0.0-cqadupstack-programmers.splade-pp-ed.onnx.md | 4 ++-- ...v1.0.0-cqadupstack-programmers.unicoil-noexp.cached.md | 4 +++- ...s-beir-v1.0.0-cqadupstack-stats.splade-pp-ed.cached.md | 6 +++--- ...ons-beir-v1.0.0-cqadupstack-stats.splade-pp-ed.onnx.md | 4 ++-- ...-beir-v1.0.0-cqadupstack-stats.unicoil-noexp.cached.md | 4 +++- ...ons-beir-v1.0.0-cqadupstack-tex.splade-pp-ed.cached.md | 6 +++--- ...sions-beir-v1.0.0-cqadupstack-tex.splade-pp-ed.onnx.md | 4 ++-- ...ns-beir-v1.0.0-cqadupstack-tex.unicoil-noexp.cached.md | 4 +++- ...ns-beir-v1.0.0-cqadupstack-unix.splade-pp-ed.cached.md | 6 +++--- ...ions-beir-v1.0.0-cqadupstack-unix.splade-pp-ed.onnx.md | 4 ++-- ...s-beir-v1.0.0-cqadupstack-unix.unicoil-noexp.cached.md | 4 +++- ...r-v1.0.0-cqadupstack-webmasters.splade-pp-ed.cached.md | 6 +++--- ...eir-v1.0.0-cqadupstack-webmasters.splade-pp-ed.onnx.md | 4 ++-- ...-v1.0.0-cqadupstack-webmasters.unicoil-noexp.cached.md | 4 +++- ...ir-v1.0.0-cqadupstack-wordpress.splade-pp-ed.cached.md | 6 +++--- ...beir-v1.0.0-cqadupstack-wordpress.splade-pp-ed.onnx.md | 4 ++-- ...r-v1.0.0-cqadupstack-wordpress.unicoil-noexp.cached.md | 4 +++- ...ions-beir-v1.0.0-dbpedia-entity.splade-pp-ed.cached.md | 6 +++--- ...ssions-beir-v1.0.0-dbpedia-entity.splade-pp-ed.onnx.md | 4 ++-- ...ons-beir-v1.0.0-dbpedia-entity.unicoil-noexp.cached.md | 4 +++- .../regressions-beir-v1.0.0-fever.splade-pp-ed.cached.md | 6 +++--- .../regressions-beir-v1.0.0-fever.splade-pp-ed.onnx.md | 4 ++-- .../regressions-beir-v1.0.0-fever.unicoil-noexp.cached.md | 4 +++- .../regressions-beir-v1.0.0-fiqa.splade-pp-ed.cached.md | 6 +++--- .../regressions-beir-v1.0.0-fiqa.splade-pp-ed.onnx.md | 4 ++-- .../regressions-beir-v1.0.0-fiqa.unicoil-noexp.cached.md | 4 +++- ...egressions-beir-v1.0.0-hotpotqa.splade-pp-ed.cached.md | 6 +++--- .../regressions-beir-v1.0.0-hotpotqa.splade-pp-ed.onnx.md | 4 ++-- ...gressions-beir-v1.0.0-hotpotqa.unicoil-noexp.cached.md | 4 +++- ...egressions-beir-v1.0.0-nfcorpus.splade-pp-ed.cached.md | 8 ++++---- .../regressions-beir-v1.0.0-nfcorpus.splade-pp-ed.onnx.md | 4 ++-- ...gressions-beir-v1.0.0-nfcorpus.unicoil-noexp.cached.md | 4 +++- .../regressions-beir-v1.0.0-nq.splade-pp-ed.cached.md | 6 +++--- .../regressions-beir-v1.0.0-nq.splade-pp-ed.onnx.md | 4 ++-- .../regressions-beir-v1.0.0-nq.unicoil-noexp.cached.md | 4 +++- .../regressions-beir-v1.0.0-quora.splade-pp-ed.cached.md | 6 +++--- .../regressions-beir-v1.0.0-quora.splade-pp-ed.onnx.md | 4 ++-- .../regressions-beir-v1.0.0-quora.unicoil-noexp.cached.md | 4 +++- ...egressions-beir-v1.0.0-robust04.splade-pp-ed.cached.md | 6 +++--- .../regressions-beir-v1.0.0-robust04.splade-pp-ed.onnx.md | 4 ++-- ...gressions-beir-v1.0.0-robust04.unicoil-noexp.cached.md | 4 +++- ...regressions-beir-v1.0.0-scidocs.splade-pp-ed.cached.md | 6 +++--- .../regressions-beir-v1.0.0-scidocs.splade-pp-ed.onnx.md | 4 ++-- ...egressions-beir-v1.0.0-scidocs.unicoil-noexp.cached.md | 4 +++- ...regressions-beir-v1.0.0-scifact.splade-pp-ed.cached.md | 6 +++--- .../regressions-beir-v1.0.0-scifact.splade-pp-ed.onnx.md | 4 ++-- ...egressions-beir-v1.0.0-scifact.unicoil-noexp.cached.md | 4 +++- ...egressions-beir-v1.0.0-signal1m.splade-pp-ed.cached.md | 6 +++--- .../regressions-beir-v1.0.0-signal1m.splade-pp-ed.onnx.md | 4 ++-- ...gressions-beir-v1.0.0-signal1m.unicoil-noexp.cached.md | 4 +++- ...ressions-beir-v1.0.0-trec-covid.splade-pp-ed.cached.md | 6 +++--- ...egressions-beir-v1.0.0-trec-covid.splade-pp-ed.onnx.md | 4 ++-- ...essions-beir-v1.0.0-trec-covid.unicoil-noexp.cached.md | 4 +++- ...gressions-beir-v1.0.0-trec-news.splade-pp-ed.cached.md | 6 +++--- ...regressions-beir-v1.0.0-trec-news.splade-pp-ed.onnx.md | 4 ++-- ...ressions-beir-v1.0.0-trec-news.unicoil-noexp.cached.md | 4 +++- ...ns-beir-v1.0.0-webis-touche2020.splade-pp-ed.cached.md | 6 +++--- ...ions-beir-v1.0.0-webis-touche2020.splade-pp-ed.onnx.md | 4 ++-- ...s-beir-v1.0.0-webis-touche2020.unicoil-noexp.cached.md | 4 +++- ...ions-dl19-passage.bge-base-en-v1.5.hnsw-int8.cached.md | 6 +++--- ...ssions-dl19-passage.bge-base-en-v1.5.hnsw-int8.onnx.md | 4 ++-- ...gressions-dl19-passage.bge-base-en-v1.5.hnsw.cached.md | 4 ++-- ...-passage.cohere-embed-english-v3.0.hnsw-int8.cached.md | 6 +++--- ...-dl19-passage.cohere-embed-english-v3.0.hnsw.cached.md | 4 ++-- .../regressions-dl19-passage.cos-dpr-distil.fw.md | 4 ++-- ...ssions-dl19-passage.cos-dpr-distil.hnsw-int8.cached.md | 6 +++--- ...ressions-dl19-passage.cos-dpr-distil.hnsw-int8.onnx.md | 4 ++-- ...regressions-dl19-passage.cos-dpr-distil.hnsw.cached.md | 4 ++-- .../regressions-dl19-passage.cos-dpr-distil.lexlsh.md | 4 ++-- ...gressions-dl19-passage.openai-ada2.hnsw-int8.cached.md | 6 +++--- .../regressions-dl19-passage.openai-ada2.hnsw.cached.md | 4 ++-- .../regressions-dl19-passage.splade-pp-ed.cached.md | 4 ++-- .../regressions-dl19-passage.splade-pp-sd.cached.md | 4 ++-- .../regressions-dl19-passage.unicoil-noexp.cached.md | 4 +++- .../regressions-dl19-passage.unicoil.cached.md | 4 +++- ...ions-dl20-passage.bge-base-en-v1.5.hnsw-int8.cached.md | 6 +++--- ...ssions-dl20-passage.bge-base-en-v1.5.hnsw-int8.onnx.md | 4 ++-- ...gressions-dl20-passage.bge-base-en-v1.5.hnsw.cached.md | 4 ++-- ...-passage.cohere-embed-english-v3.0.hnsw-int8.cached.md | 6 +++--- ...-dl20-passage.cohere-embed-english-v3.0.hnsw.cached.md | 4 ++-- .../regressions-dl20-passage.cos-dpr-distil.fw.md | 4 ++-- ...ssions-dl20-passage.cos-dpr-distil.hnsw-int8.cached.md | 6 +++--- ...ressions-dl20-passage.cos-dpr-distil.hnsw-int8.onnx.md | 4 ++-- ...regressions-dl20-passage.cos-dpr-distil.hnsw.cached.md | 4 ++-- .../regressions-dl20-passage.cos-dpr-distil.lexlsh.md | 4 ++-- ...gressions-dl20-passage.openai-ada2.hnsw-int8.cached.md | 6 +++--- .../regressions-dl20-passage.openai-ada2.hnsw.cached.md | 4 ++-- .../regressions-dl20-passage.splade-pp-ed.cached.md | 4 ++-- .../regressions-dl20-passage.splade-pp-sd.cached.md | 4 ++-- .../regressions-dl20-passage.unicoil-noexp.cached.md | 4 +++- .../regressions-dl20-passage.unicoil.cached.md | 4 +++- .../regressions-dl21-passage.splade-pp-ed.cached.md | 4 ++-- .../regressions-dl21-passage.splade-pp-sd.cached.md | 4 ++-- .../regressions-dl21-passage.unicoil-0shot.cached.md | 2 +- ...regressions-dl21-passage.unicoil-noexp-0shot.cached.md | 2 +- .../regressions-dl22-passage.splade-pp-ed.cached.md | 4 ++-- .../regressions-dl22-passage.splade-pp-sd.cached.md | 4 ++-- .../regressions-dl23-passage.splade-pp-ed.cached.md | 4 ++-- .../regressions-dl23-passage.splade-pp-sd.cached.md | 4 ++-- ...smarco-v1-passage.bge-base-en-v1.5.hnsw-int8.cached.md | 6 +++--- ...-msmarco-v1-passage.bge-base-en-v1.5.hnsw-int8.onnx.md | 4 ++-- ...ons-msmarco-v1-passage.bge-base-en-v1.5.hnsw.cached.md | 4 ++-- ...-passage.cohere-embed-english-v3.0.hnsw-int8.cached.md | 6 +++--- ...co-v1-passage.cohere-embed-english-v3.0.hnsw.cached.md | 4 ++-- .../regressions-msmarco-v1-passage.cos-dpr-distil.fw.md | 4 ++-- ...-msmarco-v1-passage.cos-dpr-distil.hnsw-int8.cached.md | 6 +++--- ...ns-msmarco-v1-passage.cos-dpr-distil.hnsw-int8.onnx.md | 4 ++-- ...sions-msmarco-v1-passage.cos-dpr-distil.hnsw.cached.md | 4 ++-- ...egressions-msmarco-v1-passage.cos-dpr-distil.lexlsh.md | 4 ++-- .../regressions-msmarco-v1-passage.deepimpact.cached.md | 4 +++- ...ssions-msmarco-v1-passage.distill-splade-max.cached.md | 4 +++- ...ons-msmarco-v1-passage.openai-ada2.hnsw-int8.cached.md | 6 +++--- ...ressions-msmarco-v1-passage.openai-ada2.hnsw.cached.md | 4 ++-- .../regressions-msmarco-v1-passage.splade-pp-ed.cached.md | 4 ++-- .../regressions-msmarco-v1-passage.splade-pp-sd.cached.md | 4 ++-- ...regressions-msmarco-v1-passage.unicoil-noexp.cached.md | 4 +++- ...s-msmarco-v1-passage.unicoil-tilde-expansion.cached.md | 4 +++- .../regressions-msmarco-v1-passage.unicoil.cached.md | 4 +++- .../regressions-msmarco-v2-passage.splade-pp-ed.cached.md | 4 ++-- .../regressions-msmarco-v2-passage.splade-pp-sd.cached.md | 4 ++-- .../beir-v1.0.0-arguana.splade-pp-ed.cached.template | 6 +++--- .../beir-v1.0.0-arguana.splade-pp-ed.onnx.template | 4 ++-- .../beir-v1.0.0-arguana.unicoil-noexp.cached.template | 4 +++- .../beir-v1.0.0-bioasq.splade-pp-ed.cached.template | 6 +++--- .../beir-v1.0.0-bioasq.splade-pp-ed.onnx.template | 4 ++-- .../beir-v1.0.0-bioasq.unicoil-noexp.cached.template | 4 +++- ...beir-v1.0.0-climate-fever.splade-pp-ed.cached.template | 6 +++--- .../beir-v1.0.0-climate-fever.splade-pp-ed.onnx.template | 4 ++-- ...eir-v1.0.0-climate-fever.unicoil-noexp.cached.template | 4 +++- ...1.0.0-cqadupstack-android.splade-pp-ed.cached.template | 6 +++--- ...-v1.0.0-cqadupstack-android.splade-pp-ed.onnx.template | 4 ++-- ....0.0-cqadupstack-android.unicoil-noexp.cached.template | 4 +++- ...1.0.0-cqadupstack-english.splade-pp-ed.cached.template | 6 +++--- ...-v1.0.0-cqadupstack-english.splade-pp-ed.onnx.template | 4 ++-- ....0.0-cqadupstack-english.unicoil-noexp.cached.template | 4 +++- ...v1.0.0-cqadupstack-gaming.splade-pp-ed.cached.template | 6 +++--- ...r-v1.0.0-cqadupstack-gaming.splade-pp-ed.onnx.template | 4 ++-- ...1.0.0-cqadupstack-gaming.unicoil-noexp.cached.template | 4 +++- ...ir-v1.0.0-cqadupstack-gis.splade-pp-ed.cached.template | 6 +++--- ...beir-v1.0.0-cqadupstack-gis.splade-pp-ed.onnx.template | 4 ++-- ...r-v1.0.0-cqadupstack-gis.unicoil-noexp.cached.template | 4 +++- ...0-cqadupstack-mathematica.splade-pp-ed.cached.template | 6 +++--- ...0.0-cqadupstack-mathematica.splade-pp-ed.onnx.template | 4 ++-- ...-cqadupstack-mathematica.unicoil-noexp.cached.template | 4 +++- ...1.0.0-cqadupstack-physics.splade-pp-ed.cached.template | 6 +++--- ...-v1.0.0-cqadupstack-physics.splade-pp-ed.onnx.template | 4 ++-- ....0.0-cqadupstack-physics.unicoil-noexp.cached.template | 4 +++- ...0-cqadupstack-programmers.splade-pp-ed.cached.template | 6 +++--- ...0.0-cqadupstack-programmers.splade-pp-ed.onnx.template | 4 ++-- ...-cqadupstack-programmers.unicoil-noexp.cached.template | 4 +++- ...-v1.0.0-cqadupstack-stats.splade-pp-ed.cached.template | 6 +++--- ...ir-v1.0.0-cqadupstack-stats.splade-pp-ed.onnx.template | 4 ++-- ...v1.0.0-cqadupstack-stats.unicoil-noexp.cached.template | 4 +++- ...ir-v1.0.0-cqadupstack-tex.splade-pp-ed.cached.template | 6 +++--- ...beir-v1.0.0-cqadupstack-tex.splade-pp-ed.onnx.template | 4 ++-- ...r-v1.0.0-cqadupstack-tex.unicoil-noexp.cached.template | 4 +++- ...r-v1.0.0-cqadupstack-unix.splade-pp-ed.cached.template | 6 +++--- ...eir-v1.0.0-cqadupstack-unix.splade-pp-ed.onnx.template | 4 ++-- ...-v1.0.0-cqadupstack-unix.unicoil-noexp.cached.template | 4 +++- ....0-cqadupstack-webmasters.splade-pp-ed.cached.template | 6 +++--- ....0.0-cqadupstack-webmasters.splade-pp-ed.onnx.template | 4 ++-- ...0-cqadupstack-webmasters.unicoil-noexp.cached.template | 4 +++- ...0.0-cqadupstack-wordpress.splade-pp-ed.cached.template | 6 +++--- ...1.0.0-cqadupstack-wordpress.splade-pp-ed.onnx.template | 4 ++-- ....0-cqadupstack-wordpress.unicoil-noexp.cached.template | 4 +++- ...eir-v1.0.0-dbpedia-entity.splade-pp-ed.cached.template | 6 +++--- .../beir-v1.0.0-dbpedia-entity.splade-pp-ed.onnx.template | 4 ++-- ...ir-v1.0.0-dbpedia-entity.unicoil-noexp.cached.template | 4 +++- .../beir-v1.0.0-fever.splade-pp-ed.cached.template | 6 +++--- .../beir-v1.0.0-fever.splade-pp-ed.onnx.template | 4 ++-- .../beir-v1.0.0-fever.unicoil-noexp.cached.template | 4 +++- .../beir-v1.0.0-fiqa.splade-pp-ed.cached.template | 6 +++--- .../templates/beir-v1.0.0-fiqa.splade-pp-ed.onnx.template | 4 ++-- .../beir-v1.0.0-fiqa.unicoil-noexp.cached.template | 4 +++- .../beir-v1.0.0-hotpotqa.splade-pp-ed.cached.template | 6 +++--- .../beir-v1.0.0-hotpotqa.splade-pp-ed.onnx.template | 4 ++-- .../beir-v1.0.0-hotpotqa.unicoil-noexp.cached.template | 4 +++- .../beir-v1.0.0-nfcorpus.splade-pp-ed.cached.template | 8 ++++---- .../beir-v1.0.0-nfcorpus.splade-pp-ed.onnx.template | 4 ++-- .../beir-v1.0.0-nfcorpus.unicoil-noexp.cached.template | 4 +++- .../templates/beir-v1.0.0-nq.splade-pp-ed.cached.template | 6 +++--- .../templates/beir-v1.0.0-nq.splade-pp-ed.onnx.template | 4 ++-- .../beir-v1.0.0-nq.unicoil-noexp.cached.template | 4 +++- .../beir-v1.0.0-quora.splade-pp-ed.cached.template | 6 +++--- .../beir-v1.0.0-quora.splade-pp-ed.onnx.template | 4 ++-- .../beir-v1.0.0-quora.unicoil-noexp.cached.template | 4 +++- .../beir-v1.0.0-robust04.splade-pp-ed.cached.template | 6 +++--- .../beir-v1.0.0-robust04.splade-pp-ed.onnx.template | 4 ++-- .../beir-v1.0.0-robust04.unicoil-noexp.cached.template | 4 +++- .../beir-v1.0.0-scidocs.splade-pp-ed.cached.template | 6 +++--- .../beir-v1.0.0-scidocs.splade-pp-ed.onnx.template | 4 ++-- .../beir-v1.0.0-scidocs.unicoil-noexp.cached.template | 4 +++- .../beir-v1.0.0-scifact.splade-pp-ed.cached.template | 6 +++--- .../beir-v1.0.0-scifact.splade-pp-ed.onnx.template | 4 ++-- .../beir-v1.0.0-scifact.unicoil-noexp.cached.template | 4 +++- .../beir-v1.0.0-signal1m.splade-pp-ed.cached.template | 6 +++--- .../beir-v1.0.0-signal1m.splade-pp-ed.onnx.template | 4 ++-- .../beir-v1.0.0-signal1m.unicoil-noexp.cached.template | 4 +++- .../beir-v1.0.0-trec-covid.splade-pp-ed.cached.template | 6 +++--- .../beir-v1.0.0-trec-covid.splade-pp-ed.onnx.template | 4 ++-- .../beir-v1.0.0-trec-covid.unicoil-noexp.cached.template | 4 +++- .../beir-v1.0.0-trec-news.splade-pp-ed.cached.template | 6 +++--- .../beir-v1.0.0-trec-news.splade-pp-ed.onnx.template | 4 ++-- .../beir-v1.0.0-trec-news.unicoil-noexp.cached.template | 4 +++- ...r-v1.0.0-webis-touche2020.splade-pp-ed.cached.template | 6 +++--- ...eir-v1.0.0-webis-touche2020.splade-pp-ed.onnx.template | 4 ++-- ...-v1.0.0-webis-touche2020.unicoil-noexp.cached.template | 4 +++- ...l19-passage.bge-base-en-v1.5.hnsw-int8.cached.template | 6 +++--- .../dl19-passage.bge-base-en-v1.5.hnsw-int8.onnx.template | 4 ++-- .../dl19-passage.bge-base-en-v1.5.hnsw.cached.template | 4 ++-- ...ge.cohere-embed-english-v3.0.hnsw-int8.cached.template | 6 +++--- ...passage.cohere-embed-english-v3.0.hnsw.cached.template | 4 ++-- .../templates/dl19-passage.cos-dpr-distil.fw.template | 4 ++-- .../dl19-passage.cos-dpr-distil.hnsw-int8.cached.template | 6 +++--- .../dl19-passage.cos-dpr-distil.hnsw-int8.onnx.template | 4 ++-- .../dl19-passage.cos-dpr-distil.hnsw.cached.template | 4 ++-- .../templates/dl19-passage.cos-dpr-distil.lexlsh.template | 4 ++-- .../dl19-passage.openai-ada2.hnsw-int8.cached.template | 6 +++--- .../dl19-passage.openai-ada2.hnsw.cached.template | 4 ++-- .../templates/dl19-passage.splade-pp-ed.cached.template | 4 ++-- .../templates/dl19-passage.splade-pp-sd.cached.template | 4 ++-- .../templates/dl19-passage.unicoil-noexp.cached.template | 4 +++- .../docgen/templates/dl19-passage.unicoil.cached.template | 4 +++- ...l20-passage.bge-base-en-v1.5.hnsw-int8.cached.template | 6 +++--- .../dl20-passage.bge-base-en-v1.5.hnsw-int8.onnx.template | 4 ++-- .../dl20-passage.bge-base-en-v1.5.hnsw.cached.template | 4 ++-- ...ge.cohere-embed-english-v3.0.hnsw-int8.cached.template | 6 +++--- ...passage.cohere-embed-english-v3.0.hnsw.cached.template | 4 ++-- .../templates/dl20-passage.cos-dpr-distil.fw.template | 4 ++-- .../dl20-passage.cos-dpr-distil.hnsw-int8.cached.template | 6 +++--- .../dl20-passage.cos-dpr-distil.hnsw-int8.onnx.template | 4 ++-- .../dl20-passage.cos-dpr-distil.hnsw.cached.template | 4 ++-- .../templates/dl20-passage.cos-dpr-distil.lexlsh.template | 4 ++-- .../dl20-passage.openai-ada2.hnsw-int8.cached.template | 6 +++--- .../dl20-passage.openai-ada2.hnsw.cached.template | 4 ++-- .../templates/dl20-passage.splade-pp-ed.cached.template | 4 ++-- .../templates/dl20-passage.splade-pp-sd.cached.template | 4 ++-- .../templates/dl20-passage.unicoil-noexp.cached.template | 4 +++- .../docgen/templates/dl20-passage.unicoil.cached.template | 4 +++- .../templates/dl21-passage.splade-pp-ed.cached.template | 4 ++-- .../templates/dl21-passage.splade-pp-sd.cached.template | 4 ++-- .../templates/dl21-passage.unicoil-0shot.cached.template | 2 +- .../dl21-passage.unicoil-noexp-0shot.cached.template | 2 +- .../templates/dl22-passage.splade-pp-ed.cached.template | 4 ++-- .../templates/dl22-passage.splade-pp-sd.cached.template | 4 ++-- .../templates/dl23-passage.splade-pp-ed.cached.template | 4 ++-- .../templates/dl23-passage.splade-pp-sd.cached.template | 4 ++-- ...-v1-passage.bge-base-en-v1.5.hnsw-int8.cached.template | 6 +++--- ...co-v1-passage.bge-base-en-v1.5.hnsw-int8.onnx.template | 4 ++-- ...marco-v1-passage.bge-base-en-v1.5.hnsw.cached.template | 4 ++-- ...ge.cohere-embed-english-v3.0.hnsw-int8.cached.template | 6 +++--- ...passage.cohere-embed-english-v3.0.hnsw.cached.template | 4 ++-- .../msmarco-v1-passage.cos-dpr-distil.fw.template | 4 ++-- ...co-v1-passage.cos-dpr-distil.hnsw-int8.cached.template | 6 +++--- ...arco-v1-passage.cos-dpr-distil.hnsw-int8.onnx.template | 4 ++-- ...msmarco-v1-passage.cos-dpr-distil.hnsw.cached.template | 4 ++-- .../msmarco-v1-passage.cos-dpr-distil.lexlsh.template | 4 ++-- .../msmarco-v1-passage.deepimpact.cached.template | 4 +++- .../msmarco-v1-passage.distill-splade-max.cached.template | 4 +++- ...marco-v1-passage.openai-ada2.hnsw-int8.cached.template | 6 +++--- .../msmarco-v1-passage.openai-ada2.hnsw.cached.template | 4 ++-- .../msmarco-v1-passage.splade-pp-ed.cached.template | 4 ++-- .../msmarco-v1-passage.splade-pp-sd.cached.template | 4 ++-- .../msmarco-v1-passage.unicoil-noexp.cached.template | 4 +++- ...rco-v1-passage.unicoil-tilde-expansion.cached.template | 4 +++- .../templates/msmarco-v1-passage.unicoil.cached.template | 4 +++- .../msmarco-v2-passage.splade-pp-ed.cached.template | 4 ++-- .../msmarco-v2-passage.splade-pp-sd.cached.template | 4 ++-- 296 files changed, 748 insertions(+), 596 deletions(-) diff --git a/docs/regressions/regressions-beir-v1.0.0-arguana.splade-pp-ed.cached.md b/docs/regressions/regressions-beir-v1.0.0-arguana.splade-pp-ed.cached.md index 58ac89c424..30bf17e7db 100644 --- a/docs/regressions/regressions-beir-v1.0.0-arguana.splade-pp-ed.cached.md +++ b/docs/regressions/regressions-beir-v1.0.0-arguana.splade-pp-ed.cached.md @@ -1,14 +1,14 @@ # Anserini Regressions: BEIR (v1.0.0) — ArguAna -**Model**: [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) (using pre-encoded queries) +**Model**: [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) (using cached queries) -This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — ArguAna](http://beir.ai/). +This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — ArguAna](http://beir.ai/). The model itself can be download [here](https://huggingface.co/naver/splade-cocondenser-ensembledistil). See the [official SPLADE repo](/~https://github.com/naver/splade) and the following paper for more details: > Thibault Formal, Carlos Lassance, Benjamin Piwowarski, and Stéphane Clinchant. [From Distillation to Hard Negative Sampling: Making Sparse Neural IR Models More Effective.](https://dl.acm.org/doi/10.1145/3477495.3531857) _Proceedings of the 45th International ACM SIGIR Conference on Research and Development in Information Retrieval_, pages 2353–2359. -In these experiments, we are using pre-encoded queries (i.e., cached results of query encoding). +In these experiments, we are using cached queries (i.e., cached results of query encoding). The exact configurations for these regressions are stored in [this YAML file](../../src/main/resources/regression/beir-v1.0.0-arguana.splade-pp-ed.cached.yaml). Note that this page is automatically generated from [this template](../../src/main/resources/docgen/templates/beir-v1.0.0-arguana.splade-pp-ed.cached.template) as part of Anserini's regression pipeline, so do not modify this page directly; modify the template instead and then run `bin/build.sh` to rebuild the documentation. diff --git a/docs/regressions/regressions-beir-v1.0.0-arguana.splade-pp-ed.onnx.md b/docs/regressions/regressions-beir-v1.0.0-arguana.splade-pp-ed.onnx.md index 0b908beaac..ce9781315d 100644 --- a/docs/regressions/regressions-beir-v1.0.0-arguana.splade-pp-ed.onnx.md +++ b/docs/regressions/regressions-beir-v1.0.0-arguana.splade-pp-ed.onnx.md @@ -1,8 +1,8 @@ # Anserini Regressions: BEIR (v1.0.0) — ArguAna -**Model**: [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) (using ONNX for on-the-fly query encoding) +**Model**: [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) (using ONNX for on-the-fly query encoding) -This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — ArguAna](http://beir.ai/). +This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — ArguAna](http://beir.ai/). The model itself can be download [here](https://huggingface.co/naver/splade-cocondenser-ensembledistil). See the [official SPLADE repo](/~https://github.com/naver/splade) and the following paper for more details: diff --git a/docs/regressions/regressions-beir-v1.0.0-arguana.unicoil-noexp.cached.md b/docs/regressions/regressions-beir-v1.0.0-arguana.unicoil-noexp.cached.md index 656affe1e1..c036329d4a 100644 --- a/docs/regressions/regressions-beir-v1.0.0-arguana.unicoil-noexp.cached.md +++ b/docs/regressions/regressions-beir-v1.0.0-arguana.unicoil-noexp.cached.md @@ -1,12 +1,14 @@ # Anserini Regressions: BEIR (v1.0.0) — ArguAna -**Model**: uniCOIL (without any expansions) +**Model**: uniCOIL without any expansions (using cached queries) This page describes regression experiments, integrated into Anserini's regression testing framework, using uniCOIL (without any expansions) on [BEIR (v1.0.0) — ArguAna](http://beir.ai/). 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.](https://arxiv.org/abs/2106.14807) _arXiv:2106.14807_. +In these experiments, we are using cached queries (i.e., cached results of query encoding). + The exact configurations for these regressions are stored in [this YAML file](../../src/main/resources/regression/beir-v1.0.0-arguana.unicoil-noexp.cached.yaml). Note that this page is automatically generated from [this template](../../src/main/resources/docgen/templates/beir-v1.0.0-arguana.unicoil-noexp.cached.template) as part of Anserini's regression pipeline, so do not modify this page directly; modify the template instead. diff --git a/docs/regressions/regressions-beir-v1.0.0-bioasq.splade-pp-ed.cached.md b/docs/regressions/regressions-beir-v1.0.0-bioasq.splade-pp-ed.cached.md index 05071947a6..5100e1eacf 100644 --- a/docs/regressions/regressions-beir-v1.0.0-bioasq.splade-pp-ed.cached.md +++ b/docs/regressions/regressions-beir-v1.0.0-bioasq.splade-pp-ed.cached.md @@ -1,14 +1,14 @@ # Anserini Regressions: BEIR (v1.0.0) — BioASQ -**Model**: [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) (using pre-encoded queries) +**Model**: [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) (using cached queries) -This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — BioASQ](http://beir.ai/). +This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — BioASQ](http://beir.ai/). The model itself can be download [here](https://huggingface.co/naver/splade-cocondenser-ensembledistil). See the [official SPLADE repo](/~https://github.com/naver/splade) and the following paper for more details: > Thibault Formal, Carlos Lassance, Benjamin Piwowarski, and Stéphane Clinchant. [From Distillation to Hard Negative Sampling: Making Sparse Neural IR Models More Effective.](https://dl.acm.org/doi/10.1145/3477495.3531857) _Proceedings of the 45th International ACM SIGIR Conference on Research and Development in Information Retrieval_, pages 2353–2359. -In these experiments, we are using pre-encoded queries (i.e., cached results of query encoding). +In these experiments, we are using cached queries (i.e., cached results of query encoding). The exact configurations for these regressions are stored in [this YAML file](../../src/main/resources/regression/beir-v1.0.0-bioasq.splade-pp-ed.cached.yaml). Note that this page is automatically generated from [this template](../../src/main/resources/docgen/templates/beir-v1.0.0-bioasq.splade-pp-ed.cached.template) as part of Anserini's regression pipeline, so do not modify this page directly; modify the template instead and then run `bin/build.sh` to rebuild the documentation. diff --git a/docs/regressions/regressions-beir-v1.0.0-bioasq.splade-pp-ed.onnx.md b/docs/regressions/regressions-beir-v1.0.0-bioasq.splade-pp-ed.onnx.md index 82024d25a5..4a0a53a596 100644 --- a/docs/regressions/regressions-beir-v1.0.0-bioasq.splade-pp-ed.onnx.md +++ b/docs/regressions/regressions-beir-v1.0.0-bioasq.splade-pp-ed.onnx.md @@ -1,8 +1,8 @@ # Anserini Regressions: BEIR (v1.0.0) — BioASQ -**Model**: [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) (using ONNX for on-the-fly query encoding) +**Model**: [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) (using ONNX for on-the-fly query encoding) -This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — BioASQ](http://beir.ai/). +This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — BioASQ](http://beir.ai/). The model itself can be download [here](https://huggingface.co/naver/splade-cocondenser-ensembledistil). See the [official SPLADE repo](/~https://github.com/naver/splade) and the following paper for more details: diff --git a/docs/regressions/regressions-beir-v1.0.0-bioasq.unicoil-noexp.cached.md b/docs/regressions/regressions-beir-v1.0.0-bioasq.unicoil-noexp.cached.md index 99dc6f9af5..8c28260e4b 100644 --- a/docs/regressions/regressions-beir-v1.0.0-bioasq.unicoil-noexp.cached.md +++ b/docs/regressions/regressions-beir-v1.0.0-bioasq.unicoil-noexp.cached.md @@ -1,12 +1,14 @@ # Anserini Regressions: BEIR (v1.0.0) — BioASQ -**Model**: uniCOIL (without any expansions) +**Model**: uniCOIL without any expansions (using cached queries) This page describes regression experiments, integrated into Anserini's regression testing framework, using uniCOIL (without any expansions) on [BEIR (v1.0.0) — BioASQ](http://beir.ai/). 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.](https://arxiv.org/abs/2106.14807) _arXiv:2106.14807_. +In these experiments, we are using cached queries (i.e., cached results of query encoding). + The exact configurations for these regressions are stored in [this YAML file](../../src/main/resources/regression/beir-v1.0.0-bioasq.unicoil-noexp.cached.yaml). Note that this page is automatically generated from [this template](../../src/main/resources/docgen/templates/beir-v1.0.0-bioasq.unicoil-noexp.cached.template) as part of Anserini's regression pipeline, so do not modify this page directly; modify the template instead. diff --git a/docs/regressions/regressions-beir-v1.0.0-climate-fever.splade-pp-ed.cached.md b/docs/regressions/regressions-beir-v1.0.0-climate-fever.splade-pp-ed.cached.md index 7082bd4d2b..83ee798358 100644 --- a/docs/regressions/regressions-beir-v1.0.0-climate-fever.splade-pp-ed.cached.md +++ b/docs/regressions/regressions-beir-v1.0.0-climate-fever.splade-pp-ed.cached.md @@ -1,14 +1,14 @@ # Anserini Regressions: BEIR (v1.0.0) — Climate-FEVER -**Model**: [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) (using pre-encoded queries) +**Model**: [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) (using cached queries) -This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — Climate-FEVER](http://beir.ai/). +This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — Climate-FEVER](http://beir.ai/). The model itself can be download [here](https://huggingface.co/naver/splade-cocondenser-ensembledistil). See the [official SPLADE repo](/~https://github.com/naver/splade) and the following paper for more details: > Thibault Formal, Carlos Lassance, Benjamin Piwowarski, and Stéphane Clinchant. [From Distillation to Hard Negative Sampling: Making Sparse Neural IR Models More Effective.](https://dl.acm.org/doi/10.1145/3477495.3531857) _Proceedings of the 45th International ACM SIGIR Conference on Research and Development in Information Retrieval_, pages 2353–2359. -In these experiments, we are using pre-encoded queries (i.e., cached results of query encoding). +In these experiments, we are using cached queries (i.e., cached results of query encoding). The exact configurations for these regressions are stored in [this YAML file](../../src/main/resources/regression/beir-v1.0.0-climate-fever.splade-pp-ed.cached.yaml). Note that this page is automatically generated from [this template](../../src/main/resources/docgen/templates/beir-v1.0.0-climate-fever.splade-pp-ed.cached.template) as part of Anserini's regression pipeline, so do not modify this page directly; modify the template instead and then run `bin/build.sh` to rebuild the documentation. diff --git a/docs/regressions/regressions-beir-v1.0.0-climate-fever.splade-pp-ed.onnx.md b/docs/regressions/regressions-beir-v1.0.0-climate-fever.splade-pp-ed.onnx.md index 737fb14a03..e2ba3388ab 100644 --- a/docs/regressions/regressions-beir-v1.0.0-climate-fever.splade-pp-ed.onnx.md +++ b/docs/regressions/regressions-beir-v1.0.0-climate-fever.splade-pp-ed.onnx.md @@ -1,8 +1,8 @@ # Anserini Regressions: BEIR (v1.0.0) — Climate-FEVER -**Model**: [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) (using ONNX for on-the-fly query encoding) +**Model**: [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) (using ONNX for on-the-fly query encoding) -This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — Climate-FEVER](http://beir.ai/). +This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — Climate-FEVER](http://beir.ai/). The model itself can be download [here](https://huggingface.co/naver/splade-cocondenser-ensembledistil). See the [official SPLADE repo](/~https://github.com/naver/splade) and the following paper for more details: diff --git a/docs/regressions/regressions-beir-v1.0.0-climate-fever.unicoil-noexp.cached.md b/docs/regressions/regressions-beir-v1.0.0-climate-fever.unicoil-noexp.cached.md index a89d905a9f..3f4cc62862 100644 --- a/docs/regressions/regressions-beir-v1.0.0-climate-fever.unicoil-noexp.cached.md +++ b/docs/regressions/regressions-beir-v1.0.0-climate-fever.unicoil-noexp.cached.md @@ -1,12 +1,14 @@ # Anserini Regressions: BEIR (v1.0.0) — Climate-FEVER -**Model**: uniCOIL (without any expansions) +**Model**: uniCOIL without any expansions (using cached queries) This page describes regression experiments, integrated into Anserini's regression testing framework, using uniCOIL (without any expansions) on [BEIR (v1.0.0) — Climate-FEVER](http://beir.ai/). 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.](https://arxiv.org/abs/2106.14807) _arXiv:2106.14807_. +In these experiments, we are using cached queries (i.e., cached results of query encoding). + The exact configurations for these regressions are stored in [this YAML file](../../src/main/resources/regression/beir-v1.0.0-climate-fever.unicoil-noexp.cached.yaml). Note that this page is automatically generated from [this template](../../src/main/resources/docgen/templates/beir-v1.0.0-climate-fever.unicoil-noexp.cached.template) as part of Anserini's regression pipeline, so do not modify this page directly; modify the template instead. diff --git a/docs/regressions/regressions-beir-v1.0.0-cqadupstack-android.splade-pp-ed.cached.md b/docs/regressions/regressions-beir-v1.0.0-cqadupstack-android.splade-pp-ed.cached.md index 959c5db61e..912449b55f 100644 --- a/docs/regressions/regressions-beir-v1.0.0-cqadupstack-android.splade-pp-ed.cached.md +++ b/docs/regressions/regressions-beir-v1.0.0-cqadupstack-android.splade-pp-ed.cached.md @@ -1,14 +1,14 @@ # Anserini Regressions: BEIR (v1.0.0) — CQADupStack-android -**Model**: [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) (using pre-encoded queries) +**Model**: [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) (using cached queries) -This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — CQADupStack-android](http://beir.ai/). +This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — CQADupStack-android](http://beir.ai/). The model itself can be download [here](https://huggingface.co/naver/splade-cocondenser-ensembledistil). See the [official SPLADE repo](/~https://github.com/naver/splade) and the following paper for more details: > Thibault Formal, Carlos Lassance, Benjamin Piwowarski, and Stéphane Clinchant. [From Distillation to Hard Negative Sampling: Making Sparse Neural IR Models More Effective.](https://dl.acm.org/doi/10.1145/3477495.3531857) _Proceedings of the 45th International ACM SIGIR Conference on Research and Development in Information Retrieval_, pages 2353–2359. -In these experiments, we are using pre-encoded queries (i.e., cached results of query encoding). +In these experiments, we are using cached queries (i.e., cached results of query encoding). The exact configurations for these regressions are stored in [this YAML file](../../src/main/resources/regression/beir-v1.0.0-cqadupstack-android.splade-pp-ed.cached.yaml). Note that this page is automatically generated from [this template](../../src/main/resources/docgen/templates/beir-v1.0.0-cqadupstack-android.splade-pp-ed.cached.template) as part of Anserini's regression pipeline, so do not modify this page directly; modify the template instead and then run `bin/build.sh` to rebuild the documentation. diff --git a/docs/regressions/regressions-beir-v1.0.0-cqadupstack-android.splade-pp-ed.onnx.md b/docs/regressions/regressions-beir-v1.0.0-cqadupstack-android.splade-pp-ed.onnx.md index 4029c9a339..4dd5b1841e 100644 --- a/docs/regressions/regressions-beir-v1.0.0-cqadupstack-android.splade-pp-ed.onnx.md +++ b/docs/regressions/regressions-beir-v1.0.0-cqadupstack-android.splade-pp-ed.onnx.md @@ -1,8 +1,8 @@ # Anserini Regressions: BEIR (v1.0.0) — CQADupStack-android -**Model**: [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) (using ONNX for on-the-fly query encoding) +**Model**: [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) (using ONNX for on-the-fly query encoding) -This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — CQADupStack-android](http://beir.ai/). +This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — CQADupStack-android](http://beir.ai/). The model itself can be download [here](https://huggingface.co/naver/splade-cocondenser-ensembledistil). See the [official SPLADE repo](/~https://github.com/naver/splade) and the following paper for more details: diff --git a/docs/regressions/regressions-beir-v1.0.0-cqadupstack-android.unicoil-noexp.cached.md b/docs/regressions/regressions-beir-v1.0.0-cqadupstack-android.unicoil-noexp.cached.md index 7cba66d968..d3f3e27ce6 100644 --- a/docs/regressions/regressions-beir-v1.0.0-cqadupstack-android.unicoil-noexp.cached.md +++ b/docs/regressions/regressions-beir-v1.0.0-cqadupstack-android.unicoil-noexp.cached.md @@ -1,12 +1,14 @@ # Anserini Regressions: BEIR (v1.0.0) — CQADupStack-android -**Model**: uniCOIL (without any expansions) +**Model**: uniCOIL without any expansions (using cached queries) This page describes regression experiments, integrated into Anserini's regression testing framework, using uniCOIL (without any expansions) on [BEIR (v1.0.0) — CQADupStack-android](http://beir.ai/). 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.](https://arxiv.org/abs/2106.14807) _arXiv:2106.14807_. +In these experiments, we are using cached queries (i.e., cached results of query encoding). + The exact configurations for these regressions are stored in [this YAML file](../../src/main/resources/regression/beir-v1.0.0-cqadupstack-android.unicoil-noexp.cached.yaml). Note that this page is automatically generated from [this template](../../src/main/resources/docgen/templates/beir-v1.0.0-cqadupstack-android.unicoil-noexp.cached.template) as part of Anserini's regression pipeline, so do not modify this page directly; modify the template instead. diff --git a/docs/regressions/regressions-beir-v1.0.0-cqadupstack-english.splade-pp-ed.cached.md b/docs/regressions/regressions-beir-v1.0.0-cqadupstack-english.splade-pp-ed.cached.md index b6d5cd7b14..e151116747 100644 --- a/docs/regressions/regressions-beir-v1.0.0-cqadupstack-english.splade-pp-ed.cached.md +++ b/docs/regressions/regressions-beir-v1.0.0-cqadupstack-english.splade-pp-ed.cached.md @@ -1,14 +1,14 @@ # Anserini Regressions: BEIR (v1.0.0) — CQADupStack-english -**Model**: [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) (using pre-encoded queries) +**Model**: [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) (using cached queries) -This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — CQADupStack-english](http://beir.ai/). +This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — CQADupStack-english](http://beir.ai/). The model itself can be download [here](https://huggingface.co/naver/splade-cocondenser-ensembledistil). See the [official SPLADE repo](/~https://github.com/naver/splade) and the following paper for more details: > Thibault Formal, Carlos Lassance, Benjamin Piwowarski, and Stéphane Clinchant. [From Distillation to Hard Negative Sampling: Making Sparse Neural IR Models More Effective.](https://dl.acm.org/doi/10.1145/3477495.3531857) _Proceedings of the 45th International ACM SIGIR Conference on Research and Development in Information Retrieval_, pages 2353–2359. -In these experiments, we are using pre-encoded queries (i.e., cached results of query encoding). +In these experiments, we are using cached queries (i.e., cached results of query encoding). The exact configurations for these regressions are stored in [this YAML file](../../src/main/resources/regression/beir-v1.0.0-cqadupstack-english.splade-pp-ed.cached.yaml). Note that this page is automatically generated from [this template](../../src/main/resources/docgen/templates/beir-v1.0.0-cqadupstack-english.splade-pp-ed.cached.template) as part of Anserini's regression pipeline, so do not modify this page directly; modify the template instead and then run `bin/build.sh` to rebuild the documentation. diff --git a/docs/regressions/regressions-beir-v1.0.0-cqadupstack-english.splade-pp-ed.onnx.md b/docs/regressions/regressions-beir-v1.0.0-cqadupstack-english.splade-pp-ed.onnx.md index 0017124d6f..f6970dc40b 100644 --- a/docs/regressions/regressions-beir-v1.0.0-cqadupstack-english.splade-pp-ed.onnx.md +++ b/docs/regressions/regressions-beir-v1.0.0-cqadupstack-english.splade-pp-ed.onnx.md @@ -1,8 +1,8 @@ # Anserini Regressions: BEIR (v1.0.0) — CQADupStack-english -**Model**: [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) (using ONNX for on-the-fly query encoding) +**Model**: [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) (using ONNX for on-the-fly query encoding) -This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — CQADupStack-english](http://beir.ai/). +This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — CQADupStack-english](http://beir.ai/). The model itself can be download [here](https://huggingface.co/naver/splade-cocondenser-ensembledistil). See the [official SPLADE repo](/~https://github.com/naver/splade) and the following paper for more details: diff --git a/docs/regressions/regressions-beir-v1.0.0-cqadupstack-english.unicoil-noexp.cached.md b/docs/regressions/regressions-beir-v1.0.0-cqadupstack-english.unicoil-noexp.cached.md index 8c74bdf8d4..6dc184385b 100644 --- a/docs/regressions/regressions-beir-v1.0.0-cqadupstack-english.unicoil-noexp.cached.md +++ b/docs/regressions/regressions-beir-v1.0.0-cqadupstack-english.unicoil-noexp.cached.md @@ -1,12 +1,14 @@ # Anserini Regressions: BEIR (v1.0.0) — CQADupStack-english -**Model**: uniCOIL (without any expansions) +**Model**: uniCOIL without any expansions (using cached queries) This page describes regression experiments, integrated into Anserini's regression testing framework, using uniCOIL (without any expansions) on [BEIR (v1.0.0) — CQADupStack-english](http://beir.ai/). 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.](https://arxiv.org/abs/2106.14807) _arXiv:2106.14807_. +In these experiments, we are using cached queries (i.e., cached results of query encoding). + The exact configurations for these regressions are stored in [this YAML file](../../src/main/resources/regression/beir-v1.0.0-cqadupstack-english.unicoil-noexp.cached.yaml). Note that this page is automatically generated from [this template](../../src/main/resources/docgen/templates/beir-v1.0.0-cqadupstack-english.unicoil-noexp.cached.template) as part of Anserini's regression pipeline, so do not modify this page directly; modify the template instead. diff --git a/docs/regressions/regressions-beir-v1.0.0-cqadupstack-gaming.splade-pp-ed.cached.md b/docs/regressions/regressions-beir-v1.0.0-cqadupstack-gaming.splade-pp-ed.cached.md index b48c354c59..cc30dd40f2 100644 --- a/docs/regressions/regressions-beir-v1.0.0-cqadupstack-gaming.splade-pp-ed.cached.md +++ b/docs/regressions/regressions-beir-v1.0.0-cqadupstack-gaming.splade-pp-ed.cached.md @@ -1,14 +1,14 @@ # Anserini Regressions: BEIR (v1.0.0) — CQADupStack-gaming -**Model**: [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) (using pre-encoded queries) +**Model**: [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) (using cached queries) -This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — CQADupStack-gaming](http://beir.ai/). +This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — CQADupStack-gaming](http://beir.ai/). The model itself can be download [here](https://huggingface.co/naver/splade-cocondenser-ensembledistil). See the [official SPLADE repo](/~https://github.com/naver/splade) and the following paper for more details: > Thibault Formal, Carlos Lassance, Benjamin Piwowarski, and Stéphane Clinchant. [From Distillation to Hard Negative Sampling: Making Sparse Neural IR Models More Effective.](https://dl.acm.org/doi/10.1145/3477495.3531857) _Proceedings of the 45th International ACM SIGIR Conference on Research and Development in Information Retrieval_, pages 2353–2359. -In these experiments, we are using pre-encoded queries (i.e., cached results of query encoding). +In these experiments, we are using cached queries (i.e., cached results of query encoding). The exact configurations for these regressions are stored in [this YAML file](../../src/main/resources/regression/beir-v1.0.0-cqadupstack-gaming.splade-pp-ed.cached.yaml). Note that this page is automatically generated from [this template](../../src/main/resources/docgen/templates/beir-v1.0.0-cqadupstack-gaming.splade-pp-ed.cached.template) as part of Anserini's regression pipeline, so do not modify this page directly; modify the template instead and then run `bin/build.sh` to rebuild the documentation. diff --git a/docs/regressions/regressions-beir-v1.0.0-cqadupstack-gaming.splade-pp-ed.onnx.md b/docs/regressions/regressions-beir-v1.0.0-cqadupstack-gaming.splade-pp-ed.onnx.md index 59b288c095..9b9ab24839 100644 --- a/docs/regressions/regressions-beir-v1.0.0-cqadupstack-gaming.splade-pp-ed.onnx.md +++ b/docs/regressions/regressions-beir-v1.0.0-cqadupstack-gaming.splade-pp-ed.onnx.md @@ -1,8 +1,8 @@ # Anserini Regressions: BEIR (v1.0.0) — CQADupStack-gaming -**Model**: [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) (using ONNX for on-the-fly query encoding) +**Model**: [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) (using ONNX for on-the-fly query encoding) -This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — CQADupStack-gaming](http://beir.ai/). +This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — CQADupStack-gaming](http://beir.ai/). The model itself can be download [here](https://huggingface.co/naver/splade-cocondenser-ensembledistil). See the [official SPLADE repo](/~https://github.com/naver/splade) and the following paper for more details: diff --git a/docs/regressions/regressions-beir-v1.0.0-cqadupstack-gaming.unicoil-noexp.cached.md b/docs/regressions/regressions-beir-v1.0.0-cqadupstack-gaming.unicoil-noexp.cached.md index 31716db32e..41a09c99ba 100644 --- a/docs/regressions/regressions-beir-v1.0.0-cqadupstack-gaming.unicoil-noexp.cached.md +++ b/docs/regressions/regressions-beir-v1.0.0-cqadupstack-gaming.unicoil-noexp.cached.md @@ -1,12 +1,14 @@ # Anserini Regressions: BEIR (v1.0.0) — CQADupStack-gaming -**Model**: uniCOIL (without any expansions) +**Model**: uniCOIL without any expansions (using cached queries) This page describes regression experiments, integrated into Anserini's regression testing framework, using uniCOIL (without any expansions) on [BEIR (v1.0.0) — CQADupStack-gaming](http://beir.ai/). 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.](https://arxiv.org/abs/2106.14807) _arXiv:2106.14807_. +In these experiments, we are using cached queries (i.e., cached results of query encoding). + The exact configurations for these regressions are stored in [this YAML file](../../src/main/resources/regression/beir-v1.0.0-cqadupstack-gaming.unicoil-noexp.cached.yaml). Note that this page is automatically generated from [this template](../../src/main/resources/docgen/templates/beir-v1.0.0-cqadupstack-gaming.unicoil-noexp.cached.template) as part of Anserini's regression pipeline, so do not modify this page directly; modify the template instead. diff --git a/docs/regressions/regressions-beir-v1.0.0-cqadupstack-gis.splade-pp-ed.cached.md b/docs/regressions/regressions-beir-v1.0.0-cqadupstack-gis.splade-pp-ed.cached.md index e2b18f286c..2440942c39 100644 --- a/docs/regressions/regressions-beir-v1.0.0-cqadupstack-gis.splade-pp-ed.cached.md +++ b/docs/regressions/regressions-beir-v1.0.0-cqadupstack-gis.splade-pp-ed.cached.md @@ -1,14 +1,14 @@ # Anserini Regressions: BEIR (v1.0.0) — CQADupStack-gis -**Model**: [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) (using pre-encoded queries) +**Model**: [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) (using cached queries) -This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — CQADupStack-gis](http://beir.ai/). +This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — CQADupStack-gis](http://beir.ai/). The model itself can be download [here](https://huggingface.co/naver/splade-cocondenser-ensembledistil). See the [official SPLADE repo](/~https://github.com/naver/splade) and the following paper for more details: > Thibault Formal, Carlos Lassance, Benjamin Piwowarski, and Stéphane Clinchant. [From Distillation to Hard Negative Sampling: Making Sparse Neural IR Models More Effective.](https://dl.acm.org/doi/10.1145/3477495.3531857) _Proceedings of the 45th International ACM SIGIR Conference on Research and Development in Information Retrieval_, pages 2353–2359. -In these experiments, we are using pre-encoded queries (i.e., cached results of query encoding). +In these experiments, we are using cached queries (i.e., cached results of query encoding). The exact configurations for these regressions are stored in [this YAML file](../../src/main/resources/regression/beir-v1.0.0-cqadupstack-gis.splade-pp-ed.cached.yaml). Note that this page is automatically generated from [this template](../../src/main/resources/docgen/templates/beir-v1.0.0-cqadupstack-gis.splade-pp-ed.cached.template) as part of Anserini's regression pipeline, so do not modify this page directly; modify the template instead and then run `bin/build.sh` to rebuild the documentation. diff --git a/docs/regressions/regressions-beir-v1.0.0-cqadupstack-gis.splade-pp-ed.onnx.md b/docs/regressions/regressions-beir-v1.0.0-cqadupstack-gis.splade-pp-ed.onnx.md index c9de4a091c..6a805764d9 100644 --- a/docs/regressions/regressions-beir-v1.0.0-cqadupstack-gis.splade-pp-ed.onnx.md +++ b/docs/regressions/regressions-beir-v1.0.0-cqadupstack-gis.splade-pp-ed.onnx.md @@ -1,8 +1,8 @@ # Anserini Regressions: BEIR (v1.0.0) — CQADupStack-gis -**Model**: [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) (using ONNX for on-the-fly query encoding) +**Model**: [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) (using ONNX for on-the-fly query encoding) -This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — CQADupStack-gis](http://beir.ai/). +This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — CQADupStack-gis](http://beir.ai/). The model itself can be download [here](https://huggingface.co/naver/splade-cocondenser-ensembledistil). See the [official SPLADE repo](/~https://github.com/naver/splade) and the following paper for more details: diff --git a/docs/regressions/regressions-beir-v1.0.0-cqadupstack-gis.unicoil-noexp.cached.md b/docs/regressions/regressions-beir-v1.0.0-cqadupstack-gis.unicoil-noexp.cached.md index 09ec2040c3..9b66791631 100644 --- a/docs/regressions/regressions-beir-v1.0.0-cqadupstack-gis.unicoil-noexp.cached.md +++ b/docs/regressions/regressions-beir-v1.0.0-cqadupstack-gis.unicoil-noexp.cached.md @@ -1,12 +1,14 @@ # Anserini Regressions: BEIR (v1.0.0) — CQADupStack-gis -**Model**: uniCOIL (without any expansions) +**Model**: uniCOIL without any expansions (using cached queries) This page describes regression experiments, integrated into Anserini's regression testing framework, using uniCOIL (without any expansions) on [BEIR (v1.0.0) — CQADupStack-gis](http://beir.ai/). 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.](https://arxiv.org/abs/2106.14807) _arXiv:2106.14807_. +In these experiments, we are using cached queries (i.e., cached results of query encoding). + The exact configurations for these regressions are stored in [this YAML file](../../src/main/resources/regression/beir-v1.0.0-cqadupstack-gis.unicoil-noexp.cached.yaml). Note that this page is automatically generated from [this template](../../src/main/resources/docgen/templates/beir-v1.0.0-cqadupstack-gis.unicoil-noexp.cached.template) as part of Anserini's regression pipeline, so do not modify this page directly; modify the template instead. diff --git a/docs/regressions/regressions-beir-v1.0.0-cqadupstack-mathematica.splade-pp-ed.cached.md b/docs/regressions/regressions-beir-v1.0.0-cqadupstack-mathematica.splade-pp-ed.cached.md index 326701b38a..b05e6d50bd 100644 --- a/docs/regressions/regressions-beir-v1.0.0-cqadupstack-mathematica.splade-pp-ed.cached.md +++ b/docs/regressions/regressions-beir-v1.0.0-cqadupstack-mathematica.splade-pp-ed.cached.md @@ -1,14 +1,14 @@ # Anserini Regressions: BEIR (v1.0.0) — CQADupStack-mathematica -**Model**: [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) (using pre-encoded queries) +**Model**: [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) (using cached queries) -This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — CQADupStack-mathematica](http://beir.ai/). +This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — CQADupStack-mathematica](http://beir.ai/). The model itself can be download [here](https://huggingface.co/naver/splade-cocondenser-ensembledistil). See the [official SPLADE repo](/~https://github.com/naver/splade) and the following paper for more details: > Thibault Formal, Carlos Lassance, Benjamin Piwowarski, and Stéphane Clinchant. [From Distillation to Hard Negative Sampling: Making Sparse Neural IR Models More Effective.](https://dl.acm.org/doi/10.1145/3477495.3531857) _Proceedings of the 45th International ACM SIGIR Conference on Research and Development in Information Retrieval_, pages 2353–2359. -In these experiments, we are using pre-encoded queries (i.e., cached results of query encoding). +In these experiments, we are using cached queries (i.e., cached results of query encoding). The exact configurations for these regressions are stored in [this YAML file](../../src/main/resources/regression/beir-v1.0.0-cqadupstack-mathematica.splade-pp-ed.cached.yaml). Note that this page is automatically generated from [this template](../../src/main/resources/docgen/templates/beir-v1.0.0-cqadupstack-mathematica.splade-pp-ed.cached.template) as part of Anserini's regression pipeline, so do not modify this page directly; modify the template instead and then run `bin/build.sh` to rebuild the documentation. diff --git a/docs/regressions/regressions-beir-v1.0.0-cqadupstack-mathematica.splade-pp-ed.onnx.md b/docs/regressions/regressions-beir-v1.0.0-cqadupstack-mathematica.splade-pp-ed.onnx.md index dcdbf1f348..c626bffb50 100644 --- a/docs/regressions/regressions-beir-v1.0.0-cqadupstack-mathematica.splade-pp-ed.onnx.md +++ b/docs/regressions/regressions-beir-v1.0.0-cqadupstack-mathematica.splade-pp-ed.onnx.md @@ -1,8 +1,8 @@ # Anserini Regressions: BEIR (v1.0.0) — CQADupStack-mathematica -**Model**: [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) (using ONNX for on-the-fly query encoding) +**Model**: [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) (using ONNX for on-the-fly query encoding) -This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — CQADupStack-mathematica](http://beir.ai/). +This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — CQADupStack-mathematica](http://beir.ai/). The model itself can be download [here](https://huggingface.co/naver/splade-cocondenser-ensembledistil). See the [official SPLADE repo](/~https://github.com/naver/splade) and the following paper for more details: diff --git a/docs/regressions/regressions-beir-v1.0.0-cqadupstack-mathematica.unicoil-noexp.cached.md b/docs/regressions/regressions-beir-v1.0.0-cqadupstack-mathematica.unicoil-noexp.cached.md index ba65beb5a7..bdc84b71ce 100644 --- a/docs/regressions/regressions-beir-v1.0.0-cqadupstack-mathematica.unicoil-noexp.cached.md +++ b/docs/regressions/regressions-beir-v1.0.0-cqadupstack-mathematica.unicoil-noexp.cached.md @@ -1,12 +1,14 @@ # Anserini Regressions: BEIR (v1.0.0) — CQADupStack-mathematica -**Model**: uniCOIL (without any expansions) +**Model**: uniCOIL without any expansions (using cached queries) This page describes regression experiments, integrated into Anserini's regression testing framework, using uniCOIL (without any expansions) on [BEIR (v1.0.0) — CQADupStack-mathematica](http://beir.ai/). 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.](https://arxiv.org/abs/2106.14807) _arXiv:2106.14807_. +In these experiments, we are using cached queries (i.e., cached results of query encoding). + The exact configurations for these regressions are stored in [this YAML file](../../src/main/resources/regression/beir-v1.0.0-cqadupstack-mathematica.unicoil-noexp.cached.yaml). Note that this page is automatically generated from [this template](../../src/main/resources/docgen/templates/beir-v1.0.0-cqadupstack-mathematica.unicoil-noexp.cached.template) as part of Anserini's regression pipeline, so do not modify this page directly; modify the template instead. diff --git a/docs/regressions/regressions-beir-v1.0.0-cqadupstack-physics.splade-pp-ed.cached.md b/docs/regressions/regressions-beir-v1.0.0-cqadupstack-physics.splade-pp-ed.cached.md index 1c5b90ce07..f2c7cd6e96 100644 --- a/docs/regressions/regressions-beir-v1.0.0-cqadupstack-physics.splade-pp-ed.cached.md +++ b/docs/regressions/regressions-beir-v1.0.0-cqadupstack-physics.splade-pp-ed.cached.md @@ -1,14 +1,14 @@ # Anserini Regressions: BEIR (v1.0.0) — CQADupStack-physics -**Model**: [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) (using pre-encoded queries) +**Model**: [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) (using cached queries) -This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — CQADupStack-physics](http://beir.ai/). +This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — CQADupStack-physics](http://beir.ai/). The model itself can be download [here](https://huggingface.co/naver/splade-cocondenser-ensembledistil). See the [official SPLADE repo](/~https://github.com/naver/splade) and the following paper for more details: > Thibault Formal, Carlos Lassance, Benjamin Piwowarski, and Stéphane Clinchant. [From Distillation to Hard Negative Sampling: Making Sparse Neural IR Models More Effective.](https://dl.acm.org/doi/10.1145/3477495.3531857) _Proceedings of the 45th International ACM SIGIR Conference on Research and Development in Information Retrieval_, pages 2353–2359. -In these experiments, we are using pre-encoded queries (i.e., cached results of query encoding). +In these experiments, we are using cached queries (i.e., cached results of query encoding). The exact configurations for these regressions are stored in [this YAML file](../../src/main/resources/regression/beir-v1.0.0-cqadupstack-physics.splade-pp-ed.cached.yaml). Note that this page is automatically generated from [this template](../../src/main/resources/docgen/templates/beir-v1.0.0-cqadupstack-physics.splade-pp-ed.cached.template) as part of Anserini's regression pipeline, so do not modify this page directly; modify the template instead and then run `bin/build.sh` to rebuild the documentation. diff --git a/docs/regressions/regressions-beir-v1.0.0-cqadupstack-physics.splade-pp-ed.onnx.md b/docs/regressions/regressions-beir-v1.0.0-cqadupstack-physics.splade-pp-ed.onnx.md index 6a22df25ac..efbfde8d21 100644 --- a/docs/regressions/regressions-beir-v1.0.0-cqadupstack-physics.splade-pp-ed.onnx.md +++ b/docs/regressions/regressions-beir-v1.0.0-cqadupstack-physics.splade-pp-ed.onnx.md @@ -1,8 +1,8 @@ # Anserini Regressions: BEIR (v1.0.0) — CQADupStack-physics -**Model**: [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) (using ONNX for on-the-fly query encoding) +**Model**: [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) (using ONNX for on-the-fly query encoding) -This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — CQADupStack-physics](http://beir.ai/). +This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — CQADupStack-physics](http://beir.ai/). The model itself can be download [here](https://huggingface.co/naver/splade-cocondenser-ensembledistil). See the [official SPLADE repo](/~https://github.com/naver/splade) and the following paper for more details: diff --git a/docs/regressions/regressions-beir-v1.0.0-cqadupstack-physics.unicoil-noexp.cached.md b/docs/regressions/regressions-beir-v1.0.0-cqadupstack-physics.unicoil-noexp.cached.md index df5399cebf..a7d21e6dc9 100644 --- a/docs/regressions/regressions-beir-v1.0.0-cqadupstack-physics.unicoil-noexp.cached.md +++ b/docs/regressions/regressions-beir-v1.0.0-cqadupstack-physics.unicoil-noexp.cached.md @@ -1,12 +1,14 @@ # Anserini Regressions: BEIR (v1.0.0) — CQADupStack-physics -**Model**: uniCOIL (without any expansions) +**Model**: uniCOIL without any expansions (using cached queries) This page describes regression experiments, integrated into Anserini's regression testing framework, using uniCOIL (without any expansions) on [BEIR (v1.0.0) — CQADupStack-physics](http://beir.ai/). 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.](https://arxiv.org/abs/2106.14807) _arXiv:2106.14807_. +In these experiments, we are using cached queries (i.e., cached results of query encoding). + The exact configurations for these regressions are stored in [this YAML file](../../src/main/resources/regression/beir-v1.0.0-cqadupstack-physics.unicoil-noexp.cached.yaml). Note that this page is automatically generated from [this template](../../src/main/resources/docgen/templates/beir-v1.0.0-cqadupstack-physics.unicoil-noexp.cached.template) as part of Anserini's regression pipeline, so do not modify this page directly; modify the template instead. diff --git a/docs/regressions/regressions-beir-v1.0.0-cqadupstack-programmers.splade-pp-ed.cached.md b/docs/regressions/regressions-beir-v1.0.0-cqadupstack-programmers.splade-pp-ed.cached.md index 9dd215f579..d8ad28d089 100644 --- a/docs/regressions/regressions-beir-v1.0.0-cqadupstack-programmers.splade-pp-ed.cached.md +++ b/docs/regressions/regressions-beir-v1.0.0-cqadupstack-programmers.splade-pp-ed.cached.md @@ -1,14 +1,14 @@ # Anserini Regressions: BEIR (v1.0.0) — CQADupStack-programmers -**Model**: [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) (using pre-encoded queries) +**Model**: [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) (using cached queries) -This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — CQADupStack-programmers](http://beir.ai/). +This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — CQADupStack-programmers](http://beir.ai/). The model itself can be download [here](https://huggingface.co/naver/splade-cocondenser-ensembledistil). See the [official SPLADE repo](/~https://github.com/naver/splade) and the following paper for more details: > Thibault Formal, Carlos Lassance, Benjamin Piwowarski, and Stéphane Clinchant. [From Distillation to Hard Negative Sampling: Making Sparse Neural IR Models More Effective.](https://dl.acm.org/doi/10.1145/3477495.3531857) _Proceedings of the 45th International ACM SIGIR Conference on Research and Development in Information Retrieval_, pages 2353–2359. -In these experiments, we are using pre-encoded queries (i.e., cached results of query encoding). +In these experiments, we are using cached queries (i.e., cached results of query encoding). The exact configurations for these regressions are stored in [this YAML file](../../src/main/resources/regression/beir-v1.0.0-cqadupstack-programmers.splade-pp-ed.cached.yaml). Note that this page is automatically generated from [this template](../../src/main/resources/docgen/templates/beir-v1.0.0-cqadupstack-programmers.splade-pp-ed.cached.template) as part of Anserini's regression pipeline, so do not modify this page directly; modify the template instead and then run `bin/build.sh` to rebuild the documentation. diff --git a/docs/regressions/regressions-beir-v1.0.0-cqadupstack-programmers.splade-pp-ed.onnx.md b/docs/regressions/regressions-beir-v1.0.0-cqadupstack-programmers.splade-pp-ed.onnx.md index 4b114ccfe5..36679cc64b 100644 --- a/docs/regressions/regressions-beir-v1.0.0-cqadupstack-programmers.splade-pp-ed.onnx.md +++ b/docs/regressions/regressions-beir-v1.0.0-cqadupstack-programmers.splade-pp-ed.onnx.md @@ -1,8 +1,8 @@ # Anserini Regressions: BEIR (v1.0.0) — CQADupStack-programmers -**Model**: [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) (using ONNX for on-the-fly query encoding) +**Model**: [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) (using ONNX for on-the-fly query encoding) -This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — CQADupStack-programmers](http://beir.ai/). +This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — CQADupStack-programmers](http://beir.ai/). The model itself can be download [here](https://huggingface.co/naver/splade-cocondenser-ensembledistil). See the [official SPLADE repo](/~https://github.com/naver/splade) and the following paper for more details: diff --git a/docs/regressions/regressions-beir-v1.0.0-cqadupstack-programmers.unicoil-noexp.cached.md b/docs/regressions/regressions-beir-v1.0.0-cqadupstack-programmers.unicoil-noexp.cached.md index fa97165d43..28aeded250 100644 --- a/docs/regressions/regressions-beir-v1.0.0-cqadupstack-programmers.unicoil-noexp.cached.md +++ b/docs/regressions/regressions-beir-v1.0.0-cqadupstack-programmers.unicoil-noexp.cached.md @@ -1,12 +1,14 @@ # Anserini Regressions: BEIR (v1.0.0) — CQADupStack-programmers -**Model**: uniCOIL (without any expansions) +**Model**: uniCOIL without any expansions (using cached queries) This page describes regression experiments, integrated into Anserini's regression testing framework, using uniCOIL (without any expansions) on [BEIR (v1.0.0) — CQADupStack-programmers](http://beir.ai/). 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.](https://arxiv.org/abs/2106.14807) _arXiv:2106.14807_. +In these experiments, we are using cached queries (i.e., cached results of query encoding). + The exact configurations for these regressions are stored in [this YAML file](../../src/main/resources/regression/beir-v1.0.0-cqadupstack-programmers.unicoil-noexp.cached.yaml). Note that this page is automatically generated from [this template](../../src/main/resources/docgen/templates/beir-v1.0.0-cqadupstack-programmers.unicoil-noexp.cached.template) as part of Anserini's regression pipeline, so do not modify this page directly; modify the template instead. diff --git a/docs/regressions/regressions-beir-v1.0.0-cqadupstack-stats.splade-pp-ed.cached.md b/docs/regressions/regressions-beir-v1.0.0-cqadupstack-stats.splade-pp-ed.cached.md index ed30488f15..65c6991163 100644 --- a/docs/regressions/regressions-beir-v1.0.0-cqadupstack-stats.splade-pp-ed.cached.md +++ b/docs/regressions/regressions-beir-v1.0.0-cqadupstack-stats.splade-pp-ed.cached.md @@ -1,14 +1,14 @@ # Anserini Regressions: BEIR (v1.0.0) — CQADupStack-stats -**Model**: [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) (using pre-encoded queries) +**Model**: [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) (using cached queries) -This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — CQADupStack-stats](http://beir.ai/). +This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — CQADupStack-stats](http://beir.ai/). The model itself can be download [here](https://huggingface.co/naver/splade-cocondenser-ensembledistil). See the [official SPLADE repo](/~https://github.com/naver/splade) and the following paper for more details: > Thibault Formal, Carlos Lassance, Benjamin Piwowarski, and Stéphane Clinchant. [From Distillation to Hard Negative Sampling: Making Sparse Neural IR Models More Effective.](https://dl.acm.org/doi/10.1145/3477495.3531857) _Proceedings of the 45th International ACM SIGIR Conference on Research and Development in Information Retrieval_, pages 2353–2359. -In these experiments, we are using pre-encoded queries (i.e., cached results of query encoding). +In these experiments, we are using cached queries (i.e., cached results of query encoding). The exact configurations for these regressions are stored in [this YAML file](../../src/main/resources/regression/beir-v1.0.0-cqadupstack-stats.splade-pp-ed.cached.yaml). Note that this page is automatically generated from [this template](../../src/main/resources/docgen/templates/beir-v1.0.0-cqadupstack-stats.splade-pp-ed.cached.template) as part of Anserini's regression pipeline, so do not modify this page directly; modify the template instead and then run `bin/build.sh` to rebuild the documentation. diff --git a/docs/regressions/regressions-beir-v1.0.0-cqadupstack-stats.splade-pp-ed.onnx.md b/docs/regressions/regressions-beir-v1.0.0-cqadupstack-stats.splade-pp-ed.onnx.md index 46f605b064..46a47d9408 100644 --- a/docs/regressions/regressions-beir-v1.0.0-cqadupstack-stats.splade-pp-ed.onnx.md +++ b/docs/regressions/regressions-beir-v1.0.0-cqadupstack-stats.splade-pp-ed.onnx.md @@ -1,8 +1,8 @@ # Anserini Regressions: BEIR (v1.0.0) — CQADupStack-stats -**Model**: [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) (using ONNX for on-the-fly query encoding) +**Model**: [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) (using ONNX for on-the-fly query encoding) -This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — CQADupStack-stats](http://beir.ai/). +This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — CQADupStack-stats](http://beir.ai/). The model itself can be download [here](https://huggingface.co/naver/splade-cocondenser-ensembledistil). See the [official SPLADE repo](/~https://github.com/naver/splade) and the following paper for more details: diff --git a/docs/regressions/regressions-beir-v1.0.0-cqadupstack-stats.unicoil-noexp.cached.md b/docs/regressions/regressions-beir-v1.0.0-cqadupstack-stats.unicoil-noexp.cached.md index 41914da227..11c0915a6b 100644 --- a/docs/regressions/regressions-beir-v1.0.0-cqadupstack-stats.unicoil-noexp.cached.md +++ b/docs/regressions/regressions-beir-v1.0.0-cqadupstack-stats.unicoil-noexp.cached.md @@ -1,12 +1,14 @@ # Anserini Regressions: BEIR (v1.0.0) — CQADupStack-stats -**Model**: uniCOIL (without any expansions) +**Model**: uniCOIL without any expansions (using cached queries) This page describes regression experiments, integrated into Anserini's regression testing framework, using uniCOIL (without any expansions) on [BEIR (v1.0.0) — CQADupStack-stats](http://beir.ai/). 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.](https://arxiv.org/abs/2106.14807) _arXiv:2106.14807_. +In these experiments, we are using cached queries (i.e., cached results of query encoding). + The exact configurations for these regressions are stored in [this YAML file](../../src/main/resources/regression/beir-v1.0.0-cqadupstack-stats.unicoil-noexp.cached.yaml). Note that this page is automatically generated from [this template](../../src/main/resources/docgen/templates/beir-v1.0.0-cqadupstack-stats.unicoil-noexp.cached.template) as part of Anserini's regression pipeline, so do not modify this page directly; modify the template instead. diff --git a/docs/regressions/regressions-beir-v1.0.0-cqadupstack-tex.splade-pp-ed.cached.md b/docs/regressions/regressions-beir-v1.0.0-cqadupstack-tex.splade-pp-ed.cached.md index 10d13511d1..666c46f3da 100644 --- a/docs/regressions/regressions-beir-v1.0.0-cqadupstack-tex.splade-pp-ed.cached.md +++ b/docs/regressions/regressions-beir-v1.0.0-cqadupstack-tex.splade-pp-ed.cached.md @@ -1,14 +1,14 @@ # Anserini Regressions: BEIR (v1.0.0) — CQADupStack-tex -**Model**: [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) (using pre-encoded queries) +**Model**: [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) (using cached queries) -This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — CQADupStack-tex](http://beir.ai/). +This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — CQADupStack-tex](http://beir.ai/). The model itself can be download [here](https://huggingface.co/naver/splade-cocondenser-ensembledistil). See the [official SPLADE repo](/~https://github.com/naver/splade) and the following paper for more details: > Thibault Formal, Carlos Lassance, Benjamin Piwowarski, and Stéphane Clinchant. [From Distillation to Hard Negative Sampling: Making Sparse Neural IR Models More Effective.](https://dl.acm.org/doi/10.1145/3477495.3531857) _Proceedings of the 45th International ACM SIGIR Conference on Research and Development in Information Retrieval_, pages 2353–2359. -In these experiments, we are using pre-encoded queries (i.e., cached results of query encoding). +In these experiments, we are using cached queries (i.e., cached results of query encoding). The exact configurations for these regressions are stored in [this YAML file](../../src/main/resources/regression/beir-v1.0.0-cqadupstack-tex.splade-pp-ed.cached.yaml). Note that this page is automatically generated from [this template](../../src/main/resources/docgen/templates/beir-v1.0.0-cqadupstack-tex.splade-pp-ed.cached.template) as part of Anserini's regression pipeline, so do not modify this page directly; modify the template instead and then run `bin/build.sh` to rebuild the documentation. diff --git a/docs/regressions/regressions-beir-v1.0.0-cqadupstack-tex.splade-pp-ed.onnx.md b/docs/regressions/regressions-beir-v1.0.0-cqadupstack-tex.splade-pp-ed.onnx.md index b22cb73b5c..bee68234b7 100644 --- a/docs/regressions/regressions-beir-v1.0.0-cqadupstack-tex.splade-pp-ed.onnx.md +++ b/docs/regressions/regressions-beir-v1.0.0-cqadupstack-tex.splade-pp-ed.onnx.md @@ -1,8 +1,8 @@ # Anserini Regressions: BEIR (v1.0.0) — CQADupStack-tex -**Model**: [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) (using ONNX for on-the-fly query encoding) +**Model**: [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) (using ONNX for on-the-fly query encoding) -This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — CQADupStack-tex](http://beir.ai/). +This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — CQADupStack-tex](http://beir.ai/). The model itself can be download [here](https://huggingface.co/naver/splade-cocondenser-ensembledistil). See the [official SPLADE repo](/~https://github.com/naver/splade) and the following paper for more details: diff --git a/docs/regressions/regressions-beir-v1.0.0-cqadupstack-tex.unicoil-noexp.cached.md b/docs/regressions/regressions-beir-v1.0.0-cqadupstack-tex.unicoil-noexp.cached.md index 85e0c4b546..9032ba93cb 100644 --- a/docs/regressions/regressions-beir-v1.0.0-cqadupstack-tex.unicoil-noexp.cached.md +++ b/docs/regressions/regressions-beir-v1.0.0-cqadupstack-tex.unicoil-noexp.cached.md @@ -1,12 +1,14 @@ # Anserini Regressions: BEIR (v1.0.0) — CQADupStack-tex -**Model**: uniCOIL (without any expansions) +**Model**: uniCOIL without any expansions (using cached queries) This page describes regression experiments, integrated into Anserini's regression testing framework, using uniCOIL (without any expansions) on [BEIR (v1.0.0) — CQADupStack-tex](http://beir.ai/). 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.](https://arxiv.org/abs/2106.14807) _arXiv:2106.14807_. +In these experiments, we are using cached queries (i.e., cached results of query encoding). + The exact configurations for these regressions are stored in [this YAML file](../../src/main/resources/regression/beir-v1.0.0-cqadupstack-tex.unicoil-noexp.cached.yaml). Note that this page is automatically generated from [this template](../../src/main/resources/docgen/templates/beir-v1.0.0-cqadupstack-tex.unicoil-noexp.cached.template) as part of Anserini's regression pipeline, so do not modify this page directly; modify the template instead. diff --git a/docs/regressions/regressions-beir-v1.0.0-cqadupstack-unix.splade-pp-ed.cached.md b/docs/regressions/regressions-beir-v1.0.0-cqadupstack-unix.splade-pp-ed.cached.md index ab6a7fe0c4..631d9bfe08 100644 --- a/docs/regressions/regressions-beir-v1.0.0-cqadupstack-unix.splade-pp-ed.cached.md +++ b/docs/regressions/regressions-beir-v1.0.0-cqadupstack-unix.splade-pp-ed.cached.md @@ -1,14 +1,14 @@ # Anserini Regressions: BEIR (v1.0.0) — CQADupStack-unix -**Model**: [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) (using pre-encoded queries) +**Model**: [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) (using cached queries) -This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — CQADupStack-unix](http://beir.ai/). +This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — CQADupStack-unix](http://beir.ai/). The model itself can be download [here](https://huggingface.co/naver/splade-cocondenser-ensembledistil). See the [official SPLADE repo](/~https://github.com/naver/splade) and the following paper for more details: > Thibault Formal, Carlos Lassance, Benjamin Piwowarski, and Stéphane Clinchant. [From Distillation to Hard Negative Sampling: Making Sparse Neural IR Models More Effective.](https://dl.acm.org/doi/10.1145/3477495.3531857) _Proceedings of the 45th International ACM SIGIR Conference on Research and Development in Information Retrieval_, pages 2353–2359. -In these experiments, we are using pre-encoded queries (i.e., cached results of query encoding). +In these experiments, we are using cached queries (i.e., cached results of query encoding). The exact configurations for these regressions are stored in [this YAML file](../../src/main/resources/regression/beir-v1.0.0-cqadupstack-unix.splade-pp-ed.cached.yaml). Note that this page is automatically generated from [this template](../../src/main/resources/docgen/templates/beir-v1.0.0-cqadupstack-unix.splade-pp-ed.cached.template) as part of Anserini's regression pipeline, so do not modify this page directly; modify the template instead and then run `bin/build.sh` to rebuild the documentation. diff --git a/docs/regressions/regressions-beir-v1.0.0-cqadupstack-unix.splade-pp-ed.onnx.md b/docs/regressions/regressions-beir-v1.0.0-cqadupstack-unix.splade-pp-ed.onnx.md index 6b808210fe..5f4b7c4bd5 100644 --- a/docs/regressions/regressions-beir-v1.0.0-cqadupstack-unix.splade-pp-ed.onnx.md +++ b/docs/regressions/regressions-beir-v1.0.0-cqadupstack-unix.splade-pp-ed.onnx.md @@ -1,8 +1,8 @@ # Anserini Regressions: BEIR (v1.0.0) — CQADupStack-unix -**Model**: [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) (using ONNX for on-the-fly query encoding) +**Model**: [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) (using ONNX for on-the-fly query encoding) -This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — CQADupStack-unix](http://beir.ai/). +This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — CQADupStack-unix](http://beir.ai/). The model itself can be download [here](https://huggingface.co/naver/splade-cocondenser-ensembledistil). See the [official SPLADE repo](/~https://github.com/naver/splade) and the following paper for more details: diff --git a/docs/regressions/regressions-beir-v1.0.0-cqadupstack-unix.unicoil-noexp.cached.md b/docs/regressions/regressions-beir-v1.0.0-cqadupstack-unix.unicoil-noexp.cached.md index e85d39c82d..070328eada 100644 --- a/docs/regressions/regressions-beir-v1.0.0-cqadupstack-unix.unicoil-noexp.cached.md +++ b/docs/regressions/regressions-beir-v1.0.0-cqadupstack-unix.unicoil-noexp.cached.md @@ -1,12 +1,14 @@ # Anserini Regressions: BEIR (v1.0.0) — CQADupStack-unix -**Model**: uniCOIL (without any expansions) +**Model**: uniCOIL without any expansions (using cached queries) This page describes regression experiments, integrated into Anserini's regression testing framework, using uniCOIL (without any expansions) on [BEIR (v1.0.0) — CQADupStack-unix](http://beir.ai/). 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.](https://arxiv.org/abs/2106.14807) _arXiv:2106.14807_. +In these experiments, we are using cached queries (i.e., cached results of query encoding). + The exact configurations for these regressions are stored in [this YAML file](../../src/main/resources/regression/beir-v1.0.0-cqadupstack-unix.unicoil-noexp.cached.yaml). Note that this page is automatically generated from [this template](../../src/main/resources/docgen/templates/beir-v1.0.0-cqadupstack-unix.unicoil-noexp.cached.template) as part of Anserini's regression pipeline, so do not modify this page directly; modify the template instead. diff --git a/docs/regressions/regressions-beir-v1.0.0-cqadupstack-webmasters.splade-pp-ed.cached.md b/docs/regressions/regressions-beir-v1.0.0-cqadupstack-webmasters.splade-pp-ed.cached.md index 420f2fc6f4..8f61195b99 100644 --- a/docs/regressions/regressions-beir-v1.0.0-cqadupstack-webmasters.splade-pp-ed.cached.md +++ b/docs/regressions/regressions-beir-v1.0.0-cqadupstack-webmasters.splade-pp-ed.cached.md @@ -1,14 +1,14 @@ # Anserini Regressions: BEIR (v1.0.0) — CQADupStack-webmasters -**Model**: [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) (using pre-encoded queries) +**Model**: [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) (using cached queries) -This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — CQADupStack-webmasters](http://beir.ai/). +This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — CQADupStack-webmasters](http://beir.ai/). The model itself can be download [here](https://huggingface.co/naver/splade-cocondenser-ensembledistil). See the [official SPLADE repo](/~https://github.com/naver/splade) and the following paper for more details: > Thibault Formal, Carlos Lassance, Benjamin Piwowarski, and Stéphane Clinchant. [From Distillation to Hard Negative Sampling: Making Sparse Neural IR Models More Effective.](https://dl.acm.org/doi/10.1145/3477495.3531857) _Proceedings of the 45th International ACM SIGIR Conference on Research and Development in Information Retrieval_, pages 2353–2359. -In these experiments, we are using pre-encoded queries (i.e., cached results of query encoding). +In these experiments, we are using cached queries (i.e., cached results of query encoding). The exact configurations for these regressions are stored in [this YAML file](../../src/main/resources/regression/beir-v1.0.0-cqadupstack-webmasters.splade-pp-ed.cached.yaml). Note that this page is automatically generated from [this template](../../src/main/resources/docgen/templates/beir-v1.0.0-cqadupstack-webmasters.splade-pp-ed.cached.template) as part of Anserini's regression pipeline, so do not modify this page directly; modify the template instead and then run `bin/build.sh` to rebuild the documentation. diff --git a/docs/regressions/regressions-beir-v1.0.0-cqadupstack-webmasters.splade-pp-ed.onnx.md b/docs/regressions/regressions-beir-v1.0.0-cqadupstack-webmasters.splade-pp-ed.onnx.md index fa7827ceb7..e111240324 100644 --- a/docs/regressions/regressions-beir-v1.0.0-cqadupstack-webmasters.splade-pp-ed.onnx.md +++ b/docs/regressions/regressions-beir-v1.0.0-cqadupstack-webmasters.splade-pp-ed.onnx.md @@ -1,8 +1,8 @@ # Anserini Regressions: BEIR (v1.0.0) — CQADupStack-webmasters -**Model**: [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) (using ONNX for on-the-fly query encoding) +**Model**: [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) (using ONNX for on-the-fly query encoding) -This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — CQADupStack-webmasters](http://beir.ai/). +This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — CQADupStack-webmasters](http://beir.ai/). The model itself can be download [here](https://huggingface.co/naver/splade-cocondenser-ensembledistil). See the [official SPLADE repo](/~https://github.com/naver/splade) and the following paper for more details: diff --git a/docs/regressions/regressions-beir-v1.0.0-cqadupstack-webmasters.unicoil-noexp.cached.md b/docs/regressions/regressions-beir-v1.0.0-cqadupstack-webmasters.unicoil-noexp.cached.md index 277ec41bf8..4fd3735576 100644 --- a/docs/regressions/regressions-beir-v1.0.0-cqadupstack-webmasters.unicoil-noexp.cached.md +++ b/docs/regressions/regressions-beir-v1.0.0-cqadupstack-webmasters.unicoil-noexp.cached.md @@ -1,12 +1,14 @@ # Anserini Regressions: BEIR (v1.0.0) — CQADupStack-webmasters -**Model**: uniCOIL (without any expansions) +**Model**: uniCOIL without any expansions (using cached queries) This page describes regression experiments, integrated into Anserini's regression testing framework, using uniCOIL (without any expansions) on [BEIR (v1.0.0) — CQADupStack-webmasters](http://beir.ai/). 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.](https://arxiv.org/abs/2106.14807) _arXiv:2106.14807_. +In these experiments, we are using cached queries (i.e., cached results of query encoding). + The exact configurations for these regressions are stored in [this YAML file](../../src/main/resources/regression/beir-v1.0.0-cqadupstack-webmasters.unicoil-noexp.cached.yaml). Note that this page is automatically generated from [this template](../../src/main/resources/docgen/templates/beir-v1.0.0-cqadupstack-webmasters.unicoil-noexp.cached.template) as part of Anserini's regression pipeline, so do not modify this page directly; modify the template instead. diff --git a/docs/regressions/regressions-beir-v1.0.0-cqadupstack-wordpress.splade-pp-ed.cached.md b/docs/regressions/regressions-beir-v1.0.0-cqadupstack-wordpress.splade-pp-ed.cached.md index 412d19556a..54cc8950d8 100644 --- a/docs/regressions/regressions-beir-v1.0.0-cqadupstack-wordpress.splade-pp-ed.cached.md +++ b/docs/regressions/regressions-beir-v1.0.0-cqadupstack-wordpress.splade-pp-ed.cached.md @@ -1,14 +1,14 @@ # Anserini Regressions: BEIR (v1.0.0) — CQADupStack-wordpress -**Model**: [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) (using pre-encoded queries) +**Model**: [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) (using cached queries) -This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — CQADupStack-wordpress](http://beir.ai/). +This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — CQADupStack-wordpress](http://beir.ai/). The model itself can be download [here](https://huggingface.co/naver/splade-cocondenser-ensembledistil). See the [official SPLADE repo](/~https://github.com/naver/splade) and the following paper for more details: > Thibault Formal, Carlos Lassance, Benjamin Piwowarski, and Stéphane Clinchant. [From Distillation to Hard Negative Sampling: Making Sparse Neural IR Models More Effective.](https://dl.acm.org/doi/10.1145/3477495.3531857) _Proceedings of the 45th International ACM SIGIR Conference on Research and Development in Information Retrieval_, pages 2353–2359. -In these experiments, we are using pre-encoded queries (i.e., cached results of query encoding). +In these experiments, we are using cached queries (i.e., cached results of query encoding). The exact configurations for these regressions are stored in [this YAML file](../../src/main/resources/regression/beir-v1.0.0-cqadupstack-wordpress.splade-pp-ed.cached.yaml). Note that this page is automatically generated from [this template](../../src/main/resources/docgen/templates/beir-v1.0.0-cqadupstack-wordpress.splade-pp-ed.cached.template) as part of Anserini's regression pipeline, so do not modify this page directly; modify the template instead and then run `bin/build.sh` to rebuild the documentation. diff --git a/docs/regressions/regressions-beir-v1.0.0-cqadupstack-wordpress.splade-pp-ed.onnx.md b/docs/regressions/regressions-beir-v1.0.0-cqadupstack-wordpress.splade-pp-ed.onnx.md index 6720b1f7ce..2708e46abb 100644 --- a/docs/regressions/regressions-beir-v1.0.0-cqadupstack-wordpress.splade-pp-ed.onnx.md +++ b/docs/regressions/regressions-beir-v1.0.0-cqadupstack-wordpress.splade-pp-ed.onnx.md @@ -1,8 +1,8 @@ # Anserini Regressions: BEIR (v1.0.0) — CQADupStack-wordpress -**Model**: [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) (using ONNX for on-the-fly query encoding) +**Model**: [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) (using ONNX for on-the-fly query encoding) -This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — CQADupStack-wordpress](http://beir.ai/). +This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — CQADupStack-wordpress](http://beir.ai/). The model itself can be download [here](https://huggingface.co/naver/splade-cocondenser-ensembledistil). See the [official SPLADE repo](/~https://github.com/naver/splade) and the following paper for more details: diff --git a/docs/regressions/regressions-beir-v1.0.0-cqadupstack-wordpress.unicoil-noexp.cached.md b/docs/regressions/regressions-beir-v1.0.0-cqadupstack-wordpress.unicoil-noexp.cached.md index ae652933da..f5d3d90903 100644 --- a/docs/regressions/regressions-beir-v1.0.0-cqadupstack-wordpress.unicoil-noexp.cached.md +++ b/docs/regressions/regressions-beir-v1.0.0-cqadupstack-wordpress.unicoil-noexp.cached.md @@ -1,12 +1,14 @@ # Anserini Regressions: BEIR (v1.0.0) — CQADupStack-wordpress -**Model**: uniCOIL (without any expansions) +**Model**: uniCOIL without any expansions (using cached queries) This page describes regression experiments, integrated into Anserini's regression testing framework, using uniCOIL (without any expansions) on [BEIR (v1.0.0) — CQADupStack-wordpress](http://beir.ai/). 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.](https://arxiv.org/abs/2106.14807) _arXiv:2106.14807_. +In these experiments, we are using cached queries (i.e., cached results of query encoding). + The exact configurations for these regressions are stored in [this YAML file](../../src/main/resources/regression/beir-v1.0.0-cqadupstack-wordpress.unicoil-noexp.cached.yaml). Note that this page is automatically generated from [this template](../../src/main/resources/docgen/templates/beir-v1.0.0-cqadupstack-wordpress.unicoil-noexp.cached.template) as part of Anserini's regression pipeline, so do not modify this page directly; modify the template instead. diff --git a/docs/regressions/regressions-beir-v1.0.0-dbpedia-entity.splade-pp-ed.cached.md b/docs/regressions/regressions-beir-v1.0.0-dbpedia-entity.splade-pp-ed.cached.md index 6bc70a9e2e..16db4c01c1 100644 --- a/docs/regressions/regressions-beir-v1.0.0-dbpedia-entity.splade-pp-ed.cached.md +++ b/docs/regressions/regressions-beir-v1.0.0-dbpedia-entity.splade-pp-ed.cached.md @@ -1,14 +1,14 @@ # Anserini Regressions: BEIR (v1.0.0) — DBPedia -**Model**: [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) (using pre-encoded queries) +**Model**: [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) (using cached queries) -This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — DBPedia](http://beir.ai/). +This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — DBPedia](http://beir.ai/). The model itself can be download [here](https://huggingface.co/naver/splade-cocondenser-ensembledistil). See the [official SPLADE repo](/~https://github.com/naver/splade) and the following paper for more details: > Thibault Formal, Carlos Lassance, Benjamin Piwowarski, and Stéphane Clinchant. [From Distillation to Hard Negative Sampling: Making Sparse Neural IR Models More Effective.](https://dl.acm.org/doi/10.1145/3477495.3531857) _Proceedings of the 45th International ACM SIGIR Conference on Research and Development in Information Retrieval_, pages 2353–2359. -In these experiments, we are using pre-encoded queries (i.e., cached results of query encoding). +In these experiments, we are using cached queries (i.e., cached results of query encoding). The exact configurations for these regressions are stored in [this YAML file](../../src/main/resources/regression/beir-v1.0.0-dbpedia-entity.splade-pp-ed.cached.yaml). Note that this page is automatically generated from [this template](../../src/main/resources/docgen/templates/beir-v1.0.0-dbpedia-entity.splade-pp-ed.cached.template) as part of Anserini's regression pipeline, so do not modify this page directly; modify the template instead and then run `bin/build.sh` to rebuild the documentation. diff --git a/docs/regressions/regressions-beir-v1.0.0-dbpedia-entity.splade-pp-ed.onnx.md b/docs/regressions/regressions-beir-v1.0.0-dbpedia-entity.splade-pp-ed.onnx.md index b9aa13f10c..f8d4e50a85 100644 --- a/docs/regressions/regressions-beir-v1.0.0-dbpedia-entity.splade-pp-ed.onnx.md +++ b/docs/regressions/regressions-beir-v1.0.0-dbpedia-entity.splade-pp-ed.onnx.md @@ -1,8 +1,8 @@ # Anserini Regressions: BEIR (v1.0.0) — DBPedia -**Model**: [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) (using ONNX for on-the-fly query encoding) +**Model**: [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) (using ONNX for on-the-fly query encoding) -This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — DBPedia](http://beir.ai/). +This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — DBPedia](http://beir.ai/). The model itself can be download [here](https://huggingface.co/naver/splade-cocondenser-ensembledistil). See the [official SPLADE repo](/~https://github.com/naver/splade) and the following paper for more details: diff --git a/docs/regressions/regressions-beir-v1.0.0-dbpedia-entity.unicoil-noexp.cached.md b/docs/regressions/regressions-beir-v1.0.0-dbpedia-entity.unicoil-noexp.cached.md index 3bf793dd50..f57eaf5cc7 100644 --- a/docs/regressions/regressions-beir-v1.0.0-dbpedia-entity.unicoil-noexp.cached.md +++ b/docs/regressions/regressions-beir-v1.0.0-dbpedia-entity.unicoil-noexp.cached.md @@ -1,12 +1,14 @@ # Anserini Regressions: BEIR (v1.0.0) — DBPedia -**Model**: uniCOIL (without any expansions) +**Model**: uniCOIL without any expansions (using cached queries) This page describes regression experiments, integrated into Anserini's regression testing framework, using uniCOIL (without any expansions) on [BEIR (v1.0.0) — DBPedia](http://beir.ai/). 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.](https://arxiv.org/abs/2106.14807) _arXiv:2106.14807_. +In these experiments, we are using cached queries (i.e., cached results of query encoding). + The exact configurations for these regressions are stored in [this YAML file](../../src/main/resources/regression/beir-v1.0.0-dbpedia-entity.unicoil-noexp.cached.yaml). Note that this page is automatically generated from [this template](../../src/main/resources/docgen/templates/beir-v1.0.0-dbpedia-entity.unicoil-noexp.cached.template) as part of Anserini's regression pipeline, so do not modify this page directly; modify the template instead. diff --git a/docs/regressions/regressions-beir-v1.0.0-fever.splade-pp-ed.cached.md b/docs/regressions/regressions-beir-v1.0.0-fever.splade-pp-ed.cached.md index dd76c9efba..bd6f863bc3 100644 --- a/docs/regressions/regressions-beir-v1.0.0-fever.splade-pp-ed.cached.md +++ b/docs/regressions/regressions-beir-v1.0.0-fever.splade-pp-ed.cached.md @@ -1,14 +1,14 @@ # Anserini Regressions: BEIR (v1.0.0) — FEVER -**Model**: [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) (using pre-encoded queries) +**Model**: [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) (using cached queries) -This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — FEVER](http://beir.ai/). +This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — FEVER](http://beir.ai/). The model itself can be download [here](https://huggingface.co/naver/splade-cocondenser-ensembledistil). See the [official SPLADE repo](/~https://github.com/naver/splade) and the following paper for more details: > Thibault Formal, Carlos Lassance, Benjamin Piwowarski, and Stéphane Clinchant. [From Distillation to Hard Negative Sampling: Making Sparse Neural IR Models More Effective.](https://dl.acm.org/doi/10.1145/3477495.3531857) _Proceedings of the 45th International ACM SIGIR Conference on Research and Development in Information Retrieval_, pages 2353–2359. -In these experiments, we are using pre-encoded queries (i.e., cached results of query encoding). +In these experiments, we are using cached queries (i.e., cached results of query encoding). The exact configurations for these regressions are stored in [this YAML file](../../src/main/resources/regression/beir-v1.0.0-fever.splade-pp-ed.cached.yaml). Note that this page is automatically generated from [this template](../../src/main/resources/docgen/templates/beir-v1.0.0-fever.splade-pp-ed.cached.template) as part of Anserini's regression pipeline, so do not modify this page directly; modify the template instead and then run `bin/build.sh` to rebuild the documentation. diff --git a/docs/regressions/regressions-beir-v1.0.0-fever.splade-pp-ed.onnx.md b/docs/regressions/regressions-beir-v1.0.0-fever.splade-pp-ed.onnx.md index 0f840705a1..5161c56246 100644 --- a/docs/regressions/regressions-beir-v1.0.0-fever.splade-pp-ed.onnx.md +++ b/docs/regressions/regressions-beir-v1.0.0-fever.splade-pp-ed.onnx.md @@ -1,8 +1,8 @@ # Anserini Regressions: BEIR (v1.0.0) — FEVER -**Model**: [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) (using ONNX for on-the-fly query encoding) +**Model**: [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) (using ONNX for on-the-fly query encoding) -This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — FEVER](http://beir.ai/). +This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — FEVER](http://beir.ai/). The model itself can be download [here](https://huggingface.co/naver/splade-cocondenser-ensembledistil). See the [official SPLADE repo](/~https://github.com/naver/splade) and the following paper for more details: diff --git a/docs/regressions/regressions-beir-v1.0.0-fever.unicoil-noexp.cached.md b/docs/regressions/regressions-beir-v1.0.0-fever.unicoil-noexp.cached.md index 69a38b4342..831af08a57 100644 --- a/docs/regressions/regressions-beir-v1.0.0-fever.unicoil-noexp.cached.md +++ b/docs/regressions/regressions-beir-v1.0.0-fever.unicoil-noexp.cached.md @@ -1,12 +1,14 @@ # Anserini Regressions: BEIR (v1.0.0) — FEVER -**Model**: uniCOIL (without any expansions) +**Model**: uniCOIL without any expansions (using cached queries) This page describes regression experiments, integrated into Anserini's regression testing framework, using uniCOIL (without any expansions) on [BEIR (v1.0.0) — FEVER](http://beir.ai/). 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.](https://arxiv.org/abs/2106.14807) _arXiv:2106.14807_. +In these experiments, we are using cached queries (i.e., cached results of query encoding). + The exact configurations for these regressions are stored in [this YAML file](../../src/main/resources/regression/beir-v1.0.0-fever.unicoil-noexp.cached.yaml). Note that this page is automatically generated from [this template](../../src/main/resources/docgen/templates/beir-v1.0.0-fever.unicoil-noexp.cached.template) as part of Anserini's regression pipeline, so do not modify this page directly; modify the template instead. diff --git a/docs/regressions/regressions-beir-v1.0.0-fiqa.splade-pp-ed.cached.md b/docs/regressions/regressions-beir-v1.0.0-fiqa.splade-pp-ed.cached.md index ce4553ad77..90acedb3e4 100644 --- a/docs/regressions/regressions-beir-v1.0.0-fiqa.splade-pp-ed.cached.md +++ b/docs/regressions/regressions-beir-v1.0.0-fiqa.splade-pp-ed.cached.md @@ -1,14 +1,14 @@ # Anserini Regressions: BEIR (v1.0.0) — FiQA-2018 -**Model**: [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) (using pre-encoded queries) +**Model**: [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) (using cached queries) -This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — FiQA-2018](http://beir.ai/). +This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — FiQA-2018](http://beir.ai/). The model itself can be download [here](https://huggingface.co/naver/splade-cocondenser-ensembledistil). See the [official SPLADE repo](/~https://github.com/naver/splade) and the following paper for more details: > Thibault Formal, Carlos Lassance, Benjamin Piwowarski, and Stéphane Clinchant. [From Distillation to Hard Negative Sampling: Making Sparse Neural IR Models More Effective.](https://dl.acm.org/doi/10.1145/3477495.3531857) _Proceedings of the 45th International ACM SIGIR Conference on Research and Development in Information Retrieval_, pages 2353–2359. -In these experiments, we are using pre-encoded queries (i.e., cached results of query encoding). +In these experiments, we are using cached queries (i.e., cached results of query encoding). The exact configurations for these regressions are stored in [this YAML file](../../src/main/resources/regression/beir-v1.0.0-fiqa.splade-pp-ed.cached.yaml). Note that this page is automatically generated from [this template](../../src/main/resources/docgen/templates/beir-v1.0.0-fiqa.splade-pp-ed.cached.template) as part of Anserini's regression pipeline, so do not modify this page directly; modify the template instead and then run `bin/build.sh` to rebuild the documentation. diff --git a/docs/regressions/regressions-beir-v1.0.0-fiqa.splade-pp-ed.onnx.md b/docs/regressions/regressions-beir-v1.0.0-fiqa.splade-pp-ed.onnx.md index 7422b96233..73b8a0cc93 100644 --- a/docs/regressions/regressions-beir-v1.0.0-fiqa.splade-pp-ed.onnx.md +++ b/docs/regressions/regressions-beir-v1.0.0-fiqa.splade-pp-ed.onnx.md @@ -1,8 +1,8 @@ # Anserini Regressions: BEIR (v1.0.0) — FiQA-2018 -**Model**: [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) (using ONNX for on-the-fly query encoding) +**Model**: [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) (using ONNX for on-the-fly query encoding) -This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — FiQA-2018](http://beir.ai/). +This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — FiQA-2018](http://beir.ai/). The model itself can be download [here](https://huggingface.co/naver/splade-cocondenser-ensembledistil). See the [official SPLADE repo](/~https://github.com/naver/splade) and the following paper for more details: diff --git a/docs/regressions/regressions-beir-v1.0.0-fiqa.unicoil-noexp.cached.md b/docs/regressions/regressions-beir-v1.0.0-fiqa.unicoil-noexp.cached.md index ba6ad91535..16764262b6 100644 --- a/docs/regressions/regressions-beir-v1.0.0-fiqa.unicoil-noexp.cached.md +++ b/docs/regressions/regressions-beir-v1.0.0-fiqa.unicoil-noexp.cached.md @@ -1,12 +1,14 @@ # Anserini Regressions: BEIR (v1.0.0) — FiQA-2018 -**Model**: uniCOIL (without any expansions) +**Model**: uniCOIL without any expansions (using cached queries) This page describes regression experiments, integrated into Anserini's regression testing framework, using uniCOIL (without any expansions) on [BEIR (v1.0.0) — FiQA-2018](http://beir.ai/). 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.](https://arxiv.org/abs/2106.14807) _arXiv:2106.14807_. +In these experiments, we are using cached queries (i.e., cached results of query encoding). + The exact configurations for these regressions are stored in [this YAML file](../../src/main/resources/regression/beir-v1.0.0-fiqa.unicoil-noexp.cached.yaml). Note that this page is automatically generated from [this template](../../src/main/resources/docgen/templates/beir-v1.0.0-fiqa.unicoil-noexp.cached.template) as part of Anserini's regression pipeline, so do not modify this page directly; modify the template instead. diff --git a/docs/regressions/regressions-beir-v1.0.0-hotpotqa.splade-pp-ed.cached.md b/docs/regressions/regressions-beir-v1.0.0-hotpotqa.splade-pp-ed.cached.md index 9945aa60af..f93ba9ad5e 100644 --- a/docs/regressions/regressions-beir-v1.0.0-hotpotqa.splade-pp-ed.cached.md +++ b/docs/regressions/regressions-beir-v1.0.0-hotpotqa.splade-pp-ed.cached.md @@ -1,14 +1,14 @@ # Anserini Regressions: BEIR (v1.0.0) — HotpotQA -**Model**: [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) (using pre-encoded queries) +**Model**: [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) (using cached queries) -This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — HotpotQA](http://beir.ai/). +This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — HotpotQA](http://beir.ai/). The model itself can be download [here](https://huggingface.co/naver/splade-cocondenser-ensembledistil). See the [official SPLADE repo](/~https://github.com/naver/splade) and the following paper for more details: > Thibault Formal, Carlos Lassance, Benjamin Piwowarski, and Stéphane Clinchant. [From Distillation to Hard Negative Sampling: Making Sparse Neural IR Models More Effective.](https://dl.acm.org/doi/10.1145/3477495.3531857) _Proceedings of the 45th International ACM SIGIR Conference on Research and Development in Information Retrieval_, pages 2353–2359. -In these experiments, we are using pre-encoded queries (i.e., cached results of query encoding). +In these experiments, we are using cached queries (i.e., cached results of query encoding). The exact configurations for these regressions are stored in [this YAML file](../../src/main/resources/regression/beir-v1.0.0-hotpotqa.splade-pp-ed.cached.yaml). Note that this page is automatically generated from [this template](../../src/main/resources/docgen/templates/beir-v1.0.0-hotpotqa.splade-pp-ed.cached.template) as part of Anserini's regression pipeline, so do not modify this page directly; modify the template instead and then run `bin/build.sh` to rebuild the documentation. diff --git a/docs/regressions/regressions-beir-v1.0.0-hotpotqa.splade-pp-ed.onnx.md b/docs/regressions/regressions-beir-v1.0.0-hotpotqa.splade-pp-ed.onnx.md index 7b9cf5b1f3..2efb222d5c 100644 --- a/docs/regressions/regressions-beir-v1.0.0-hotpotqa.splade-pp-ed.onnx.md +++ b/docs/regressions/regressions-beir-v1.0.0-hotpotqa.splade-pp-ed.onnx.md @@ -1,8 +1,8 @@ # Anserini Regressions: BEIR (v1.0.0) — HotpotQA -**Model**: [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) (using ONNX for on-the-fly query encoding) +**Model**: [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) (using ONNX for on-the-fly query encoding) -This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — HotpotQA](http://beir.ai/). +This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — HotpotQA](http://beir.ai/). The model itself can be download [here](https://huggingface.co/naver/splade-cocondenser-ensembledistil). See the [official SPLADE repo](/~https://github.com/naver/splade) and the following paper for more details: diff --git a/docs/regressions/regressions-beir-v1.0.0-hotpotqa.unicoil-noexp.cached.md b/docs/regressions/regressions-beir-v1.0.0-hotpotqa.unicoil-noexp.cached.md index 08d58b5fa6..14c83a4d99 100644 --- a/docs/regressions/regressions-beir-v1.0.0-hotpotqa.unicoil-noexp.cached.md +++ b/docs/regressions/regressions-beir-v1.0.0-hotpotqa.unicoil-noexp.cached.md @@ -1,12 +1,14 @@ # Anserini Regressions: BEIR (v1.0.0) — HotpotQA -**Model**: uniCOIL (without any expansions) +**Model**: uniCOIL without any expansions (using cached queries) This page describes regression experiments, integrated into Anserini's regression testing framework, using uniCOIL (without any expansions) on [BEIR (v1.0.0) — HotpotQA](http://beir.ai/). 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.](https://arxiv.org/abs/2106.14807) _arXiv:2106.14807_. +In these experiments, we are using cached queries (i.e., cached results of query encoding). + The exact configurations for these regressions are stored in [this YAML file](../../src/main/resources/regression/beir-v1.0.0-hotpotqa.unicoil-noexp.cached.yaml). Note that this page is automatically generated from [this template](../../src/main/resources/docgen/templates/beir-v1.0.0-hotpotqa.unicoil-noexp.cached.template) as part of Anserini's regression pipeline, so do not modify this page directly; modify the template instead. diff --git a/docs/regressions/regressions-beir-v1.0.0-nfcorpus.splade-pp-ed.cached.md b/docs/regressions/regressions-beir-v1.0.0-nfcorpus.splade-pp-ed.cached.md index e1292bb586..1bc5713d7f 100644 --- a/docs/regressions/regressions-beir-v1.0.0-nfcorpus.splade-pp-ed.cached.md +++ b/docs/regressions/regressions-beir-v1.0.0-nfcorpus.splade-pp-ed.cached.md @@ -1,14 +1,14 @@ -# Anserini Regressions: BEIR (v1.0.0) — NFCorpus +# Anserini Regressions: BEIR (v1.0.0) — NCorpus -**Model**: [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) (using pre-encoded queries) +**Model**: [SPLADE++ (CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) (using cached queries) -This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — NFCorpus](http://beir.ai/). +This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — NFCorpus](http://beir.ai/). The model itself can be download [here](https://huggingface.co/naver/splade-cocondenser-ensembledistil). See the [official SPLADE repo](/~https://github.com/naver/splade) and the following paper for more details: > Thibault Formal, Carlos Lassance, Benjamin Piwowarski, and Stéphane Clinchant. [From Distillation to Hard Negative Sampling: Making Sparse Neural IR Models More Effective.](https://dl.acm.org/doi/10.1145/3477495.3531857) _Proceedings of the 45th International ACM SIGIR Conference on Research and Development in Information Retrieval_, pages 2353–2359. -In these experiments, we are using pre-encoded queries (i.e., cached results of query encoding). +In these experiments, we are using cached queries (i.e., cached results of query encoding). The exact configurations for these regressions are stored in [this YAML file](../../src/main/resources/regression/beir-v1.0.0-nfcorpus.splade-pp-ed.cached.yaml). Note that this page is automatically generated from [this template](../../src/main/resources/docgen/templates/beir-v1.0.0-nfcorpus.splade-pp-ed.cached.template) as part of Anserini's regression pipeline, so do not modify this page directly; modify the template instead and then run `bin/build.sh` to rebuild the documentation. diff --git a/docs/regressions/regressions-beir-v1.0.0-nfcorpus.splade-pp-ed.onnx.md b/docs/regressions/regressions-beir-v1.0.0-nfcorpus.splade-pp-ed.onnx.md index 9dcaa564f1..e20e85613d 100644 --- a/docs/regressions/regressions-beir-v1.0.0-nfcorpus.splade-pp-ed.onnx.md +++ b/docs/regressions/regressions-beir-v1.0.0-nfcorpus.splade-pp-ed.onnx.md @@ -1,8 +1,8 @@ # Anserini Regressions: BEIR (v1.0.0) — NFCorpus -**Model**: [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) (using ONNX for on-the-fly query encoding) +**Model**: [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) (using ONNX for on-the-fly query encoding) -This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — NFCorpus](http://beir.ai/). +This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — NFCorpus](http://beir.ai/). The model itself can be download [here](https://huggingface.co/naver/splade-cocondenser-ensembledistil). See the [official SPLADE repo](/~https://github.com/naver/splade) and the following paper for more details: diff --git a/docs/regressions/regressions-beir-v1.0.0-nfcorpus.unicoil-noexp.cached.md b/docs/regressions/regressions-beir-v1.0.0-nfcorpus.unicoil-noexp.cached.md index ccd3dccc0c..fc9695bdba 100644 --- a/docs/regressions/regressions-beir-v1.0.0-nfcorpus.unicoil-noexp.cached.md +++ b/docs/regressions/regressions-beir-v1.0.0-nfcorpus.unicoil-noexp.cached.md @@ -1,12 +1,14 @@ # Anserini Regressions: BEIR (v1.0.0) — NFCorpus -**Model**: uniCOIL (without any expansions) +**Model**: uniCOIL without any expansions (using cached queries) This page describes regression experiments, integrated into Anserini's regression testing framework, using uniCOIL (without any expansions) on [BEIR (v1.0.0) — NFCorpus](http://beir.ai/). 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.](https://arxiv.org/abs/2106.14807) _arXiv:2106.14807_. +In these experiments, we are using cached queries (i.e., cached results of query encoding). + The exact configurations for these regressions are stored in [this YAML file](../../src/main/resources/regression/beir-v1.0.0-nfcorpus.unicoil-noexp.cached.yaml). Note that this page is automatically generated from [this template](../../src/main/resources/docgen/templates/beir-v1.0.0-nfcorpus.unicoil-noexp.cached.template) as part of Anserini's regression pipeline, so do not modify this page directly; modify the template instead. diff --git a/docs/regressions/regressions-beir-v1.0.0-nq.splade-pp-ed.cached.md b/docs/regressions/regressions-beir-v1.0.0-nq.splade-pp-ed.cached.md index 3cbeed4d04..f8aa35d052 100644 --- a/docs/regressions/regressions-beir-v1.0.0-nq.splade-pp-ed.cached.md +++ b/docs/regressions/regressions-beir-v1.0.0-nq.splade-pp-ed.cached.md @@ -1,14 +1,14 @@ # Anserini Regressions: BEIR (v1.0.0) — NQ -**Model**: [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) (using pre-encoded queries) +**Model**: [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) (using cached queries) -This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — NQ](http://beir.ai/). +This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — NQ](http://beir.ai/). The model itself can be download [here](https://huggingface.co/naver/splade-cocondenser-ensembledistil). See the [official SPLADE repo](/~https://github.com/naver/splade) and the following paper for more details: > Thibault Formal, Carlos Lassance, Benjamin Piwowarski, and Stéphane Clinchant. [From Distillation to Hard Negative Sampling: Making Sparse Neural IR Models More Effective.](https://dl.acm.org/doi/10.1145/3477495.3531857) _Proceedings of the 45th International ACM SIGIR Conference on Research and Development in Information Retrieval_, pages 2353–2359. -In these experiments, we are using pre-encoded queries (i.e., cached results of query encoding). +In these experiments, we are using cached queries (i.e., cached results of query encoding). The exact configurations for these regressions are stored in [this YAML file](../../src/main/resources/regression/beir-v1.0.0-nq.splade-pp-ed.cached.yaml). Note that this page is automatically generated from [this template](../../src/main/resources/docgen/templates/beir-v1.0.0-nq.splade-pp-ed.cached.template) as part of Anserini's regression pipeline, so do not modify this page directly; modify the template instead and then run `bin/build.sh` to rebuild the documentation. diff --git a/docs/regressions/regressions-beir-v1.0.0-nq.splade-pp-ed.onnx.md b/docs/regressions/regressions-beir-v1.0.0-nq.splade-pp-ed.onnx.md index 10085404c4..ea03b81de6 100644 --- a/docs/regressions/regressions-beir-v1.0.0-nq.splade-pp-ed.onnx.md +++ b/docs/regressions/regressions-beir-v1.0.0-nq.splade-pp-ed.onnx.md @@ -1,8 +1,8 @@ # Anserini Regressions: BEIR (v1.0.0) — NQ -**Model**: [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) (using ONNX for on-the-fly query encoding) +**Model**: [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) (using ONNX for on-the-fly query encoding) -This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — NQ](http://beir.ai/). +This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — NQ](http://beir.ai/). The model itself can be download [here](https://huggingface.co/naver/splade-cocondenser-ensembledistil). See the [official SPLADE repo](/~https://github.com/naver/splade) and the following paper for more details: diff --git a/docs/regressions/regressions-beir-v1.0.0-nq.unicoil-noexp.cached.md b/docs/regressions/regressions-beir-v1.0.0-nq.unicoil-noexp.cached.md index 64578c23e9..312ef826a7 100644 --- a/docs/regressions/regressions-beir-v1.0.0-nq.unicoil-noexp.cached.md +++ b/docs/regressions/regressions-beir-v1.0.0-nq.unicoil-noexp.cached.md @@ -1,12 +1,14 @@ # Anserini Regressions: BEIR (v1.0.0) — NQ -**Model**: uniCOIL (without any expansions) +**Model**: uniCOIL without any expansions (using cached queries) This page describes regression experiments, integrated into Anserini's regression testing framework, using uniCOIL (without any expansions) on [BEIR (v1.0.0) — NQ](http://beir.ai/). 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.](https://arxiv.org/abs/2106.14807) _arXiv:2106.14807_. +In these experiments, we are using cached queries (i.e., cached results of query encoding). + The exact configurations for these regressions are stored in [this YAML file](../../src/main/resources/regression/beir-v1.0.0-nq.unicoil-noexp.cached.yaml). Note that this page is automatically generated from [this template](../../src/main/resources/docgen/templates/beir-v1.0.0-nq.unicoil-noexp.cached.template) as part of Anserini's regression pipeline, so do not modify this page directly; modify the template instead. diff --git a/docs/regressions/regressions-beir-v1.0.0-quora.splade-pp-ed.cached.md b/docs/regressions/regressions-beir-v1.0.0-quora.splade-pp-ed.cached.md index fe50cd786d..aa1dd6ea18 100644 --- a/docs/regressions/regressions-beir-v1.0.0-quora.splade-pp-ed.cached.md +++ b/docs/regressions/regressions-beir-v1.0.0-quora.splade-pp-ed.cached.md @@ -1,14 +1,14 @@ # Anserini Regressions: BEIR (v1.0.0) — Quora -**Model**: [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) (using pre-encoded queries) +**Model**: [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) (using cached queries) -This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — Quora](http://beir.ai/). +This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — Quora](http://beir.ai/). The model itself can be download [here](https://huggingface.co/naver/splade-cocondenser-ensembledistil). See the [official SPLADE repo](/~https://github.com/naver/splade) and the following paper for more details: > Thibault Formal, Carlos Lassance, Benjamin Piwowarski, and Stéphane Clinchant. [From Distillation to Hard Negative Sampling: Making Sparse Neural IR Models More Effective.](https://dl.acm.org/doi/10.1145/3477495.3531857) _Proceedings of the 45th International ACM SIGIR Conference on Research and Development in Information Retrieval_, pages 2353–2359. -In these experiments, we are using pre-encoded queries (i.e., cached results of query encoding). +In these experiments, we are using cached queries (i.e., cached results of query encoding). The exact configurations for these regressions are stored in [this YAML file](../../src/main/resources/regression/beir-v1.0.0-quora.splade-pp-ed.cached.yaml). Note that this page is automatically generated from [this template](../../src/main/resources/docgen/templates/beir-v1.0.0-quora.splade-pp-ed.cached.template) as part of Anserini's regression pipeline, so do not modify this page directly; modify the template instead and then run `bin/build.sh` to rebuild the documentation. diff --git a/docs/regressions/regressions-beir-v1.0.0-quora.splade-pp-ed.onnx.md b/docs/regressions/regressions-beir-v1.0.0-quora.splade-pp-ed.onnx.md index e7705ed6bf..77abe7fbf0 100644 --- a/docs/regressions/regressions-beir-v1.0.0-quora.splade-pp-ed.onnx.md +++ b/docs/regressions/regressions-beir-v1.0.0-quora.splade-pp-ed.onnx.md @@ -1,8 +1,8 @@ # Anserini Regressions: BEIR (v1.0.0) — Quora -**Model**: [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) (using ONNX for on-the-fly query encoding) +**Model**: [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) (using ONNX for on-the-fly query encoding) -This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — Quora](http://beir.ai/). +This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — Quora](http://beir.ai/). The model itself can be download [here](https://huggingface.co/naver/splade-cocondenser-ensembledistil). See the [official SPLADE repo](/~https://github.com/naver/splade) and the following paper for more details: diff --git a/docs/regressions/regressions-beir-v1.0.0-quora.unicoil-noexp.cached.md b/docs/regressions/regressions-beir-v1.0.0-quora.unicoil-noexp.cached.md index 32627a3f48..3a214ba841 100644 --- a/docs/regressions/regressions-beir-v1.0.0-quora.unicoil-noexp.cached.md +++ b/docs/regressions/regressions-beir-v1.0.0-quora.unicoil-noexp.cached.md @@ -1,12 +1,14 @@ # Anserini Regressions: BEIR (v1.0.0) — Quora -**Model**: uniCOIL (without any expansions) +**Model**: uniCOIL without any expansions (using cached queries) This page describes regression experiments, integrated into Anserini's regression testing framework, using uniCOIL (without any expansions) on [BEIR (v1.0.0) — Quora](http://beir.ai/). 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.](https://arxiv.org/abs/2106.14807) _arXiv:2106.14807_. +In these experiments, we are using cached queries (i.e., cached results of query encoding). + The exact configurations for these regressions are stored in [this YAML file](../../src/main/resources/regression/beir-v1.0.0-quora.unicoil-noexp.cached.yaml). Note that this page is automatically generated from [this template](../../src/main/resources/docgen/templates/beir-v1.0.0-quora.unicoil-noexp.cached.template) as part of Anserini's regression pipeline, so do not modify this page directly; modify the template instead. diff --git a/docs/regressions/regressions-beir-v1.0.0-robust04.splade-pp-ed.cached.md b/docs/regressions/regressions-beir-v1.0.0-robust04.splade-pp-ed.cached.md index 35ff859fa4..d6892cfd66 100644 --- a/docs/regressions/regressions-beir-v1.0.0-robust04.splade-pp-ed.cached.md +++ b/docs/regressions/regressions-beir-v1.0.0-robust04.splade-pp-ed.cached.md @@ -1,14 +1,14 @@ # Anserini Regressions: BEIR (v1.0.0) — Robust04 -**Model**: [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) (using pre-encoded queries) +**Model**: [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) (using cached queries) -This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — Robust04](http://beir.ai/). +This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — Robust04](http://beir.ai/). The model itself can be download [here](https://huggingface.co/naver/splade-cocondenser-ensembledistil). See the [official SPLADE repo](/~https://github.com/naver/splade) and the following paper for more details: > Thibault Formal, Carlos Lassance, Benjamin Piwowarski, and Stéphane Clinchant. [From Distillation to Hard Negative Sampling: Making Sparse Neural IR Models More Effective.](https://dl.acm.org/doi/10.1145/3477495.3531857) _Proceedings of the 45th International ACM SIGIR Conference on Research and Development in Information Retrieval_, pages 2353–2359. -In these experiments, we are using pre-encoded queries (i.e., cached results of query encoding). +In these experiments, we are using cached queries (i.e., cached results of query encoding). The exact configurations for these regressions are stored in [this YAML file](../../src/main/resources/regression/beir-v1.0.0-robust04.splade-pp-ed.cached.yaml). Note that this page is automatically generated from [this template](../../src/main/resources/docgen/templates/beir-v1.0.0-robust04.splade-pp-ed.cached.template) as part of Anserini's regression pipeline, so do not modify this page directly; modify the template instead and then run `bin/build.sh` to rebuild the documentation. diff --git a/docs/regressions/regressions-beir-v1.0.0-robust04.splade-pp-ed.onnx.md b/docs/regressions/regressions-beir-v1.0.0-robust04.splade-pp-ed.onnx.md index 96a993bde5..c87303bcbf 100644 --- a/docs/regressions/regressions-beir-v1.0.0-robust04.splade-pp-ed.onnx.md +++ b/docs/regressions/regressions-beir-v1.0.0-robust04.splade-pp-ed.onnx.md @@ -1,8 +1,8 @@ # Anserini Regressions: BEIR (v1.0.0) — Robust04 -**Model**: [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) (using ONNX for on-the-fly query encoding) +**Model**: [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) (using ONNX for on-the-fly query encoding) -This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — Robust04](http://beir.ai/). +This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — Robust04](http://beir.ai/). The model itself can be download [here](https://huggingface.co/naver/splade-cocondenser-ensembledistil). See the [official SPLADE repo](/~https://github.com/naver/splade) and the following paper for more details: diff --git a/docs/regressions/regressions-beir-v1.0.0-robust04.unicoil-noexp.cached.md b/docs/regressions/regressions-beir-v1.0.0-robust04.unicoil-noexp.cached.md index d744648bfc..de97566bb3 100644 --- a/docs/regressions/regressions-beir-v1.0.0-robust04.unicoil-noexp.cached.md +++ b/docs/regressions/regressions-beir-v1.0.0-robust04.unicoil-noexp.cached.md @@ -1,12 +1,14 @@ # Anserini Regressions: BEIR (v1.0.0) — Robust04 -**Model**: uniCOIL (without any expansions) +**Model**: uniCOIL without any expansions (using cached queries) This page describes regression experiments, integrated into Anserini's regression testing framework, using uniCOIL (without any expansions) on [BEIR (v1.0.0) — Robust04](http://beir.ai/). 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.](https://arxiv.org/abs/2106.14807) _arXiv:2106.14807_. +In these experiments, we are using cached queries (i.e., cached results of query encoding). + The exact configurations for these regressions are stored in [this YAML file](../../src/main/resources/regression/beir-v1.0.0-robust04.unicoil-noexp.cached.yaml). Note that this page is automatically generated from [this template](../../src/main/resources/docgen/templates/beir-v1.0.0-robust04.unicoil-noexp.cached.template) as part of Anserini's regression pipeline, so do not modify this page directly; modify the template instead. diff --git a/docs/regressions/regressions-beir-v1.0.0-scidocs.splade-pp-ed.cached.md b/docs/regressions/regressions-beir-v1.0.0-scidocs.splade-pp-ed.cached.md index 49793a82f4..e8c94e255c 100644 --- a/docs/regressions/regressions-beir-v1.0.0-scidocs.splade-pp-ed.cached.md +++ b/docs/regressions/regressions-beir-v1.0.0-scidocs.splade-pp-ed.cached.md @@ -1,14 +1,14 @@ # Anserini Regressions: BEIR (v1.0.0) — SCIDOCS -**Model**: [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) (using pre-encoded queries) +**Model**: [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) (using cached queries) -This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — SCIDOCS](http://beir.ai/). +This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — SCIDOCS](http://beir.ai/). The model itself can be download [here](https://huggingface.co/naver/splade-cocondenser-ensembledistil). See the [official SPLADE repo](/~https://github.com/naver/splade) and the following paper for more details: > Thibault Formal, Carlos Lassance, Benjamin Piwowarski, and Stéphane Clinchant. [From Distillation to Hard Negative Sampling: Making Sparse Neural IR Models More Effective.](https://dl.acm.org/doi/10.1145/3477495.3531857) _Proceedings of the 45th International ACM SIGIR Conference on Research and Development in Information Retrieval_, pages 2353–2359. -In these experiments, we are using pre-encoded queries (i.e., cached results of query encoding). +In these experiments, we are using cached queries (i.e., cached results of query encoding). The exact configurations for these regressions are stored in [this YAML file](../../src/main/resources/regression/beir-v1.0.0-scidocs.splade-pp-ed.cached.yaml). Note that this page is automatically generated from [this template](../../src/main/resources/docgen/templates/beir-v1.0.0-scidocs.splade-pp-ed.cached.template) as part of Anserini's regression pipeline, so do not modify this page directly; modify the template instead and then run `bin/build.sh` to rebuild the documentation. diff --git a/docs/regressions/regressions-beir-v1.0.0-scidocs.splade-pp-ed.onnx.md b/docs/regressions/regressions-beir-v1.0.0-scidocs.splade-pp-ed.onnx.md index e3669fed76..51fba27e85 100644 --- a/docs/regressions/regressions-beir-v1.0.0-scidocs.splade-pp-ed.onnx.md +++ b/docs/regressions/regressions-beir-v1.0.0-scidocs.splade-pp-ed.onnx.md @@ -1,8 +1,8 @@ # Anserini Regressions: BEIR (v1.0.0) — SCIDOCS -**Model**: [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) (using ONNX for on-the-fly query encoding) +**Model**: [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) (using ONNX for on-the-fly query encoding) -This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — SCIDOCS](http://beir.ai/). +This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — SCIDOCS](http://beir.ai/). The model itself can be download [here](https://huggingface.co/naver/splade-cocondenser-ensembledistil). See the [official SPLADE repo](/~https://github.com/naver/splade) and the following paper for more details: diff --git a/docs/regressions/regressions-beir-v1.0.0-scidocs.unicoil-noexp.cached.md b/docs/regressions/regressions-beir-v1.0.0-scidocs.unicoil-noexp.cached.md index 3675c82160..58a02134a8 100644 --- a/docs/regressions/regressions-beir-v1.0.0-scidocs.unicoil-noexp.cached.md +++ b/docs/regressions/regressions-beir-v1.0.0-scidocs.unicoil-noexp.cached.md @@ -1,12 +1,14 @@ # Anserini Regressions: BEIR (v1.0.0) — SCIDOCS -**Model**: uniCOIL (without any expansions) +**Model**: uniCOIL without any expansions (using cached queries) This page describes regression experiments, integrated into Anserini's regression testing framework, using uniCOIL (without any expansions) on [BEIR (v1.0.0) — SCIDOCS](http://beir.ai/). 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.](https://arxiv.org/abs/2106.14807) _arXiv:2106.14807_. +In these experiments, we are using cached queries (i.e., cached results of query encoding). + The exact configurations for these regressions are stored in [this YAML file](../../src/main/resources/regression/beir-v1.0.0-scidocs.unicoil-noexp.cached.yaml). Note that this page is automatically generated from [this template](../../src/main/resources/docgen/templates/beir-v1.0.0-scidocs.unicoil-noexp.cached.template) as part of Anserini's regression pipeline, so do not modify this page directly; modify the template instead. diff --git a/docs/regressions/regressions-beir-v1.0.0-scifact.splade-pp-ed.cached.md b/docs/regressions/regressions-beir-v1.0.0-scifact.splade-pp-ed.cached.md index 9fadd1cf60..4c40d394b4 100644 --- a/docs/regressions/regressions-beir-v1.0.0-scifact.splade-pp-ed.cached.md +++ b/docs/regressions/regressions-beir-v1.0.0-scifact.splade-pp-ed.cached.md @@ -1,14 +1,14 @@ # Anserini Regressions: BEIR (v1.0.0) — SciFact -**Model**: [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) (using pre-encoded queries) +**Model**: [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) (using cached queries) -This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — SciFact](http://beir.ai/). +This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — SciFact](http://beir.ai/). The model itself can be download [here](https://huggingface.co/naver/splade-cocondenser-ensembledistil). See the [official SPLADE repo](/~https://github.com/naver/splade) and the following paper for more details: > Thibault Formal, Carlos Lassance, Benjamin Piwowarski, and Stéphane Clinchant. [From Distillation to Hard Negative Sampling: Making Sparse Neural IR Models More Effective.](https://dl.acm.org/doi/10.1145/3477495.3531857) _Proceedings of the 45th International ACM SIGIR Conference on Research and Development in Information Retrieval_, pages 2353–2359. -In these experiments, we are using pre-encoded queries (i.e., cached results of query encoding). +In these experiments, we are using cached queries (i.e., cached results of query encoding). The exact configurations for these regressions are stored in [this YAML file](../../src/main/resources/regression/beir-v1.0.0-scifact.splade-pp-ed.cached.yaml). Note that this page is automatically generated from [this template](../../src/main/resources/docgen/templates/beir-v1.0.0-scifact.splade-pp-ed.cached.template) as part of Anserini's regression pipeline, so do not modify this page directly; modify the template instead and then run `bin/build.sh` to rebuild the documentation. diff --git a/docs/regressions/regressions-beir-v1.0.0-scifact.splade-pp-ed.onnx.md b/docs/regressions/regressions-beir-v1.0.0-scifact.splade-pp-ed.onnx.md index e17435a832..603d657c1c 100644 --- a/docs/regressions/regressions-beir-v1.0.0-scifact.splade-pp-ed.onnx.md +++ b/docs/regressions/regressions-beir-v1.0.0-scifact.splade-pp-ed.onnx.md @@ -1,8 +1,8 @@ # Anserini Regressions: BEIR (v1.0.0) — SciFact -**Model**: [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) (using ONNX for on-the-fly query encoding) +**Model**: [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) (using ONNX for on-the-fly query encoding) -This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — SciFact](http://beir.ai/). +This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — SciFact](http://beir.ai/). The model itself can be download [here](https://huggingface.co/naver/splade-cocondenser-ensembledistil). See the [official SPLADE repo](/~https://github.com/naver/splade) and the following paper for more details: diff --git a/docs/regressions/regressions-beir-v1.0.0-scifact.unicoil-noexp.cached.md b/docs/regressions/regressions-beir-v1.0.0-scifact.unicoil-noexp.cached.md index 3dcacad1a6..d71c6016eb 100644 --- a/docs/regressions/regressions-beir-v1.0.0-scifact.unicoil-noexp.cached.md +++ b/docs/regressions/regressions-beir-v1.0.0-scifact.unicoil-noexp.cached.md @@ -1,12 +1,14 @@ # Anserini Regressions: BEIR (v1.0.0) — SciFact -**Model**: uniCOIL (without any expansions) +**Model**: uniCOIL without any expansions (using cached queries) This page describes regression experiments, integrated into Anserini's regression testing framework, using uniCOIL (without any expansions) on [BEIR (v1.0.0) — SciFact](http://beir.ai/). 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.](https://arxiv.org/abs/2106.14807) _arXiv:2106.14807_. +In these experiments, we are using cached queries (i.e., cached results of query encoding). + The exact configurations for these regressions are stored in [this YAML file](../../src/main/resources/regression/beir-v1.0.0-scifact.unicoil-noexp.cached.yaml). Note that this page is automatically generated from [this template](../../src/main/resources/docgen/templates/beir-v1.0.0-scifact.unicoil-noexp.cached.template) as part of Anserini's regression pipeline, so do not modify this page directly; modify the template instead. diff --git a/docs/regressions/regressions-beir-v1.0.0-signal1m.splade-pp-ed.cached.md b/docs/regressions/regressions-beir-v1.0.0-signal1m.splade-pp-ed.cached.md index 46dba2873e..fc85d4e3ea 100644 --- a/docs/regressions/regressions-beir-v1.0.0-signal1m.splade-pp-ed.cached.md +++ b/docs/regressions/regressions-beir-v1.0.0-signal1m.splade-pp-ed.cached.md @@ -1,14 +1,14 @@ # Anserini Regressions: BEIR (v1.0.0) — Signal-1M -**Model**: [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) (using pre-encoded queries) +**Model**: [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) (using cached queries) -This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — Signal-1M](http://beir.ai/). +This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — Signal-1M](http://beir.ai/). The model itself can be download [here](https://huggingface.co/naver/splade-cocondenser-ensembledistil). See the [official SPLADE repo](/~https://github.com/naver/splade) and the following paper for more details: > Thibault Formal, Carlos Lassance, Benjamin Piwowarski, and Stéphane Clinchant. [From Distillation to Hard Negative Sampling: Making Sparse Neural IR Models More Effective.](https://dl.acm.org/doi/10.1145/3477495.3531857) _Proceedings of the 45th International ACM SIGIR Conference on Research and Development in Information Retrieval_, pages 2353–2359. -In these experiments, we are using pre-encoded queries (i.e., cached results of query encoding). +In these experiments, we are using cached queries (i.e., cached results of query encoding). The exact configurations for these regressions are stored in [this YAML file](../../src/main/resources/regression/beir-v1.0.0-signal1m.splade-pp-ed.cached.yaml). Note that this page is automatically generated from [this template](../../src/main/resources/docgen/templates/beir-v1.0.0-signal1m.splade-pp-ed.cached.template) as part of Anserini's regression pipeline, so do not modify this page directly; modify the template instead and then run `bin/build.sh` to rebuild the documentation. diff --git a/docs/regressions/regressions-beir-v1.0.0-signal1m.splade-pp-ed.onnx.md b/docs/regressions/regressions-beir-v1.0.0-signal1m.splade-pp-ed.onnx.md index 44ee0a9258..922e3792fa 100644 --- a/docs/regressions/regressions-beir-v1.0.0-signal1m.splade-pp-ed.onnx.md +++ b/docs/regressions/regressions-beir-v1.0.0-signal1m.splade-pp-ed.onnx.md @@ -1,8 +1,8 @@ # Anserini Regressions: BEIR (v1.0.0) — Signal-1M -**Model**: [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) (using ONNX for on-the-fly query encoding) +**Model**: [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) (using ONNX for on-the-fly query encoding) -This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — Signal-1M](http://beir.ai/). +This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — Signal-1M](http://beir.ai/). The model itself can be download [here](https://huggingface.co/naver/splade-cocondenser-ensembledistil). See the [official SPLADE repo](/~https://github.com/naver/splade) and the following paper for more details: diff --git a/docs/regressions/regressions-beir-v1.0.0-signal1m.unicoil-noexp.cached.md b/docs/regressions/regressions-beir-v1.0.0-signal1m.unicoil-noexp.cached.md index c29163b3f7..61c0509fd6 100644 --- a/docs/regressions/regressions-beir-v1.0.0-signal1m.unicoil-noexp.cached.md +++ b/docs/regressions/regressions-beir-v1.0.0-signal1m.unicoil-noexp.cached.md @@ -1,12 +1,14 @@ # Anserini Regressions: BEIR (v1.0.0) — Signal-1M -**Model**: uniCOIL (without any expansions) +**Model**: uniCOIL without any expansions (using cached queries) This page describes regression experiments, integrated into Anserini's regression testing framework, using uniCOIL (without any expansions) on [BEIR (v1.0.0) — Signal-1M](http://beir.ai/). 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.](https://arxiv.org/abs/2106.14807) _arXiv:2106.14807_. +In these experiments, we are using cached queries (i.e., cached results of query encoding). + The exact configurations for these regressions are stored in [this YAML file](../../src/main/resources/regression/beir-v1.0.0-signal1m.unicoil-noexp.cached.yaml). Note that this page is automatically generated from [this template](../../src/main/resources/docgen/templates/beir-v1.0.0-signal1m.unicoil-noexp.cached.template) as part of Anserini's regression pipeline, so do not modify this page directly; modify the template instead. diff --git a/docs/regressions/regressions-beir-v1.0.0-trec-covid.splade-pp-ed.cached.md b/docs/regressions/regressions-beir-v1.0.0-trec-covid.splade-pp-ed.cached.md index 51abbcd9ef..8477e54874 100644 --- a/docs/regressions/regressions-beir-v1.0.0-trec-covid.splade-pp-ed.cached.md +++ b/docs/regressions/regressions-beir-v1.0.0-trec-covid.splade-pp-ed.cached.md @@ -1,14 +1,14 @@ # Anserini Regressions: BEIR (v1.0.0) — TREC-COVID -**Model**: [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) (using pre-encoded queries) +**Model**: [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) (using cached queries) -This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — TREC-COVID](http://beir.ai/). +This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — TREC-COVID](http://beir.ai/). The model itself can be download [here](https://huggingface.co/naver/splade-cocondenser-ensembledistil). See the [official SPLADE repo](/~https://github.com/naver/splade) and the following paper for more details: > Thibault Formal, Carlos Lassance, Benjamin Piwowarski, and Stéphane Clinchant. [From Distillation to Hard Negative Sampling: Making Sparse Neural IR Models More Effective.](https://dl.acm.org/doi/10.1145/3477495.3531857) _Proceedings of the 45th International ACM SIGIR Conference on Research and Development in Information Retrieval_, pages 2353–2359. -In these experiments, we are using pre-encoded queries (i.e., cached results of query encoding). +In these experiments, we are using cached queries (i.e., cached results of query encoding). The exact configurations for these regressions are stored in [this YAML file](../../src/main/resources/regression/beir-v1.0.0-trec-covid.splade-pp-ed.cached.yaml). Note that this page is automatically generated from [this template](../../src/main/resources/docgen/templates/beir-v1.0.0-trec-covid.splade-pp-ed.cached.template) as part of Anserini's regression pipeline, so do not modify this page directly; modify the template instead and then run `bin/build.sh` to rebuild the documentation. diff --git a/docs/regressions/regressions-beir-v1.0.0-trec-covid.splade-pp-ed.onnx.md b/docs/regressions/regressions-beir-v1.0.0-trec-covid.splade-pp-ed.onnx.md index e2aa9b3675..f7448f0596 100644 --- a/docs/regressions/regressions-beir-v1.0.0-trec-covid.splade-pp-ed.onnx.md +++ b/docs/regressions/regressions-beir-v1.0.0-trec-covid.splade-pp-ed.onnx.md @@ -1,8 +1,8 @@ # Anserini Regressions: BEIR (v1.0.0) — TREC-COVID -**Model**: [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) (using ONNX for on-the-fly query encoding) +**Model**: [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) (using ONNX for on-the-fly query encoding) -This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — TREC-COVID](http://beir.ai/). +This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — TREC-COVID](http://beir.ai/). The model itself can be download [here](https://huggingface.co/naver/splade-cocondenser-ensembledistil). See the [official SPLADE repo](/~https://github.com/naver/splade) and the following paper for more details: diff --git a/docs/regressions/regressions-beir-v1.0.0-trec-covid.unicoil-noexp.cached.md b/docs/regressions/regressions-beir-v1.0.0-trec-covid.unicoil-noexp.cached.md index 53e779da96..9b10dd2b38 100644 --- a/docs/regressions/regressions-beir-v1.0.0-trec-covid.unicoil-noexp.cached.md +++ b/docs/regressions/regressions-beir-v1.0.0-trec-covid.unicoil-noexp.cached.md @@ -1,12 +1,14 @@ # Anserini Regressions: BEIR (v1.0.0) — TREC-COVID -**Model**: uniCOIL (without any expansions) +**Model**: uniCOIL without any expansions (using cached queries) This page describes regression experiments, integrated into Anserini's regression testing framework, using uniCOIL (without any expansions) on [BEIR (v1.0.0) — TREC-COVID](http://beir.ai/). 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.](https://arxiv.org/abs/2106.14807) _arXiv:2106.14807_. +In these experiments, we are using cached queries (i.e., cached results of query encoding). + The exact configurations for these regressions are stored in [this YAML file](../../src/main/resources/regression/beir-v1.0.0-trec-covid.unicoil-noexp.cached.yaml). Note that this page is automatically generated from [this template](../../src/main/resources/docgen/templates/beir-v1.0.0-trec-covid.unicoil-noexp.cached.template) as part of Anserini's regression pipeline, so do not modify this page directly; modify the template instead. diff --git a/docs/regressions/regressions-beir-v1.0.0-trec-news.splade-pp-ed.cached.md b/docs/regressions/regressions-beir-v1.0.0-trec-news.splade-pp-ed.cached.md index ee30cbb0cf..3457cc6e86 100644 --- a/docs/regressions/regressions-beir-v1.0.0-trec-news.splade-pp-ed.cached.md +++ b/docs/regressions/regressions-beir-v1.0.0-trec-news.splade-pp-ed.cached.md @@ -1,14 +1,14 @@ # Anserini Regressions: BEIR (v1.0.0) — TREC-NEWS -**Model**: [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) (using pre-encoded queries) +**Model**: [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) (using cached queries) -This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — TREC-NEWS](http://beir.ai/). +This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — TREC-NEWS](http://beir.ai/). The model itself can be download [here](https://huggingface.co/naver/splade-cocondenser-ensembledistil). See the [official SPLADE repo](/~https://github.com/naver/splade) and the following paper for more details: > Thibault Formal, Carlos Lassance, Benjamin Piwowarski, and Stéphane Clinchant. [From Distillation to Hard Negative Sampling: Making Sparse Neural IR Models More Effective.](https://dl.acm.org/doi/10.1145/3477495.3531857) _Proceedings of the 45th International ACM SIGIR Conference on Research and Development in Information Retrieval_, pages 2353–2359. -In these experiments, we are using pre-encoded queries (i.e., cached results of query encoding). +In these experiments, we are using cached queries (i.e., cached results of query encoding). The exact configurations for these regressions are stored in [this YAML file](../../src/main/resources/regression/beir-v1.0.0-trec-news.splade-pp-ed.cached.yaml). Note that this page is automatically generated from [this template](../../src/main/resources/docgen/templates/beir-v1.0.0-trec-news.splade-pp-ed.cached.template) as part of Anserini's regression pipeline, so do not modify this page directly; modify the template instead and then run `bin/build.sh` to rebuild the documentation. diff --git a/docs/regressions/regressions-beir-v1.0.0-trec-news.splade-pp-ed.onnx.md b/docs/regressions/regressions-beir-v1.0.0-trec-news.splade-pp-ed.onnx.md index 0f6de1c15d..d405d59a95 100644 --- a/docs/regressions/regressions-beir-v1.0.0-trec-news.splade-pp-ed.onnx.md +++ b/docs/regressions/regressions-beir-v1.0.0-trec-news.splade-pp-ed.onnx.md @@ -1,8 +1,8 @@ # Anserini Regressions: BEIR (v1.0.0) — TREC-NEWS -**Model**: [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) (using ONNX for on-the-fly query encoding) +**Model**: [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) (using ONNX for on-the-fly query encoding) -This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — TREC-NEWS](http://beir.ai/). +This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — TREC-NEWS](http://beir.ai/). The model itself can be download [here](https://huggingface.co/naver/splade-cocondenser-ensembledistil). See the [official SPLADE repo](/~https://github.com/naver/splade) and the following paper for more details: diff --git a/docs/regressions/regressions-beir-v1.0.0-trec-news.unicoil-noexp.cached.md b/docs/regressions/regressions-beir-v1.0.0-trec-news.unicoil-noexp.cached.md index 2dd723c4b3..57b9d7569d 100644 --- a/docs/regressions/regressions-beir-v1.0.0-trec-news.unicoil-noexp.cached.md +++ b/docs/regressions/regressions-beir-v1.0.0-trec-news.unicoil-noexp.cached.md @@ -1,12 +1,14 @@ # Anserini Regressions: BEIR (v1.0.0) — TREC-NEWS -**Model**: uniCOIL (without any expansions) +**Model**: uniCOIL without any expansions (using cached queries) This page describes regression experiments, integrated into Anserini's regression testing framework, using uniCOIL (without any expansions) on [BEIR (v1.0.0) — TREC-NEWS](http://beir.ai/). 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.](https://arxiv.org/abs/2106.14807) _arXiv:2106.14807_. +In these experiments, we are using cached queries (i.e., cached results of query encoding). + The exact configurations for these regressions are stored in [this YAML file](../../src/main/resources/regression/beir-v1.0.0-trec-news.unicoil-noexp.cached.yaml). Note that this page is automatically generated from [this template](../../src/main/resources/docgen/templates/beir-v1.0.0-trec-news.unicoil-noexp.cached.template) as part of Anserini's regression pipeline, so do not modify this page directly; modify the template instead. diff --git a/docs/regressions/regressions-beir-v1.0.0-webis-touche2020.splade-pp-ed.cached.md b/docs/regressions/regressions-beir-v1.0.0-webis-touche2020.splade-pp-ed.cached.md index 705b9e2fa1..f47038b6f4 100644 --- a/docs/regressions/regressions-beir-v1.0.0-webis-touche2020.splade-pp-ed.cached.md +++ b/docs/regressions/regressions-beir-v1.0.0-webis-touche2020.splade-pp-ed.cached.md @@ -1,14 +1,14 @@ # Anserini Regressions: BEIR (v1.0.0) — Webis-Touche2020 -**Model**: [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) (using pre-encoded queries) +**Model**: [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) (using cached queries) -This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — Webis-Touche2020](http://beir.ai/). +This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — Webis-Touche2020](http://beir.ai/). The model itself can be download [here](https://huggingface.co/naver/splade-cocondenser-ensembledistil). See the [official SPLADE repo](/~https://github.com/naver/splade) and the following paper for more details: > Thibault Formal, Carlos Lassance, Benjamin Piwowarski, and Stéphane Clinchant. [From Distillation to Hard Negative Sampling: Making Sparse Neural IR Models More Effective.](https://dl.acm.org/doi/10.1145/3477495.3531857) _Proceedings of the 45th International ACM SIGIR Conference on Research and Development in Information Retrieval_, pages 2353–2359. -In these experiments, we are using pre-encoded queries (i.e., cached results of query encoding). +In these experiments, we are using cached queries (i.e., cached results of query encoding). The exact configurations for these regressions are stored in [this YAML file](../../src/main/resources/regression/beir-v1.0.0-webis-touche2020.splade-pp-ed.cached.yaml). Note that this page is automatically generated from [this template](../../src/main/resources/docgen/templates/beir-v1.0.0-webis-touche2020.splade-pp-ed.cached.template) as part of Anserini's regression pipeline, so do not modify this page directly; modify the template instead and then run `bin/build.sh` to rebuild the documentation. diff --git a/docs/regressions/regressions-beir-v1.0.0-webis-touche2020.splade-pp-ed.onnx.md b/docs/regressions/regressions-beir-v1.0.0-webis-touche2020.splade-pp-ed.onnx.md index 61e43946e8..5612be576d 100644 --- a/docs/regressions/regressions-beir-v1.0.0-webis-touche2020.splade-pp-ed.onnx.md +++ b/docs/regressions/regressions-beir-v1.0.0-webis-touche2020.splade-pp-ed.onnx.md @@ -1,8 +1,8 @@ # Anserini Regressions: BEIR (v1.0.0) — Webis-Touche2020 -**Model**: [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) (using ONNX for on-the-fly query encoding) +**Model**: [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) (using ONNX for on-the-fly query encoding) -This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — Webis-Touche2020](http://beir.ai/). +This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — Webis-Touche2020](http://beir.ai/). The model itself can be download [here](https://huggingface.co/naver/splade-cocondenser-ensembledistil). See the [official SPLADE repo](/~https://github.com/naver/splade) and the following paper for more details: diff --git a/docs/regressions/regressions-beir-v1.0.0-webis-touche2020.unicoil-noexp.cached.md b/docs/regressions/regressions-beir-v1.0.0-webis-touche2020.unicoil-noexp.cached.md index d3c809b958..d7c28ed72d 100644 --- a/docs/regressions/regressions-beir-v1.0.0-webis-touche2020.unicoil-noexp.cached.md +++ b/docs/regressions/regressions-beir-v1.0.0-webis-touche2020.unicoil-noexp.cached.md @@ -1,12 +1,14 @@ # Anserini Regressions: BEIR (v1.0.0) — Webis-Touche2020 -**Model**: uniCOIL (without any expansions) +**Model**: uniCOIL without any expansions (using cached queries) This page describes regression experiments, integrated into Anserini's regression testing framework, using uniCOIL (without any expansions) on [BEIR (v1.0.0) — Webis-Touche2020](http://beir.ai/). 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.](https://arxiv.org/abs/2106.14807) _arXiv:2106.14807_. +In these experiments, we are using cached queries (i.e., cached results of query encoding). + The exact configurations for these regressions are stored in [this YAML file](../../src/main/resources/regression/beir-v1.0.0-webis-touche2020.unicoil-noexp.cached.yaml). Note that this page is automatically generated from [this template](../../src/main/resources/docgen/templates/beir-v1.0.0-webis-touche2020.unicoil-noexp.cached.template) as part of Anserini's regression pipeline, so do not modify this page directly; modify the template instead. diff --git a/docs/regressions/regressions-dl19-passage.bge-base-en-v1.5.hnsw-int8.cached.md b/docs/regressions/regressions-dl19-passage.bge-base-en-v1.5.hnsw-int8.cached.md index ea9d5a5de0..0f9ac0e344 100644 --- a/docs/regressions/regressions-dl19-passage.bge-base-en-v1.5.hnsw-int8.cached.md +++ b/docs/regressions/regressions-dl19-passage.bge-base-en-v1.5.hnsw-int8.cached.md @@ -1,12 +1,12 @@ # Anserini Regressions: TREC 2019 Deep Learning Track (Passage) -**Model**: [BGE-base-en-v1.5](https://huggingface.co/BAAI/bge-base-en-v1.5) with HNSW quantized indexes (using pre-encoded queries) +**Model**: [BGE-base-en-v1.5](https://huggingface.co/BAAI/bge-base-en-v1.5) with quantized HNSW indexes (using cached queries) This page describes regression experiments, integrated into Anserini's regression testing framework, using the [BGE-base-en-v1.5](https://huggingface.co/BAAI/bge-base-en-v1.5) model on the [TREC 2019 Deep Learning Track passage ranking task](https://trec.nist.gov/data/deep2019.html), as described in the following paper: > Shitao Xiao, Zheng Liu, Peitian Zhang, and Niklas Muennighoff. [C-Pack: Packaged Resources To Advance General Chinese Embedding.](https://arxiv.org/abs/2309.07597) _arXiv:2309.07597_, 2023. -In these experiments, we are using pre-encoded queries (i.e., cached results of query encoding). +In these experiments, we are using cached queries (i.e., cached results of query encoding). Note that the NIST relevance judgments provide far more relevant passages per topic, unlike the "sparse" judgments provided by Microsoft (these are sometimes called "dense" judgments to emphasize this contrast). For additional instructions on working with MS MARCO passage collection, refer to [this page](experiments-msmarco-passage.md). @@ -49,7 +49,7 @@ python src/main/python/run_regression.py --index --verify --search --regression ## Indexing -Sample indexing command, building HNSW indexes: +Sample indexing command, building quantized HNSW indexes: ```bash bin/run.sh io.anserini.index.IndexHnswDenseVectors \ diff --git a/docs/regressions/regressions-dl19-passage.bge-base-en-v1.5.hnsw-int8.onnx.md b/docs/regressions/regressions-dl19-passage.bge-base-en-v1.5.hnsw-int8.onnx.md index 934c3c9928..61e35bcda9 100644 --- a/docs/regressions/regressions-dl19-passage.bge-base-en-v1.5.hnsw-int8.onnx.md +++ b/docs/regressions/regressions-dl19-passage.bge-base-en-v1.5.hnsw-int8.onnx.md @@ -1,6 +1,6 @@ # Anserini Regressions: TREC 2019 Deep Learning Track (Passage) -**Model**: [BGE-base-en-v1.5](https://huggingface.co/BAAI/bge-base-en-v1.5) with HNSW quantized indexes (using ONNX for on-the-fly query encoding) +**Model**: [BGE-base-en-v1.5](https://huggingface.co/BAAI/bge-base-en-v1.5) with quantized HNSW indexes (using ONNX for on-the-fly query encoding) This page describes regression experiments, integrated into Anserini's regression testing framework, using the [BGE-base-en-v1.5](https://huggingface.co/BAAI/bge-base-en-v1.5) model on the [TREC 2019 Deep Learning Track passage ranking task](https://trec.nist.gov/data/deep2019.html), as described in the following paper: @@ -49,7 +49,7 @@ python src/main/python/run_regression.py --index --verify --search --regression ## Indexing -Sample indexing command, building HNSW indexes: +Sample indexing command, building quantized HNSW indexes: ```bash bin/run.sh io.anserini.index.IndexHnswDenseVectors \ diff --git a/docs/regressions/regressions-dl19-passage.bge-base-en-v1.5.hnsw.cached.md b/docs/regressions/regressions-dl19-passage.bge-base-en-v1.5.hnsw.cached.md index 121fba0b5e..68c21b41a1 100644 --- a/docs/regressions/regressions-dl19-passage.bge-base-en-v1.5.hnsw.cached.md +++ b/docs/regressions/regressions-dl19-passage.bge-base-en-v1.5.hnsw.cached.md @@ -1,12 +1,12 @@ # Anserini Regressions: TREC 2019 Deep Learning Track (Passage) -**Model**: [BGE-base-en-v1.5](https://huggingface.co/BAAI/bge-base-en-v1.5) with HNSW indexes (using pre-encoded queries) +**Model**: [BGE-base-en-v1.5](https://huggingface.co/BAAI/bge-base-en-v1.5) with HNSW indexes (using cached queries) This page describes regression experiments, integrated into Anserini's regression testing framework, using the [BGE-base-en-v1.5](https://huggingface.co/BAAI/bge-base-en-v1.5) model on the [TREC 2019 Deep Learning Track passage ranking task](https://trec.nist.gov/data/deep2019.html), as described in the following paper: > Shitao Xiao, Zheng Liu, Peitian Zhang, and Niklas Muennighoff. [C-Pack: Packaged Resources To Advance General Chinese Embedding.](https://arxiv.org/abs/2309.07597) _arXiv:2309.07597_, 2023. -In these experiments, we are using pre-encoded queries (i.e., cached results of query encoding). +In these experiments, we are using cached queries (i.e., cached results of query encoding). Note that the NIST relevance judgments provide far more relevant passages per topic, unlike the "sparse" judgments provided by Microsoft (these are sometimes called "dense" judgments to emphasize this contrast). For additional instructions on working with MS MARCO passage collection, refer to [this page](experiments-msmarco-passage.md). diff --git a/docs/regressions/regressions-dl19-passage.cohere-embed-english-v3.0.hnsw-int8.cached.md b/docs/regressions/regressions-dl19-passage.cohere-embed-english-v3.0.hnsw-int8.cached.md index 4d1611a571..0b315fcb0f 100644 --- a/docs/regressions/regressions-dl19-passage.cohere-embed-english-v3.0.hnsw-int8.cached.md +++ b/docs/regressions/regressions-dl19-passage.cohere-embed-english-v3.0.hnsw-int8.cached.md @@ -1,10 +1,10 @@ # Anserini Regressions: TREC 2019 Deep Learning Track (Passage) -**Model**: [Cohere embed-english-v3.0](https://docs.cohere.com/reference/embed) with HNSW quantized indexes (using pre-encoded queries) +**Model**: [Cohere embed-english-v3.0](https://docs.cohere.com/reference/embed) with quantized HNSW indexes (using cached queries) This page describes regression experiments, integrated into Anserini's regression testing framework, using the [Cohere embed-english-v3.0](https://docs.cohere.com/reference/embed) model on the [TREC 2019 Deep Learning Track passage ranking task](https://trec.nist.gov/data/deep2019.html). -In these experiments, we are using pre-encoded queries (i.e., cached results of query encoding). +In these experiments, we are using cached queries (i.e., cached results of query encoding). The exact configurations for these regressions are stored in [this YAML file](../../src/main/resources/regression/dl19-passage.cohere-embed-english-v3.0.hnsw-int8.cached.yaml). Note that this page is automatically generated from [this template](../../src/main/resources/docgen/templates/dl19-passage.cohere-embed-english-v3.0.hnsw-int8.cached.template) as part of Anserini's regression pipeline, so do not modify this page directly; modify the template instead and then run `bin/build.sh` to rebuild the documentation. @@ -44,7 +44,7 @@ python src/main/python/run_regression.py --index --verify --search --regression ## Indexing -Sample indexing command, building HNSW indexes: +Sample indexing command, building quantized HNSW indexes: ```bash bin/run.sh io.anserini.index.IndexHnswDenseVectors \ diff --git a/docs/regressions/regressions-dl19-passage.cohere-embed-english-v3.0.hnsw.cached.md b/docs/regressions/regressions-dl19-passage.cohere-embed-english-v3.0.hnsw.cached.md index 0ff20b8508..5790a6f827 100644 --- a/docs/regressions/regressions-dl19-passage.cohere-embed-english-v3.0.hnsw.cached.md +++ b/docs/regressions/regressions-dl19-passage.cohere-embed-english-v3.0.hnsw.cached.md @@ -1,10 +1,10 @@ # Anserini Regressions: TREC 2019 Deep Learning Track (Passage) -**Model**: [Cohere embed-english-v3.0](https://docs.cohere.com/reference/embed) with HNSW indexes (using pre-encoded queries) +**Model**: [Cohere embed-english-v3.0](https://docs.cohere.com/reference/embed) with HNSW indexes (using cached queries) This page describes regression experiments, integrated into Anserini's regression testing framework, using the [Cohere embed-english-v3.0](https://docs.cohere.com/reference/embed) model on the [TREC 2019 Deep Learning Track passage ranking task](https://trec.nist.gov/data/deep2019.html). -In these experiments, we are using pre-encoded queries (i.e., cached results of query encoding). +In these experiments, we are using cached queries (i.e., cached results of query encoding). The exact configurations for these regressions are stored in [this YAML file](../../src/main/resources/regression/dl19-passage.cohere-embed-english-v3.0.hnsw.cached.yaml). Note that this page is automatically generated from [this template](../../src/main/resources/docgen/templates/dl19-passage.cohere-embed-english-v3.0.hnsw.cached.template) as part of Anserini's regression pipeline, so do not modify this page directly; modify the template instead and then run `bin/build.sh` to rebuild the documentation. diff --git a/docs/regressions/regressions-dl19-passage.cos-dpr-distil.fw.md b/docs/regressions/regressions-dl19-passage.cos-dpr-distil.fw.md index 2f214ce256..2dde972b80 100644 --- a/docs/regressions/regressions-dl19-passage.cos-dpr-distil.fw.md +++ b/docs/regressions/regressions-dl19-passage.cos-dpr-distil.fw.md @@ -1,9 +1,9 @@ # Anserini Regressions: TREC 2019 Deep Learning Track (Passage) -**Model**: cosDPR-distil with inverted indexes using the "fake-words" technique (q=40); pre-encoded queries +**Model**: cosDPR-distil with inverted indexes using the "fake-words" technique (q=40) using cached queries This page describes regression experiments, integrated into Anserini's regression testing framework, using the cosDPR-distil model on the [TREC 2019 Deep Learning Track passage ranking task](https://trec.nist.gov/data/deep2019.html). -In these experiments, we are using pre-encoded queries (i.e., cached results of query encoding). +In these experiments, we are using cached queries (i.e., cached results of query encoding). Note that the NIST relevance judgments provide far more relevant passages per topic, unlike the "sparse" judgments provided by Microsoft (these are sometimes called "dense" judgments to emphasize this contrast). For additional instructions on working with MS MARCO passage collection, refer to [this page](experiments-msmarco-passage.md). diff --git a/docs/regressions/regressions-dl19-passage.cos-dpr-distil.hnsw-int8.cached.md b/docs/regressions/regressions-dl19-passage.cos-dpr-distil.hnsw-int8.cached.md index c7801207d0..c2eb13d4c0 100644 --- a/docs/regressions/regressions-dl19-passage.cos-dpr-distil.hnsw-int8.cached.md +++ b/docs/regressions/regressions-dl19-passage.cos-dpr-distil.hnsw-int8.cached.md @@ -1,12 +1,12 @@ # Anserini Regressions: TREC 2019 Deep Learning Track (Passage) -**Model**: cosDPR-distil with HNSW quantized indexes (using pre-encoded queries) +**Model**: cosDPR-distil with quantized HNSW indexes (using cached queries) This page describes regression experiments, integrated into Anserini's regression testing framework, using the cosDPR-distil model on the [TREC 2019 Deep Learning Track passage ranking task](https://trec.nist.gov/data/deep2019.html), as described in the following paper: > Xueguang Ma, Tommaso Teofili, and Jimmy Lin. [Anserini Gets Dense Retrieval: Integration of Lucene's HNSW Indexes.](https://dl.acm.org/doi/10.1145/3583780.3615112) _Proceedings of the 32nd International Conference on Information and Knowledge Management (CIKM 2023)_, October 2023, pages 5366–5370, Birmingham, the United Kingdom. -In these experiments, we are using pre-encoded queries (i.e., cached results of query encoding). +In these experiments, we are using cached queries (i.e., cached results of query encoding). Note that the NIST relevance judgments provide far more relevant passages per topic, unlike the "sparse" judgments provided by Microsoft (these are sometimes called "dense" judgments to emphasize this contrast). For additional instructions on working with MS MARCO passage collection, refer to [this page](experiments-msmarco-passage.md). @@ -49,7 +49,7 @@ python src/main/python/run_regression.py --index --verify --search --regression ## Indexing -Sample indexing command, building HNSW indexes: +Sample indexing command, building quantized HNSW indexes: ```bash bin/run.sh io.anserini.index.IndexHnswDenseVectors \ diff --git a/docs/regressions/regressions-dl19-passage.cos-dpr-distil.hnsw-int8.onnx.md b/docs/regressions/regressions-dl19-passage.cos-dpr-distil.hnsw-int8.onnx.md index 2915f66a8c..6b3b8fb2d1 100644 --- a/docs/regressions/regressions-dl19-passage.cos-dpr-distil.hnsw-int8.onnx.md +++ b/docs/regressions/regressions-dl19-passage.cos-dpr-distil.hnsw-int8.onnx.md @@ -1,6 +1,6 @@ # Anserini Regressions: TREC 2019 Deep Learning Track (Passage) -**Model**: cosDPR-distil with HNSW quantized indexes (using ONNX for on-the-fly query encoding) +**Model**: cosDPR-distil with quantized HNSW indexes (using ONNX for on-the-fly query encoding) This page describes regression experiments, integrated into Anserini's regression testing framework, using the cosDPR-distil model on the [TREC 2019 Deep Learning Track passage ranking task](https://trec.nist.gov/data/deep2019.html), as described in the following paper: @@ -49,7 +49,7 @@ python src/main/python/run_regression.py --index --verify --search --regression ## Indexing -Sample indexing command, building HNSW indexes: +Sample indexing command, building quantized HNSW indexes: ```bash bin/run.sh io.anserini.index.IndexHnswDenseVectors \ diff --git a/docs/regressions/regressions-dl19-passage.cos-dpr-distil.hnsw.cached.md b/docs/regressions/regressions-dl19-passage.cos-dpr-distil.hnsw.cached.md index b748551771..3f49b4990a 100644 --- a/docs/regressions/regressions-dl19-passage.cos-dpr-distil.hnsw.cached.md +++ b/docs/regressions/regressions-dl19-passage.cos-dpr-distil.hnsw.cached.md @@ -1,12 +1,12 @@ # Anserini Regressions: TREC 2019 Deep Learning Track (Passage) -**Model**: cosDPR-distil with HNSW indexes (using pre-encoded queries) +**Model**: cosDPR-distil with HNSW indexes (using cached queries) This page describes regression experiments, integrated into Anserini's regression testing framework, using the cosDPR-distil model on the [TREC 2019 Deep Learning Track passage ranking task](https://trec.nist.gov/data/deep2019.html), as described in the following paper: > Xueguang Ma, Tommaso Teofili, and Jimmy Lin. [Anserini Gets Dense Retrieval: Integration of Lucene's HNSW Indexes.](https://dl.acm.org/doi/10.1145/3583780.3615112) _Proceedings of the 32nd International Conference on Information and Knowledge Management (CIKM 2023)_, October 2023, pages 5366–5370, Birmingham, the United Kingdom. -In these experiments, we are using pre-encoded queries (i.e., cached results of query encoding). +In these experiments, we are using cached queries (i.e., cached results of query encoding). Note that the NIST relevance judgments provide far more relevant passages per topic, unlike the "sparse" judgments provided by Microsoft (these are sometimes called "dense" judgments to emphasize this contrast). For additional instructions on working with MS MARCO passage collection, refer to [this page](experiments-msmarco-passage.md). diff --git a/docs/regressions/regressions-dl19-passage.cos-dpr-distil.lexlsh.md b/docs/regressions/regressions-dl19-passage.cos-dpr-distil.lexlsh.md index 6ac6ed1a66..03ea30c25e 100644 --- a/docs/regressions/regressions-dl19-passage.cos-dpr-distil.lexlsh.md +++ b/docs/regressions/regressions-dl19-passage.cos-dpr-distil.lexlsh.md @@ -1,9 +1,9 @@ # Anserini Regressions: TREC 2019 Deep Learning Track (Passage) -**Model**: cosDPR-distil with inverted indexes using the "LexLSH" technique (b=600); pre-encoded queries +**Model**: cosDPR-distil with inverted indexes using the "LexLSH" technique (b=600) using cached queries This page describes regression experiments, integrated into Anserini's regression testing framework, using the cosDPR-distil model on the [TREC 2019 Deep Learning Track passage ranking task](https://trec.nist.gov/data/deep2019.html). -In these experiments, we are using pre-encoded queries (i.e., cached results of query encoding). +In these experiments, we are using cached queries (i.e., cached results of query encoding). Note that the NIST relevance judgments provide far more relevant passages per topic, unlike the "sparse" judgments provided by Microsoft (these are sometimes called "dense" judgments to emphasize this contrast). For additional instructions on working with MS MARCO passage collection, refer to [this page](experiments-msmarco-passage.md). diff --git a/docs/regressions/regressions-dl19-passage.openai-ada2.hnsw-int8.cached.md b/docs/regressions/regressions-dl19-passage.openai-ada2.hnsw-int8.cached.md index 9b0807d589..f9dc73b958 100644 --- a/docs/regressions/regressions-dl19-passage.openai-ada2.hnsw-int8.cached.md +++ b/docs/regressions/regressions-dl19-passage.openai-ada2.hnsw-int8.cached.md @@ -1,12 +1,12 @@ # Anserini Regressions: TREC 2019 Deep Learning Track (Passage) -**Model**: OpenAI-ada2 embeddings (using pre-encoded queries) with HNSW quantized indexes +**Model**: OpenAI-ada2 embeddings with quantized HNSW indexes (using cached queries) This page describes regression experiments, integrated into Anserini's regression testing framework, using OpenAI-ada2 embeddings on the [TREC 2019 Deep Learning Track passage ranking task](https://trec.nist.gov/data/deep2019.html), as described in the following paper: > Jimmy Lin, Ronak Pradeep, Tommaso Teofili, and Jasper Xian. [Vector Search with OpenAI Embeddings: Lucene Is All You Need.](https://arxiv.org/abs/2308.14963) _arXiv:2308.14963_, 2023. -In these experiments, we are using pre-encoded queries (i.e., cached results of query encoding). +In these experiments, we are using cached queries (i.e., cached results of query encoding). Note that the NIST relevance judgments provide far more relevant passages per topic, unlike the "sparse" judgments provided by Microsoft (these are sometimes called "dense" judgments to emphasize this contrast). For additional instructions on working with MS MARCO passage collection, refer to [this page](experiments-msmarco-passage.md). @@ -49,7 +49,7 @@ python src/main/python/run_regression.py --index --verify --search --regression ## Indexing -Sample indexing command, building HNSW indexes: +Sample indexing command, building quantized HNSW indexes: ```bash bin/run.sh io.anserini.index.IndexHnswDenseVectors \ diff --git a/docs/regressions/regressions-dl19-passage.openai-ada2.hnsw.cached.md b/docs/regressions/regressions-dl19-passage.openai-ada2.hnsw.cached.md index 36fb78172d..e962fb3cae 100644 --- a/docs/regressions/regressions-dl19-passage.openai-ada2.hnsw.cached.md +++ b/docs/regressions/regressions-dl19-passage.openai-ada2.hnsw.cached.md @@ -1,12 +1,12 @@ # Anserini Regressions: TREC 2019 Deep Learning Track (Passage) -**Model**: OpenAI-ada2 embeddings (using pre-encoded queries) with HNSW indexes +**Model**: OpenAI-ada2 embeddings with HNSW indexes (using cached queries) This page describes regression experiments, integrated into Anserini's regression testing framework, using OpenAI-ada2 embeddings on the [TREC 2019 Deep Learning Track passage ranking task](https://trec.nist.gov/data/deep2019.html), as described in the following paper: > Jimmy Lin, Ronak Pradeep, Tommaso Teofili, and Jasper Xian. [Vector Search with OpenAI Embeddings: Lucene Is All You Need.](https://arxiv.org/abs/2308.14963) _arXiv:2308.14963_, 2023. -In these experiments, we are using pre-encoded queries (i.e., cached results of query encoding). +In these experiments, we are using cached queries (i.e., cached results of query encoding). Note that the NIST relevance judgments provide far more relevant passages per topic, unlike the "sparse" judgments provided by Microsoft (these are sometimes called "dense" judgments to emphasize this contrast). For additional instructions on working with MS MARCO passage collection, refer to [this page](experiments-msmarco-passage.md). diff --git a/docs/regressions/regressions-dl19-passage.splade-pp-ed.cached.md b/docs/regressions/regressions-dl19-passage.splade-pp-ed.cached.md index f5a136f1d9..74dcdeaa35 100644 --- a/docs/regressions/regressions-dl19-passage.splade-pp-ed.cached.md +++ b/docs/regressions/regressions-dl19-passage.splade-pp-ed.cached.md @@ -1,12 +1,12 @@ # Anserini Regressions: TREC 2019 Deep Learning Track (Passage) -**Model**: SPLADE++ CoCondenser-EnsembleDistil (using pre-encoded queries) +**Model**: SPLADE++ CoCondenser-EnsembleDistil (using cached queries) This page describes regression experiments, integrated into Anserini's regression testing framework, using the [SPLADE++ CoCondenser-EnsembleDistil](https://huggingface.co/naver/splade-cocondenser-ensembledistil) model on the [TREC 2019 Deep Learning Track passage ranking task](https://trec.nist.gov/data/deep2019.html), as described in the following paper: > Thibault Formal, Carlos Lassance, Benjamin Piwowarski, and Stéphane Clinchant. [From Distillation to Hard Negative Sampling: Making Sparse Neural IR Models More Effective.](https://dl.acm.org/doi/10.1145/3477495.3531857) _Proceedings of the 45th International ACM SIGIR Conference on Research and Development in Information Retrieval_, pages 2353–2359. -In these experiments, we are using pre-encoded queries (i.e., cached results of query encoding). +In these experiments, we are using cached queries (i.e., cached results of query encoding). Note that the NIST relevance judgments provide far more relevant passages per topic, unlike the "sparse" judgments provided by Microsoft (these are sometimes called "dense" judgments to emphasize this contrast). For additional instructions on working with MS MARCO passage collection, refer to [this page](../../docs/experiments-msmarco-passage.md). diff --git a/docs/regressions/regressions-dl19-passage.splade-pp-sd.cached.md b/docs/regressions/regressions-dl19-passage.splade-pp-sd.cached.md index 03550f00cc..cf6ff642ba 100644 --- a/docs/regressions/regressions-dl19-passage.splade-pp-sd.cached.md +++ b/docs/regressions/regressions-dl19-passage.splade-pp-sd.cached.md @@ -1,12 +1,12 @@ # Anserini Regressions: TREC 2019 Deep Learning Track (Passage) -**Model**: SPLADE++ CoCondenser-SelfDistil (using pre-encoded queries) +**Model**: SPLADE++ CoCondenser-SelfDistil (using cached queries) This page describes regression experiments, integrated into Anserini's regression testing framework, using the [SPLADE++ CoCondenser-SelfDistil](https://huggingface.co/naver/splade-cocondenser-selfdistil) model on the [TREC 2019 Deep Learning Track passage ranking task](https://trec.nist.gov/data/deep2019.html), as described in the following paper: > Thibault Formal, Carlos Lassance, Benjamin Piwowarski, and Stéphane Clinchant. [From Distillation to Hard Negative Sampling: Making Sparse Neural IR Models More Effective.](https://dl.acm.org/doi/10.1145/3477495.3531857) _Proceedings of the 45th International ACM SIGIR Conference on Research and Development in Information Retrieval_, pages 2353–2359. -In these experiments, we are using pre-encoded queries (i.e., cached results of query encoding). +In these experiments, we are using cached queries (i.e., cached results of query encoding). Note that the NIST relevance judgments provide far more relevant passages per topic, unlike the "sparse" judgments provided by Microsoft (these are sometimes called "dense" judgments to emphasize this contrast). For additional instructions on working with MS MARCO passage collection, refer to [this page](../../docs/experiments-msmarco-passage.md). diff --git a/docs/regressions/regressions-dl19-passage.unicoil-noexp.cached.md b/docs/regressions/regressions-dl19-passage.unicoil-noexp.cached.md index 97f5745d10..6ff964bd1a 100644 --- a/docs/regressions/regressions-dl19-passage.unicoil-noexp.cached.md +++ b/docs/regressions/regressions-dl19-passage.unicoil-noexp.cached.md @@ -1,6 +1,6 @@ # Anserini Regressions: TREC 2019 Deep Learning Track (Passage) -**Model**: uniCOIL (without any expansions) +**Model**: uniCOIL without any expansions (using cached queries) This page describes regression experiments, integrated into Anserini's regression testing framework, using uniCOIL (without any expansions) on the [TREC 2019 Deep Learning Track passage ranking task](https://trec.nist.gov/data/deep2019.html).. The uniCOIL model is described in the following paper: @@ -10,6 +10,8 @@ The uniCOIL model is described in the following paper: The experiments on this page are not actually reported in the paper. Here, a variant model without expansion is used. +In these experiments, we are using cached queries (i.e., cached results of query encoding). + Note that the NIST relevance judgments provide far more relevant passages per topic, unlike the "sparse" judgments provided by Microsoft (these are sometimes called "dense" judgments to emphasize this contrast). For additional instructions on working with MS MARCO passage collection, refer to [this page](../../docs/experiments-msmarco-passage.md). diff --git a/docs/regressions/regressions-dl19-passage.unicoil.cached.md b/docs/regressions/regressions-dl19-passage.unicoil.cached.md index e4f0d22fb6..e711e59bb6 100644 --- a/docs/regressions/regressions-dl19-passage.unicoil.cached.md +++ b/docs/regressions/regressions-dl19-passage.unicoil.cached.md @@ -1,6 +1,6 @@ # Anserini Regressions: TREC 2019 Deep Learning Track (Passage) -**Model**: uniCOIL (with doc2query-T5 expansions) +**Model**: uniCOIL with doc2query-T5 expansions (using cached queries) This page describes regression experiments, integrated into Anserini's regression testing framework, using uniCOIL (with doc2query-T5 expansions) on the [TREC 2019 Deep Learning Track passage ranking task](https://trec.nist.gov/data/deep2019.html). The uniCOIL model is described in the following paper: @@ -10,6 +10,8 @@ The uniCOIL model is described in the following paper: The experiments on this page are not actually reported in the paper. However, the model is the same. +In these experiments, we are using cached queries (i.e., cached results of query encoding). + Note that the NIST relevance judgments provide far more relevant passages per topic, unlike the "sparse" judgments provided by Microsoft (these are sometimes called "dense" judgments to emphasize this contrast). For additional instructions on working with MS MARCO passage collection, refer to [this page](../../docs/experiments-msmarco-passage.md). diff --git a/docs/regressions/regressions-dl20-passage.bge-base-en-v1.5.hnsw-int8.cached.md b/docs/regressions/regressions-dl20-passage.bge-base-en-v1.5.hnsw-int8.cached.md index 8e396faf37..1389133f8c 100644 --- a/docs/regressions/regressions-dl20-passage.bge-base-en-v1.5.hnsw-int8.cached.md +++ b/docs/regressions/regressions-dl20-passage.bge-base-en-v1.5.hnsw-int8.cached.md @@ -1,12 +1,12 @@ # Anserini Regressions: TREC 2020 Deep Learning Track (Passage) -**Model**: [BGE-base-en-v1.5](https://huggingface.co/BAAI/bge-base-en-v1.5) with HNSW quantized indexes (using pre-encoded queries) +**Model**: [BGE-base-en-v1.5](https://huggingface.co/BAAI/bge-base-en-v1.5) with quantized HNSW indexes (using cached queries) This page describes regression experiments, integrated into Anserini's regression testing framework, using the [BGE-base-en-v1.5](https://huggingface.co/BAAI/bge-base-en-v1.5) model on the [TREC 2020 Deep Learning Track passage ranking task](https://trec.nist.gov/data/deep2019.html), as described in the following paper: > Shitao Xiao, Zheng Liu, Peitian Zhang, and Niklas Muennighoff. [C-Pack: Packaged Resources To Advance General Chinese Embedding.](https://arxiv.org/abs/2309.07597) _arXiv:2309.07597_, 2023. -In these experiments, we are using pre-encoded queries (i.e., cached results of query encoding). +In these experiments, we are using cached queries (i.e., cached results of query encoding). Note that the NIST relevance judgments provide far more relevant passages per topic, unlike the "sparse" judgments provided by Microsoft (these are sometimes called "dense" judgments to emphasize this contrast). For additional instructions on working with MS MARCO passage collection, refer to [this page](experiments-msmarco-passage.md). @@ -49,7 +49,7 @@ python src/main/python/run_regression.py --index --verify --search --regression ## Indexing -Sample indexing command, building HNSW indexes: +Sample indexing command, building quantized HNSW indexes: ```bash bin/run.sh io.anserini.index.IndexHnswDenseVectors \ diff --git a/docs/regressions/regressions-dl20-passage.bge-base-en-v1.5.hnsw-int8.onnx.md b/docs/regressions/regressions-dl20-passage.bge-base-en-v1.5.hnsw-int8.onnx.md index 6070b96b62..8fe0ca8dfd 100644 --- a/docs/regressions/regressions-dl20-passage.bge-base-en-v1.5.hnsw-int8.onnx.md +++ b/docs/regressions/regressions-dl20-passage.bge-base-en-v1.5.hnsw-int8.onnx.md @@ -1,6 +1,6 @@ # Anserini Regressions: TREC 2020 Deep Learning Track (Passage) -**Model**: [BGE-base-en-v1.5](https://huggingface.co/BAAI/bge-base-en-v1.5) with HNSW quantized indexes (using ONNX for on-the-fly query encoding) +**Model**: [BGE-base-en-v1.5](https://huggingface.co/BAAI/bge-base-en-v1.5) with quantized HNSW indexes (using ONNX for on-the-fly query encoding) This page describes regression experiments, integrated into Anserini's regression testing framework, using the [BGE-base-en-v1.5](https://huggingface.co/BAAI/bge-base-en-v1.5) model on the [TREC 2020 Deep Learning Track passage ranking task](https://trec.nist.gov/data/deep2019.html), as described in the following paper: @@ -49,7 +49,7 @@ python src/main/python/run_regression.py --index --verify --search --regression ## Indexing -Sample indexing command, building HNSW indexes: +Sample indexing command, building quantized HNSW indexes: ```bash bin/run.sh io.anserini.index.IndexHnswDenseVectors \ diff --git a/docs/regressions/regressions-dl20-passage.bge-base-en-v1.5.hnsw.cached.md b/docs/regressions/regressions-dl20-passage.bge-base-en-v1.5.hnsw.cached.md index 0d5755f660..0f1eb7dd85 100644 --- a/docs/regressions/regressions-dl20-passage.bge-base-en-v1.5.hnsw.cached.md +++ b/docs/regressions/regressions-dl20-passage.bge-base-en-v1.5.hnsw.cached.md @@ -1,12 +1,12 @@ # Anserini Regressions: TREC 2020 Deep Learning Track (Passage) -**Model**: [BGE-base-en-v1.5](https://huggingface.co/BAAI/bge-base-en-v1.5) with HNSW indexes (using pre-encoded queries) +**Model**: [BGE-base-en-v1.5](https://huggingface.co/BAAI/bge-base-en-v1.5) with HNSW indexes (using cached queries) This page describes regression experiments, integrated into Anserini's regression testing framework, using the [BGE-base-en-v1.5](https://huggingface.co/BAAI/bge-base-en-v1.5) model on the [TREC 2020 Deep Learning Track passage ranking task](https://trec.nist.gov/data/deep2019.html), as described in the following paper: > Shitao Xiao, Zheng Liu, Peitian Zhang, and Niklas Muennighoff. [C-Pack: Packaged Resources To Advance General Chinese Embedding.](https://arxiv.org/abs/2309.07597) _arXiv:2309.07597_, 2023. -In these experiments, we are using pre-encoded queries (i.e., cached results of query encoding). +In these experiments, we are using cached queries (i.e., cached results of query encoding). Note that the NIST relevance judgments provide far more relevant passages per topic, unlike the "sparse" judgments provided by Microsoft (these are sometimes called "dense" judgments to emphasize this contrast). For additional instructions on working with MS MARCO passage collection, refer to [this page](experiments-msmarco-passage.md). diff --git a/docs/regressions/regressions-dl20-passage.cohere-embed-english-v3.0.hnsw-int8.cached.md b/docs/regressions/regressions-dl20-passage.cohere-embed-english-v3.0.hnsw-int8.cached.md index e78da982df..fc24f69cf3 100644 --- a/docs/regressions/regressions-dl20-passage.cohere-embed-english-v3.0.hnsw-int8.cached.md +++ b/docs/regressions/regressions-dl20-passage.cohere-embed-english-v3.0.hnsw-int8.cached.md @@ -1,10 +1,10 @@ # Anserini Regressions: TREC 2020 Deep Learning Track (Passage) -**Model**: [Cohere embed-english-v3.0](https://docs.cohere.com/reference/embed) with HNSW quantized indexes (using pre-encoded queries) +**Model**: [Cohere embed-english-v3.0](https://docs.cohere.com/reference/embed) with quantized HNSW indexes (using cached queries) This page describes regression experiments, integrated into Anserini's regression testing framework, using the [Cohere embed-english-v3.0](https://docs.cohere.com/reference/embed) model on the [TREC 2020 Deep Learning Track passage ranking task](https://trec.nist.gov/data/deep2020.html). -In these experiments, we are using pre-encoded queries (i.e., cached results of query encoding). +In these experiments, we are using cached queries (i.e., cached results of query encoding). The exact configurations for these regressions are stored in [this YAML file](../../src/main/resources/regression/dl20-passage.cohere-embed-english-v3.0.hnsw-int8.cached.yaml). Note that this page is automatically generated from [this template](../../src/main/resources/docgen/templates/dl20-passage.cohere-embed-english-v3.0.hnsw-int8.cached.template) as part of Anserini's regression pipeline, so do not modify this page directly; modify the template instead and then run `bin/build.sh` to rebuild the documentation. @@ -44,7 +44,7 @@ python src/main/python/run_regression.py --index --verify --search --regression ## Indexing -Sample indexing command, building HNSW indexes: +Sample indexing command, building quantized HNSW indexes: ```bash bin/run.sh io.anserini.index.IndexHnswDenseVectors \ diff --git a/docs/regressions/regressions-dl20-passage.cohere-embed-english-v3.0.hnsw.cached.md b/docs/regressions/regressions-dl20-passage.cohere-embed-english-v3.0.hnsw.cached.md index fbcbc21945..718b72144d 100644 --- a/docs/regressions/regressions-dl20-passage.cohere-embed-english-v3.0.hnsw.cached.md +++ b/docs/regressions/regressions-dl20-passage.cohere-embed-english-v3.0.hnsw.cached.md @@ -1,10 +1,10 @@ # Anserini Regressions: TREC 2020 Deep Learning Track (Passage) -**Model**: [Cohere embed-english-v3.0](https://docs.cohere.com/reference/embed) with HNSW indexes (using pre-encoded queries) +**Model**: [Cohere embed-english-v3.0](https://docs.cohere.com/reference/embed) with HNSW indexes (using cached queries) This page describes regression experiments, integrated into Anserini's regression testing framework, using the [Cohere embed-english-v3.0](https://docs.cohere.com/reference/embed) model on the [TREC 2020 Deep Learning Track passage ranking task](https://trec.nist.gov/data/deep2020.html). -In these experiments, we are using pre-encoded queries (i.e., cached results of query encoding). +In these experiments, we are using cached queries (i.e., cached results of query encoding). The exact configurations for these regressions are stored in [this YAML file](../../src/main/resources/regression/dl20-passage.cohere-embed-english-v3.0.hnsw.cached.yaml). Note that this page is automatically generated from [this template](../../src/main/resources/docgen/templates/dl20-passage.cohere-embed-english-v3.0.hnsw.cached.template) as part of Anserini's regression pipeline, so do not modify this page directly; modify the template instead and then run `bin/build.sh` to rebuild the documentation. diff --git a/docs/regressions/regressions-dl20-passage.cos-dpr-distil.fw.md b/docs/regressions/regressions-dl20-passage.cos-dpr-distil.fw.md index 6b10fbb8cd..b1123e4190 100644 --- a/docs/regressions/regressions-dl20-passage.cos-dpr-distil.fw.md +++ b/docs/regressions/regressions-dl20-passage.cos-dpr-distil.fw.md @@ -1,9 +1,9 @@ # Anserini Regressions: TREC 2020 Deep Learning Track (Passage) -**Model**: cosDPR-distil with inverted indexes using the "fake-words" technique (q=40); pre-encoded queries +**Model**: cosDPR-distil with inverted indexes using the "fake-words" technique (q=40) using cached queries This page describes regression experiments, integrated into Anserini's regression testing framework, using the cosDPR-distil model on the [TREC 2020 Deep Learning Track passage ranking task](https://trec.nist.gov/data/deep2019.html). -In these experiments, we are using pre-encoded queries (i.e., cached results of query encoding). +In these experiments, we are using cached queries (i.e., cached results of query encoding). Note that the NIST relevance judgments provide far more relevant passages per topic, unlike the "sparse" judgments provided by Microsoft (these are sometimes called "dense" judgments to emphasize this contrast). For additional instructions on working with MS MARCO passage collection, refer to [this page](experiments-msmarco-passage.md). diff --git a/docs/regressions/regressions-dl20-passage.cos-dpr-distil.hnsw-int8.cached.md b/docs/regressions/regressions-dl20-passage.cos-dpr-distil.hnsw-int8.cached.md index ce2ed33445..a66f2fbc4e 100644 --- a/docs/regressions/regressions-dl20-passage.cos-dpr-distil.hnsw-int8.cached.md +++ b/docs/regressions/regressions-dl20-passage.cos-dpr-distil.hnsw-int8.cached.md @@ -1,12 +1,12 @@ # Anserini Regressions: TREC 2020 Deep Learning Track (Passage) -**Model**: cosDPR-distil with HNSW quantized indexes (using pre-encoded queries) +**Model**: cosDPR-distil with quantized HNSW indexes (using cached queries) This page describes regression experiments, integrated into Anserini's regression testing framework, using the cosDPR-distil model on the [TREC 2020 Deep Learning Track passage ranking task](https://trec.nist.gov/data/deep2019.html), as described in the following paper: > Xueguang Ma, Tommaso Teofili, and Jimmy Lin. [Anserini Gets Dense Retrieval: Integration of Lucene's HNSW Indexes.](https://dl.acm.org/doi/10.1145/3583780.3615112) _Proceedings of the 32nd International Conference on Information and Knowledge Management (CIKM 2023)_, October 2023, pages 5366–5370, Birmingham, the United Kingdom. -In these experiments, we are using pre-encoded queries (i.e., cached results of query encoding). +In these experiments, we are using cached queries (i.e., cached results of query encoding). Note that the NIST relevance judgments provide far more relevant passages per topic, unlike the "sparse" judgments provided by Microsoft (these are sometimes called "dense" judgments to emphasize this contrast). For additional instructions on working with MS MARCO passage collection, refer to [this page](experiments-msmarco-passage.md). @@ -49,7 +49,7 @@ python src/main/python/run_regression.py --index --verify --search --regression ## Indexing -Sample indexing command, building HNSW indexes: +Sample indexing command, building quantized HNSW indexes: ```bash bin/run.sh io.anserini.index.IndexHnswDenseVectors \ diff --git a/docs/regressions/regressions-dl20-passage.cos-dpr-distil.hnsw-int8.onnx.md b/docs/regressions/regressions-dl20-passage.cos-dpr-distil.hnsw-int8.onnx.md index 99d8208b33..0a5031af13 100644 --- a/docs/regressions/regressions-dl20-passage.cos-dpr-distil.hnsw-int8.onnx.md +++ b/docs/regressions/regressions-dl20-passage.cos-dpr-distil.hnsw-int8.onnx.md @@ -1,6 +1,6 @@ # Anserini Regressions: TREC 2020 Deep Learning Track (Passage) -**Model**: cosDPR-distil with HNSW quantized indexes (using ONNX for on-the-fly query encoding) +**Model**: cosDPR-distil with quantized HNSW indexes (using ONNX for on-the-fly query encoding) This page describes regression experiments, integrated into Anserini's regression testing framework, using the cosDPR-distil model on the [TREC 2020 Deep Learning Track passage ranking task](https://trec.nist.gov/data/deep2019.html), as described in the following paper: @@ -49,7 +49,7 @@ python src/main/python/run_regression.py --index --verify --search --regression ## Indexing -Sample indexing command, building HNSW indexes: +Sample indexing command, building quantized HNSW indexes: ```bash bin/run.sh io.anserini.index.IndexHnswDenseVectors \ diff --git a/docs/regressions/regressions-dl20-passage.cos-dpr-distil.hnsw.cached.md b/docs/regressions/regressions-dl20-passage.cos-dpr-distil.hnsw.cached.md index 98773ce4a2..f90a87fc8c 100644 --- a/docs/regressions/regressions-dl20-passage.cos-dpr-distil.hnsw.cached.md +++ b/docs/regressions/regressions-dl20-passage.cos-dpr-distil.hnsw.cached.md @@ -1,12 +1,12 @@ # Anserini Regressions: TREC 2020 Deep Learning Track (Passage) -**Model**: cosDPR-distil with HNSW indexes (using pre-encoded queries) +**Model**: cosDPR-distil with HNSW indexes (using cached queries) This page describes regression experiments, integrated into Anserini's regression testing framework, using the cosDPR-distil model on the [TREC 2020 Deep Learning Track passage ranking task](https://trec.nist.gov/data/deep2019.html), as described in the following paper: > Xueguang Ma, Tommaso Teofili, and Jimmy Lin. [Anserini Gets Dense Retrieval: Integration of Lucene's HNSW Indexes.](https://dl.acm.org/doi/10.1145/3583780.3615112) _Proceedings of the 32nd International Conference on Information and Knowledge Management (CIKM 2023)_, October 2023, pages 5366–5370, Birmingham, the United Kingdom. -In these experiments, we are using pre-encoded queries (i.e., cached results of query encoding). +In these experiments, we are using cached queries (i.e., cached results of query encoding). Note that the NIST relevance judgments provide far more relevant passages per topic, unlike the "sparse" judgments provided by Microsoft (these are sometimes called "dense" judgments to emphasize this contrast). For additional instructions on working with MS MARCO passage collection, refer to [this page](experiments-msmarco-passage.md). diff --git a/docs/regressions/regressions-dl20-passage.cos-dpr-distil.lexlsh.md b/docs/regressions/regressions-dl20-passage.cos-dpr-distil.lexlsh.md index c44cc8dae9..57524142b0 100644 --- a/docs/regressions/regressions-dl20-passage.cos-dpr-distil.lexlsh.md +++ b/docs/regressions/regressions-dl20-passage.cos-dpr-distil.lexlsh.md @@ -1,9 +1,9 @@ # Anserini Regressions: TREC 2020 Deep Learning Track (Passage) -**Model**: cosDPR-distil with inverted indexes using the "LexLSH" technique (b=600); pre-encoded queries +**Model**: cosDPR-distil with inverted indexes using the "LexLSH" technique (b=600) using cached queries This page describes regression experiments, integrated into Anserini's regression testing framework, using the cosDPR-distil model on the [TREC 2020 Deep Learning Track passage ranking task](https://trec.nist.gov/data/deep2019.html). -In these experiments, we are using pre-encoded queries (i.e., cached results of query encoding). +In these experiments, we are using cached queries (i.e., cached results of query encoding). Note that the NIST relevance judgments provide far more relevant passages per topic, unlike the "sparse" judgments provided by Microsoft (these are sometimes called "dense" judgments to emphasize this contrast). For additional instructions on working with MS MARCO passage collection, refer to [this page](experiments-msmarco-passage.md). diff --git a/docs/regressions/regressions-dl20-passage.openai-ada2.hnsw-int8.cached.md b/docs/regressions/regressions-dl20-passage.openai-ada2.hnsw-int8.cached.md index e949565140..08cb273595 100644 --- a/docs/regressions/regressions-dl20-passage.openai-ada2.hnsw-int8.cached.md +++ b/docs/regressions/regressions-dl20-passage.openai-ada2.hnsw-int8.cached.md @@ -1,12 +1,12 @@ # Anserini Regressions: TREC 2020 Deep Learning Track (Passage) -**Model**: OpenAI-ada2 embeddings (using pre-encoded queries) with HNSW quantized indexes +**Model**: OpenAI-ada2 embeddings with quantized HNSW indexes (using cached queries) This page describes regression experiments, integrated into Anserini's regression testing framework, using OpenAI-ada2 embeddings on the [TREC 2020 Deep Learning Track passage ranking task](https://trec.nist.gov/data/deep2019.html), as described in the following paper: > Jimmy Lin, Ronak Pradeep, Tommaso Teofili, and Jasper Xian. [Vector Search with OpenAI Embeddings: Lucene Is All You Need.](https://arxiv.org/abs/2308.14963) _arXiv:2308.14963_, 2023. -In these experiments, we are using pre-encoded queries (i.e., cached results of query encoding). +In these experiments, we are using cached queries (i.e., cached results of query encoding). Note that the NIST relevance judgments provide far more relevant passages per topic, unlike the "sparse" judgments provided by Microsoft (these are sometimes called "dense" judgments to emphasize this contrast). For additional instructions on working with MS MARCO passage collection, refer to [this page](experiments-msmarco-passage.md). @@ -49,7 +49,7 @@ python src/main/python/run_regression.py --index --verify --search --regression ## Indexing -Sample indexing command, building HNSW indexes: +Sample indexing command, building quantized HNSW indexes: ```bash bin/run.sh io.anserini.index.IndexHnswDenseVectors \ diff --git a/docs/regressions/regressions-dl20-passage.openai-ada2.hnsw.cached.md b/docs/regressions/regressions-dl20-passage.openai-ada2.hnsw.cached.md index 9364dca153..72ad94978d 100644 --- a/docs/regressions/regressions-dl20-passage.openai-ada2.hnsw.cached.md +++ b/docs/regressions/regressions-dl20-passage.openai-ada2.hnsw.cached.md @@ -1,12 +1,12 @@ # Anserini Regressions: TREC 2020 Deep Learning Track (Passage) -**Model**: OpenAI-ada2 embeddings (using pre-encoded queries) with HNSW indexes +**Model**: OpenAI-ada2 embeddings with HNSW indexes (using cached queries) This page describes regression experiments, integrated into Anserini's regression testing framework, using OpenAI-ada2 embeddings on the [TREC 2020 Deep Learning Track passage ranking task](https://trec.nist.gov/data/deep2019.html), as described in the following paper: > Jimmy Lin, Ronak Pradeep, Tommaso Teofili, and Jasper Xian. [Vector Search with OpenAI Embeddings: Lucene Is All You Need.](https://arxiv.org/abs/2308.14963) _arXiv:2308.14963_, 2023. -In these experiments, we are using pre-encoded queries (i.e., cached results of query encoding). +In these experiments, we are using cached queries (i.e., cached results of query encoding). Note that the NIST relevance judgments provide far more relevant passages per topic, unlike the "sparse" judgments provided by Microsoft (these are sometimes called "dense" judgments to emphasize this contrast). For additional instructions on working with MS MARCO passage collection, refer to [this page](experiments-msmarco-passage.md). diff --git a/docs/regressions/regressions-dl20-passage.splade-pp-ed.cached.md b/docs/regressions/regressions-dl20-passage.splade-pp-ed.cached.md index 3830420597..5a89bc5d53 100644 --- a/docs/regressions/regressions-dl20-passage.splade-pp-ed.cached.md +++ b/docs/regressions/regressions-dl20-passage.splade-pp-ed.cached.md @@ -1,12 +1,12 @@ # Anserini Regressions: TREC 2020 Deep Learning Track (Passage) -**Model**: SPLADE++ CoCondenser-EnsembleDistil (using pre-encoded queries) +**Model**: SPLADE++ CoCondenser-EnsembleDistil (using cached queries) This page describes regression experiments, integrated into Anserini's regression testing framework, using the [SPLADE++ CoCondenser-EnsembleDistil](https://huggingface.co/naver/splade-cocondenser-ensembledistil) model on the [TREC 2020 Deep Learning Track passage ranking task](https://trec.nist.gov/data/deep2019.html), as described in the following paper: > Thibault Formal, Carlos Lassance, Benjamin Piwowarski, and Stéphane Clinchant. [From Distillation to Hard Negative Sampling: Making Sparse Neural IR Models More Effective.](https://dl.acm.org/doi/10.1145/3477495.3531857) _Proceedings of the 45th International ACM SIGIR Conference on Research and Development in Information Retrieval_, pages 2353–2359. -In these experiments, we are using pre-encoded queries (i.e., cached results of query encoding). +In these experiments, we are using cached queries (i.e., cached results of query encoding). Note that the NIST relevance judgments provide far more relevant passages per topic, unlike the "sparse" judgments provided by Microsoft (these are sometimes called "dense" judgments to emphasize this contrast). For additional instructions on working with MS MARCO passage collection, refer to [this page](../../docs/experiments-msmarco-passage.md). diff --git a/docs/regressions/regressions-dl20-passage.splade-pp-sd.cached.md b/docs/regressions/regressions-dl20-passage.splade-pp-sd.cached.md index 43349eaff1..fc8d3b9531 100644 --- a/docs/regressions/regressions-dl20-passage.splade-pp-sd.cached.md +++ b/docs/regressions/regressions-dl20-passage.splade-pp-sd.cached.md @@ -1,12 +1,12 @@ # Anserini Regressions: TREC 2020 Deep Learning Track (Passage) -**Model**: SPLADE++ CoCondenser-SelfDistil (using pre-encoded queries) +**Model**: SPLADE++ CoCondenser-SelfDistil (using cached queries) This page describes regression experiments, integrated into Anserini's regression testing framework, using the [SPLADE++ CoCondenser-SelfDistil](https://huggingface.co/naver/splade-cocondenser-selfdistil) model on the [TREC 2020 Deep Learning Track passage ranking task](https://trec.nist.gov/data/deep2019.html), as described in the following paper: > Thibault Formal, Carlos Lassance, Benjamin Piwowarski, and Stéphane Clinchant. [From Distillation to Hard Negative Sampling: Making Sparse Neural IR Models More Effective.](https://dl.acm.org/doi/10.1145/3477495.3531857) _Proceedings of the 45th International ACM SIGIR Conference on Research and Development in Information Retrieval_, pages 2353–2359. -In these experiments, we are using pre-encoded queries (i.e., cached results of query encoding). +In these experiments, we are using cached queries (i.e., cached results of query encoding). Note that the NIST relevance judgments provide far more relevant passages per topic, unlike the "sparse" judgments provided by Microsoft (these are sometimes called "dense" judgments to emphasize this contrast). For additional instructions on working with MS MARCO passage collection, refer to [this page](../../docs/experiments-msmarco-passage.md). diff --git a/docs/regressions/regressions-dl20-passage.unicoil-noexp.cached.md b/docs/regressions/regressions-dl20-passage.unicoil-noexp.cached.md index d0bfd696e0..91a121928f 100644 --- a/docs/regressions/regressions-dl20-passage.unicoil-noexp.cached.md +++ b/docs/regressions/regressions-dl20-passage.unicoil-noexp.cached.md @@ -1,6 +1,6 @@ # Anserini Regressions: TREC 2020 Deep Learning Track (Passage) -**Model**: uniCOIL (without any expansions) +**Model**: uniCOIL without any expansions (using cached queries) This page describes regression experiments, integrated into Anserini's regression testing framework, using uniCOIL (without any expansions) on the [TREC 2020 Deep Learning Track passage ranking task](https://trec.nist.gov/data/deep2020.html). The uniCOIL model is described in the following paper: @@ -10,6 +10,8 @@ The uniCOIL model is described in the following paper: The experiments on this page are not actually reported in the paper. Here, a variant model without expansion is used. +In these experiments, we are using cached queries (i.e., cached results of query encoding). + Note that the NIST relevance judgments provide far more relevant passages per topic, unlike the "sparse" judgments provided by Microsoft (these are sometimes called "dense" judgments to emphasize this contrast). For additional instructions on working with MS MARCO passage collection, refer to [this page](../../docs/experiments-msmarco-passage.md). diff --git a/docs/regressions/regressions-dl20-passage.unicoil.cached.md b/docs/regressions/regressions-dl20-passage.unicoil.cached.md index eef6ba3a49..5c17bcb314 100644 --- a/docs/regressions/regressions-dl20-passage.unicoil.cached.md +++ b/docs/regressions/regressions-dl20-passage.unicoil.cached.md @@ -1,6 +1,6 @@ # Anserini Regressions: TREC 2020 Deep Learning Track (Passage) -**Model**: uniCOIL (with doc2query-T5 expansions) +**Model**: uniCOIL with doc2query-T5 expansions (using cached queries) This page describes regression experiments, integrated into Anserini's regression testing framework, using uniCOIL (with doc2query-T5 expansions) on the [TREC 2020 Deep Learning Track passage ranking task](https://trec.nist.gov/data/deep2020.html). The uniCOIL model is described in the following paper: @@ -10,6 +10,8 @@ The uniCOIL model is described in the following paper: The experiments on this page are not actually reported in the paper. However, the model is the same. +In these experiments, we are using cached queries (i.e., cached results of query encoding). + Note that the NIST relevance judgments provide far more relevant passages per topic, unlike the "sparse" judgments provided by Microsoft (these are sometimes called "dense" judgments to emphasize this contrast). For additional instructions on working with MS MARCO passage collection, refer to [this page](../../docs/experiments-msmarco-passage.md). diff --git a/docs/regressions/regressions-dl21-passage.splade-pp-ed.cached.md b/docs/regressions/regressions-dl21-passage.splade-pp-ed.cached.md index b6e8122bd7..3db77b6add 100644 --- a/docs/regressions/regressions-dl21-passage.splade-pp-ed.cached.md +++ b/docs/regressions/regressions-dl21-passage.splade-pp-ed.cached.md @@ -1,9 +1,9 @@ # Anserini Regressions: TREC 2021 Deep Learning Track (Passage) -**Model**: SPLADE++ CoCondenser-EnsembleDistil (using pre-encoded queries) +**Model**: SPLADE++ CoCondenser-EnsembleDistil (using cached queries) This page describes regression experiments, integrated into Anserini's regression testing framework, applying the [SPLADE++ CoCondenser-EnsembleDistil](https://huggingface.co/naver/splade-cocondenser-ensembledistil) model to the MS MARCO V2 passage corpus. -Here, we evaluate on the [TREC 2021 Deep Learning Track passage ranking task](https://trec.nist.gov/data/deep2021.html), using pre-encoded queries (i.e., cached results of query encoding). +Here, we evaluate on the [TREC 2021 Deep Learning Track passage ranking task](https://trec.nist.gov/data/deep2021.html), using cached queries (i.e., cached results of query encoding). The model is described in the following paper: diff --git a/docs/regressions/regressions-dl21-passage.splade-pp-sd.cached.md b/docs/regressions/regressions-dl21-passage.splade-pp-sd.cached.md index e8caea1b86..962f07391b 100644 --- a/docs/regressions/regressions-dl21-passage.splade-pp-sd.cached.md +++ b/docs/regressions/regressions-dl21-passage.splade-pp-sd.cached.md @@ -1,9 +1,9 @@ # Anserini Regressions: TREC 2021 Deep Learning Track (Passage) -**Model**: SPLADE++ CoCondenser-SelfDistil (using pre-encoded queries) +**Model**: SPLADE++ CoCondenser-SelfDistil (using cached queries) This page describes regression experiments, integrated into Anserini's regression testing framework, applying the [SPLADE++ CoCondenser-SelfDistil](https://huggingface.co/naver/splade-cocondenser-selfdistil) model to the MS MARCO V2 passage corpus. -Here, we evaluate on the [TREC 2021 Deep Learning Track passage ranking task](https://trec.nist.gov/data/deep2021.html), using pre-encoded queries (i.e., cached results of query encoding). +Here, we evaluate on the [TREC 2021 Deep Learning Track passage ranking task](https://trec.nist.gov/data/deep2021.html), using cached queries (i.e., cached results of query encoding). The model is described in the following paper: diff --git a/docs/regressions/regressions-dl21-passage.unicoil-0shot.cached.md b/docs/regressions/regressions-dl21-passage.unicoil-0shot.cached.md index 1a60846a88..0a59bac94d 100644 --- a/docs/regressions/regressions-dl21-passage.unicoil-0shot.cached.md +++ b/docs/regressions/regressions-dl21-passage.unicoil-0shot.cached.md @@ -146,4 +146,4 @@ With the above commands, you should be able to reproduce the following results: | [DL21 (Passage)](https://microsoft.github.io/msmarco/TREC-Deep-Learning) | 0.7551 | 0.7889 | 0.8096 | This run roughly corresponds to run `d_unicoil0` submitted to the TREC 2021 Deep Learning Track under the "baseline" group. -The difference is that here we are using pre-encoded queries, whereas the official submission performed query encoding on the fly. +The difference is that here we are using cached queries (i.e., cached results of query encoding), whereas the official submission performed query encoding on the fly. diff --git a/docs/regressions/regressions-dl21-passage.unicoil-noexp-0shot.cached.md b/docs/regressions/regressions-dl21-passage.unicoil-noexp-0shot.cached.md index d884f37c59..fee17270b4 100644 --- a/docs/regressions/regressions-dl21-passage.unicoil-noexp-0shot.cached.md +++ b/docs/regressions/regressions-dl21-passage.unicoil-noexp-0shot.cached.md @@ -146,4 +146,4 @@ With the above commands, you should be able to reproduce the following results: | [DL21 (Passage)](https://microsoft.github.io/msmarco/TREC-Deep-Learning) | 0.6897 | 0.7309 | 0.7509 | This run roughly corresponds to run `d_unicoil0` submitted to the TREC 2021 Deep Learning Track under the "baseline" group. -The difference is that here we are using pre-encoded queries, whereas the official submission performed query encoding on the fly. +The difference is that here we are using cached queries (i.e., cached results of query encoding), whereas the official submission performed query encoding on the fly. diff --git a/docs/regressions/regressions-dl22-passage.splade-pp-ed.cached.md b/docs/regressions/regressions-dl22-passage.splade-pp-ed.cached.md index 66ab715451..ab3b025e85 100644 --- a/docs/regressions/regressions-dl22-passage.splade-pp-ed.cached.md +++ b/docs/regressions/regressions-dl22-passage.splade-pp-ed.cached.md @@ -1,9 +1,9 @@ # Anserini Regressions: TREC 2022 Deep Learning Track (Passage) -**Model**: SPLADE++ CoCondenser-EnsembleDistil (using pre-encoded queries) +**Model**: SPLADE++ CoCondenser-EnsembleDistil (using cached queries) This page describes regression experiments, integrated into Anserini's regression testing framework, applying the [SPLADE++ CoCondenser-EnsembleDistil](https://huggingface.co/naver/splade-cocondenser-ensembledistil) model to the MS MARCO V2 passage corpus. -Here, we evaluate on the [TREC 2022 Deep Learning Track passage ranking task](https://trec.nist.gov/data/deep2022.html), using pre-encoded queries (i.e., cached results of query encoding). +Here, we evaluate on the [TREC 2022 Deep Learning Track passage ranking task](https://trec.nist.gov/data/deep2022.html), using cached queries (i.e., cached results of query encoding). The model is described in the following paper: diff --git a/docs/regressions/regressions-dl22-passage.splade-pp-sd.cached.md b/docs/regressions/regressions-dl22-passage.splade-pp-sd.cached.md index 4074ca415e..2d12e890a9 100644 --- a/docs/regressions/regressions-dl22-passage.splade-pp-sd.cached.md +++ b/docs/regressions/regressions-dl22-passage.splade-pp-sd.cached.md @@ -1,9 +1,9 @@ # Anserini Regressions: TREC 2022 Deep Learning Track (Passage) -**Model**: SPLADE++ CoCondenser-SelfDistil (using pre-encoded queries) +**Model**: SPLADE++ CoCondenser-SelfDistil (using cached queries) This page describes regression experiments, integrated into Anserini's regression testing framework, applying the [SPLADE++ CoCondenser-SelfDistil](https://huggingface.co/naver/splade-cocondenser-selfdistil) model to the MS MARCO V2 passage corpus. -Here, we evaluate on the [TREC 2022 Deep Learning Track passage ranking task](https://trec.nist.gov/data/deep2022.html), using pre-encoded queries (i.e., cached results of query encoding). +Here, we evaluate on the [TREC 2022 Deep Learning Track passage ranking task](https://trec.nist.gov/data/deep2022.html), using cached queries (i.e., cached results of query encoding). The model is described in the following paper: diff --git a/docs/regressions/regressions-dl23-passage.splade-pp-ed.cached.md b/docs/regressions/regressions-dl23-passage.splade-pp-ed.cached.md index 94e3d11f2c..4af25d1dee 100644 --- a/docs/regressions/regressions-dl23-passage.splade-pp-ed.cached.md +++ b/docs/regressions/regressions-dl23-passage.splade-pp-ed.cached.md @@ -1,9 +1,9 @@ # Anserini Regressions: TREC 2023 Deep Learning Track (Passage) -**Model**: SPLADE++ CoCondenser-EnsembleDistil (using pre-encoded queries) +**Model**: SPLADE++ CoCondenser-EnsembleDistil (using cached queries) This page describes regression experiments, integrated into Anserini's regression testing framework, applying the [SPLADE++ CoCondenser-EnsembleDistil](https://huggingface.co/naver/splade-cocondenser-ensembledistil) model to the MS MARCO V2 passage corpus. -Here, we evaluate on the [TREC 2023 Deep Learning Track passage ranking task](https://trec.nist.gov/data/deep2023.html), using pre-encoded queries (i.e., cached results of query encoding). +Here, we evaluate on the [TREC 2023 Deep Learning Track passage ranking task](https://trec.nist.gov/data/deep2023.html), using cached queries (i.e., cached results of query encoding). The model is described in the following paper: diff --git a/docs/regressions/regressions-dl23-passage.splade-pp-sd.cached.md b/docs/regressions/regressions-dl23-passage.splade-pp-sd.cached.md index 7c9980faa2..e2f2caba3b 100644 --- a/docs/regressions/regressions-dl23-passage.splade-pp-sd.cached.md +++ b/docs/regressions/regressions-dl23-passage.splade-pp-sd.cached.md @@ -1,9 +1,9 @@ # Anserini Regressions: TREC 2023 Deep Learning Track (Passage) -**Model**: SPLADE++ CoCondenser-SelfDistil (using pre-encoded queries) +**Model**: SPLADE++ CoCondenser-SelfDistil (using cached queries) This page describes regression experiments, integrated into Anserini's regression testing framework, applying the [SPLADE++ CoCondenser-SelfDistil](https://huggingface.co/naver/splade-cocondenser-selfdistil) model to the MS MARCO V2 passage corpus. -Here, we evaluate on the [TREC 2023 Deep Learning Track passage ranking task](https://trec.nist.gov/data/deep2023.html), using pre-encoded queries (i.e., cached results of query encoding). +Here, we evaluate on the [TREC 2023 Deep Learning Track passage ranking task](https://trec.nist.gov/data/deep2023.html), using cached queries (i.e., cached results of query encoding). The model is described in the following paper: diff --git a/docs/regressions/regressions-msmarco-v1-passage.bge-base-en-v1.5.hnsw-int8.cached.md b/docs/regressions/regressions-msmarco-v1-passage.bge-base-en-v1.5.hnsw-int8.cached.md index 33347257f2..3241e99e40 100644 --- a/docs/regressions/regressions-msmarco-v1-passage.bge-base-en-v1.5.hnsw-int8.cached.md +++ b/docs/regressions/regressions-msmarco-v1-passage.bge-base-en-v1.5.hnsw-int8.cached.md @@ -1,12 +1,12 @@ # Anserini Regressions: MS MARCO Passage Ranking -**Model**: [BGE-base-en-v1.5](https://huggingface.co/BAAI/bge-base-en-v1.5) with HNSW quantized indexes (using pre-encoded queries) +**Model**: [BGE-base-en-v1.5](https://huggingface.co/BAAI/bge-base-en-v1.5) with quantized HNSW indexes (using cached queries) This page describes regression experiments, integrated into Anserini's regression testing framework, using the [BGE-base-en-v1.5](https://huggingface.co/BAAI/bge-base-en-v1.5) model on the [MS MARCO passage ranking task](/~https://github.com/microsoft/MSMARCO-Passage-Ranking), as described in the following paper: > Shitao Xiao, Zheng Liu, Peitian Zhang, and Niklas Muennighoff. [C-Pack: Packaged Resources To Advance General Chinese Embedding.](https://arxiv.org/abs/2309.07597) _arXiv:2309.07597_, 2023. -In these experiments, we are using pre-encoded queries (i.e., cached results of query encoding). +In these experiments, we are using cached queries (i.e., cached results of query encoding). The exact configurations for these regressions are stored in [this YAML file](../../src/main/resources/regression/msmarco-v1-passage.bge-base-en-v1.5.hnsw-int8.cached.yaml). Note that this page is automatically generated from [this template](../../src/main/resources/docgen/templates/msmarco-v1-passage.bge-base-en-v1.5.hnsw-int8.cached.template) as part of Anserini's regression pipeline, so do not modify this page directly; modify the template instead and then run `bin/build.sh` to rebuild the documentation. @@ -46,7 +46,7 @@ python src/main/python/run_regression.py --index --verify --search --regression ## Indexing -Sample indexing command, building HNSW indexes: +Sample indexing command, building quantized HNSW indexes: ```bash bin/run.sh io.anserini.index.IndexHnswDenseVectors \ diff --git a/docs/regressions/regressions-msmarco-v1-passage.bge-base-en-v1.5.hnsw-int8.onnx.md b/docs/regressions/regressions-msmarco-v1-passage.bge-base-en-v1.5.hnsw-int8.onnx.md index 2483a681e5..02083f3abe 100644 --- a/docs/regressions/regressions-msmarco-v1-passage.bge-base-en-v1.5.hnsw-int8.onnx.md +++ b/docs/regressions/regressions-msmarco-v1-passage.bge-base-en-v1.5.hnsw-int8.onnx.md @@ -1,6 +1,6 @@ # Anserini Regressions: MS MARCO Passage Ranking -**Model**: [BGE-base-en-v1.5](https://huggingface.co/BAAI/bge-base-en-v1.5) with HNSW quantized indexes (using ONNX for on-the-fly query encoding) +**Model**: [BGE-base-en-v1.5](https://huggingface.co/BAAI/bge-base-en-v1.5) with quantized HNSW indexes (using ONNX for on-the-fly query encoding) This page describes regression experiments, integrated into Anserini's regression testing framework, using the [BGE-base-en-v1.5](https://huggingface.co/BAAI/bge-base-en-v1.5) model on the [MS MARCO passage ranking task](/~https://github.com/microsoft/MSMARCO-Passage-Ranking), as described in the following paper: @@ -46,7 +46,7 @@ python src/main/python/run_regression.py --index --verify --search --regression ## Indexing -Sample indexing command, building HNSW indexes: +Sample indexing command, building quantized HNSW indexes: ```bash bin/run.sh io.anserini.index.IndexHnswDenseVectors \ diff --git a/docs/regressions/regressions-msmarco-v1-passage.bge-base-en-v1.5.hnsw.cached.md b/docs/regressions/regressions-msmarco-v1-passage.bge-base-en-v1.5.hnsw.cached.md index 71a92d9ca4..9ba8bf2c07 100644 --- a/docs/regressions/regressions-msmarco-v1-passage.bge-base-en-v1.5.hnsw.cached.md +++ b/docs/regressions/regressions-msmarco-v1-passage.bge-base-en-v1.5.hnsw.cached.md @@ -1,12 +1,12 @@ # Anserini Regressions: MS MARCO Passage Ranking -**Model**: [BGE-base-en-v1.5](https://huggingface.co/BAAI/bge-base-en-v1.5) with HNSW indexes (using pre-encoded queries) +**Model**: [BGE-base-en-v1.5](https://huggingface.co/BAAI/bge-base-en-v1.5) with HNSW indexes (using cached queries) This page describes regression experiments, integrated into Anserini's regression testing framework, using the [BGE-base-en-v1.5](https://huggingface.co/BAAI/bge-base-en-v1.5) model on the [MS MARCO passage ranking task](/~https://github.com/microsoft/MSMARCO-Passage-Ranking), as described in the following paper: > Shitao Xiao, Zheng Liu, Peitian Zhang, and Niklas Muennighoff. [C-Pack: Packaged Resources To Advance General Chinese Embedding.](https://arxiv.org/abs/2309.07597) _arXiv:2309.07597_, 2023. -In these experiments, we are using pre-encoded queries (i.e., cached results of query encoding). +In these experiments, we are using cached queries (i.e., cached results of query encoding). The exact configurations for these regressions are stored in [this YAML file](../../src/main/resources/regression/msmarco-v1-passage.bge-base-en-v1.5.hnsw.cached.yaml). Note that this page is automatically generated from [this template](../../src/main/resources/docgen/templates/msmarco-v1-passage.bge-base-en-v1.5.hnsw.cached.template) as part of Anserini's regression pipeline, so do not modify this page directly; modify the template instead and then run `bin/build.sh` to rebuild the documentation. diff --git a/docs/regressions/regressions-msmarco-v1-passage.cohere-embed-english-v3.0.hnsw-int8.cached.md b/docs/regressions/regressions-msmarco-v1-passage.cohere-embed-english-v3.0.hnsw-int8.cached.md index 84a22d83dc..80c68013f6 100644 --- a/docs/regressions/regressions-msmarco-v1-passage.cohere-embed-english-v3.0.hnsw-int8.cached.md +++ b/docs/regressions/regressions-msmarco-v1-passage.cohere-embed-english-v3.0.hnsw-int8.cached.md @@ -1,10 +1,10 @@ # Anserini Regressions: MS MARCO Passage Ranking -**Model**: [Cohere embed-english-v3.0](https://docs.cohere.com/reference/embed) with HNSW quantized indexes (using pre-encoded queries) +**Model**: [Cohere embed-english-v3.0](https://docs.cohere.com/reference/embed) with quantized HNSW indexes (using cached queries) This page describes regression experiments, integrated into Anserini's regression testing framework, using the [Cohere embed-english-v3.0](https://docs.cohere.com/reference/embed) model on the [MS MARCO passage ranking task](/~https://github.com/microsoft/MSMARCO-Passage-Ranking). -In these experiments, we are using pre-encoded queries (i.e., cached results of query encoding). +In these experiments, we are using cached queries (i.e., cached results of query encoding). The exact configurations for these regressions are stored in [this YAML file](../../src/main/resources/regression/msmarco-v1-passage.cohere-embed-english-v3.0.hnsw-int8.cached.yaml). Note that this page is automatically generated from [this template](../../src/main/resources/docgen/templates/msmarco-v1-passage.cohere-embed-english-v3.0.hnsw-int8.cached.template) as part of Anserini's regression pipeline, so do not modify this page directly; modify the template instead and then run `bin/build.sh` to rebuild the documentation. @@ -44,7 +44,7 @@ python src/main/python/run_regression.py --index --verify --search --regression ## Indexing -Sample indexing command, building HNSW indexes: +Sample indexing command, building quantized HNSW indexes: ```bash bin/run.sh io.anserini.index.IndexHnswDenseVectors \ diff --git a/docs/regressions/regressions-msmarco-v1-passage.cohere-embed-english-v3.0.hnsw.cached.md b/docs/regressions/regressions-msmarco-v1-passage.cohere-embed-english-v3.0.hnsw.cached.md index 05696275ae..8b7e862312 100644 --- a/docs/regressions/regressions-msmarco-v1-passage.cohere-embed-english-v3.0.hnsw.cached.md +++ b/docs/regressions/regressions-msmarco-v1-passage.cohere-embed-english-v3.0.hnsw.cached.md @@ -1,10 +1,10 @@ # Anserini Regressions: MS MARCO Passage Ranking -**Model**: [Cohere embed-english-v3.0](https://docs.cohere.com/reference/embed) with HNSW indexes (using pre-encoded queries) +**Model**: [Cohere embed-english-v3.0](https://docs.cohere.com/reference/embed) with HNSW indexes (using cached queries) This page describes regression experiments, integrated into Anserini's regression testing framework, using the [Cohere embed-english-v3.0](https://docs.cohere.com/reference/embed) model on the [MS MARCO passage ranking task](/~https://github.com/microsoft/MSMARCO-Passage-Ranking). -In these experiments, we are using pre-encoded queries (i.e., cached results of query encoding). +In these experiments, we are using cached queries (i.e., cached results of query encoding). The exact configurations for these regressions are stored in [this YAML file](../../src/main/resources/regression/msmarco-v1-passage.cohere-embed-english-v3.0.hnsw.cached.yaml). Note that this page is automatically generated from [this template](../../src/main/resources/docgen/templates/msmarco-v1-passage.cohere-embed-english-v3.0.hnsw.cached.template) as part of Anserini's regression pipeline, so do not modify this page directly; modify the template instead and then run `bin/build.sh` to rebuild the documentation. diff --git a/docs/regressions/regressions-msmarco-v1-passage.cos-dpr-distil.fw.md b/docs/regressions/regressions-msmarco-v1-passage.cos-dpr-distil.fw.md index 6649e46dae..4439c21b1c 100644 --- a/docs/regressions/regressions-msmarco-v1-passage.cos-dpr-distil.fw.md +++ b/docs/regressions/regressions-msmarco-v1-passage.cos-dpr-distil.fw.md @@ -1,9 +1,9 @@ # Anserini Regressions: MS MARCO Passage Ranking -**Model**: cosDPR-distil with inverted indexes using the "fake-words" technique (q=40); pre-encoded queries +**Model**: cosDPR-distil with inverted indexes using the "fake-words" technique (q=40) using cached queries This page describes regression experiments, integrated into Anserini's regression testing framework, using the cosDPR-distil model on the [MS MARCO passage ranking task](/~https://github.com/microsoft/MSMARCO-Passage-Ranking). -In these experiments, we are using pre-encoded queries (i.e., cached results of query encoding). +In these experiments, we are using cached queries (i.e., cached results of query encoding). The exact configurations for these regressions are stored in [this YAML file](../../src/main/resources/regression/msmarco-v1-passage.cos-dpr-distil.fw.yaml). Note that this page is automatically generated from [this template](../../src/main/resources/docgen/templates/msmarco-v1-passage.cos-dpr-distil.fw.template) as part of Anserini's regression pipeline, so do not modify this page directly; modify the template instead and then run `bin/build.sh` to rebuild the documentation. diff --git a/docs/regressions/regressions-msmarco-v1-passage.cos-dpr-distil.hnsw-int8.cached.md b/docs/regressions/regressions-msmarco-v1-passage.cos-dpr-distil.hnsw-int8.cached.md index e4ad1fbd6b..1ff6066552 100644 --- a/docs/regressions/regressions-msmarco-v1-passage.cos-dpr-distil.hnsw-int8.cached.md +++ b/docs/regressions/regressions-msmarco-v1-passage.cos-dpr-distil.hnsw-int8.cached.md @@ -1,12 +1,12 @@ # Anserini Regressions: MS MARCO Passage Ranking -**Model**: cosDPR-distil with HNSW quantized indexes (using pre-encoded queries) +**Model**: cosDPR-distil with quantized HNSW indexes (using cached queries) This page describes regression experiments, integrated into Anserini's regression testing framework, using the cosDPR-distil model on the [MS MARCO passage ranking task](/~https://github.com/microsoft/MSMARCO-Passage-Ranking), as described in the following paper: > Xueguang Ma, Tommaso Teofili, and Jimmy Lin. [Anserini Gets Dense Retrieval: Integration of Lucene's HNSW Indexes.](https://dl.acm.org/doi/10.1145/3583780.3615112) _Proceedings of the 32nd International Conference on Information and Knowledge Management (CIKM 2023)_, October 2023, pages 5366–5370, Birmingham, the United Kingdom. -In these experiments, we are using pre-encoded queries (i.e., cached results of query encoding). +In these experiments, we are using cached queries (i.e., cached results of query encoding). The exact configurations for these regressions are stored in [this YAML file](../../src/main/resources/regression/msmarco-v1-passage.cos-dpr-distil.hnsw-int8.cached.yaml). Note that this page is automatically generated from [this template](../../src/main/resources/docgen/templates/msmarco-v1-passage.cos-dpr-distil.hnsw-int8.cached.template) as part of Anserini's regression pipeline, so do not modify this page directly; modify the template instead and then run `bin/build.sh` to rebuild the documentation. @@ -46,7 +46,7 @@ python src/main/python/run_regression.py --index --verify --search --regression ## Indexing -Sample indexing command, building HNSW indexes: +Sample indexing command, building quantized HNSW indexes: ```bash bin/run.sh io.anserini.index.IndexHnswDenseVectors \ diff --git a/docs/regressions/regressions-msmarco-v1-passage.cos-dpr-distil.hnsw-int8.onnx.md b/docs/regressions/regressions-msmarco-v1-passage.cos-dpr-distil.hnsw-int8.onnx.md index f83f7c072d..baa03a30a1 100644 --- a/docs/regressions/regressions-msmarco-v1-passage.cos-dpr-distil.hnsw-int8.onnx.md +++ b/docs/regressions/regressions-msmarco-v1-passage.cos-dpr-distil.hnsw-int8.onnx.md @@ -1,6 +1,6 @@ # Anserini Regressions: MS MARCO Passage Ranking -**Model**: cosDPR-distil with HNSW quantized indexes (using ONNX for on-the-fly query encoding) +**Model**: cosDPR-distil with quantized HNSW indexes (using ONNX for on-the-fly query encoding) This page describes regression experiments, integrated into Anserini's regression testing framework, using the cosDPR-distil model on the [MS MARCO passage ranking task](/~https://github.com/microsoft/MSMARCO-Passage-Ranking), as described in the following paper: @@ -46,7 +46,7 @@ python src/main/python/run_regression.py --index --verify --search --regression ## Indexing -Sample indexing command, building HNSW indexes: +Sample indexing command, building quantized HNSW indexes: ```bash bin/run.sh io.anserini.index.IndexHnswDenseVectors \ diff --git a/docs/regressions/regressions-msmarco-v1-passage.cos-dpr-distil.hnsw.cached.md b/docs/regressions/regressions-msmarco-v1-passage.cos-dpr-distil.hnsw.cached.md index 50e74c205d..b04cf54cc7 100644 --- a/docs/regressions/regressions-msmarco-v1-passage.cos-dpr-distil.hnsw.cached.md +++ b/docs/regressions/regressions-msmarco-v1-passage.cos-dpr-distil.hnsw.cached.md @@ -1,12 +1,12 @@ # Anserini Regressions: MS MARCO Passage Ranking -**Model**: cosDPR-distil with HNSW indexes (using pre-encoded queries) +**Model**: cosDPR-distil with HNSW indexes (using cached queries) This page describes regression experiments, integrated into Anserini's regression testing framework, using the cosDPR-distil model on the [MS MARCO passage ranking task](/~https://github.com/microsoft/MSMARCO-Passage-Ranking), as described in the following paper: > Xueguang Ma, Tommaso Teofili, and Jimmy Lin. [Anserini Gets Dense Retrieval: Integration of Lucene's HNSW Indexes.](https://dl.acm.org/doi/10.1145/3583780.3615112) _Proceedings of the 32nd International Conference on Information and Knowledge Management (CIKM 2023)_, October 2023, pages 5366–5370, Birmingham, the United Kingdom. -In these experiments, we are using pre-encoded queries (i.e., cached results of query encoding). +In these experiments, we are using cached queries (i.e., cached results of query encoding). The exact configurations for these regressions are stored in [this YAML file](../../src/main/resources/regression/msmarco-v1-passage.cos-dpr-distil.hnsw.cached.yaml). Note that this page is automatically generated from [this template](../../src/main/resources/docgen/templates/msmarco-v1-passage.cos-dpr-distil.hnsw.cached.template) as part of Anserini's regression pipeline, so do not modify this page directly; modify the template instead and then run `bin/build.sh` to rebuild the documentation. diff --git a/docs/regressions/regressions-msmarco-v1-passage.cos-dpr-distil.lexlsh.md b/docs/regressions/regressions-msmarco-v1-passage.cos-dpr-distil.lexlsh.md index 46334e6f17..cda211f78f 100644 --- a/docs/regressions/regressions-msmarco-v1-passage.cos-dpr-distil.lexlsh.md +++ b/docs/regressions/regressions-msmarco-v1-passage.cos-dpr-distil.lexlsh.md @@ -1,9 +1,9 @@ # Anserini Regressions: MS MARCO Passage Ranking -**Model**: cosDPR-distil with inverted indexes using the "LexLSH" technique (b=600); pre-encoded queries +**Model**: cosDPR-distil with inverted indexes using the "LexLSH" technique (b=600) using cached queries This page describes regression experiments, integrated into Anserini's regression testing framework, using the cosDPR-distil model on the [MS MARCO passage ranking task](/~https://github.com/microsoft/MSMARCO-Passage-Ranking). -In these experiments, we are using pre-encoded queries (i.e., cached results of query encoding). +In these experiments, we are using cached queries (i.e., cached results of query encoding). The exact configurations for these regressions are stored in [this YAML file](../../src/main/resources/regression/msmarco-v1-passage.cos-dpr-distil.lexlsh.yaml). Note that this page is automatically generated from [this template](../../src/main/resources/docgen/templates/msmarco-v1-passage.cos-dpr-distil.lexlsh.template) as part of Anserini's regression pipeline, so do not modify this page directly; modify the template instead and then run `bin/build.sh` to rebuild the documentation. diff --git a/docs/regressions/regressions-msmarco-v1-passage.deepimpact.cached.md b/docs/regressions/regressions-msmarco-v1-passage.deepimpact.cached.md index 78f3efbc44..01bad446b0 100644 --- a/docs/regressions/regressions-msmarco-v1-passage.deepimpact.cached.md +++ b/docs/regressions/regressions-msmarco-v1-passage.deepimpact.cached.md @@ -1,12 +1,14 @@ # Anserini Regressions: MS MARCO Passage Ranking -**Model**: DeepImpact +**Model**: DeepImpact (using cached queries) This page describes regression experiments, integrated into Anserini's regression testing framework, using DeepImpact on the [MS MARCO passage ranking task](/~https://github.com/microsoft/MSMARCO-Passage-Ranking). The DeepImpact model is described in the following paper: > Antonio Mallia, Omar Khattab, Nicola Tonellotto, and Torsten Suel. [Learning Passage Impacts for Inverted Indexes.](https://dl.acm.org/doi/10.1145/3404835.3463030) _SIGIR 2021_. +In these experiments, we are using cached queries (i.e., cached results of query encoding). + The exact configurations for these regressions are stored in [this YAML file](../../src/main/resources/regression/msmarco-v1-passage.deepimpact.cached.yaml). Note that this page is automatically generated from [this template](../../src/main/resources/docgen/templates/msmarco-v1-passage.deepimpact.cached.template) as part of Anserini's regression pipeline, so do not modify this page directly; modify the template instead. diff --git a/docs/regressions/regressions-msmarco-v1-passage.distill-splade-max.cached.md b/docs/regressions/regressions-msmarco-v1-passage.distill-splade-max.cached.md index b9b4250833..155e884cda 100644 --- a/docs/regressions/regressions-msmarco-v1-passage.distill-splade-max.cached.md +++ b/docs/regressions/regressions-msmarco-v1-passage.distill-splade-max.cached.md @@ -1,12 +1,14 @@ # Anserini Regressions: MS MARCO Passage Ranking -**Model**: DistilSPLADE-max +**Model**: DistilSPLADE-max (using cached queries) This page describes regression experiments, integrated into Anserini's regression testing framework, using the DistilSPLADE-max model from SPLADEv2 on the [MS MARCO passage ranking task](/~https://github.com/microsoft/MSMARCO-Passage-Ranking). The DistilSPLADE-max model is described in the following paper: > Thibault Formal, Carlos Lassance, Benjamin Piwowarski, Stéphane Clinchant. [SPLADE v2: Sparse Lexical and Expansion Model for Information Retrieval.](https://arxiv.org/abs/2109.10086) _arXiv:2109.10086_. +In these experiments, we are using cached queries (i.e., cached results of query encoding). + The exact configurations for these regressions are stored in [this YAML file](../../src/main/resources/regression/msmarco-v1-passage.distill-splade-max.cached.yaml). Note that this page is automatically generated from [this template](../../src/main/resources/docgen/templates/msmarco-v1-passage.distill-splade-max.cached.template) as part of Anserini's regression pipeline, so do not modify this page directly; modify the template instead. diff --git a/docs/regressions/regressions-msmarco-v1-passage.openai-ada2.hnsw-int8.cached.md b/docs/regressions/regressions-msmarco-v1-passage.openai-ada2.hnsw-int8.cached.md index decb5d1010..a8086ddfc1 100644 --- a/docs/regressions/regressions-msmarco-v1-passage.openai-ada2.hnsw-int8.cached.md +++ b/docs/regressions/regressions-msmarco-v1-passage.openai-ada2.hnsw-int8.cached.md @@ -1,12 +1,12 @@ # Anserini Regressions: MS MARCO Passage Ranking -**Model**: OpenAI-ada2 embeddings (using pre-encoded queries) with HNSW quantized indexes +**Model**: OpenAI-ada2 embeddings with quantized HNSW indexes (using cached queries) This page describes regression experiments, integrated into Anserini's regression testing framework, using OpenAI-ada2 embeddings on the [MS MARCO passage ranking task](/~https://github.com/microsoft/MSMARCO-Passage-Ranking), as described in the following paper: > Jimmy Lin, Ronak Pradeep, Tommaso Teofili, and Jasper Xian. [Vector Search with OpenAI Embeddings: Lucene Is All You Need.](https://arxiv.org/abs/2308.14963) _arXiv:2308.14963_, 2023. -In these experiments, we are using pre-encoded queries (i.e., cached results of query encoding). +In these experiments, we are using cached queries (i.e., cached results of query encoding). The exact configurations for these regressions are stored in [this YAML file](../../src/main/resources/regression/msmarco-v1-passage.openai-ada2.hnsw-int8.cached.yaml). Note that this page is automatically generated from [this template](../../src/main/resources/docgen/templates/msmarco-v1-passage.openai-ada2.hnsw-int8.cached.template) as part of Anserini's regression pipeline, so do not modify this page directly; modify the template instead and then run `bin/build.sh` to rebuild the documentation. @@ -46,7 +46,7 @@ python src/main/python/run_regression.py --index --verify --search --regression ## Indexing -Sample indexing command, building HNSW indexes: +Sample indexing command, building quantized HNSW indexes: ```bash bin/run.sh io.anserini.index.IndexHnswDenseVectors \ diff --git a/docs/regressions/regressions-msmarco-v1-passage.openai-ada2.hnsw.cached.md b/docs/regressions/regressions-msmarco-v1-passage.openai-ada2.hnsw.cached.md index 5ae70e245e..d36b9cfdee 100644 --- a/docs/regressions/regressions-msmarco-v1-passage.openai-ada2.hnsw.cached.md +++ b/docs/regressions/regressions-msmarco-v1-passage.openai-ada2.hnsw.cached.md @@ -1,12 +1,12 @@ # Anserini Regressions: MS MARCO Passage Ranking -**Model**: OpenAI-ada2 embeddings (using pre-encoded queries) with HNSW indexes +**Model**: OpenAI-ada2 embeddings with HNSW indexes (using cached queries) This page describes regression experiments, integrated into Anserini's regression testing framework, using OpenAI-ada2 embeddings on the [MS MARCO passage ranking task](/~https://github.com/microsoft/MSMARCO-Passage-Ranking), as described in the following paper: > Jimmy Lin, Ronak Pradeep, Tommaso Teofili, and Jasper Xian. [Vector Search with OpenAI Embeddings: Lucene Is All You Need.](https://arxiv.org/abs/2308.14963) _arXiv:2308.14963_, 2023. -In these experiments, we are using pre-encoded queries (i.e., cached results of query encoding). +In these experiments, we are using cached queries (i.e., cached results of query encoding). The exact configurations for these regressions are stored in [this YAML file](../../src/main/resources/regression/msmarco-v1-passage.openai-ada2.hnsw.cached.yaml). Note that this page is automatically generated from [this template](../../src/main/resources/docgen/templates/msmarco-v1-passage.openai-ada2.hnsw.cached.template) as part of Anserini's regression pipeline, so do not modify this page directly; modify the template instead and then run `bin/build.sh` to rebuild the documentation. diff --git a/docs/regressions/regressions-msmarco-v1-passage.splade-pp-ed.cached.md b/docs/regressions/regressions-msmarco-v1-passage.splade-pp-ed.cached.md index 1a01e405ad..97b895b007 100644 --- a/docs/regressions/regressions-msmarco-v1-passage.splade-pp-ed.cached.md +++ b/docs/regressions/regressions-msmarco-v1-passage.splade-pp-ed.cached.md @@ -1,12 +1,12 @@ # Anserini Regressions: MS MARCO Passage Ranking -**Model**: SPLADE++ CoCondenser-EnsembleDistil (using pre-encoded queries) +**Model**: SPLADE++ CoCondenser-EnsembleDistil (using cached queries) This page describes regression experiments, integrated into Anserini's regression testing framework, using the [SPLADE++ CoCondenser-EnsembleDistil](https://huggingface.co/naver/splade-cocondenser-ensembledistil) model on the [MS MARCO passage ranking task](/~https://github.com/microsoft/MSMARCO-Passage-Ranking), as described in the following paper: > Thibault Formal, Carlos Lassance, Benjamin Piwowarski, and Stéphane Clinchant. [From Distillation to Hard Negative Sampling: Making Sparse Neural IR Models More Effective.](https://dl.acm.org/doi/10.1145/3477495.3531857) _Proceedings of the 45th International ACM SIGIR Conference on Research and Development in Information Retrieval_, pages 2353–2359. -In these experiments, we are using pre-encoded queries (i.e., cached results of query encoding). +In these experiments, we are using cached queries (i.e., cached results of query encoding). The exact configurations for these regressions are stored in [this YAML file](../../src/main/resources/regression/msmarco-v1-passage.splade-pp-ed.cached.yaml). Note that this page is automatically generated from [this template](../../src/main/resources/docgen/templates/msmarco-v1-passage.splade-pp-ed.cached.template) as part of Anserini's regression pipeline, so do not modify this page directly; modify the template instead and then run `bin/build.sh` to rebuild the documentation. diff --git a/docs/regressions/regressions-msmarco-v1-passage.splade-pp-sd.cached.md b/docs/regressions/regressions-msmarco-v1-passage.splade-pp-sd.cached.md index 7e58b01676..48da048a04 100644 --- a/docs/regressions/regressions-msmarco-v1-passage.splade-pp-sd.cached.md +++ b/docs/regressions/regressions-msmarco-v1-passage.splade-pp-sd.cached.md @@ -1,12 +1,12 @@ # Anserini Regressions: MS MARCO Passage Ranking -**Model**: SPLADE++ CoCondenser-SelfDistil (using pre-encoded queries) +**Model**: SPLADE++ CoCondenser-SelfDistil (using cached queries) This page describes regression experiments, integrated into Anserini's regression testing framework, using the [SPLADE++ CoCondenser-SelfDistil](https://huggingface.co/naver/splade-cocondenser-selfdistil) model on the [MS MARCO passage ranking task](/~https://github.com/microsoft/MSMARCO-Passage-Ranking), as described in the following paper: > Thibault Formal, Carlos Lassance, Benjamin Piwowarski, and Stéphane Clinchant. [From Distillation to Hard Negative Sampling: Making Sparse Neural IR Models More Effective.](https://dl.acm.org/doi/10.1145/3477495.3531857) _Proceedings of the 45th International ACM SIGIR Conference on Research and Development in Information Retrieval_, pages 2353–2359. -In these experiments, we are using pre-encoded queries (i.e., cached results of query encoding). +In these experiments, we are using cached queries (i.e., cached results of query encoding). The exact configurations for these regressions are stored in [this YAML file](../../src/main/resources/regression/msmarco-v1-passage.splade-pp-sd.cached.yaml). Note that this page is automatically generated from [this template](../../src/main/resources/docgen/templates/msmarco-v1-passage.splade-pp-sd.cached.template) as part of Anserini's regression pipeline, so do not modify this page directly; modify the template instead and then run `bin/build.sh` to rebuild the documentation. diff --git a/docs/regressions/regressions-msmarco-v1-passage.unicoil-noexp.cached.md b/docs/regressions/regressions-msmarco-v1-passage.unicoil-noexp.cached.md index b3acf1baec..cb66f336ef 100644 --- a/docs/regressions/regressions-msmarco-v1-passage.unicoil-noexp.cached.md +++ b/docs/regressions/regressions-msmarco-v1-passage.unicoil-noexp.cached.md @@ -1,6 +1,6 @@ # Anserini Regressions: MS MARCO Passage Ranking -**Model**: uniCOIL (without any expansions) +**Model**: uniCOIL without any expansions (using cached queries) This page describes regression experiments, integrated into Anserini's regression testing framework, using uniCOIL (without any expansions) on the [MS MARCO passage ranking task](/~https://github.com/microsoft/MSMARCO-Passage-Ranking). The uniCOIL model is described in the following paper: @@ -10,6 +10,8 @@ The uniCOIL model is described in the following paper: The experiments on this page are not actually reported in the paper. Here, a variant model without expansion is used. +In these experiments, we are using cached queries (i.e., cached results of query encoding). + The exact configurations for these regressions are stored in [this YAML file](../../src/main/resources/regression/msmarco-v1-passage.unicoil-noexp.cached.yaml). Note that this page is automatically generated from [this template](../../src/main/resources/docgen/templates/msmarco-v1-passage.unicoil-noexp.cached.template) as part of Anserini's regression pipeline, so do not modify this page directly; modify the template instead and then run `bin/build.sh` to rebuild the documentation. diff --git a/docs/regressions/regressions-msmarco-v1-passage.unicoil-tilde-expansion.cached.md b/docs/regressions/regressions-msmarco-v1-passage.unicoil-tilde-expansion.cached.md index 30f82fd619..a6ef5bab79 100644 --- a/docs/regressions/regressions-msmarco-v1-passage.unicoil-tilde-expansion.cached.md +++ b/docs/regressions/regressions-msmarco-v1-passage.unicoil-tilde-expansion.cached.md @@ -1,12 +1,14 @@ # Anserini Regressions: MS MARCO Passage Ranking -**Model**: uniCOIL (with TILDE expansions) +**Model**: uniCOIL with TILDE expansions (using cached queries) This page describes regression experiments, integrated into Anserini's regression testing framework, using uniCOIL (with TILDE expansions) on the [MS MARCO passage ranking task](/~https://github.com/microsoft/MSMARCO-Passage-Ranking). The uniCOIL+TILDE model is described in the following paper: > Shengyao Zhuang and Guido Zuccon. [Fast Passage Re-ranking with Contextualized Exact Term Matching and Efficient Passage Expansion.](https://arxiv.org/pdf/2108.08513) _arXiv:2108.08513_. +In these experiments, we are using cached queries (i.e., cached results of query encoding). + The exact configurations for these regressions are stored in [this YAML file](../../src/main/resources/regression/msmarco-v1-passage.unicoil-tilde-expansion.cached.yaml). Note that this page is automatically generated from [this template](../../src/main/resources/docgen/templates/msmarco-v1-passage.unicoil-tilde-expansion.cached.template) as part of Anserini's regression pipeline, so do not modify this page directly; modify the template instead. diff --git a/docs/regressions/regressions-msmarco-v1-passage.unicoil.cached.md b/docs/regressions/regressions-msmarco-v1-passage.unicoil.cached.md index 1463240f39..5cf684a05f 100644 --- a/docs/regressions/regressions-msmarco-v1-passage.unicoil.cached.md +++ b/docs/regressions/regressions-msmarco-v1-passage.unicoil.cached.md @@ -1,12 +1,14 @@ # Anserini Regressions: MS MARCO Passage Ranking -**Model**: uniCOIL (with doc2query-T5 expansions) +**Model**: uniCOIL with doc2query-T5 expansions (using cached queries) This page describes regression experiments, integrated into Anserini's regression testing framework, using uniCOIL (with doc2query-T5 expansions) on the [MS MARCO passage ranking task](/~https://github.com/microsoft/MSMARCO-Passage-Ranking). 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.](https://arxiv.org/abs/2106.14807) _arXiv:2106.14807_. +In these experiments, we are using cached queries (i.e., cached results of query encoding). + The exact configurations for these regressions are stored in [this YAML file](../../src/main/resources/regression/msmarco-v1-passage.unicoil.cached.yaml). Note that this page is automatically generated from [this template](../../src/main/resources/docgen/templates/msmarco-v1-passage.unicoil.cached.template) as part of Anserini's regression pipeline, so do not modify this page directly; modify the template instead and then run `bin/build.sh` to rebuild the documentation. diff --git a/docs/regressions/regressions-msmarco-v2-passage.splade-pp-ed.cached.md b/docs/regressions/regressions-msmarco-v2-passage.splade-pp-ed.cached.md index 04f90051fb..c337b0f7a3 100644 --- a/docs/regressions/regressions-msmarco-v2-passage.splade-pp-ed.cached.md +++ b/docs/regressions/regressions-msmarco-v2-passage.splade-pp-ed.cached.md @@ -1,9 +1,9 @@ # Anserini Regressions: MS MARCO (V2) Passage Ranking -**Model**: SPLADE++ CoCondenser-EnsembleDistil (using pre-encoded queries) +**Model**: SPLADE++ CoCondenser-EnsembleDistil (using cached queries) This page describes regression experiments, integrated into Anserini's regression testing framework, applying the [SPLADE++ CoCondenser-EnsembleDistil](https://huggingface.co/naver/splade-cocondenser-ensembledistil) model to the MS MARCO V2 passage corpus. -Here, we evaluate on the dev queries, using pre-encoded queries (i.e., cached results of query encoding). +Here, we evaluate on the dev queries, using cached queries (i.e., cached results of query encoding). The model is described in the following paper: diff --git a/docs/regressions/regressions-msmarco-v2-passage.splade-pp-sd.cached.md b/docs/regressions/regressions-msmarco-v2-passage.splade-pp-sd.cached.md index 76f003ab69..499a7241a9 100644 --- a/docs/regressions/regressions-msmarco-v2-passage.splade-pp-sd.cached.md +++ b/docs/regressions/regressions-msmarco-v2-passage.splade-pp-sd.cached.md @@ -1,9 +1,9 @@ # Anserini Regressions: MS MARCO (V2) Passage Ranking -**Model**: SPLADE++ CoCondenser-SelfDistil (using pre-encoded queries) +**Model**: SPLADE++ CoCondenser-SelfDistil (using cached queries) This page describes regression experiments, integrated into Anserini's regression testing framework, applying the [SPLADE++ CoCondenser-SelfDistil](https://huggingface.co/naver/splade-cocondenser-selfdistil) model to the MS MARCO V2 passage corpus. -Here, we evaluate on the dev queries, using pre-encoded queries (i.e., cached results of query encoding). +Here, we evaluate on the dev queries, using cached queries (i.e., cached results of query encoding). The model is described in the following paper: diff --git a/src/main/resources/docgen/templates/beir-v1.0.0-arguana.splade-pp-ed.cached.template b/src/main/resources/docgen/templates/beir-v1.0.0-arguana.splade-pp-ed.cached.template index 5110060670..444ef4640c 100644 --- a/src/main/resources/docgen/templates/beir-v1.0.0-arguana.splade-pp-ed.cached.template +++ b/src/main/resources/docgen/templates/beir-v1.0.0-arguana.splade-pp-ed.cached.template @@ -1,14 +1,14 @@ # Anserini Regressions: BEIR (v1.0.0) — ArguAna -**Model**: [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) (using pre-encoded queries) +**Model**: [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) (using cached queries) -This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — ArguAna](http://beir.ai/). +This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — ArguAna](http://beir.ai/). The model itself can be download [here](https://huggingface.co/naver/splade-cocondenser-ensembledistil). See the [official SPLADE repo](/~https://github.com/naver/splade) and the following paper for more details: > Thibault Formal, Carlos Lassance, Benjamin Piwowarski, and Stéphane Clinchant. [From Distillation to Hard Negative Sampling: Making Sparse Neural IR Models More Effective.](https://dl.acm.org/doi/10.1145/3477495.3531857) _Proceedings of the 45th International ACM SIGIR Conference on Research and Development in Information Retrieval_, pages 2353–2359. -In these experiments, we are using pre-encoded queries (i.e., cached results of query encoding). +In these experiments, we are using cached queries (i.e., cached results of query encoding). The exact configurations for these regressions are stored in [this YAML file](${yaml}). Note that this page is automatically generated from [this template](${template}) as part of Anserini's regression pipeline, so do not modify this page directly; modify the template instead and then run `bin/build.sh` to rebuild the documentation. diff --git a/src/main/resources/docgen/templates/beir-v1.0.0-arguana.splade-pp-ed.onnx.template b/src/main/resources/docgen/templates/beir-v1.0.0-arguana.splade-pp-ed.onnx.template index a8ceb7550b..267e94937e 100644 --- a/src/main/resources/docgen/templates/beir-v1.0.0-arguana.splade-pp-ed.onnx.template +++ b/src/main/resources/docgen/templates/beir-v1.0.0-arguana.splade-pp-ed.onnx.template @@ -1,8 +1,8 @@ # Anserini Regressions: BEIR (v1.0.0) — ArguAna -**Model**: [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) (using ONNX for on-the-fly query encoding) +**Model**: [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) (using ONNX for on-the-fly query encoding) -This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — ArguAna](http://beir.ai/). +This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — ArguAna](http://beir.ai/). The model itself can be download [here](https://huggingface.co/naver/splade-cocondenser-ensembledistil). See the [official SPLADE repo](/~https://github.com/naver/splade) and the following paper for more details: diff --git a/src/main/resources/docgen/templates/beir-v1.0.0-arguana.unicoil-noexp.cached.template b/src/main/resources/docgen/templates/beir-v1.0.0-arguana.unicoil-noexp.cached.template index 9a49bf09b4..69743e0051 100644 --- a/src/main/resources/docgen/templates/beir-v1.0.0-arguana.unicoil-noexp.cached.template +++ b/src/main/resources/docgen/templates/beir-v1.0.0-arguana.unicoil-noexp.cached.template @@ -1,12 +1,14 @@ # Anserini Regressions: BEIR (v1.0.0) — ArguAna -**Model**: uniCOIL (without any expansions) +**Model**: uniCOIL without any expansions (using cached queries) This page describes regression experiments, integrated into Anserini's regression testing framework, using uniCOIL (without any expansions) on [BEIR (v1.0.0) — ArguAna](http://beir.ai/). 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.](https://arxiv.org/abs/2106.14807) _arXiv:2106.14807_. +In these experiments, we are using cached queries (i.e., cached results of query encoding). + The exact configurations for these regressions are stored in [this YAML file](${yaml}). Note that this page is automatically generated from [this template](${template}) as part of Anserini's regression pipeline, so do not modify this page directly; modify the template instead. diff --git a/src/main/resources/docgen/templates/beir-v1.0.0-bioasq.splade-pp-ed.cached.template b/src/main/resources/docgen/templates/beir-v1.0.0-bioasq.splade-pp-ed.cached.template index 1d3fe31b46..f472645297 100644 --- a/src/main/resources/docgen/templates/beir-v1.0.0-bioasq.splade-pp-ed.cached.template +++ b/src/main/resources/docgen/templates/beir-v1.0.0-bioasq.splade-pp-ed.cached.template @@ -1,14 +1,14 @@ # Anserini Regressions: BEIR (v1.0.0) — BioASQ -**Model**: [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) (using pre-encoded queries) +**Model**: [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) (using cached queries) -This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — BioASQ](http://beir.ai/). +This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — BioASQ](http://beir.ai/). The model itself can be download [here](https://huggingface.co/naver/splade-cocondenser-ensembledistil). See the [official SPLADE repo](/~https://github.com/naver/splade) and the following paper for more details: > Thibault Formal, Carlos Lassance, Benjamin Piwowarski, and Stéphane Clinchant. [From Distillation to Hard Negative Sampling: Making Sparse Neural IR Models More Effective.](https://dl.acm.org/doi/10.1145/3477495.3531857) _Proceedings of the 45th International ACM SIGIR Conference on Research and Development in Information Retrieval_, pages 2353–2359. -In these experiments, we are using pre-encoded queries (i.e., cached results of query encoding). +In these experiments, we are using cached queries (i.e., cached results of query encoding). The exact configurations for these regressions are stored in [this YAML file](${yaml}). Note that this page is automatically generated from [this template](${template}) as part of Anserini's regression pipeline, so do not modify this page directly; modify the template instead and then run `bin/build.sh` to rebuild the documentation. diff --git a/src/main/resources/docgen/templates/beir-v1.0.0-bioasq.splade-pp-ed.onnx.template b/src/main/resources/docgen/templates/beir-v1.0.0-bioasq.splade-pp-ed.onnx.template index f3d75cde7a..a1d2fcff32 100644 --- a/src/main/resources/docgen/templates/beir-v1.0.0-bioasq.splade-pp-ed.onnx.template +++ b/src/main/resources/docgen/templates/beir-v1.0.0-bioasq.splade-pp-ed.onnx.template @@ -1,8 +1,8 @@ # Anserini Regressions: BEIR (v1.0.0) — BioASQ -**Model**: [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) (using ONNX for on-the-fly query encoding) +**Model**: [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) (using ONNX for on-the-fly query encoding) -This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — BioASQ](http://beir.ai/). +This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — BioASQ](http://beir.ai/). The model itself can be download [here](https://huggingface.co/naver/splade-cocondenser-ensembledistil). See the [official SPLADE repo](/~https://github.com/naver/splade) and the following paper for more details: diff --git a/src/main/resources/docgen/templates/beir-v1.0.0-bioasq.unicoil-noexp.cached.template b/src/main/resources/docgen/templates/beir-v1.0.0-bioasq.unicoil-noexp.cached.template index 6af96dd5d6..a4d3d2cca3 100644 --- a/src/main/resources/docgen/templates/beir-v1.0.0-bioasq.unicoil-noexp.cached.template +++ b/src/main/resources/docgen/templates/beir-v1.0.0-bioasq.unicoil-noexp.cached.template @@ -1,12 +1,14 @@ # Anserini Regressions: BEIR (v1.0.0) — BioASQ -**Model**: uniCOIL (without any expansions) +**Model**: uniCOIL without any expansions (using cached queries) This page describes regression experiments, integrated into Anserini's regression testing framework, using uniCOIL (without any expansions) on [BEIR (v1.0.0) — BioASQ](http://beir.ai/). 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.](https://arxiv.org/abs/2106.14807) _arXiv:2106.14807_. +In these experiments, we are using cached queries (i.e., cached results of query encoding). + The exact configurations for these regressions are stored in [this YAML file](${yaml}). Note that this page is automatically generated from [this template](${template}) as part of Anserini's regression pipeline, so do not modify this page directly; modify the template instead. diff --git a/src/main/resources/docgen/templates/beir-v1.0.0-climate-fever.splade-pp-ed.cached.template b/src/main/resources/docgen/templates/beir-v1.0.0-climate-fever.splade-pp-ed.cached.template index 0f7c975488..b05b45b150 100644 --- a/src/main/resources/docgen/templates/beir-v1.0.0-climate-fever.splade-pp-ed.cached.template +++ b/src/main/resources/docgen/templates/beir-v1.0.0-climate-fever.splade-pp-ed.cached.template @@ -1,14 +1,14 @@ # Anserini Regressions: BEIR (v1.0.0) — Climate-FEVER -**Model**: [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) (using pre-encoded queries) +**Model**: [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) (using cached queries) -This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — Climate-FEVER](http://beir.ai/). +This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — Climate-FEVER](http://beir.ai/). The model itself can be download [here](https://huggingface.co/naver/splade-cocondenser-ensembledistil). See the [official SPLADE repo](/~https://github.com/naver/splade) and the following paper for more details: > Thibault Formal, Carlos Lassance, Benjamin Piwowarski, and Stéphane Clinchant. [From Distillation to Hard Negative Sampling: Making Sparse Neural IR Models More Effective.](https://dl.acm.org/doi/10.1145/3477495.3531857) _Proceedings of the 45th International ACM SIGIR Conference on Research and Development in Information Retrieval_, pages 2353–2359. -In these experiments, we are using pre-encoded queries (i.e., cached results of query encoding). +In these experiments, we are using cached queries (i.e., cached results of query encoding). The exact configurations for these regressions are stored in [this YAML file](${yaml}). Note that this page is automatically generated from [this template](${template}) as part of Anserini's regression pipeline, so do not modify this page directly; modify the template instead and then run `bin/build.sh` to rebuild the documentation. diff --git a/src/main/resources/docgen/templates/beir-v1.0.0-climate-fever.splade-pp-ed.onnx.template b/src/main/resources/docgen/templates/beir-v1.0.0-climate-fever.splade-pp-ed.onnx.template index 804d2d7292..04b715ea6a 100644 --- a/src/main/resources/docgen/templates/beir-v1.0.0-climate-fever.splade-pp-ed.onnx.template +++ b/src/main/resources/docgen/templates/beir-v1.0.0-climate-fever.splade-pp-ed.onnx.template @@ -1,8 +1,8 @@ # Anserini Regressions: BEIR (v1.0.0) — Climate-FEVER -**Model**: [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) (using ONNX for on-the-fly query encoding) +**Model**: [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) (using ONNX for on-the-fly query encoding) -This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — Climate-FEVER](http://beir.ai/). +This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — Climate-FEVER](http://beir.ai/). The model itself can be download [here](https://huggingface.co/naver/splade-cocondenser-ensembledistil). See the [official SPLADE repo](/~https://github.com/naver/splade) and the following paper for more details: diff --git a/src/main/resources/docgen/templates/beir-v1.0.0-climate-fever.unicoil-noexp.cached.template b/src/main/resources/docgen/templates/beir-v1.0.0-climate-fever.unicoil-noexp.cached.template index 3b10d56eb8..4c8e5f76ed 100644 --- a/src/main/resources/docgen/templates/beir-v1.0.0-climate-fever.unicoil-noexp.cached.template +++ b/src/main/resources/docgen/templates/beir-v1.0.0-climate-fever.unicoil-noexp.cached.template @@ -1,12 +1,14 @@ # Anserini Regressions: BEIR (v1.0.0) — Climate-FEVER -**Model**: uniCOIL (without any expansions) +**Model**: uniCOIL without any expansions (using cached queries) This page describes regression experiments, integrated into Anserini's regression testing framework, using uniCOIL (without any expansions) on [BEIR (v1.0.0) — Climate-FEVER](http://beir.ai/). 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.](https://arxiv.org/abs/2106.14807) _arXiv:2106.14807_. +In these experiments, we are using cached queries (i.e., cached results of query encoding). + The exact configurations for these regressions are stored in [this YAML file](${yaml}). Note that this page is automatically generated from [this template](${template}) as part of Anserini's regression pipeline, so do not modify this page directly; modify the template instead. diff --git a/src/main/resources/docgen/templates/beir-v1.0.0-cqadupstack-android.splade-pp-ed.cached.template b/src/main/resources/docgen/templates/beir-v1.0.0-cqadupstack-android.splade-pp-ed.cached.template index a7048efd82..11de3b42ad 100644 --- a/src/main/resources/docgen/templates/beir-v1.0.0-cqadupstack-android.splade-pp-ed.cached.template +++ b/src/main/resources/docgen/templates/beir-v1.0.0-cqadupstack-android.splade-pp-ed.cached.template @@ -1,14 +1,14 @@ # Anserini Regressions: BEIR (v1.0.0) — CQADupStack-android -**Model**: [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) (using pre-encoded queries) +**Model**: [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) (using cached queries) -This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — CQADupStack-android](http://beir.ai/). +This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — CQADupStack-android](http://beir.ai/). The model itself can be download [here](https://huggingface.co/naver/splade-cocondenser-ensembledistil). See the [official SPLADE repo](/~https://github.com/naver/splade) and the following paper for more details: > Thibault Formal, Carlos Lassance, Benjamin Piwowarski, and Stéphane Clinchant. [From Distillation to Hard Negative Sampling: Making Sparse Neural IR Models More Effective.](https://dl.acm.org/doi/10.1145/3477495.3531857) _Proceedings of the 45th International ACM SIGIR Conference on Research and Development in Information Retrieval_, pages 2353–2359. -In these experiments, we are using pre-encoded queries (i.e., cached results of query encoding). +In these experiments, we are using cached queries (i.e., cached results of query encoding). The exact configurations for these regressions are stored in [this YAML file](${yaml}). Note that this page is automatically generated from [this template](${template}) as part of Anserini's regression pipeline, so do not modify this page directly; modify the template instead and then run `bin/build.sh` to rebuild the documentation. diff --git a/src/main/resources/docgen/templates/beir-v1.0.0-cqadupstack-android.splade-pp-ed.onnx.template b/src/main/resources/docgen/templates/beir-v1.0.0-cqadupstack-android.splade-pp-ed.onnx.template index 3546e9c775..b4c8daccd0 100644 --- a/src/main/resources/docgen/templates/beir-v1.0.0-cqadupstack-android.splade-pp-ed.onnx.template +++ b/src/main/resources/docgen/templates/beir-v1.0.0-cqadupstack-android.splade-pp-ed.onnx.template @@ -1,8 +1,8 @@ # Anserini Regressions: BEIR (v1.0.0) — CQADupStack-android -**Model**: [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) (using ONNX for on-the-fly query encoding) +**Model**: [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) (using ONNX for on-the-fly query encoding) -This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — CQADupStack-android](http://beir.ai/). +This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — CQADupStack-android](http://beir.ai/). The model itself can be download [here](https://huggingface.co/naver/splade-cocondenser-ensembledistil). See the [official SPLADE repo](/~https://github.com/naver/splade) and the following paper for more details: diff --git a/src/main/resources/docgen/templates/beir-v1.0.0-cqadupstack-android.unicoil-noexp.cached.template b/src/main/resources/docgen/templates/beir-v1.0.0-cqadupstack-android.unicoil-noexp.cached.template index 82aa18004e..2499b5f939 100644 --- a/src/main/resources/docgen/templates/beir-v1.0.0-cqadupstack-android.unicoil-noexp.cached.template +++ b/src/main/resources/docgen/templates/beir-v1.0.0-cqadupstack-android.unicoil-noexp.cached.template @@ -1,12 +1,14 @@ # Anserini Regressions: BEIR (v1.0.0) — CQADupStack-android -**Model**: uniCOIL (without any expansions) +**Model**: uniCOIL without any expansions (using cached queries) This page describes regression experiments, integrated into Anserini's regression testing framework, using uniCOIL (without any expansions) on [BEIR (v1.0.0) — CQADupStack-android](http://beir.ai/). 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.](https://arxiv.org/abs/2106.14807) _arXiv:2106.14807_. +In these experiments, we are using cached queries (i.e., cached results of query encoding). + The exact configurations for these regressions are stored in [this YAML file](${yaml}). Note that this page is automatically generated from [this template](${template}) as part of Anserini's regression pipeline, so do not modify this page directly; modify the template instead. diff --git a/src/main/resources/docgen/templates/beir-v1.0.0-cqadupstack-english.splade-pp-ed.cached.template b/src/main/resources/docgen/templates/beir-v1.0.0-cqadupstack-english.splade-pp-ed.cached.template index a65ec9986d..ee21b864c0 100644 --- a/src/main/resources/docgen/templates/beir-v1.0.0-cqadupstack-english.splade-pp-ed.cached.template +++ b/src/main/resources/docgen/templates/beir-v1.0.0-cqadupstack-english.splade-pp-ed.cached.template @@ -1,14 +1,14 @@ # Anserini Regressions: BEIR (v1.0.0) — CQADupStack-english -**Model**: [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) (using pre-encoded queries) +**Model**: [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) (using cached queries) -This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — CQADupStack-english](http://beir.ai/). +This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — CQADupStack-english](http://beir.ai/). The model itself can be download [here](https://huggingface.co/naver/splade-cocondenser-ensembledistil). See the [official SPLADE repo](/~https://github.com/naver/splade) and the following paper for more details: > Thibault Formal, Carlos Lassance, Benjamin Piwowarski, and Stéphane Clinchant. [From Distillation to Hard Negative Sampling: Making Sparse Neural IR Models More Effective.](https://dl.acm.org/doi/10.1145/3477495.3531857) _Proceedings of the 45th International ACM SIGIR Conference on Research and Development in Information Retrieval_, pages 2353–2359. -In these experiments, we are using pre-encoded queries (i.e., cached results of query encoding). +In these experiments, we are using cached queries (i.e., cached results of query encoding). The exact configurations for these regressions are stored in [this YAML file](${yaml}). Note that this page is automatically generated from [this template](${template}) as part of Anserini's regression pipeline, so do not modify this page directly; modify the template instead and then run `bin/build.sh` to rebuild the documentation. diff --git a/src/main/resources/docgen/templates/beir-v1.0.0-cqadupstack-english.splade-pp-ed.onnx.template b/src/main/resources/docgen/templates/beir-v1.0.0-cqadupstack-english.splade-pp-ed.onnx.template index e87394ffa5..647042c080 100644 --- a/src/main/resources/docgen/templates/beir-v1.0.0-cqadupstack-english.splade-pp-ed.onnx.template +++ b/src/main/resources/docgen/templates/beir-v1.0.0-cqadupstack-english.splade-pp-ed.onnx.template @@ -1,8 +1,8 @@ # Anserini Regressions: BEIR (v1.0.0) — CQADupStack-english -**Model**: [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) (using ONNX for on-the-fly query encoding) +**Model**: [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) (using ONNX for on-the-fly query encoding) -This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — CQADupStack-english](http://beir.ai/). +This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — CQADupStack-english](http://beir.ai/). The model itself can be download [here](https://huggingface.co/naver/splade-cocondenser-ensembledistil). See the [official SPLADE repo](/~https://github.com/naver/splade) and the following paper for more details: diff --git a/src/main/resources/docgen/templates/beir-v1.0.0-cqadupstack-english.unicoil-noexp.cached.template b/src/main/resources/docgen/templates/beir-v1.0.0-cqadupstack-english.unicoil-noexp.cached.template index 465e45cf78..6ee1aeada7 100644 --- a/src/main/resources/docgen/templates/beir-v1.0.0-cqadupstack-english.unicoil-noexp.cached.template +++ b/src/main/resources/docgen/templates/beir-v1.0.0-cqadupstack-english.unicoil-noexp.cached.template @@ -1,12 +1,14 @@ # Anserini Regressions: BEIR (v1.0.0) — CQADupStack-english -**Model**: uniCOIL (without any expansions) +**Model**: uniCOIL without any expansions (using cached queries) This page describes regression experiments, integrated into Anserini's regression testing framework, using uniCOIL (without any expansions) on [BEIR (v1.0.0) — CQADupStack-english](http://beir.ai/). 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.](https://arxiv.org/abs/2106.14807) _arXiv:2106.14807_. +In these experiments, we are using cached queries (i.e., cached results of query encoding). + The exact configurations for these regressions are stored in [this YAML file](${yaml}). Note that this page is automatically generated from [this template](${template}) as part of Anserini's regression pipeline, so do not modify this page directly; modify the template instead. diff --git a/src/main/resources/docgen/templates/beir-v1.0.0-cqadupstack-gaming.splade-pp-ed.cached.template b/src/main/resources/docgen/templates/beir-v1.0.0-cqadupstack-gaming.splade-pp-ed.cached.template index 5b737600ae..1a09a6e097 100644 --- a/src/main/resources/docgen/templates/beir-v1.0.0-cqadupstack-gaming.splade-pp-ed.cached.template +++ b/src/main/resources/docgen/templates/beir-v1.0.0-cqadupstack-gaming.splade-pp-ed.cached.template @@ -1,14 +1,14 @@ # Anserini Regressions: BEIR (v1.0.0) — CQADupStack-gaming -**Model**: [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) (using pre-encoded queries) +**Model**: [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) (using cached queries) -This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — CQADupStack-gaming](http://beir.ai/). +This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — CQADupStack-gaming](http://beir.ai/). The model itself can be download [here](https://huggingface.co/naver/splade-cocondenser-ensembledistil). See the [official SPLADE repo](/~https://github.com/naver/splade) and the following paper for more details: > Thibault Formal, Carlos Lassance, Benjamin Piwowarski, and Stéphane Clinchant. [From Distillation to Hard Negative Sampling: Making Sparse Neural IR Models More Effective.](https://dl.acm.org/doi/10.1145/3477495.3531857) _Proceedings of the 45th International ACM SIGIR Conference on Research and Development in Information Retrieval_, pages 2353–2359. -In these experiments, we are using pre-encoded queries (i.e., cached results of query encoding). +In these experiments, we are using cached queries (i.e., cached results of query encoding). The exact configurations for these regressions are stored in [this YAML file](${yaml}). Note that this page is automatically generated from [this template](${template}) as part of Anserini's regression pipeline, so do not modify this page directly; modify the template instead and then run `bin/build.sh` to rebuild the documentation. diff --git a/src/main/resources/docgen/templates/beir-v1.0.0-cqadupstack-gaming.splade-pp-ed.onnx.template b/src/main/resources/docgen/templates/beir-v1.0.0-cqadupstack-gaming.splade-pp-ed.onnx.template index 632e8e5db4..b58fe0c81f 100644 --- a/src/main/resources/docgen/templates/beir-v1.0.0-cqadupstack-gaming.splade-pp-ed.onnx.template +++ b/src/main/resources/docgen/templates/beir-v1.0.0-cqadupstack-gaming.splade-pp-ed.onnx.template @@ -1,8 +1,8 @@ # Anserini Regressions: BEIR (v1.0.0) — CQADupStack-gaming -**Model**: [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) (using ONNX for on-the-fly query encoding) +**Model**: [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) (using ONNX for on-the-fly query encoding) -This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — CQADupStack-gaming](http://beir.ai/). +This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — CQADupStack-gaming](http://beir.ai/). The model itself can be download [here](https://huggingface.co/naver/splade-cocondenser-ensembledistil). See the [official SPLADE repo](/~https://github.com/naver/splade) and the following paper for more details: diff --git a/src/main/resources/docgen/templates/beir-v1.0.0-cqadupstack-gaming.unicoil-noexp.cached.template b/src/main/resources/docgen/templates/beir-v1.0.0-cqadupstack-gaming.unicoil-noexp.cached.template index 752940fbae..7ad7cf5d9a 100644 --- a/src/main/resources/docgen/templates/beir-v1.0.0-cqadupstack-gaming.unicoil-noexp.cached.template +++ b/src/main/resources/docgen/templates/beir-v1.0.0-cqadupstack-gaming.unicoil-noexp.cached.template @@ -1,12 +1,14 @@ # Anserini Regressions: BEIR (v1.0.0) — CQADupStack-gaming -**Model**: uniCOIL (without any expansions) +**Model**: uniCOIL without any expansions (using cached queries) This page describes regression experiments, integrated into Anserini's regression testing framework, using uniCOIL (without any expansions) on [BEIR (v1.0.0) — CQADupStack-gaming](http://beir.ai/). 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.](https://arxiv.org/abs/2106.14807) _arXiv:2106.14807_. +In these experiments, we are using cached queries (i.e., cached results of query encoding). + The exact configurations for these regressions are stored in [this YAML file](${yaml}). Note that this page is automatically generated from [this template](${template}) as part of Anserini's regression pipeline, so do not modify this page directly; modify the template instead. diff --git a/src/main/resources/docgen/templates/beir-v1.0.0-cqadupstack-gis.splade-pp-ed.cached.template b/src/main/resources/docgen/templates/beir-v1.0.0-cqadupstack-gis.splade-pp-ed.cached.template index dd16f03a1a..d436355263 100644 --- a/src/main/resources/docgen/templates/beir-v1.0.0-cqadupstack-gis.splade-pp-ed.cached.template +++ b/src/main/resources/docgen/templates/beir-v1.0.0-cqadupstack-gis.splade-pp-ed.cached.template @@ -1,14 +1,14 @@ # Anserini Regressions: BEIR (v1.0.0) — CQADupStack-gis -**Model**: [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) (using pre-encoded queries) +**Model**: [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) (using cached queries) -This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — CQADupStack-gis](http://beir.ai/). +This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — CQADupStack-gis](http://beir.ai/). The model itself can be download [here](https://huggingface.co/naver/splade-cocondenser-ensembledistil). See the [official SPLADE repo](/~https://github.com/naver/splade) and the following paper for more details: > Thibault Formal, Carlos Lassance, Benjamin Piwowarski, and Stéphane Clinchant. [From Distillation to Hard Negative Sampling: Making Sparse Neural IR Models More Effective.](https://dl.acm.org/doi/10.1145/3477495.3531857) _Proceedings of the 45th International ACM SIGIR Conference on Research and Development in Information Retrieval_, pages 2353–2359. -In these experiments, we are using pre-encoded queries (i.e., cached results of query encoding). +In these experiments, we are using cached queries (i.e., cached results of query encoding). The exact configurations for these regressions are stored in [this YAML file](${yaml}). Note that this page is automatically generated from [this template](${template}) as part of Anserini's regression pipeline, so do not modify this page directly; modify the template instead and then run `bin/build.sh` to rebuild the documentation. diff --git a/src/main/resources/docgen/templates/beir-v1.0.0-cqadupstack-gis.splade-pp-ed.onnx.template b/src/main/resources/docgen/templates/beir-v1.0.0-cqadupstack-gis.splade-pp-ed.onnx.template index 6b3bee9fbf..d56f9eabb1 100644 --- a/src/main/resources/docgen/templates/beir-v1.0.0-cqadupstack-gis.splade-pp-ed.onnx.template +++ b/src/main/resources/docgen/templates/beir-v1.0.0-cqadupstack-gis.splade-pp-ed.onnx.template @@ -1,8 +1,8 @@ # Anserini Regressions: BEIR (v1.0.0) — CQADupStack-gis -**Model**: [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) (using ONNX for on-the-fly query encoding) +**Model**: [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) (using ONNX for on-the-fly query encoding) -This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — CQADupStack-gis](http://beir.ai/). +This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — CQADupStack-gis](http://beir.ai/). The model itself can be download [here](https://huggingface.co/naver/splade-cocondenser-ensembledistil). See the [official SPLADE repo](/~https://github.com/naver/splade) and the following paper for more details: diff --git a/src/main/resources/docgen/templates/beir-v1.0.0-cqadupstack-gis.unicoil-noexp.cached.template b/src/main/resources/docgen/templates/beir-v1.0.0-cqadupstack-gis.unicoil-noexp.cached.template index 14a9d91633..63d3ded3a4 100644 --- a/src/main/resources/docgen/templates/beir-v1.0.0-cqadupstack-gis.unicoil-noexp.cached.template +++ b/src/main/resources/docgen/templates/beir-v1.0.0-cqadupstack-gis.unicoil-noexp.cached.template @@ -1,12 +1,14 @@ # Anserini Regressions: BEIR (v1.0.0) — CQADupStack-gis -**Model**: uniCOIL (without any expansions) +**Model**: uniCOIL without any expansions (using cached queries) This page describes regression experiments, integrated into Anserini's regression testing framework, using uniCOIL (without any expansions) on [BEIR (v1.0.0) — CQADupStack-gis](http://beir.ai/). 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.](https://arxiv.org/abs/2106.14807) _arXiv:2106.14807_. +In these experiments, we are using cached queries (i.e., cached results of query encoding). + The exact configurations for these regressions are stored in [this YAML file](${yaml}). Note that this page is automatically generated from [this template](${template}) as part of Anserini's regression pipeline, so do not modify this page directly; modify the template instead. diff --git a/src/main/resources/docgen/templates/beir-v1.0.0-cqadupstack-mathematica.splade-pp-ed.cached.template b/src/main/resources/docgen/templates/beir-v1.0.0-cqadupstack-mathematica.splade-pp-ed.cached.template index d0dfa471ef..d776c6576f 100644 --- a/src/main/resources/docgen/templates/beir-v1.0.0-cqadupstack-mathematica.splade-pp-ed.cached.template +++ b/src/main/resources/docgen/templates/beir-v1.0.0-cqadupstack-mathematica.splade-pp-ed.cached.template @@ -1,14 +1,14 @@ # Anserini Regressions: BEIR (v1.0.0) — CQADupStack-mathematica -**Model**: [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) (using pre-encoded queries) +**Model**: [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) (using cached queries) -This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — CQADupStack-mathematica](http://beir.ai/). +This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — CQADupStack-mathematica](http://beir.ai/). The model itself can be download [here](https://huggingface.co/naver/splade-cocondenser-ensembledistil). See the [official SPLADE repo](/~https://github.com/naver/splade) and the following paper for more details: > Thibault Formal, Carlos Lassance, Benjamin Piwowarski, and Stéphane Clinchant. [From Distillation to Hard Negative Sampling: Making Sparse Neural IR Models More Effective.](https://dl.acm.org/doi/10.1145/3477495.3531857) _Proceedings of the 45th International ACM SIGIR Conference on Research and Development in Information Retrieval_, pages 2353–2359. -In these experiments, we are using pre-encoded queries (i.e., cached results of query encoding). +In these experiments, we are using cached queries (i.e., cached results of query encoding). The exact configurations for these regressions are stored in [this YAML file](${yaml}). Note that this page is automatically generated from [this template](${template}) as part of Anserini's regression pipeline, so do not modify this page directly; modify the template instead and then run `bin/build.sh` to rebuild the documentation. diff --git a/src/main/resources/docgen/templates/beir-v1.0.0-cqadupstack-mathematica.splade-pp-ed.onnx.template b/src/main/resources/docgen/templates/beir-v1.0.0-cqadupstack-mathematica.splade-pp-ed.onnx.template index 35f2eb943e..ef74638eb3 100644 --- a/src/main/resources/docgen/templates/beir-v1.0.0-cqadupstack-mathematica.splade-pp-ed.onnx.template +++ b/src/main/resources/docgen/templates/beir-v1.0.0-cqadupstack-mathematica.splade-pp-ed.onnx.template @@ -1,8 +1,8 @@ # Anserini Regressions: BEIR (v1.0.0) — CQADupStack-mathematica -**Model**: [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) (using ONNX for on-the-fly query encoding) +**Model**: [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) (using ONNX for on-the-fly query encoding) -This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — CQADupStack-mathematica](http://beir.ai/). +This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — CQADupStack-mathematica](http://beir.ai/). The model itself can be download [here](https://huggingface.co/naver/splade-cocondenser-ensembledistil). See the [official SPLADE repo](/~https://github.com/naver/splade) and the following paper for more details: diff --git a/src/main/resources/docgen/templates/beir-v1.0.0-cqadupstack-mathematica.unicoil-noexp.cached.template b/src/main/resources/docgen/templates/beir-v1.0.0-cqadupstack-mathematica.unicoil-noexp.cached.template index cef0268ccb..44c2fc4383 100644 --- a/src/main/resources/docgen/templates/beir-v1.0.0-cqadupstack-mathematica.unicoil-noexp.cached.template +++ b/src/main/resources/docgen/templates/beir-v1.0.0-cqadupstack-mathematica.unicoil-noexp.cached.template @@ -1,12 +1,14 @@ # Anserini Regressions: BEIR (v1.0.0) — CQADupStack-mathematica -**Model**: uniCOIL (without any expansions) +**Model**: uniCOIL without any expansions (using cached queries) This page describes regression experiments, integrated into Anserini's regression testing framework, using uniCOIL (without any expansions) on [BEIR (v1.0.0) — CQADupStack-mathematica](http://beir.ai/). 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.](https://arxiv.org/abs/2106.14807) _arXiv:2106.14807_. +In these experiments, we are using cached queries (i.e., cached results of query encoding). + The exact configurations for these regressions are stored in [this YAML file](${yaml}). Note that this page is automatically generated from [this template](${template}) as part of Anserini's regression pipeline, so do not modify this page directly; modify the template instead. diff --git a/src/main/resources/docgen/templates/beir-v1.0.0-cqadupstack-physics.splade-pp-ed.cached.template b/src/main/resources/docgen/templates/beir-v1.0.0-cqadupstack-physics.splade-pp-ed.cached.template index 4b76c1d7e5..7e7d5ac5e0 100644 --- a/src/main/resources/docgen/templates/beir-v1.0.0-cqadupstack-physics.splade-pp-ed.cached.template +++ b/src/main/resources/docgen/templates/beir-v1.0.0-cqadupstack-physics.splade-pp-ed.cached.template @@ -1,14 +1,14 @@ # Anserini Regressions: BEIR (v1.0.0) — CQADupStack-physics -**Model**: [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) (using pre-encoded queries) +**Model**: [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) (using cached queries) -This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — CQADupStack-physics](http://beir.ai/). +This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — CQADupStack-physics](http://beir.ai/). The model itself can be download [here](https://huggingface.co/naver/splade-cocondenser-ensembledistil). See the [official SPLADE repo](/~https://github.com/naver/splade) and the following paper for more details: > Thibault Formal, Carlos Lassance, Benjamin Piwowarski, and Stéphane Clinchant. [From Distillation to Hard Negative Sampling: Making Sparse Neural IR Models More Effective.](https://dl.acm.org/doi/10.1145/3477495.3531857) _Proceedings of the 45th International ACM SIGIR Conference on Research and Development in Information Retrieval_, pages 2353–2359. -In these experiments, we are using pre-encoded queries (i.e., cached results of query encoding). +In these experiments, we are using cached queries (i.e., cached results of query encoding). The exact configurations for these regressions are stored in [this YAML file](${yaml}). Note that this page is automatically generated from [this template](${template}) as part of Anserini's regression pipeline, so do not modify this page directly; modify the template instead and then run `bin/build.sh` to rebuild the documentation. diff --git a/src/main/resources/docgen/templates/beir-v1.0.0-cqadupstack-physics.splade-pp-ed.onnx.template b/src/main/resources/docgen/templates/beir-v1.0.0-cqadupstack-physics.splade-pp-ed.onnx.template index 1764e580ab..40ffaa2824 100644 --- a/src/main/resources/docgen/templates/beir-v1.0.0-cqadupstack-physics.splade-pp-ed.onnx.template +++ b/src/main/resources/docgen/templates/beir-v1.0.0-cqadupstack-physics.splade-pp-ed.onnx.template @@ -1,8 +1,8 @@ # Anserini Regressions: BEIR (v1.0.0) — CQADupStack-physics -**Model**: [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) (using ONNX for on-the-fly query encoding) +**Model**: [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) (using ONNX for on-the-fly query encoding) -This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — CQADupStack-physics](http://beir.ai/). +This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — CQADupStack-physics](http://beir.ai/). The model itself can be download [here](https://huggingface.co/naver/splade-cocondenser-ensembledistil). See the [official SPLADE repo](/~https://github.com/naver/splade) and the following paper for more details: diff --git a/src/main/resources/docgen/templates/beir-v1.0.0-cqadupstack-physics.unicoil-noexp.cached.template b/src/main/resources/docgen/templates/beir-v1.0.0-cqadupstack-physics.unicoil-noexp.cached.template index 0a1b94d143..b5135134d5 100644 --- a/src/main/resources/docgen/templates/beir-v1.0.0-cqadupstack-physics.unicoil-noexp.cached.template +++ b/src/main/resources/docgen/templates/beir-v1.0.0-cqadupstack-physics.unicoil-noexp.cached.template @@ -1,12 +1,14 @@ # Anserini Regressions: BEIR (v1.0.0) — CQADupStack-physics -**Model**: uniCOIL (without any expansions) +**Model**: uniCOIL without any expansions (using cached queries) This page describes regression experiments, integrated into Anserini's regression testing framework, using uniCOIL (without any expansions) on [BEIR (v1.0.0) — CQADupStack-physics](http://beir.ai/). 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.](https://arxiv.org/abs/2106.14807) _arXiv:2106.14807_. +In these experiments, we are using cached queries (i.e., cached results of query encoding). + The exact configurations for these regressions are stored in [this YAML file](${yaml}). Note that this page is automatically generated from [this template](${template}) as part of Anserini's regression pipeline, so do not modify this page directly; modify the template instead. diff --git a/src/main/resources/docgen/templates/beir-v1.0.0-cqadupstack-programmers.splade-pp-ed.cached.template b/src/main/resources/docgen/templates/beir-v1.0.0-cqadupstack-programmers.splade-pp-ed.cached.template index d798e23ae7..0f80bdc835 100644 --- a/src/main/resources/docgen/templates/beir-v1.0.0-cqadupstack-programmers.splade-pp-ed.cached.template +++ b/src/main/resources/docgen/templates/beir-v1.0.0-cqadupstack-programmers.splade-pp-ed.cached.template @@ -1,14 +1,14 @@ # Anserini Regressions: BEIR (v1.0.0) — CQADupStack-programmers -**Model**: [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) (using pre-encoded queries) +**Model**: [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) (using cached queries) -This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — CQADupStack-programmers](http://beir.ai/). +This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — CQADupStack-programmers](http://beir.ai/). The model itself can be download [here](https://huggingface.co/naver/splade-cocondenser-ensembledistil). See the [official SPLADE repo](/~https://github.com/naver/splade) and the following paper for more details: > Thibault Formal, Carlos Lassance, Benjamin Piwowarski, and Stéphane Clinchant. [From Distillation to Hard Negative Sampling: Making Sparse Neural IR Models More Effective.](https://dl.acm.org/doi/10.1145/3477495.3531857) _Proceedings of the 45th International ACM SIGIR Conference on Research and Development in Information Retrieval_, pages 2353–2359. -In these experiments, we are using pre-encoded queries (i.e., cached results of query encoding). +In these experiments, we are using cached queries (i.e., cached results of query encoding). The exact configurations for these regressions are stored in [this YAML file](${yaml}). Note that this page is automatically generated from [this template](${template}) as part of Anserini's regression pipeline, so do not modify this page directly; modify the template instead and then run `bin/build.sh` to rebuild the documentation. diff --git a/src/main/resources/docgen/templates/beir-v1.0.0-cqadupstack-programmers.splade-pp-ed.onnx.template b/src/main/resources/docgen/templates/beir-v1.0.0-cqadupstack-programmers.splade-pp-ed.onnx.template index 79640d05ea..9be5dd0bc1 100644 --- a/src/main/resources/docgen/templates/beir-v1.0.0-cqadupstack-programmers.splade-pp-ed.onnx.template +++ b/src/main/resources/docgen/templates/beir-v1.0.0-cqadupstack-programmers.splade-pp-ed.onnx.template @@ -1,8 +1,8 @@ # Anserini Regressions: BEIR (v1.0.0) — CQADupStack-programmers -**Model**: [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) (using ONNX for on-the-fly query encoding) +**Model**: [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) (using ONNX for on-the-fly query encoding) -This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — CQADupStack-programmers](http://beir.ai/). +This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — CQADupStack-programmers](http://beir.ai/). The model itself can be download [here](https://huggingface.co/naver/splade-cocondenser-ensembledistil). See the [official SPLADE repo](/~https://github.com/naver/splade) and the following paper for more details: diff --git a/src/main/resources/docgen/templates/beir-v1.0.0-cqadupstack-programmers.unicoil-noexp.cached.template b/src/main/resources/docgen/templates/beir-v1.0.0-cqadupstack-programmers.unicoil-noexp.cached.template index 33c39a8c18..e608383744 100644 --- a/src/main/resources/docgen/templates/beir-v1.0.0-cqadupstack-programmers.unicoil-noexp.cached.template +++ b/src/main/resources/docgen/templates/beir-v1.0.0-cqadupstack-programmers.unicoil-noexp.cached.template @@ -1,12 +1,14 @@ # Anserini Regressions: BEIR (v1.0.0) — CQADupStack-programmers -**Model**: uniCOIL (without any expansions) +**Model**: uniCOIL without any expansions (using cached queries) This page describes regression experiments, integrated into Anserini's regression testing framework, using uniCOIL (without any expansions) on [BEIR (v1.0.0) — CQADupStack-programmers](http://beir.ai/). 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.](https://arxiv.org/abs/2106.14807) _arXiv:2106.14807_. +In these experiments, we are using cached queries (i.e., cached results of query encoding). + The exact configurations for these regressions are stored in [this YAML file](${yaml}). Note that this page is automatically generated from [this template](${template}) as part of Anserini's regression pipeline, so do not modify this page directly; modify the template instead. diff --git a/src/main/resources/docgen/templates/beir-v1.0.0-cqadupstack-stats.splade-pp-ed.cached.template b/src/main/resources/docgen/templates/beir-v1.0.0-cqadupstack-stats.splade-pp-ed.cached.template index 708d971069..af327a8b24 100644 --- a/src/main/resources/docgen/templates/beir-v1.0.0-cqadupstack-stats.splade-pp-ed.cached.template +++ b/src/main/resources/docgen/templates/beir-v1.0.0-cqadupstack-stats.splade-pp-ed.cached.template @@ -1,14 +1,14 @@ # Anserini Regressions: BEIR (v1.0.0) — CQADupStack-stats -**Model**: [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) (using pre-encoded queries) +**Model**: [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) (using cached queries) -This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — CQADupStack-stats](http://beir.ai/). +This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — CQADupStack-stats](http://beir.ai/). The model itself can be download [here](https://huggingface.co/naver/splade-cocondenser-ensembledistil). See the [official SPLADE repo](/~https://github.com/naver/splade) and the following paper for more details: > Thibault Formal, Carlos Lassance, Benjamin Piwowarski, and Stéphane Clinchant. [From Distillation to Hard Negative Sampling: Making Sparse Neural IR Models More Effective.](https://dl.acm.org/doi/10.1145/3477495.3531857) _Proceedings of the 45th International ACM SIGIR Conference on Research and Development in Information Retrieval_, pages 2353–2359. -In these experiments, we are using pre-encoded queries (i.e., cached results of query encoding). +In these experiments, we are using cached queries (i.e., cached results of query encoding). The exact configurations for these regressions are stored in [this YAML file](${yaml}). Note that this page is automatically generated from [this template](${template}) as part of Anserini's regression pipeline, so do not modify this page directly; modify the template instead and then run `bin/build.sh` to rebuild the documentation. diff --git a/src/main/resources/docgen/templates/beir-v1.0.0-cqadupstack-stats.splade-pp-ed.onnx.template b/src/main/resources/docgen/templates/beir-v1.0.0-cqadupstack-stats.splade-pp-ed.onnx.template index 8fa30cfc9b..c4f43c54da 100644 --- a/src/main/resources/docgen/templates/beir-v1.0.0-cqadupstack-stats.splade-pp-ed.onnx.template +++ b/src/main/resources/docgen/templates/beir-v1.0.0-cqadupstack-stats.splade-pp-ed.onnx.template @@ -1,8 +1,8 @@ # Anserini Regressions: BEIR (v1.0.0) — CQADupStack-stats -**Model**: [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) (using ONNX for on-the-fly query encoding) +**Model**: [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) (using ONNX for on-the-fly query encoding) -This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — CQADupStack-stats](http://beir.ai/). +This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — CQADupStack-stats](http://beir.ai/). The model itself can be download [here](https://huggingface.co/naver/splade-cocondenser-ensembledistil). See the [official SPLADE repo](/~https://github.com/naver/splade) and the following paper for more details: diff --git a/src/main/resources/docgen/templates/beir-v1.0.0-cqadupstack-stats.unicoil-noexp.cached.template b/src/main/resources/docgen/templates/beir-v1.0.0-cqadupstack-stats.unicoil-noexp.cached.template index 2349351ff6..7e5b30c148 100644 --- a/src/main/resources/docgen/templates/beir-v1.0.0-cqadupstack-stats.unicoil-noexp.cached.template +++ b/src/main/resources/docgen/templates/beir-v1.0.0-cqadupstack-stats.unicoil-noexp.cached.template @@ -1,12 +1,14 @@ # Anserini Regressions: BEIR (v1.0.0) — CQADupStack-stats -**Model**: uniCOIL (without any expansions) +**Model**: uniCOIL without any expansions (using cached queries) This page describes regression experiments, integrated into Anserini's regression testing framework, using uniCOIL (without any expansions) on [BEIR (v1.0.0) — CQADupStack-stats](http://beir.ai/). 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.](https://arxiv.org/abs/2106.14807) _arXiv:2106.14807_. +In these experiments, we are using cached queries (i.e., cached results of query encoding). + The exact configurations for these regressions are stored in [this YAML file](${yaml}). Note that this page is automatically generated from [this template](${template}) as part of Anserini's regression pipeline, so do not modify this page directly; modify the template instead. diff --git a/src/main/resources/docgen/templates/beir-v1.0.0-cqadupstack-tex.splade-pp-ed.cached.template b/src/main/resources/docgen/templates/beir-v1.0.0-cqadupstack-tex.splade-pp-ed.cached.template index 2b75720fc2..4494f24aad 100644 --- a/src/main/resources/docgen/templates/beir-v1.0.0-cqadupstack-tex.splade-pp-ed.cached.template +++ b/src/main/resources/docgen/templates/beir-v1.0.0-cqadupstack-tex.splade-pp-ed.cached.template @@ -1,14 +1,14 @@ # Anserini Regressions: BEIR (v1.0.0) — CQADupStack-tex -**Model**: [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) (using pre-encoded queries) +**Model**: [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) (using cached queries) -This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — CQADupStack-tex](http://beir.ai/). +This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — CQADupStack-tex](http://beir.ai/). The model itself can be download [here](https://huggingface.co/naver/splade-cocondenser-ensembledistil). See the [official SPLADE repo](/~https://github.com/naver/splade) and the following paper for more details: > Thibault Formal, Carlos Lassance, Benjamin Piwowarski, and Stéphane Clinchant. [From Distillation to Hard Negative Sampling: Making Sparse Neural IR Models More Effective.](https://dl.acm.org/doi/10.1145/3477495.3531857) _Proceedings of the 45th International ACM SIGIR Conference on Research and Development in Information Retrieval_, pages 2353–2359. -In these experiments, we are using pre-encoded queries (i.e., cached results of query encoding). +In these experiments, we are using cached queries (i.e., cached results of query encoding). The exact configurations for these regressions are stored in [this YAML file](${yaml}). Note that this page is automatically generated from [this template](${template}) as part of Anserini's regression pipeline, so do not modify this page directly; modify the template instead and then run `bin/build.sh` to rebuild the documentation. diff --git a/src/main/resources/docgen/templates/beir-v1.0.0-cqadupstack-tex.splade-pp-ed.onnx.template b/src/main/resources/docgen/templates/beir-v1.0.0-cqadupstack-tex.splade-pp-ed.onnx.template index f2378eda8b..2b7f411d83 100644 --- a/src/main/resources/docgen/templates/beir-v1.0.0-cqadupstack-tex.splade-pp-ed.onnx.template +++ b/src/main/resources/docgen/templates/beir-v1.0.0-cqadupstack-tex.splade-pp-ed.onnx.template @@ -1,8 +1,8 @@ # Anserini Regressions: BEIR (v1.0.0) — CQADupStack-tex -**Model**: [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) (using ONNX for on-the-fly query encoding) +**Model**: [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) (using ONNX for on-the-fly query encoding) -This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — CQADupStack-tex](http://beir.ai/). +This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — CQADupStack-tex](http://beir.ai/). The model itself can be download [here](https://huggingface.co/naver/splade-cocondenser-ensembledistil). See the [official SPLADE repo](/~https://github.com/naver/splade) and the following paper for more details: diff --git a/src/main/resources/docgen/templates/beir-v1.0.0-cqadupstack-tex.unicoil-noexp.cached.template b/src/main/resources/docgen/templates/beir-v1.0.0-cqadupstack-tex.unicoil-noexp.cached.template index 39a392e655..c5415d7b58 100644 --- a/src/main/resources/docgen/templates/beir-v1.0.0-cqadupstack-tex.unicoil-noexp.cached.template +++ b/src/main/resources/docgen/templates/beir-v1.0.0-cqadupstack-tex.unicoil-noexp.cached.template @@ -1,12 +1,14 @@ # Anserini Regressions: BEIR (v1.0.0) — CQADupStack-tex -**Model**: uniCOIL (without any expansions) +**Model**: uniCOIL without any expansions (using cached queries) This page describes regression experiments, integrated into Anserini's regression testing framework, using uniCOIL (without any expansions) on [BEIR (v1.0.0) — CQADupStack-tex](http://beir.ai/). 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.](https://arxiv.org/abs/2106.14807) _arXiv:2106.14807_. +In these experiments, we are using cached queries (i.e., cached results of query encoding). + The exact configurations for these regressions are stored in [this YAML file](${yaml}). Note that this page is automatically generated from [this template](${template}) as part of Anserini's regression pipeline, so do not modify this page directly; modify the template instead. diff --git a/src/main/resources/docgen/templates/beir-v1.0.0-cqadupstack-unix.splade-pp-ed.cached.template b/src/main/resources/docgen/templates/beir-v1.0.0-cqadupstack-unix.splade-pp-ed.cached.template index a0e10a80ee..9a0262bbc0 100644 --- a/src/main/resources/docgen/templates/beir-v1.0.0-cqadupstack-unix.splade-pp-ed.cached.template +++ b/src/main/resources/docgen/templates/beir-v1.0.0-cqadupstack-unix.splade-pp-ed.cached.template @@ -1,14 +1,14 @@ # Anserini Regressions: BEIR (v1.0.0) — CQADupStack-unix -**Model**: [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) (using pre-encoded queries) +**Model**: [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) (using cached queries) -This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — CQADupStack-unix](http://beir.ai/). +This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — CQADupStack-unix](http://beir.ai/). The model itself can be download [here](https://huggingface.co/naver/splade-cocondenser-ensembledistil). See the [official SPLADE repo](/~https://github.com/naver/splade) and the following paper for more details: > Thibault Formal, Carlos Lassance, Benjamin Piwowarski, and Stéphane Clinchant. [From Distillation to Hard Negative Sampling: Making Sparse Neural IR Models More Effective.](https://dl.acm.org/doi/10.1145/3477495.3531857) _Proceedings of the 45th International ACM SIGIR Conference on Research and Development in Information Retrieval_, pages 2353–2359. -In these experiments, we are using pre-encoded queries (i.e., cached results of query encoding). +In these experiments, we are using cached queries (i.e., cached results of query encoding). The exact configurations for these regressions are stored in [this YAML file](${yaml}). Note that this page is automatically generated from [this template](${template}) as part of Anserini's regression pipeline, so do not modify this page directly; modify the template instead and then run `bin/build.sh` to rebuild the documentation. diff --git a/src/main/resources/docgen/templates/beir-v1.0.0-cqadupstack-unix.splade-pp-ed.onnx.template b/src/main/resources/docgen/templates/beir-v1.0.0-cqadupstack-unix.splade-pp-ed.onnx.template index 7b7ae8263b..7d87face0d 100644 --- a/src/main/resources/docgen/templates/beir-v1.0.0-cqadupstack-unix.splade-pp-ed.onnx.template +++ b/src/main/resources/docgen/templates/beir-v1.0.0-cqadupstack-unix.splade-pp-ed.onnx.template @@ -1,8 +1,8 @@ # Anserini Regressions: BEIR (v1.0.0) — CQADupStack-unix -**Model**: [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) (using ONNX for on-the-fly query encoding) +**Model**: [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) (using ONNX for on-the-fly query encoding) -This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — CQADupStack-unix](http://beir.ai/). +This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — CQADupStack-unix](http://beir.ai/). The model itself can be download [here](https://huggingface.co/naver/splade-cocondenser-ensembledistil). See the [official SPLADE repo](/~https://github.com/naver/splade) and the following paper for more details: diff --git a/src/main/resources/docgen/templates/beir-v1.0.0-cqadupstack-unix.unicoil-noexp.cached.template b/src/main/resources/docgen/templates/beir-v1.0.0-cqadupstack-unix.unicoil-noexp.cached.template index 40a8ce96ad..da0f98c08f 100644 --- a/src/main/resources/docgen/templates/beir-v1.0.0-cqadupstack-unix.unicoil-noexp.cached.template +++ b/src/main/resources/docgen/templates/beir-v1.0.0-cqadupstack-unix.unicoil-noexp.cached.template @@ -1,12 +1,14 @@ # Anserini Regressions: BEIR (v1.0.0) — CQADupStack-unix -**Model**: uniCOIL (without any expansions) +**Model**: uniCOIL without any expansions (using cached queries) This page describes regression experiments, integrated into Anserini's regression testing framework, using uniCOIL (without any expansions) on [BEIR (v1.0.0) — CQADupStack-unix](http://beir.ai/). 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.](https://arxiv.org/abs/2106.14807) _arXiv:2106.14807_. +In these experiments, we are using cached queries (i.e., cached results of query encoding). + The exact configurations for these regressions are stored in [this YAML file](${yaml}). Note that this page is automatically generated from [this template](${template}) as part of Anserini's regression pipeline, so do not modify this page directly; modify the template instead. diff --git a/src/main/resources/docgen/templates/beir-v1.0.0-cqadupstack-webmasters.splade-pp-ed.cached.template b/src/main/resources/docgen/templates/beir-v1.0.0-cqadupstack-webmasters.splade-pp-ed.cached.template index 5896a9e92a..d323b7479d 100644 --- a/src/main/resources/docgen/templates/beir-v1.0.0-cqadupstack-webmasters.splade-pp-ed.cached.template +++ b/src/main/resources/docgen/templates/beir-v1.0.0-cqadupstack-webmasters.splade-pp-ed.cached.template @@ -1,14 +1,14 @@ # Anserini Regressions: BEIR (v1.0.0) — CQADupStack-webmasters -**Model**: [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) (using pre-encoded queries) +**Model**: [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) (using cached queries) -This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — CQADupStack-webmasters](http://beir.ai/). +This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — CQADupStack-webmasters](http://beir.ai/). The model itself can be download [here](https://huggingface.co/naver/splade-cocondenser-ensembledistil). See the [official SPLADE repo](/~https://github.com/naver/splade) and the following paper for more details: > Thibault Formal, Carlos Lassance, Benjamin Piwowarski, and Stéphane Clinchant. [From Distillation to Hard Negative Sampling: Making Sparse Neural IR Models More Effective.](https://dl.acm.org/doi/10.1145/3477495.3531857) _Proceedings of the 45th International ACM SIGIR Conference on Research and Development in Information Retrieval_, pages 2353–2359. -In these experiments, we are using pre-encoded queries (i.e., cached results of query encoding). +In these experiments, we are using cached queries (i.e., cached results of query encoding). The exact configurations for these regressions are stored in [this YAML file](${yaml}). Note that this page is automatically generated from [this template](${template}) as part of Anserini's regression pipeline, so do not modify this page directly; modify the template instead and then run `bin/build.sh` to rebuild the documentation. diff --git a/src/main/resources/docgen/templates/beir-v1.0.0-cqadupstack-webmasters.splade-pp-ed.onnx.template b/src/main/resources/docgen/templates/beir-v1.0.0-cqadupstack-webmasters.splade-pp-ed.onnx.template index fa04f961e7..c0616e37c6 100644 --- a/src/main/resources/docgen/templates/beir-v1.0.0-cqadupstack-webmasters.splade-pp-ed.onnx.template +++ b/src/main/resources/docgen/templates/beir-v1.0.0-cqadupstack-webmasters.splade-pp-ed.onnx.template @@ -1,8 +1,8 @@ # Anserini Regressions: BEIR (v1.0.0) — CQADupStack-webmasters -**Model**: [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) (using ONNX for on-the-fly query encoding) +**Model**: [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) (using ONNX for on-the-fly query encoding) -This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — CQADupStack-webmasters](http://beir.ai/). +This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — CQADupStack-webmasters](http://beir.ai/). The model itself can be download [here](https://huggingface.co/naver/splade-cocondenser-ensembledistil). See the [official SPLADE repo](/~https://github.com/naver/splade) and the following paper for more details: diff --git a/src/main/resources/docgen/templates/beir-v1.0.0-cqadupstack-webmasters.unicoil-noexp.cached.template b/src/main/resources/docgen/templates/beir-v1.0.0-cqadupstack-webmasters.unicoil-noexp.cached.template index 67da932348..1b3d65146a 100644 --- a/src/main/resources/docgen/templates/beir-v1.0.0-cqadupstack-webmasters.unicoil-noexp.cached.template +++ b/src/main/resources/docgen/templates/beir-v1.0.0-cqadupstack-webmasters.unicoil-noexp.cached.template @@ -1,12 +1,14 @@ # Anserini Regressions: BEIR (v1.0.0) — CQADupStack-webmasters -**Model**: uniCOIL (without any expansions) +**Model**: uniCOIL without any expansions (using cached queries) This page describes regression experiments, integrated into Anserini's regression testing framework, using uniCOIL (without any expansions) on [BEIR (v1.0.0) — CQADupStack-webmasters](http://beir.ai/). 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.](https://arxiv.org/abs/2106.14807) _arXiv:2106.14807_. +In these experiments, we are using cached queries (i.e., cached results of query encoding). + The exact configurations for these regressions are stored in [this YAML file](${yaml}). Note that this page is automatically generated from [this template](${template}) as part of Anserini's regression pipeline, so do not modify this page directly; modify the template instead. diff --git a/src/main/resources/docgen/templates/beir-v1.0.0-cqadupstack-wordpress.splade-pp-ed.cached.template b/src/main/resources/docgen/templates/beir-v1.0.0-cqadupstack-wordpress.splade-pp-ed.cached.template index a7ab22c341..088b55bfcb 100644 --- a/src/main/resources/docgen/templates/beir-v1.0.0-cqadupstack-wordpress.splade-pp-ed.cached.template +++ b/src/main/resources/docgen/templates/beir-v1.0.0-cqadupstack-wordpress.splade-pp-ed.cached.template @@ -1,14 +1,14 @@ # Anserini Regressions: BEIR (v1.0.0) — CQADupStack-wordpress -**Model**: [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) (using pre-encoded queries) +**Model**: [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) (using cached queries) -This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — CQADupStack-wordpress](http://beir.ai/). +This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — CQADupStack-wordpress](http://beir.ai/). The model itself can be download [here](https://huggingface.co/naver/splade-cocondenser-ensembledistil). See the [official SPLADE repo](/~https://github.com/naver/splade) and the following paper for more details: > Thibault Formal, Carlos Lassance, Benjamin Piwowarski, and Stéphane Clinchant. [From Distillation to Hard Negative Sampling: Making Sparse Neural IR Models More Effective.](https://dl.acm.org/doi/10.1145/3477495.3531857) _Proceedings of the 45th International ACM SIGIR Conference on Research and Development in Information Retrieval_, pages 2353–2359. -In these experiments, we are using pre-encoded queries (i.e., cached results of query encoding). +In these experiments, we are using cached queries (i.e., cached results of query encoding). The exact configurations for these regressions are stored in [this YAML file](${yaml}). Note that this page is automatically generated from [this template](${template}) as part of Anserini's regression pipeline, so do not modify this page directly; modify the template instead and then run `bin/build.sh` to rebuild the documentation. diff --git a/src/main/resources/docgen/templates/beir-v1.0.0-cqadupstack-wordpress.splade-pp-ed.onnx.template b/src/main/resources/docgen/templates/beir-v1.0.0-cqadupstack-wordpress.splade-pp-ed.onnx.template index db933448b0..10e0cf40e8 100644 --- a/src/main/resources/docgen/templates/beir-v1.0.0-cqadupstack-wordpress.splade-pp-ed.onnx.template +++ b/src/main/resources/docgen/templates/beir-v1.0.0-cqadupstack-wordpress.splade-pp-ed.onnx.template @@ -1,8 +1,8 @@ # Anserini Regressions: BEIR (v1.0.0) — CQADupStack-wordpress -**Model**: [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) (using ONNX for on-the-fly query encoding) +**Model**: [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) (using ONNX for on-the-fly query encoding) -This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — CQADupStack-wordpress](http://beir.ai/). +This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — CQADupStack-wordpress](http://beir.ai/). The model itself can be download [here](https://huggingface.co/naver/splade-cocondenser-ensembledistil). See the [official SPLADE repo](/~https://github.com/naver/splade) and the following paper for more details: diff --git a/src/main/resources/docgen/templates/beir-v1.0.0-cqadupstack-wordpress.unicoil-noexp.cached.template b/src/main/resources/docgen/templates/beir-v1.0.0-cqadupstack-wordpress.unicoil-noexp.cached.template index 8e6264e4c1..e9d33724f5 100644 --- a/src/main/resources/docgen/templates/beir-v1.0.0-cqadupstack-wordpress.unicoil-noexp.cached.template +++ b/src/main/resources/docgen/templates/beir-v1.0.0-cqadupstack-wordpress.unicoil-noexp.cached.template @@ -1,12 +1,14 @@ # Anserini Regressions: BEIR (v1.0.0) — CQADupStack-wordpress -**Model**: uniCOIL (without any expansions) +**Model**: uniCOIL without any expansions (using cached queries) This page describes regression experiments, integrated into Anserini's regression testing framework, using uniCOIL (without any expansions) on [BEIR (v1.0.0) — CQADupStack-wordpress](http://beir.ai/). 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.](https://arxiv.org/abs/2106.14807) _arXiv:2106.14807_. +In these experiments, we are using cached queries (i.e., cached results of query encoding). + The exact configurations for these regressions are stored in [this YAML file](${yaml}). Note that this page is automatically generated from [this template](${template}) as part of Anserini's regression pipeline, so do not modify this page directly; modify the template instead. diff --git a/src/main/resources/docgen/templates/beir-v1.0.0-dbpedia-entity.splade-pp-ed.cached.template b/src/main/resources/docgen/templates/beir-v1.0.0-dbpedia-entity.splade-pp-ed.cached.template index c38a106e9f..ed4d98de19 100644 --- a/src/main/resources/docgen/templates/beir-v1.0.0-dbpedia-entity.splade-pp-ed.cached.template +++ b/src/main/resources/docgen/templates/beir-v1.0.0-dbpedia-entity.splade-pp-ed.cached.template @@ -1,14 +1,14 @@ # Anserini Regressions: BEIR (v1.0.0) — DBPedia -**Model**: [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) (using pre-encoded queries) +**Model**: [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) (using cached queries) -This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — DBPedia](http://beir.ai/). +This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — DBPedia](http://beir.ai/). The model itself can be download [here](https://huggingface.co/naver/splade-cocondenser-ensembledistil). See the [official SPLADE repo](/~https://github.com/naver/splade) and the following paper for more details: > Thibault Formal, Carlos Lassance, Benjamin Piwowarski, and Stéphane Clinchant. [From Distillation to Hard Negative Sampling: Making Sparse Neural IR Models More Effective.](https://dl.acm.org/doi/10.1145/3477495.3531857) _Proceedings of the 45th International ACM SIGIR Conference on Research and Development in Information Retrieval_, pages 2353–2359. -In these experiments, we are using pre-encoded queries (i.e., cached results of query encoding). +In these experiments, we are using cached queries (i.e., cached results of query encoding). The exact configurations for these regressions are stored in [this YAML file](${yaml}). Note that this page is automatically generated from [this template](${template}) as part of Anserini's regression pipeline, so do not modify this page directly; modify the template instead and then run `bin/build.sh` to rebuild the documentation. diff --git a/src/main/resources/docgen/templates/beir-v1.0.0-dbpedia-entity.splade-pp-ed.onnx.template b/src/main/resources/docgen/templates/beir-v1.0.0-dbpedia-entity.splade-pp-ed.onnx.template index 76eae87639..fd95c90e27 100644 --- a/src/main/resources/docgen/templates/beir-v1.0.0-dbpedia-entity.splade-pp-ed.onnx.template +++ b/src/main/resources/docgen/templates/beir-v1.0.0-dbpedia-entity.splade-pp-ed.onnx.template @@ -1,8 +1,8 @@ # Anserini Regressions: BEIR (v1.0.0) — DBPedia -**Model**: [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) (using ONNX for on-the-fly query encoding) +**Model**: [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) (using ONNX for on-the-fly query encoding) -This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — DBPedia](http://beir.ai/). +This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — DBPedia](http://beir.ai/). The model itself can be download [here](https://huggingface.co/naver/splade-cocondenser-ensembledistil). See the [official SPLADE repo](/~https://github.com/naver/splade) and the following paper for more details: diff --git a/src/main/resources/docgen/templates/beir-v1.0.0-dbpedia-entity.unicoil-noexp.cached.template b/src/main/resources/docgen/templates/beir-v1.0.0-dbpedia-entity.unicoil-noexp.cached.template index 289c0c5900..e2faff6af4 100644 --- a/src/main/resources/docgen/templates/beir-v1.0.0-dbpedia-entity.unicoil-noexp.cached.template +++ b/src/main/resources/docgen/templates/beir-v1.0.0-dbpedia-entity.unicoil-noexp.cached.template @@ -1,12 +1,14 @@ # Anserini Regressions: BEIR (v1.0.0) — DBPedia -**Model**: uniCOIL (without any expansions) +**Model**: uniCOIL without any expansions (using cached queries) This page describes regression experiments, integrated into Anserini's regression testing framework, using uniCOIL (without any expansions) on [BEIR (v1.0.0) — DBPedia](http://beir.ai/). 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.](https://arxiv.org/abs/2106.14807) _arXiv:2106.14807_. +In these experiments, we are using cached queries (i.e., cached results of query encoding). + The exact configurations for these regressions are stored in [this YAML file](${yaml}). Note that this page is automatically generated from [this template](${template}) as part of Anserini's regression pipeline, so do not modify this page directly; modify the template instead. diff --git a/src/main/resources/docgen/templates/beir-v1.0.0-fever.splade-pp-ed.cached.template b/src/main/resources/docgen/templates/beir-v1.0.0-fever.splade-pp-ed.cached.template index 0ef6dfd0ea..9dd34d7198 100644 --- a/src/main/resources/docgen/templates/beir-v1.0.0-fever.splade-pp-ed.cached.template +++ b/src/main/resources/docgen/templates/beir-v1.0.0-fever.splade-pp-ed.cached.template @@ -1,14 +1,14 @@ # Anserini Regressions: BEIR (v1.0.0) — FEVER -**Model**: [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) (using pre-encoded queries) +**Model**: [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) (using cached queries) -This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — FEVER](http://beir.ai/). +This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — FEVER](http://beir.ai/). The model itself can be download [here](https://huggingface.co/naver/splade-cocondenser-ensembledistil). See the [official SPLADE repo](/~https://github.com/naver/splade) and the following paper for more details: > Thibault Formal, Carlos Lassance, Benjamin Piwowarski, and Stéphane Clinchant. [From Distillation to Hard Negative Sampling: Making Sparse Neural IR Models More Effective.](https://dl.acm.org/doi/10.1145/3477495.3531857) _Proceedings of the 45th International ACM SIGIR Conference on Research and Development in Information Retrieval_, pages 2353–2359. -In these experiments, we are using pre-encoded queries (i.e., cached results of query encoding). +In these experiments, we are using cached queries (i.e., cached results of query encoding). The exact configurations for these regressions are stored in [this YAML file](${yaml}). Note that this page is automatically generated from [this template](${template}) as part of Anserini's regression pipeline, so do not modify this page directly; modify the template instead and then run `bin/build.sh` to rebuild the documentation. diff --git a/src/main/resources/docgen/templates/beir-v1.0.0-fever.splade-pp-ed.onnx.template b/src/main/resources/docgen/templates/beir-v1.0.0-fever.splade-pp-ed.onnx.template index 3fff4c0989..0f57f552d0 100644 --- a/src/main/resources/docgen/templates/beir-v1.0.0-fever.splade-pp-ed.onnx.template +++ b/src/main/resources/docgen/templates/beir-v1.0.0-fever.splade-pp-ed.onnx.template @@ -1,8 +1,8 @@ # Anserini Regressions: BEIR (v1.0.0) — FEVER -**Model**: [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) (using ONNX for on-the-fly query encoding) +**Model**: [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) (using ONNX for on-the-fly query encoding) -This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — FEVER](http://beir.ai/). +This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — FEVER](http://beir.ai/). The model itself can be download [here](https://huggingface.co/naver/splade-cocondenser-ensembledistil). See the [official SPLADE repo](/~https://github.com/naver/splade) and the following paper for more details: diff --git a/src/main/resources/docgen/templates/beir-v1.0.0-fever.unicoil-noexp.cached.template b/src/main/resources/docgen/templates/beir-v1.0.0-fever.unicoil-noexp.cached.template index 62fab6fec7..2ca101540e 100644 --- a/src/main/resources/docgen/templates/beir-v1.0.0-fever.unicoil-noexp.cached.template +++ b/src/main/resources/docgen/templates/beir-v1.0.0-fever.unicoil-noexp.cached.template @@ -1,12 +1,14 @@ # Anserini Regressions: BEIR (v1.0.0) — FEVER -**Model**: uniCOIL (without any expansions) +**Model**: uniCOIL without any expansions (using cached queries) This page describes regression experiments, integrated into Anserini's regression testing framework, using uniCOIL (without any expansions) on [BEIR (v1.0.0) — FEVER](http://beir.ai/). 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.](https://arxiv.org/abs/2106.14807) _arXiv:2106.14807_. +In these experiments, we are using cached queries (i.e., cached results of query encoding). + The exact configurations for these regressions are stored in [this YAML file](${yaml}). Note that this page is automatically generated from [this template](${template}) as part of Anserini's regression pipeline, so do not modify this page directly; modify the template instead. diff --git a/src/main/resources/docgen/templates/beir-v1.0.0-fiqa.splade-pp-ed.cached.template b/src/main/resources/docgen/templates/beir-v1.0.0-fiqa.splade-pp-ed.cached.template index b3446ef75d..b1d6d7255c 100644 --- a/src/main/resources/docgen/templates/beir-v1.0.0-fiqa.splade-pp-ed.cached.template +++ b/src/main/resources/docgen/templates/beir-v1.0.0-fiqa.splade-pp-ed.cached.template @@ -1,14 +1,14 @@ # Anserini Regressions: BEIR (v1.0.0) — FiQA-2018 -**Model**: [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) (using pre-encoded queries) +**Model**: [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) (using cached queries) -This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — FiQA-2018](http://beir.ai/). +This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — FiQA-2018](http://beir.ai/). The model itself can be download [here](https://huggingface.co/naver/splade-cocondenser-ensembledistil). See the [official SPLADE repo](/~https://github.com/naver/splade) and the following paper for more details: > Thibault Formal, Carlos Lassance, Benjamin Piwowarski, and Stéphane Clinchant. [From Distillation to Hard Negative Sampling: Making Sparse Neural IR Models More Effective.](https://dl.acm.org/doi/10.1145/3477495.3531857) _Proceedings of the 45th International ACM SIGIR Conference on Research and Development in Information Retrieval_, pages 2353–2359. -In these experiments, we are using pre-encoded queries (i.e., cached results of query encoding). +In these experiments, we are using cached queries (i.e., cached results of query encoding). The exact configurations for these regressions are stored in [this YAML file](${yaml}). Note that this page is automatically generated from [this template](${template}) as part of Anserini's regression pipeline, so do not modify this page directly; modify the template instead and then run `bin/build.sh` to rebuild the documentation. diff --git a/src/main/resources/docgen/templates/beir-v1.0.0-fiqa.splade-pp-ed.onnx.template b/src/main/resources/docgen/templates/beir-v1.0.0-fiqa.splade-pp-ed.onnx.template index 4badd7c578..9f365bbd0a 100644 --- a/src/main/resources/docgen/templates/beir-v1.0.0-fiqa.splade-pp-ed.onnx.template +++ b/src/main/resources/docgen/templates/beir-v1.0.0-fiqa.splade-pp-ed.onnx.template @@ -1,8 +1,8 @@ # Anserini Regressions: BEIR (v1.0.0) — FiQA-2018 -**Model**: [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) (using ONNX for on-the-fly query encoding) +**Model**: [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) (using ONNX for on-the-fly query encoding) -This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — FiQA-2018](http://beir.ai/). +This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — FiQA-2018](http://beir.ai/). The model itself can be download [here](https://huggingface.co/naver/splade-cocondenser-ensembledistil). See the [official SPLADE repo](/~https://github.com/naver/splade) and the following paper for more details: diff --git a/src/main/resources/docgen/templates/beir-v1.0.0-fiqa.unicoil-noexp.cached.template b/src/main/resources/docgen/templates/beir-v1.0.0-fiqa.unicoil-noexp.cached.template index 9f29b35ca5..4ccf54327d 100644 --- a/src/main/resources/docgen/templates/beir-v1.0.0-fiqa.unicoil-noexp.cached.template +++ b/src/main/resources/docgen/templates/beir-v1.0.0-fiqa.unicoil-noexp.cached.template @@ -1,12 +1,14 @@ # Anserini Regressions: BEIR (v1.0.0) — FiQA-2018 -**Model**: uniCOIL (without any expansions) +**Model**: uniCOIL without any expansions (using cached queries) This page describes regression experiments, integrated into Anserini's regression testing framework, using uniCOIL (without any expansions) on [BEIR (v1.0.0) — FiQA-2018](http://beir.ai/). 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.](https://arxiv.org/abs/2106.14807) _arXiv:2106.14807_. +In these experiments, we are using cached queries (i.e., cached results of query encoding). + The exact configurations for these regressions are stored in [this YAML file](${yaml}). Note that this page is automatically generated from [this template](${template}) as part of Anserini's regression pipeline, so do not modify this page directly; modify the template instead. diff --git a/src/main/resources/docgen/templates/beir-v1.0.0-hotpotqa.splade-pp-ed.cached.template b/src/main/resources/docgen/templates/beir-v1.0.0-hotpotqa.splade-pp-ed.cached.template index 36a9b5e500..d3117c4425 100644 --- a/src/main/resources/docgen/templates/beir-v1.0.0-hotpotqa.splade-pp-ed.cached.template +++ b/src/main/resources/docgen/templates/beir-v1.0.0-hotpotqa.splade-pp-ed.cached.template @@ -1,14 +1,14 @@ # Anserini Regressions: BEIR (v1.0.0) — HotpotQA -**Model**: [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) (using pre-encoded queries) +**Model**: [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) (using cached queries) -This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — HotpotQA](http://beir.ai/). +This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — HotpotQA](http://beir.ai/). The model itself can be download [here](https://huggingface.co/naver/splade-cocondenser-ensembledistil). See the [official SPLADE repo](/~https://github.com/naver/splade) and the following paper for more details: > Thibault Formal, Carlos Lassance, Benjamin Piwowarski, and Stéphane Clinchant. [From Distillation to Hard Negative Sampling: Making Sparse Neural IR Models More Effective.](https://dl.acm.org/doi/10.1145/3477495.3531857) _Proceedings of the 45th International ACM SIGIR Conference on Research and Development in Information Retrieval_, pages 2353–2359. -In these experiments, we are using pre-encoded queries (i.e., cached results of query encoding). +In these experiments, we are using cached queries (i.e., cached results of query encoding). The exact configurations for these regressions are stored in [this YAML file](${yaml}). Note that this page is automatically generated from [this template](${template}) as part of Anserini's regression pipeline, so do not modify this page directly; modify the template instead and then run `bin/build.sh` to rebuild the documentation. diff --git a/src/main/resources/docgen/templates/beir-v1.0.0-hotpotqa.splade-pp-ed.onnx.template b/src/main/resources/docgen/templates/beir-v1.0.0-hotpotqa.splade-pp-ed.onnx.template index 3482a27d03..eed5f5f17f 100644 --- a/src/main/resources/docgen/templates/beir-v1.0.0-hotpotqa.splade-pp-ed.onnx.template +++ b/src/main/resources/docgen/templates/beir-v1.0.0-hotpotqa.splade-pp-ed.onnx.template @@ -1,8 +1,8 @@ # Anserini Regressions: BEIR (v1.0.0) — HotpotQA -**Model**: [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) (using ONNX for on-the-fly query encoding) +**Model**: [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) (using ONNX for on-the-fly query encoding) -This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — HotpotQA](http://beir.ai/). +This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — HotpotQA](http://beir.ai/). The model itself can be download [here](https://huggingface.co/naver/splade-cocondenser-ensembledistil). See the [official SPLADE repo](/~https://github.com/naver/splade) and the following paper for more details: diff --git a/src/main/resources/docgen/templates/beir-v1.0.0-hotpotqa.unicoil-noexp.cached.template b/src/main/resources/docgen/templates/beir-v1.0.0-hotpotqa.unicoil-noexp.cached.template index 9bac77a7ac..c6546b43df 100644 --- a/src/main/resources/docgen/templates/beir-v1.0.0-hotpotqa.unicoil-noexp.cached.template +++ b/src/main/resources/docgen/templates/beir-v1.0.0-hotpotqa.unicoil-noexp.cached.template @@ -1,12 +1,14 @@ # Anserini Regressions: BEIR (v1.0.0) — HotpotQA -**Model**: uniCOIL (without any expansions) +**Model**: uniCOIL without any expansions (using cached queries) This page describes regression experiments, integrated into Anserini's regression testing framework, using uniCOIL (without any expansions) on [BEIR (v1.0.0) — HotpotQA](http://beir.ai/). 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.](https://arxiv.org/abs/2106.14807) _arXiv:2106.14807_. +In these experiments, we are using cached queries (i.e., cached results of query encoding). + The exact configurations for these regressions are stored in [this YAML file](${yaml}). Note that this page is automatically generated from [this template](${template}) as part of Anserini's regression pipeline, so do not modify this page directly; modify the template instead. diff --git a/src/main/resources/docgen/templates/beir-v1.0.0-nfcorpus.splade-pp-ed.cached.template b/src/main/resources/docgen/templates/beir-v1.0.0-nfcorpus.splade-pp-ed.cached.template index 9881ed4927..4c5829e4db 100644 --- a/src/main/resources/docgen/templates/beir-v1.0.0-nfcorpus.splade-pp-ed.cached.template +++ b/src/main/resources/docgen/templates/beir-v1.0.0-nfcorpus.splade-pp-ed.cached.template @@ -1,14 +1,14 @@ -# Anserini Regressions: BEIR (v1.0.0) — NFCorpus +# Anserini Regressions: BEIR (v1.0.0) — NCorpus -**Model**: [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) (using pre-encoded queries) +**Model**: [SPLADE++ (CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) (using cached queries) -This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — NFCorpus](http://beir.ai/). +This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — NFCorpus](http://beir.ai/). The model itself can be download [here](https://huggingface.co/naver/splade-cocondenser-ensembledistil). See the [official SPLADE repo](/~https://github.com/naver/splade) and the following paper for more details: > Thibault Formal, Carlos Lassance, Benjamin Piwowarski, and Stéphane Clinchant. [From Distillation to Hard Negative Sampling: Making Sparse Neural IR Models More Effective.](https://dl.acm.org/doi/10.1145/3477495.3531857) _Proceedings of the 45th International ACM SIGIR Conference on Research and Development in Information Retrieval_, pages 2353–2359. -In these experiments, we are using pre-encoded queries (i.e., cached results of query encoding). +In these experiments, we are using cached queries (i.e., cached results of query encoding). The exact configurations for these regressions are stored in [this YAML file](${yaml}). Note that this page is automatically generated from [this template](${template}) as part of Anserini's regression pipeline, so do not modify this page directly; modify the template instead and then run `bin/build.sh` to rebuild the documentation. diff --git a/src/main/resources/docgen/templates/beir-v1.0.0-nfcorpus.splade-pp-ed.onnx.template b/src/main/resources/docgen/templates/beir-v1.0.0-nfcorpus.splade-pp-ed.onnx.template index 34fa1be91a..6694163005 100644 --- a/src/main/resources/docgen/templates/beir-v1.0.0-nfcorpus.splade-pp-ed.onnx.template +++ b/src/main/resources/docgen/templates/beir-v1.0.0-nfcorpus.splade-pp-ed.onnx.template @@ -1,8 +1,8 @@ # Anserini Regressions: BEIR (v1.0.0) — NFCorpus -**Model**: [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) (using ONNX for on-the-fly query encoding) +**Model**: [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) (using ONNX for on-the-fly query encoding) -This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — NFCorpus](http://beir.ai/). +This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — NFCorpus](http://beir.ai/). The model itself can be download [here](https://huggingface.co/naver/splade-cocondenser-ensembledistil). See the [official SPLADE repo](/~https://github.com/naver/splade) and the following paper for more details: diff --git a/src/main/resources/docgen/templates/beir-v1.0.0-nfcorpus.unicoil-noexp.cached.template b/src/main/resources/docgen/templates/beir-v1.0.0-nfcorpus.unicoil-noexp.cached.template index 02ac462121..72ef339096 100644 --- a/src/main/resources/docgen/templates/beir-v1.0.0-nfcorpus.unicoil-noexp.cached.template +++ b/src/main/resources/docgen/templates/beir-v1.0.0-nfcorpus.unicoil-noexp.cached.template @@ -1,12 +1,14 @@ # Anserini Regressions: BEIR (v1.0.0) — NFCorpus -**Model**: uniCOIL (without any expansions) +**Model**: uniCOIL without any expansions (using cached queries) This page describes regression experiments, integrated into Anserini's regression testing framework, using uniCOIL (without any expansions) on [BEIR (v1.0.0) — NFCorpus](http://beir.ai/). 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.](https://arxiv.org/abs/2106.14807) _arXiv:2106.14807_. +In these experiments, we are using cached queries (i.e., cached results of query encoding). + The exact configurations for these regressions are stored in [this YAML file](${yaml}). Note that this page is automatically generated from [this template](${template}) as part of Anserini's regression pipeline, so do not modify this page directly; modify the template instead. diff --git a/src/main/resources/docgen/templates/beir-v1.0.0-nq.splade-pp-ed.cached.template b/src/main/resources/docgen/templates/beir-v1.0.0-nq.splade-pp-ed.cached.template index 3c217b64f0..3d26230aa6 100644 --- a/src/main/resources/docgen/templates/beir-v1.0.0-nq.splade-pp-ed.cached.template +++ b/src/main/resources/docgen/templates/beir-v1.0.0-nq.splade-pp-ed.cached.template @@ -1,14 +1,14 @@ # Anserini Regressions: BEIR (v1.0.0) — NQ -**Model**: [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) (using pre-encoded queries) +**Model**: [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) (using cached queries) -This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — NQ](http://beir.ai/). +This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — NQ](http://beir.ai/). The model itself can be download [here](https://huggingface.co/naver/splade-cocondenser-ensembledistil). See the [official SPLADE repo](/~https://github.com/naver/splade) and the following paper for more details: > Thibault Formal, Carlos Lassance, Benjamin Piwowarski, and Stéphane Clinchant. [From Distillation to Hard Negative Sampling: Making Sparse Neural IR Models More Effective.](https://dl.acm.org/doi/10.1145/3477495.3531857) _Proceedings of the 45th International ACM SIGIR Conference on Research and Development in Information Retrieval_, pages 2353–2359. -In these experiments, we are using pre-encoded queries (i.e., cached results of query encoding). +In these experiments, we are using cached queries (i.e., cached results of query encoding). The exact configurations for these regressions are stored in [this YAML file](${yaml}). Note that this page is automatically generated from [this template](${template}) as part of Anserini's regression pipeline, so do not modify this page directly; modify the template instead and then run `bin/build.sh` to rebuild the documentation. diff --git a/src/main/resources/docgen/templates/beir-v1.0.0-nq.splade-pp-ed.onnx.template b/src/main/resources/docgen/templates/beir-v1.0.0-nq.splade-pp-ed.onnx.template index fce0339adc..05b0dd1b5b 100644 --- a/src/main/resources/docgen/templates/beir-v1.0.0-nq.splade-pp-ed.onnx.template +++ b/src/main/resources/docgen/templates/beir-v1.0.0-nq.splade-pp-ed.onnx.template @@ -1,8 +1,8 @@ # Anserini Regressions: BEIR (v1.0.0) — NQ -**Model**: [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) (using ONNX for on-the-fly query encoding) +**Model**: [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) (using ONNX for on-the-fly query encoding) -This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — NQ](http://beir.ai/). +This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — NQ](http://beir.ai/). The model itself can be download [here](https://huggingface.co/naver/splade-cocondenser-ensembledistil). See the [official SPLADE repo](/~https://github.com/naver/splade) and the following paper for more details: diff --git a/src/main/resources/docgen/templates/beir-v1.0.0-nq.unicoil-noexp.cached.template b/src/main/resources/docgen/templates/beir-v1.0.0-nq.unicoil-noexp.cached.template index 1e069ab9b2..a83400d9eb 100644 --- a/src/main/resources/docgen/templates/beir-v1.0.0-nq.unicoil-noexp.cached.template +++ b/src/main/resources/docgen/templates/beir-v1.0.0-nq.unicoil-noexp.cached.template @@ -1,12 +1,14 @@ # Anserini Regressions: BEIR (v1.0.0) — NQ -**Model**: uniCOIL (without any expansions) +**Model**: uniCOIL without any expansions (using cached queries) This page describes regression experiments, integrated into Anserini's regression testing framework, using uniCOIL (without any expansions) on [BEIR (v1.0.0) — NQ](http://beir.ai/). 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.](https://arxiv.org/abs/2106.14807) _arXiv:2106.14807_. +In these experiments, we are using cached queries (i.e., cached results of query encoding). + The exact configurations for these regressions are stored in [this YAML file](${yaml}). Note that this page is automatically generated from [this template](${template}) as part of Anserini's regression pipeline, so do not modify this page directly; modify the template instead. diff --git a/src/main/resources/docgen/templates/beir-v1.0.0-quora.splade-pp-ed.cached.template b/src/main/resources/docgen/templates/beir-v1.0.0-quora.splade-pp-ed.cached.template index da8db9a4b8..bc619e0703 100644 --- a/src/main/resources/docgen/templates/beir-v1.0.0-quora.splade-pp-ed.cached.template +++ b/src/main/resources/docgen/templates/beir-v1.0.0-quora.splade-pp-ed.cached.template @@ -1,14 +1,14 @@ # Anserini Regressions: BEIR (v1.0.0) — Quora -**Model**: [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) (using pre-encoded queries) +**Model**: [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) (using cached queries) -This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — Quora](http://beir.ai/). +This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — Quora](http://beir.ai/). The model itself can be download [here](https://huggingface.co/naver/splade-cocondenser-ensembledistil). See the [official SPLADE repo](/~https://github.com/naver/splade) and the following paper for more details: > Thibault Formal, Carlos Lassance, Benjamin Piwowarski, and Stéphane Clinchant. [From Distillation to Hard Negative Sampling: Making Sparse Neural IR Models More Effective.](https://dl.acm.org/doi/10.1145/3477495.3531857) _Proceedings of the 45th International ACM SIGIR Conference on Research and Development in Information Retrieval_, pages 2353–2359. -In these experiments, we are using pre-encoded queries (i.e., cached results of query encoding). +In these experiments, we are using cached queries (i.e., cached results of query encoding). The exact configurations for these regressions are stored in [this YAML file](${yaml}). Note that this page is automatically generated from [this template](${template}) as part of Anserini's regression pipeline, so do not modify this page directly; modify the template instead and then run `bin/build.sh` to rebuild the documentation. diff --git a/src/main/resources/docgen/templates/beir-v1.0.0-quora.splade-pp-ed.onnx.template b/src/main/resources/docgen/templates/beir-v1.0.0-quora.splade-pp-ed.onnx.template index e1802930a8..419d4a0532 100644 --- a/src/main/resources/docgen/templates/beir-v1.0.0-quora.splade-pp-ed.onnx.template +++ b/src/main/resources/docgen/templates/beir-v1.0.0-quora.splade-pp-ed.onnx.template @@ -1,8 +1,8 @@ # Anserini Regressions: BEIR (v1.0.0) — Quora -**Model**: [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) (using ONNX for on-the-fly query encoding) +**Model**: [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) (using ONNX for on-the-fly query encoding) -This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — Quora](http://beir.ai/). +This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — Quora](http://beir.ai/). The model itself can be download [here](https://huggingface.co/naver/splade-cocondenser-ensembledistil). See the [official SPLADE repo](/~https://github.com/naver/splade) and the following paper for more details: diff --git a/src/main/resources/docgen/templates/beir-v1.0.0-quora.unicoil-noexp.cached.template b/src/main/resources/docgen/templates/beir-v1.0.0-quora.unicoil-noexp.cached.template index 8fd04ae393..1939017567 100644 --- a/src/main/resources/docgen/templates/beir-v1.0.0-quora.unicoil-noexp.cached.template +++ b/src/main/resources/docgen/templates/beir-v1.0.0-quora.unicoil-noexp.cached.template @@ -1,12 +1,14 @@ # Anserini Regressions: BEIR (v1.0.0) — Quora -**Model**: uniCOIL (without any expansions) +**Model**: uniCOIL without any expansions (using cached queries) This page describes regression experiments, integrated into Anserini's regression testing framework, using uniCOIL (without any expansions) on [BEIR (v1.0.0) — Quora](http://beir.ai/). 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.](https://arxiv.org/abs/2106.14807) _arXiv:2106.14807_. +In these experiments, we are using cached queries (i.e., cached results of query encoding). + The exact configurations for these regressions are stored in [this YAML file](${yaml}). Note that this page is automatically generated from [this template](${template}) as part of Anserini's regression pipeline, so do not modify this page directly; modify the template instead. diff --git a/src/main/resources/docgen/templates/beir-v1.0.0-robust04.splade-pp-ed.cached.template b/src/main/resources/docgen/templates/beir-v1.0.0-robust04.splade-pp-ed.cached.template index bcb19f9f06..d7e03177b8 100644 --- a/src/main/resources/docgen/templates/beir-v1.0.0-robust04.splade-pp-ed.cached.template +++ b/src/main/resources/docgen/templates/beir-v1.0.0-robust04.splade-pp-ed.cached.template @@ -1,14 +1,14 @@ # Anserini Regressions: BEIR (v1.0.0) — Robust04 -**Model**: [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) (using pre-encoded queries) +**Model**: [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) (using cached queries) -This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — Robust04](http://beir.ai/). +This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — Robust04](http://beir.ai/). The model itself can be download [here](https://huggingface.co/naver/splade-cocondenser-ensembledistil). See the [official SPLADE repo](/~https://github.com/naver/splade) and the following paper for more details: > Thibault Formal, Carlos Lassance, Benjamin Piwowarski, and Stéphane Clinchant. [From Distillation to Hard Negative Sampling: Making Sparse Neural IR Models More Effective.](https://dl.acm.org/doi/10.1145/3477495.3531857) _Proceedings of the 45th International ACM SIGIR Conference on Research and Development in Information Retrieval_, pages 2353–2359. -In these experiments, we are using pre-encoded queries (i.e., cached results of query encoding). +In these experiments, we are using cached queries (i.e., cached results of query encoding). The exact configurations for these regressions are stored in [this YAML file](${yaml}). Note that this page is automatically generated from [this template](${template}) as part of Anserini's regression pipeline, so do not modify this page directly; modify the template instead and then run `bin/build.sh` to rebuild the documentation. diff --git a/src/main/resources/docgen/templates/beir-v1.0.0-robust04.splade-pp-ed.onnx.template b/src/main/resources/docgen/templates/beir-v1.0.0-robust04.splade-pp-ed.onnx.template index 749f2bc039..6ed13dd977 100644 --- a/src/main/resources/docgen/templates/beir-v1.0.0-robust04.splade-pp-ed.onnx.template +++ b/src/main/resources/docgen/templates/beir-v1.0.0-robust04.splade-pp-ed.onnx.template @@ -1,8 +1,8 @@ # Anserini Regressions: BEIR (v1.0.0) — Robust04 -**Model**: [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) (using ONNX for on-the-fly query encoding) +**Model**: [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) (using ONNX for on-the-fly query encoding) -This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — Robust04](http://beir.ai/). +This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — Robust04](http://beir.ai/). The model itself can be download [here](https://huggingface.co/naver/splade-cocondenser-ensembledistil). See the [official SPLADE repo](/~https://github.com/naver/splade) and the following paper for more details: diff --git a/src/main/resources/docgen/templates/beir-v1.0.0-robust04.unicoil-noexp.cached.template b/src/main/resources/docgen/templates/beir-v1.0.0-robust04.unicoil-noexp.cached.template index 39daec88b4..3bdb306658 100644 --- a/src/main/resources/docgen/templates/beir-v1.0.0-robust04.unicoil-noexp.cached.template +++ b/src/main/resources/docgen/templates/beir-v1.0.0-robust04.unicoil-noexp.cached.template @@ -1,12 +1,14 @@ # Anserini Regressions: BEIR (v1.0.0) — Robust04 -**Model**: uniCOIL (without any expansions) +**Model**: uniCOIL without any expansions (using cached queries) This page describes regression experiments, integrated into Anserini's regression testing framework, using uniCOIL (without any expansions) on [BEIR (v1.0.0) — Robust04](http://beir.ai/). 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.](https://arxiv.org/abs/2106.14807) _arXiv:2106.14807_. +In these experiments, we are using cached queries (i.e., cached results of query encoding). + The exact configurations for these regressions are stored in [this YAML file](${yaml}). Note that this page is automatically generated from [this template](${template}) as part of Anserini's regression pipeline, so do not modify this page directly; modify the template instead. diff --git a/src/main/resources/docgen/templates/beir-v1.0.0-scidocs.splade-pp-ed.cached.template b/src/main/resources/docgen/templates/beir-v1.0.0-scidocs.splade-pp-ed.cached.template index 50dca54b5d..f6bc71930b 100644 --- a/src/main/resources/docgen/templates/beir-v1.0.0-scidocs.splade-pp-ed.cached.template +++ b/src/main/resources/docgen/templates/beir-v1.0.0-scidocs.splade-pp-ed.cached.template @@ -1,14 +1,14 @@ # Anserini Regressions: BEIR (v1.0.0) — SCIDOCS -**Model**: [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) (using pre-encoded queries) +**Model**: [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) (using cached queries) -This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — SCIDOCS](http://beir.ai/). +This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — SCIDOCS](http://beir.ai/). The model itself can be download [here](https://huggingface.co/naver/splade-cocondenser-ensembledistil). See the [official SPLADE repo](/~https://github.com/naver/splade) and the following paper for more details: > Thibault Formal, Carlos Lassance, Benjamin Piwowarski, and Stéphane Clinchant. [From Distillation to Hard Negative Sampling: Making Sparse Neural IR Models More Effective.](https://dl.acm.org/doi/10.1145/3477495.3531857) _Proceedings of the 45th International ACM SIGIR Conference on Research and Development in Information Retrieval_, pages 2353–2359. -In these experiments, we are using pre-encoded queries (i.e., cached results of query encoding). +In these experiments, we are using cached queries (i.e., cached results of query encoding). The exact configurations for these regressions are stored in [this YAML file](${yaml}). Note that this page is automatically generated from [this template](${template}) as part of Anserini's regression pipeline, so do not modify this page directly; modify the template instead and then run `bin/build.sh` to rebuild the documentation. diff --git a/src/main/resources/docgen/templates/beir-v1.0.0-scidocs.splade-pp-ed.onnx.template b/src/main/resources/docgen/templates/beir-v1.0.0-scidocs.splade-pp-ed.onnx.template index aee90cce28..77bd38c882 100644 --- a/src/main/resources/docgen/templates/beir-v1.0.0-scidocs.splade-pp-ed.onnx.template +++ b/src/main/resources/docgen/templates/beir-v1.0.0-scidocs.splade-pp-ed.onnx.template @@ -1,8 +1,8 @@ # Anserini Regressions: BEIR (v1.0.0) — SCIDOCS -**Model**: [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) (using ONNX for on-the-fly query encoding) +**Model**: [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) (using ONNX for on-the-fly query encoding) -This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — SCIDOCS](http://beir.ai/). +This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — SCIDOCS](http://beir.ai/). The model itself can be download [here](https://huggingface.co/naver/splade-cocondenser-ensembledistil). See the [official SPLADE repo](/~https://github.com/naver/splade) and the following paper for more details: diff --git a/src/main/resources/docgen/templates/beir-v1.0.0-scidocs.unicoil-noexp.cached.template b/src/main/resources/docgen/templates/beir-v1.0.0-scidocs.unicoil-noexp.cached.template index 2d2a4502a3..972595d38e 100644 --- a/src/main/resources/docgen/templates/beir-v1.0.0-scidocs.unicoil-noexp.cached.template +++ b/src/main/resources/docgen/templates/beir-v1.0.0-scidocs.unicoil-noexp.cached.template @@ -1,12 +1,14 @@ # Anserini Regressions: BEIR (v1.0.0) — SCIDOCS -**Model**: uniCOIL (without any expansions) +**Model**: uniCOIL without any expansions (using cached queries) This page describes regression experiments, integrated into Anserini's regression testing framework, using uniCOIL (without any expansions) on [BEIR (v1.0.0) — SCIDOCS](http://beir.ai/). 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.](https://arxiv.org/abs/2106.14807) _arXiv:2106.14807_. +In these experiments, we are using cached queries (i.e., cached results of query encoding). + The exact configurations for these regressions are stored in [this YAML file](${yaml}). Note that this page is automatically generated from [this template](${template}) as part of Anserini's regression pipeline, so do not modify this page directly; modify the template instead. diff --git a/src/main/resources/docgen/templates/beir-v1.0.0-scifact.splade-pp-ed.cached.template b/src/main/resources/docgen/templates/beir-v1.0.0-scifact.splade-pp-ed.cached.template index 1a83d67668..8a41138c76 100644 --- a/src/main/resources/docgen/templates/beir-v1.0.0-scifact.splade-pp-ed.cached.template +++ b/src/main/resources/docgen/templates/beir-v1.0.0-scifact.splade-pp-ed.cached.template @@ -1,14 +1,14 @@ # Anserini Regressions: BEIR (v1.0.0) — SciFact -**Model**: [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) (using pre-encoded queries) +**Model**: [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) (using cached queries) -This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — SciFact](http://beir.ai/). +This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — SciFact](http://beir.ai/). The model itself can be download [here](https://huggingface.co/naver/splade-cocondenser-ensembledistil). See the [official SPLADE repo](/~https://github.com/naver/splade) and the following paper for more details: > Thibault Formal, Carlos Lassance, Benjamin Piwowarski, and Stéphane Clinchant. [From Distillation to Hard Negative Sampling: Making Sparse Neural IR Models More Effective.](https://dl.acm.org/doi/10.1145/3477495.3531857) _Proceedings of the 45th International ACM SIGIR Conference on Research and Development in Information Retrieval_, pages 2353–2359. -In these experiments, we are using pre-encoded queries (i.e., cached results of query encoding). +In these experiments, we are using cached queries (i.e., cached results of query encoding). The exact configurations for these regressions are stored in [this YAML file](${yaml}). Note that this page is automatically generated from [this template](${template}) as part of Anserini's regression pipeline, so do not modify this page directly; modify the template instead and then run `bin/build.sh` to rebuild the documentation. diff --git a/src/main/resources/docgen/templates/beir-v1.0.0-scifact.splade-pp-ed.onnx.template b/src/main/resources/docgen/templates/beir-v1.0.0-scifact.splade-pp-ed.onnx.template index c8606a593f..5da59994cf 100644 --- a/src/main/resources/docgen/templates/beir-v1.0.0-scifact.splade-pp-ed.onnx.template +++ b/src/main/resources/docgen/templates/beir-v1.0.0-scifact.splade-pp-ed.onnx.template @@ -1,8 +1,8 @@ # Anserini Regressions: BEIR (v1.0.0) — SciFact -**Model**: [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) (using ONNX for on-the-fly query encoding) +**Model**: [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) (using ONNX for on-the-fly query encoding) -This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — SciFact](http://beir.ai/). +This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — SciFact](http://beir.ai/). The model itself can be download [here](https://huggingface.co/naver/splade-cocondenser-ensembledistil). See the [official SPLADE repo](/~https://github.com/naver/splade) and the following paper for more details: diff --git a/src/main/resources/docgen/templates/beir-v1.0.0-scifact.unicoil-noexp.cached.template b/src/main/resources/docgen/templates/beir-v1.0.0-scifact.unicoil-noexp.cached.template index b5c1ba0cc0..a0e1c54cab 100644 --- a/src/main/resources/docgen/templates/beir-v1.0.0-scifact.unicoil-noexp.cached.template +++ b/src/main/resources/docgen/templates/beir-v1.0.0-scifact.unicoil-noexp.cached.template @@ -1,12 +1,14 @@ # Anserini Regressions: BEIR (v1.0.0) — SciFact -**Model**: uniCOIL (without any expansions) +**Model**: uniCOIL without any expansions (using cached queries) This page describes regression experiments, integrated into Anserini's regression testing framework, using uniCOIL (without any expansions) on [BEIR (v1.0.0) — SciFact](http://beir.ai/). 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.](https://arxiv.org/abs/2106.14807) _arXiv:2106.14807_. +In these experiments, we are using cached queries (i.e., cached results of query encoding). + The exact configurations for these regressions are stored in [this YAML file](${yaml}). Note that this page is automatically generated from [this template](${template}) as part of Anserini's regression pipeline, so do not modify this page directly; modify the template instead. diff --git a/src/main/resources/docgen/templates/beir-v1.0.0-signal1m.splade-pp-ed.cached.template b/src/main/resources/docgen/templates/beir-v1.0.0-signal1m.splade-pp-ed.cached.template index b32e423286..c03db83397 100644 --- a/src/main/resources/docgen/templates/beir-v1.0.0-signal1m.splade-pp-ed.cached.template +++ b/src/main/resources/docgen/templates/beir-v1.0.0-signal1m.splade-pp-ed.cached.template @@ -1,14 +1,14 @@ # Anserini Regressions: BEIR (v1.0.0) — Signal-1M -**Model**: [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) (using pre-encoded queries) +**Model**: [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) (using cached queries) -This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — Signal-1M](http://beir.ai/). +This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — Signal-1M](http://beir.ai/). The model itself can be download [here](https://huggingface.co/naver/splade-cocondenser-ensembledistil). See the [official SPLADE repo](/~https://github.com/naver/splade) and the following paper for more details: > Thibault Formal, Carlos Lassance, Benjamin Piwowarski, and Stéphane Clinchant. [From Distillation to Hard Negative Sampling: Making Sparse Neural IR Models More Effective.](https://dl.acm.org/doi/10.1145/3477495.3531857) _Proceedings of the 45th International ACM SIGIR Conference on Research and Development in Information Retrieval_, pages 2353–2359. -In these experiments, we are using pre-encoded queries (i.e., cached results of query encoding). +In these experiments, we are using cached queries (i.e., cached results of query encoding). The exact configurations for these regressions are stored in [this YAML file](${yaml}). Note that this page is automatically generated from [this template](${template}) as part of Anserini's regression pipeline, so do not modify this page directly; modify the template instead and then run `bin/build.sh` to rebuild the documentation. diff --git a/src/main/resources/docgen/templates/beir-v1.0.0-signal1m.splade-pp-ed.onnx.template b/src/main/resources/docgen/templates/beir-v1.0.0-signal1m.splade-pp-ed.onnx.template index 875d24dc38..47ff0f165a 100644 --- a/src/main/resources/docgen/templates/beir-v1.0.0-signal1m.splade-pp-ed.onnx.template +++ b/src/main/resources/docgen/templates/beir-v1.0.0-signal1m.splade-pp-ed.onnx.template @@ -1,8 +1,8 @@ # Anserini Regressions: BEIR (v1.0.0) — Signal-1M -**Model**: [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) (using ONNX for on-the-fly query encoding) +**Model**: [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) (using ONNX for on-the-fly query encoding) -This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — Signal-1M](http://beir.ai/). +This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — Signal-1M](http://beir.ai/). The model itself can be download [here](https://huggingface.co/naver/splade-cocondenser-ensembledistil). See the [official SPLADE repo](/~https://github.com/naver/splade) and the following paper for more details: diff --git a/src/main/resources/docgen/templates/beir-v1.0.0-signal1m.unicoil-noexp.cached.template b/src/main/resources/docgen/templates/beir-v1.0.0-signal1m.unicoil-noexp.cached.template index 1069b452c8..916c93ce2d 100644 --- a/src/main/resources/docgen/templates/beir-v1.0.0-signal1m.unicoil-noexp.cached.template +++ b/src/main/resources/docgen/templates/beir-v1.0.0-signal1m.unicoil-noexp.cached.template @@ -1,12 +1,14 @@ # Anserini Regressions: BEIR (v1.0.0) — Signal-1M -**Model**: uniCOIL (without any expansions) +**Model**: uniCOIL without any expansions (using cached queries) This page describes regression experiments, integrated into Anserini's regression testing framework, using uniCOIL (without any expansions) on [BEIR (v1.0.0) — Signal-1M](http://beir.ai/). 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.](https://arxiv.org/abs/2106.14807) _arXiv:2106.14807_. +In these experiments, we are using cached queries (i.e., cached results of query encoding). + The exact configurations for these regressions are stored in [this YAML file](${yaml}). Note that this page is automatically generated from [this template](${template}) as part of Anserini's regression pipeline, so do not modify this page directly; modify the template instead. diff --git a/src/main/resources/docgen/templates/beir-v1.0.0-trec-covid.splade-pp-ed.cached.template b/src/main/resources/docgen/templates/beir-v1.0.0-trec-covid.splade-pp-ed.cached.template index b9818dbdc5..8be3d42ad7 100644 --- a/src/main/resources/docgen/templates/beir-v1.0.0-trec-covid.splade-pp-ed.cached.template +++ b/src/main/resources/docgen/templates/beir-v1.0.0-trec-covid.splade-pp-ed.cached.template @@ -1,14 +1,14 @@ # Anserini Regressions: BEIR (v1.0.0) — TREC-COVID -**Model**: [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) (using pre-encoded queries) +**Model**: [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) (using cached queries) -This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — TREC-COVID](http://beir.ai/). +This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — TREC-COVID](http://beir.ai/). The model itself can be download [here](https://huggingface.co/naver/splade-cocondenser-ensembledistil). See the [official SPLADE repo](/~https://github.com/naver/splade) and the following paper for more details: > Thibault Formal, Carlos Lassance, Benjamin Piwowarski, and Stéphane Clinchant. [From Distillation to Hard Negative Sampling: Making Sparse Neural IR Models More Effective.](https://dl.acm.org/doi/10.1145/3477495.3531857) _Proceedings of the 45th International ACM SIGIR Conference on Research and Development in Information Retrieval_, pages 2353–2359. -In these experiments, we are using pre-encoded queries (i.e., cached results of query encoding). +In these experiments, we are using cached queries (i.e., cached results of query encoding). The exact configurations for these regressions are stored in [this YAML file](${yaml}). Note that this page is automatically generated from [this template](${template}) as part of Anserini's regression pipeline, so do not modify this page directly; modify the template instead and then run `bin/build.sh` to rebuild the documentation. diff --git a/src/main/resources/docgen/templates/beir-v1.0.0-trec-covid.splade-pp-ed.onnx.template b/src/main/resources/docgen/templates/beir-v1.0.0-trec-covid.splade-pp-ed.onnx.template index 07e170061f..5d28a555d4 100644 --- a/src/main/resources/docgen/templates/beir-v1.0.0-trec-covid.splade-pp-ed.onnx.template +++ b/src/main/resources/docgen/templates/beir-v1.0.0-trec-covid.splade-pp-ed.onnx.template @@ -1,8 +1,8 @@ # Anserini Regressions: BEIR (v1.0.0) — TREC-COVID -**Model**: [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) (using ONNX for on-the-fly query encoding) +**Model**: [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) (using ONNX for on-the-fly query encoding) -This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — TREC-COVID](http://beir.ai/). +This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — TREC-COVID](http://beir.ai/). The model itself can be download [here](https://huggingface.co/naver/splade-cocondenser-ensembledistil). See the [official SPLADE repo](/~https://github.com/naver/splade) and the following paper for more details: diff --git a/src/main/resources/docgen/templates/beir-v1.0.0-trec-covid.unicoil-noexp.cached.template b/src/main/resources/docgen/templates/beir-v1.0.0-trec-covid.unicoil-noexp.cached.template index 01e8f042c7..7152fde9cb 100644 --- a/src/main/resources/docgen/templates/beir-v1.0.0-trec-covid.unicoil-noexp.cached.template +++ b/src/main/resources/docgen/templates/beir-v1.0.0-trec-covid.unicoil-noexp.cached.template @@ -1,12 +1,14 @@ # Anserini Regressions: BEIR (v1.0.0) — TREC-COVID -**Model**: uniCOIL (without any expansions) +**Model**: uniCOIL without any expansions (using cached queries) This page describes regression experiments, integrated into Anserini's regression testing framework, using uniCOIL (without any expansions) on [BEIR (v1.0.0) — TREC-COVID](http://beir.ai/). 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.](https://arxiv.org/abs/2106.14807) _arXiv:2106.14807_. +In these experiments, we are using cached queries (i.e., cached results of query encoding). + The exact configurations for these regressions are stored in [this YAML file](${yaml}). Note that this page is automatically generated from [this template](${template}) as part of Anserini's regression pipeline, so do not modify this page directly; modify the template instead. diff --git a/src/main/resources/docgen/templates/beir-v1.0.0-trec-news.splade-pp-ed.cached.template b/src/main/resources/docgen/templates/beir-v1.0.0-trec-news.splade-pp-ed.cached.template index 999fa3a017..4800b232b6 100644 --- a/src/main/resources/docgen/templates/beir-v1.0.0-trec-news.splade-pp-ed.cached.template +++ b/src/main/resources/docgen/templates/beir-v1.0.0-trec-news.splade-pp-ed.cached.template @@ -1,14 +1,14 @@ # Anserini Regressions: BEIR (v1.0.0) — TREC-NEWS -**Model**: [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) (using pre-encoded queries) +**Model**: [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) (using cached queries) -This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — TREC-NEWS](http://beir.ai/). +This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — TREC-NEWS](http://beir.ai/). The model itself can be download [here](https://huggingface.co/naver/splade-cocondenser-ensembledistil). See the [official SPLADE repo](/~https://github.com/naver/splade) and the following paper for more details: > Thibault Formal, Carlos Lassance, Benjamin Piwowarski, and Stéphane Clinchant. [From Distillation to Hard Negative Sampling: Making Sparse Neural IR Models More Effective.](https://dl.acm.org/doi/10.1145/3477495.3531857) _Proceedings of the 45th International ACM SIGIR Conference on Research and Development in Information Retrieval_, pages 2353–2359. -In these experiments, we are using pre-encoded queries (i.e., cached results of query encoding). +In these experiments, we are using cached queries (i.e., cached results of query encoding). The exact configurations for these regressions are stored in [this YAML file](${yaml}). Note that this page is automatically generated from [this template](${template}) as part of Anserini's regression pipeline, so do not modify this page directly; modify the template instead and then run `bin/build.sh` to rebuild the documentation. diff --git a/src/main/resources/docgen/templates/beir-v1.0.0-trec-news.splade-pp-ed.onnx.template b/src/main/resources/docgen/templates/beir-v1.0.0-trec-news.splade-pp-ed.onnx.template index 8429dfe597..2433e13344 100644 --- a/src/main/resources/docgen/templates/beir-v1.0.0-trec-news.splade-pp-ed.onnx.template +++ b/src/main/resources/docgen/templates/beir-v1.0.0-trec-news.splade-pp-ed.onnx.template @@ -1,8 +1,8 @@ # Anserini Regressions: BEIR (v1.0.0) — TREC-NEWS -**Model**: [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) (using ONNX for on-the-fly query encoding) +**Model**: [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) (using ONNX for on-the-fly query encoding) -This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — TREC-NEWS](http://beir.ai/). +This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — TREC-NEWS](http://beir.ai/). The model itself can be download [here](https://huggingface.co/naver/splade-cocondenser-ensembledistil). See the [official SPLADE repo](/~https://github.com/naver/splade) and the following paper for more details: diff --git a/src/main/resources/docgen/templates/beir-v1.0.0-trec-news.unicoil-noexp.cached.template b/src/main/resources/docgen/templates/beir-v1.0.0-trec-news.unicoil-noexp.cached.template index 7a3f983d2b..7dff4422c2 100644 --- a/src/main/resources/docgen/templates/beir-v1.0.0-trec-news.unicoil-noexp.cached.template +++ b/src/main/resources/docgen/templates/beir-v1.0.0-trec-news.unicoil-noexp.cached.template @@ -1,12 +1,14 @@ # Anserini Regressions: BEIR (v1.0.0) — TREC-NEWS -**Model**: uniCOIL (without any expansions) +**Model**: uniCOIL without any expansions (using cached queries) This page describes regression experiments, integrated into Anserini's regression testing framework, using uniCOIL (without any expansions) on [BEIR (v1.0.0) — TREC-NEWS](http://beir.ai/). 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.](https://arxiv.org/abs/2106.14807) _arXiv:2106.14807_. +In these experiments, we are using cached queries (i.e., cached results of query encoding). + The exact configurations for these regressions are stored in [this YAML file](${yaml}). Note that this page is automatically generated from [this template](${template}) as part of Anserini's regression pipeline, so do not modify this page directly; modify the template instead. diff --git a/src/main/resources/docgen/templates/beir-v1.0.0-webis-touche2020.splade-pp-ed.cached.template b/src/main/resources/docgen/templates/beir-v1.0.0-webis-touche2020.splade-pp-ed.cached.template index 932c0767e5..e3bbdb3787 100644 --- a/src/main/resources/docgen/templates/beir-v1.0.0-webis-touche2020.splade-pp-ed.cached.template +++ b/src/main/resources/docgen/templates/beir-v1.0.0-webis-touche2020.splade-pp-ed.cached.template @@ -1,14 +1,14 @@ # Anserini Regressions: BEIR (v1.0.0) — Webis-Touche2020 -**Model**: [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) (using pre-encoded queries) +**Model**: [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) (using cached queries) -This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — Webis-Touche2020](http://beir.ai/). +This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — Webis-Touche2020](http://beir.ai/). The model itself can be download [here](https://huggingface.co/naver/splade-cocondenser-ensembledistil). See the [official SPLADE repo](/~https://github.com/naver/splade) and the following paper for more details: > Thibault Formal, Carlos Lassance, Benjamin Piwowarski, and Stéphane Clinchant. [From Distillation to Hard Negative Sampling: Making Sparse Neural IR Models More Effective.](https://dl.acm.org/doi/10.1145/3477495.3531857) _Proceedings of the 45th International ACM SIGIR Conference on Research and Development in Information Retrieval_, pages 2353–2359. -In these experiments, we are using pre-encoded queries (i.e., cached results of query encoding). +In these experiments, we are using cached queries (i.e., cached results of query encoding). The exact configurations for these regressions are stored in [this YAML file](${yaml}). Note that this page is automatically generated from [this template](${template}) as part of Anserini's regression pipeline, so do not modify this page directly; modify the template instead and then run `bin/build.sh` to rebuild the documentation. diff --git a/src/main/resources/docgen/templates/beir-v1.0.0-webis-touche2020.splade-pp-ed.onnx.template b/src/main/resources/docgen/templates/beir-v1.0.0-webis-touche2020.splade-pp-ed.onnx.template index 3cc60687e4..2df4d62745 100644 --- a/src/main/resources/docgen/templates/beir-v1.0.0-webis-touche2020.splade-pp-ed.onnx.template +++ b/src/main/resources/docgen/templates/beir-v1.0.0-webis-touche2020.splade-pp-ed.onnx.template @@ -1,8 +1,8 @@ # Anserini Regressions: BEIR (v1.0.0) — Webis-Touche2020 -**Model**: [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) (using ONNX for on-the-fly query encoding) +**Model**: [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) (using ONNX for on-the-fly query encoding) -This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ (CoCondenser-EnsembleDistil)](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — Webis-Touche2020](http://beir.ai/). +This page describes regression experiments, integrated into Anserini's regression testing framework, using [SPLADE++ CoCondenser-EnsembleDistil](https://arxiv.org/abs/2205.04733) on [BEIR (v1.0.0) — Webis-Touche2020](http://beir.ai/). The model itself can be download [here](https://huggingface.co/naver/splade-cocondenser-ensembledistil). See the [official SPLADE repo](/~https://github.com/naver/splade) and the following paper for more details: diff --git a/src/main/resources/docgen/templates/beir-v1.0.0-webis-touche2020.unicoil-noexp.cached.template b/src/main/resources/docgen/templates/beir-v1.0.0-webis-touche2020.unicoil-noexp.cached.template index 26ed3bfd6a..d4543b2af6 100644 --- a/src/main/resources/docgen/templates/beir-v1.0.0-webis-touche2020.unicoil-noexp.cached.template +++ b/src/main/resources/docgen/templates/beir-v1.0.0-webis-touche2020.unicoil-noexp.cached.template @@ -1,12 +1,14 @@ # Anserini Regressions: BEIR (v1.0.0) — Webis-Touche2020 -**Model**: uniCOIL (without any expansions) +**Model**: uniCOIL without any expansions (using cached queries) This page describes regression experiments, integrated into Anserini's regression testing framework, using uniCOIL (without any expansions) on [BEIR (v1.0.0) — Webis-Touche2020](http://beir.ai/). 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.](https://arxiv.org/abs/2106.14807) _arXiv:2106.14807_. +In these experiments, we are using cached queries (i.e., cached results of query encoding). + The exact configurations for these regressions are stored in [this YAML file](${yaml}). Note that this page is automatically generated from [this template](${template}) as part of Anserini's regression pipeline, so do not modify this page directly; modify the template instead. diff --git a/src/main/resources/docgen/templates/dl19-passage.bge-base-en-v1.5.hnsw-int8.cached.template b/src/main/resources/docgen/templates/dl19-passage.bge-base-en-v1.5.hnsw-int8.cached.template index 41840f5ac6..684d6a68a8 100644 --- a/src/main/resources/docgen/templates/dl19-passage.bge-base-en-v1.5.hnsw-int8.cached.template +++ b/src/main/resources/docgen/templates/dl19-passage.bge-base-en-v1.5.hnsw-int8.cached.template @@ -1,12 +1,12 @@ # Anserini Regressions: TREC 2019 Deep Learning Track (Passage) -**Model**: [BGE-base-en-v1.5](https://huggingface.co/BAAI/bge-base-en-v1.5) with HNSW quantized indexes (using pre-encoded queries) +**Model**: [BGE-base-en-v1.5](https://huggingface.co/BAAI/bge-base-en-v1.5) with quantized HNSW indexes (using cached queries) This page describes regression experiments, integrated into Anserini's regression testing framework, using the [BGE-base-en-v1.5](https://huggingface.co/BAAI/bge-base-en-v1.5) model on the [TREC 2019 Deep Learning Track passage ranking task](https://trec.nist.gov/data/deep2019.html), as described in the following paper: > Shitao Xiao, Zheng Liu, Peitian Zhang, and Niklas Muennighoff. [C-Pack: Packaged Resources To Advance General Chinese Embedding.](https://arxiv.org/abs/2309.07597) _arXiv:2309.07597_, 2023. -In these experiments, we are using pre-encoded queries (i.e., cached results of query encoding). +In these experiments, we are using cached queries (i.e., cached results of query encoding). Note that the NIST relevance judgments provide far more relevant passages per topic, unlike the "sparse" judgments provided by Microsoft (these are sometimes called "dense" judgments to emphasize this contrast). For additional instructions on working with MS MARCO passage collection, refer to [this page](experiments-msmarco-passage.md). @@ -49,7 +49,7 @@ python src/main/python/run_regression.py --index --verify --search --regression ## Indexing -Sample indexing command, building HNSW indexes: +Sample indexing command, building quantized HNSW indexes: ```bash ${index_cmds} diff --git a/src/main/resources/docgen/templates/dl19-passage.bge-base-en-v1.5.hnsw-int8.onnx.template b/src/main/resources/docgen/templates/dl19-passage.bge-base-en-v1.5.hnsw-int8.onnx.template index c30489a87e..07f64b2993 100644 --- a/src/main/resources/docgen/templates/dl19-passage.bge-base-en-v1.5.hnsw-int8.onnx.template +++ b/src/main/resources/docgen/templates/dl19-passage.bge-base-en-v1.5.hnsw-int8.onnx.template @@ -1,6 +1,6 @@ # Anserini Regressions: TREC 2019 Deep Learning Track (Passage) -**Model**: [BGE-base-en-v1.5](https://huggingface.co/BAAI/bge-base-en-v1.5) with HNSW quantized indexes (using ONNX for on-the-fly query encoding) +**Model**: [BGE-base-en-v1.5](https://huggingface.co/BAAI/bge-base-en-v1.5) with quantized HNSW indexes (using ONNX for on-the-fly query encoding) This page describes regression experiments, integrated into Anserini's regression testing framework, using the [BGE-base-en-v1.5](https://huggingface.co/BAAI/bge-base-en-v1.5) model on the [TREC 2019 Deep Learning Track passage ranking task](https://trec.nist.gov/data/deep2019.html), as described in the following paper: @@ -49,7 +49,7 @@ python src/main/python/run_regression.py --index --verify --search --regression ## Indexing -Sample indexing command, building HNSW indexes: +Sample indexing command, building quantized HNSW indexes: ```bash ${index_cmds} diff --git a/src/main/resources/docgen/templates/dl19-passage.bge-base-en-v1.5.hnsw.cached.template b/src/main/resources/docgen/templates/dl19-passage.bge-base-en-v1.5.hnsw.cached.template index 999d5f7cb0..1684a43b43 100644 --- a/src/main/resources/docgen/templates/dl19-passage.bge-base-en-v1.5.hnsw.cached.template +++ b/src/main/resources/docgen/templates/dl19-passage.bge-base-en-v1.5.hnsw.cached.template @@ -1,12 +1,12 @@ # Anserini Regressions: TREC 2019 Deep Learning Track (Passage) -**Model**: [BGE-base-en-v1.5](https://huggingface.co/BAAI/bge-base-en-v1.5) with HNSW indexes (using pre-encoded queries) +**Model**: [BGE-base-en-v1.5](https://huggingface.co/BAAI/bge-base-en-v1.5) with HNSW indexes (using cached queries) This page describes regression experiments, integrated into Anserini's regression testing framework, using the [BGE-base-en-v1.5](https://huggingface.co/BAAI/bge-base-en-v1.5) model on the [TREC 2019 Deep Learning Track passage ranking task](https://trec.nist.gov/data/deep2019.html), as described in the following paper: > Shitao Xiao, Zheng Liu, Peitian Zhang, and Niklas Muennighoff. [C-Pack: Packaged Resources To Advance General Chinese Embedding.](https://arxiv.org/abs/2309.07597) _arXiv:2309.07597_, 2023. -In these experiments, we are using pre-encoded queries (i.e., cached results of query encoding). +In these experiments, we are using cached queries (i.e., cached results of query encoding). Note that the NIST relevance judgments provide far more relevant passages per topic, unlike the "sparse" judgments provided by Microsoft (these are sometimes called "dense" judgments to emphasize this contrast). For additional instructions on working with MS MARCO passage collection, refer to [this page](experiments-msmarco-passage.md). diff --git a/src/main/resources/docgen/templates/dl19-passage.cohere-embed-english-v3.0.hnsw-int8.cached.template b/src/main/resources/docgen/templates/dl19-passage.cohere-embed-english-v3.0.hnsw-int8.cached.template index 37cdf21790..c3b3877000 100644 --- a/src/main/resources/docgen/templates/dl19-passage.cohere-embed-english-v3.0.hnsw-int8.cached.template +++ b/src/main/resources/docgen/templates/dl19-passage.cohere-embed-english-v3.0.hnsw-int8.cached.template @@ -1,10 +1,10 @@ # Anserini Regressions: TREC 2019 Deep Learning Track (Passage) -**Model**: [Cohere embed-english-v3.0](https://docs.cohere.com/reference/embed) with HNSW quantized indexes (using pre-encoded queries) +**Model**: [Cohere embed-english-v3.0](https://docs.cohere.com/reference/embed) with quantized HNSW indexes (using cached queries) This page describes regression experiments, integrated into Anserini's regression testing framework, using the [Cohere embed-english-v3.0](https://docs.cohere.com/reference/embed) model on the [TREC 2019 Deep Learning Track passage ranking task](https://trec.nist.gov/data/deep2019.html). -In these experiments, we are using pre-encoded queries (i.e., cached results of query encoding). +In these experiments, we are using cached queries (i.e., cached results of query encoding). The exact configurations for these regressions are stored in [this YAML file](${yaml}). Note that this page is automatically generated from [this template](${template}) as part of Anserini's regression pipeline, so do not modify this page directly; modify the template instead and then run `bin/build.sh` to rebuild the documentation. @@ -44,7 +44,7 @@ python src/main/python/run_regression.py --index --verify --search --regression ## Indexing -Sample indexing command, building HNSW indexes: +Sample indexing command, building quantized HNSW indexes: ```bash ${index_cmds} diff --git a/src/main/resources/docgen/templates/dl19-passage.cohere-embed-english-v3.0.hnsw.cached.template b/src/main/resources/docgen/templates/dl19-passage.cohere-embed-english-v3.0.hnsw.cached.template index a0e870e653..789f388206 100644 --- a/src/main/resources/docgen/templates/dl19-passage.cohere-embed-english-v3.0.hnsw.cached.template +++ b/src/main/resources/docgen/templates/dl19-passage.cohere-embed-english-v3.0.hnsw.cached.template @@ -1,10 +1,10 @@ # Anserini Regressions: TREC 2019 Deep Learning Track (Passage) -**Model**: [Cohere embed-english-v3.0](https://docs.cohere.com/reference/embed) with HNSW indexes (using pre-encoded queries) +**Model**: [Cohere embed-english-v3.0](https://docs.cohere.com/reference/embed) with HNSW indexes (using cached queries) This page describes regression experiments, integrated into Anserini's regression testing framework, using the [Cohere embed-english-v3.0](https://docs.cohere.com/reference/embed) model on the [TREC 2019 Deep Learning Track passage ranking task](https://trec.nist.gov/data/deep2019.html). -In these experiments, we are using pre-encoded queries (i.e., cached results of query encoding). +In these experiments, we are using cached queries (i.e., cached results of query encoding). The exact configurations for these regressions are stored in [this YAML file](${yaml}). Note that this page is automatically generated from [this template](${template}) as part of Anserini's regression pipeline, so do not modify this page directly; modify the template instead and then run `bin/build.sh` to rebuild the documentation. diff --git a/src/main/resources/docgen/templates/dl19-passage.cos-dpr-distil.fw.template b/src/main/resources/docgen/templates/dl19-passage.cos-dpr-distil.fw.template index ddad733fc3..0da6e3f2cf 100644 --- a/src/main/resources/docgen/templates/dl19-passage.cos-dpr-distil.fw.template +++ b/src/main/resources/docgen/templates/dl19-passage.cos-dpr-distil.fw.template @@ -1,9 +1,9 @@ # Anserini Regressions: TREC 2019 Deep Learning Track (Passage) -**Model**: cosDPR-distil with inverted indexes using the "fake-words" technique (q=40); pre-encoded queries +**Model**: cosDPR-distil with inverted indexes using the "fake-words" technique (q=40) using cached queries This page describes regression experiments, integrated into Anserini's regression testing framework, using the cosDPR-distil model on the [TREC 2019 Deep Learning Track passage ranking task](https://trec.nist.gov/data/deep2019.html). -In these experiments, we are using pre-encoded queries (i.e., cached results of query encoding). +In these experiments, we are using cached queries (i.e., cached results of query encoding). Note that the NIST relevance judgments provide far more relevant passages per topic, unlike the "sparse" judgments provided by Microsoft (these are sometimes called "dense" judgments to emphasize this contrast). For additional instructions on working with MS MARCO passage collection, refer to [this page](experiments-msmarco-passage.md). diff --git a/src/main/resources/docgen/templates/dl19-passage.cos-dpr-distil.hnsw-int8.cached.template b/src/main/resources/docgen/templates/dl19-passage.cos-dpr-distil.hnsw-int8.cached.template index 59557ea246..751379ca58 100644 --- a/src/main/resources/docgen/templates/dl19-passage.cos-dpr-distil.hnsw-int8.cached.template +++ b/src/main/resources/docgen/templates/dl19-passage.cos-dpr-distil.hnsw-int8.cached.template @@ -1,12 +1,12 @@ # Anserini Regressions: TREC 2019 Deep Learning Track (Passage) -**Model**: cosDPR-distil with HNSW quantized indexes (using pre-encoded queries) +**Model**: cosDPR-distil with quantized HNSW indexes (using cached queries) This page describes regression experiments, integrated into Anserini's regression testing framework, using the cosDPR-distil model on the [TREC 2019 Deep Learning Track passage ranking task](https://trec.nist.gov/data/deep2019.html), as described in the following paper: > Xueguang Ma, Tommaso Teofili, and Jimmy Lin. [Anserini Gets Dense Retrieval: Integration of Lucene's HNSW Indexes.](https://dl.acm.org/doi/10.1145/3583780.3615112) _Proceedings of the 32nd International Conference on Information and Knowledge Management (CIKM 2023)_, October 2023, pages 5366–5370, Birmingham, the United Kingdom. -In these experiments, we are using pre-encoded queries (i.e., cached results of query encoding). +In these experiments, we are using cached queries (i.e., cached results of query encoding). Note that the NIST relevance judgments provide far more relevant passages per topic, unlike the "sparse" judgments provided by Microsoft (these are sometimes called "dense" judgments to emphasize this contrast). For additional instructions on working with MS MARCO passage collection, refer to [this page](experiments-msmarco-passage.md). @@ -49,7 +49,7 @@ python src/main/python/run_regression.py --index --verify --search --regression ## Indexing -Sample indexing command, building HNSW indexes: +Sample indexing command, building quantized HNSW indexes: ```bash ${index_cmds} diff --git a/src/main/resources/docgen/templates/dl19-passage.cos-dpr-distil.hnsw-int8.onnx.template b/src/main/resources/docgen/templates/dl19-passage.cos-dpr-distil.hnsw-int8.onnx.template index 0f52251833..e66a46e480 100644 --- a/src/main/resources/docgen/templates/dl19-passage.cos-dpr-distil.hnsw-int8.onnx.template +++ b/src/main/resources/docgen/templates/dl19-passage.cos-dpr-distil.hnsw-int8.onnx.template @@ -1,6 +1,6 @@ # Anserini Regressions: TREC 2019 Deep Learning Track (Passage) -**Model**: cosDPR-distil with HNSW quantized indexes (using ONNX for on-the-fly query encoding) +**Model**: cosDPR-distil with quantized HNSW indexes (using ONNX for on-the-fly query encoding) This page describes regression experiments, integrated into Anserini's regression testing framework, using the cosDPR-distil model on the [TREC 2019 Deep Learning Track passage ranking task](https://trec.nist.gov/data/deep2019.html), as described in the following paper: @@ -49,7 +49,7 @@ python src/main/python/run_regression.py --index --verify --search --regression ## Indexing -Sample indexing command, building HNSW indexes: +Sample indexing command, building quantized HNSW indexes: ```bash ${index_cmds} diff --git a/src/main/resources/docgen/templates/dl19-passage.cos-dpr-distil.hnsw.cached.template b/src/main/resources/docgen/templates/dl19-passage.cos-dpr-distil.hnsw.cached.template index 4929580f97..681c0bcd6d 100644 --- a/src/main/resources/docgen/templates/dl19-passage.cos-dpr-distil.hnsw.cached.template +++ b/src/main/resources/docgen/templates/dl19-passage.cos-dpr-distil.hnsw.cached.template @@ -1,12 +1,12 @@ # Anserini Regressions: TREC 2019 Deep Learning Track (Passage) -**Model**: cosDPR-distil with HNSW indexes (using pre-encoded queries) +**Model**: cosDPR-distil with HNSW indexes (using cached queries) This page describes regression experiments, integrated into Anserini's regression testing framework, using the cosDPR-distil model on the [TREC 2019 Deep Learning Track passage ranking task](https://trec.nist.gov/data/deep2019.html), as described in the following paper: > Xueguang Ma, Tommaso Teofili, and Jimmy Lin. [Anserini Gets Dense Retrieval: Integration of Lucene's HNSW Indexes.](https://dl.acm.org/doi/10.1145/3583780.3615112) _Proceedings of the 32nd International Conference on Information and Knowledge Management (CIKM 2023)_, October 2023, pages 5366–5370, Birmingham, the United Kingdom. -In these experiments, we are using pre-encoded queries (i.e., cached results of query encoding). +In these experiments, we are using cached queries (i.e., cached results of query encoding). Note that the NIST relevance judgments provide far more relevant passages per topic, unlike the "sparse" judgments provided by Microsoft (these are sometimes called "dense" judgments to emphasize this contrast). For additional instructions on working with MS MARCO passage collection, refer to [this page](experiments-msmarco-passage.md). diff --git a/src/main/resources/docgen/templates/dl19-passage.cos-dpr-distil.lexlsh.template b/src/main/resources/docgen/templates/dl19-passage.cos-dpr-distil.lexlsh.template index 2a3876b934..d29774ac9f 100644 --- a/src/main/resources/docgen/templates/dl19-passage.cos-dpr-distil.lexlsh.template +++ b/src/main/resources/docgen/templates/dl19-passage.cos-dpr-distil.lexlsh.template @@ -1,9 +1,9 @@ # Anserini Regressions: TREC 2019 Deep Learning Track (Passage) -**Model**: cosDPR-distil with inverted indexes using the "LexLSH" technique (b=600); pre-encoded queries +**Model**: cosDPR-distil with inverted indexes using the "LexLSH" technique (b=600) using cached queries This page describes regression experiments, integrated into Anserini's regression testing framework, using the cosDPR-distil model on the [TREC 2019 Deep Learning Track passage ranking task](https://trec.nist.gov/data/deep2019.html). -In these experiments, we are using pre-encoded queries (i.e., cached results of query encoding). +In these experiments, we are using cached queries (i.e., cached results of query encoding). Note that the NIST relevance judgments provide far more relevant passages per topic, unlike the "sparse" judgments provided by Microsoft (these are sometimes called "dense" judgments to emphasize this contrast). For additional instructions on working with MS MARCO passage collection, refer to [this page](experiments-msmarco-passage.md). diff --git a/src/main/resources/docgen/templates/dl19-passage.openai-ada2.hnsw-int8.cached.template b/src/main/resources/docgen/templates/dl19-passage.openai-ada2.hnsw-int8.cached.template index fb703a5308..7e71728323 100644 --- a/src/main/resources/docgen/templates/dl19-passage.openai-ada2.hnsw-int8.cached.template +++ b/src/main/resources/docgen/templates/dl19-passage.openai-ada2.hnsw-int8.cached.template @@ -1,12 +1,12 @@ # Anserini Regressions: TREC 2019 Deep Learning Track (Passage) -**Model**: OpenAI-ada2 embeddings (using pre-encoded queries) with HNSW quantized indexes +**Model**: OpenAI-ada2 embeddings with quantized HNSW indexes (using cached queries) This page describes regression experiments, integrated into Anserini's regression testing framework, using OpenAI-ada2 embeddings on the [TREC 2019 Deep Learning Track passage ranking task](https://trec.nist.gov/data/deep2019.html), as described in the following paper: > Jimmy Lin, Ronak Pradeep, Tommaso Teofili, and Jasper Xian. [Vector Search with OpenAI Embeddings: Lucene Is All You Need.](https://arxiv.org/abs/2308.14963) _arXiv:2308.14963_, 2023. -In these experiments, we are using pre-encoded queries (i.e., cached results of query encoding). +In these experiments, we are using cached queries (i.e., cached results of query encoding). Note that the NIST relevance judgments provide far more relevant passages per topic, unlike the "sparse" judgments provided by Microsoft (these are sometimes called "dense" judgments to emphasize this contrast). For additional instructions on working with MS MARCO passage collection, refer to [this page](experiments-msmarco-passage.md). @@ -49,7 +49,7 @@ python src/main/python/run_regression.py --index --verify --search --regression ## Indexing -Sample indexing command, building HNSW indexes: +Sample indexing command, building quantized HNSW indexes: ```bash ${index_cmds} diff --git a/src/main/resources/docgen/templates/dl19-passage.openai-ada2.hnsw.cached.template b/src/main/resources/docgen/templates/dl19-passage.openai-ada2.hnsw.cached.template index ebe92abc1c..0ffa09bb80 100644 --- a/src/main/resources/docgen/templates/dl19-passage.openai-ada2.hnsw.cached.template +++ b/src/main/resources/docgen/templates/dl19-passage.openai-ada2.hnsw.cached.template @@ -1,12 +1,12 @@ # Anserini Regressions: TREC 2019 Deep Learning Track (Passage) -**Model**: OpenAI-ada2 embeddings (using pre-encoded queries) with HNSW indexes +**Model**: OpenAI-ada2 embeddings with HNSW indexes (using cached queries) This page describes regression experiments, integrated into Anserini's regression testing framework, using OpenAI-ada2 embeddings on the [TREC 2019 Deep Learning Track passage ranking task](https://trec.nist.gov/data/deep2019.html), as described in the following paper: > Jimmy Lin, Ronak Pradeep, Tommaso Teofili, and Jasper Xian. [Vector Search with OpenAI Embeddings: Lucene Is All You Need.](https://arxiv.org/abs/2308.14963) _arXiv:2308.14963_, 2023. -In these experiments, we are using pre-encoded queries (i.e., cached results of query encoding). +In these experiments, we are using cached queries (i.e., cached results of query encoding). Note that the NIST relevance judgments provide far more relevant passages per topic, unlike the "sparse" judgments provided by Microsoft (these are sometimes called "dense" judgments to emphasize this contrast). For additional instructions on working with MS MARCO passage collection, refer to [this page](experiments-msmarco-passage.md). diff --git a/src/main/resources/docgen/templates/dl19-passage.splade-pp-ed.cached.template b/src/main/resources/docgen/templates/dl19-passage.splade-pp-ed.cached.template index 1291321753..e46ce6b2be 100644 --- a/src/main/resources/docgen/templates/dl19-passage.splade-pp-ed.cached.template +++ b/src/main/resources/docgen/templates/dl19-passage.splade-pp-ed.cached.template @@ -1,12 +1,12 @@ # Anserini Regressions: TREC 2019 Deep Learning Track (Passage) -**Model**: SPLADE++ CoCondenser-EnsembleDistil (using pre-encoded queries) +**Model**: SPLADE++ CoCondenser-EnsembleDistil (using cached queries) This page describes regression experiments, integrated into Anserini's regression testing framework, using the [SPLADE++ CoCondenser-EnsembleDistil](https://huggingface.co/naver/splade-cocondenser-ensembledistil) model on the [TREC 2019 Deep Learning Track passage ranking task](https://trec.nist.gov/data/deep2019.html), as described in the following paper: > Thibault Formal, Carlos Lassance, Benjamin Piwowarski, and Stéphane Clinchant. [From Distillation to Hard Negative Sampling: Making Sparse Neural IR Models More Effective.](https://dl.acm.org/doi/10.1145/3477495.3531857) _Proceedings of the 45th International ACM SIGIR Conference on Research and Development in Information Retrieval_, pages 2353–2359. -In these experiments, we are using pre-encoded queries (i.e., cached results of query encoding). +In these experiments, we are using cached queries (i.e., cached results of query encoding). Note that the NIST relevance judgments provide far more relevant passages per topic, unlike the "sparse" judgments provided by Microsoft (these are sometimes called "dense" judgments to emphasize this contrast). For additional instructions on working with MS MARCO passage collection, refer to [this page](${root_path}/docs/experiments-msmarco-passage.md). diff --git a/src/main/resources/docgen/templates/dl19-passage.splade-pp-sd.cached.template b/src/main/resources/docgen/templates/dl19-passage.splade-pp-sd.cached.template index 3dceed4228..97d073aa4e 100644 --- a/src/main/resources/docgen/templates/dl19-passage.splade-pp-sd.cached.template +++ b/src/main/resources/docgen/templates/dl19-passage.splade-pp-sd.cached.template @@ -1,12 +1,12 @@ # Anserini Regressions: TREC 2019 Deep Learning Track (Passage) -**Model**: SPLADE++ CoCondenser-SelfDistil (using pre-encoded queries) +**Model**: SPLADE++ CoCondenser-SelfDistil (using cached queries) This page describes regression experiments, integrated into Anserini's regression testing framework, using the [SPLADE++ CoCondenser-SelfDistil](https://huggingface.co/naver/splade-cocondenser-selfdistil) model on the [TREC 2019 Deep Learning Track passage ranking task](https://trec.nist.gov/data/deep2019.html), as described in the following paper: > Thibault Formal, Carlos Lassance, Benjamin Piwowarski, and Stéphane Clinchant. [From Distillation to Hard Negative Sampling: Making Sparse Neural IR Models More Effective.](https://dl.acm.org/doi/10.1145/3477495.3531857) _Proceedings of the 45th International ACM SIGIR Conference on Research and Development in Information Retrieval_, pages 2353–2359. -In these experiments, we are using pre-encoded queries (i.e., cached results of query encoding). +In these experiments, we are using cached queries (i.e., cached results of query encoding). Note that the NIST relevance judgments provide far more relevant passages per topic, unlike the "sparse" judgments provided by Microsoft (these are sometimes called "dense" judgments to emphasize this contrast). For additional instructions on working with MS MARCO passage collection, refer to [this page](${root_path}/docs/experiments-msmarco-passage.md). diff --git a/src/main/resources/docgen/templates/dl19-passage.unicoil-noexp.cached.template b/src/main/resources/docgen/templates/dl19-passage.unicoil-noexp.cached.template index 44838576c6..ddd6ae2704 100644 --- a/src/main/resources/docgen/templates/dl19-passage.unicoil-noexp.cached.template +++ b/src/main/resources/docgen/templates/dl19-passage.unicoil-noexp.cached.template @@ -1,6 +1,6 @@ # Anserini Regressions: TREC 2019 Deep Learning Track (Passage) -**Model**: uniCOIL (without any expansions) +**Model**: uniCOIL without any expansions (using cached queries) This page describes regression experiments, integrated into Anserini's regression testing framework, using uniCOIL (without any expansions) on the [TREC 2019 Deep Learning Track passage ranking task](https://trec.nist.gov/data/deep2019.html).. The uniCOIL model is described in the following paper: @@ -10,6 +10,8 @@ The uniCOIL model is described in the following paper: The experiments on this page are not actually reported in the paper. Here, a variant model without expansion is used. +In these experiments, we are using cached queries (i.e., cached results of query encoding). + Note that the NIST relevance judgments provide far more relevant passages per topic, unlike the "sparse" judgments provided by Microsoft (these are sometimes called "dense" judgments to emphasize this contrast). For additional instructions on working with MS MARCO passage collection, refer to [this page](${root_path}/docs/experiments-msmarco-passage.md). diff --git a/src/main/resources/docgen/templates/dl19-passage.unicoil.cached.template b/src/main/resources/docgen/templates/dl19-passage.unicoil.cached.template index 1d6523b274..443e649836 100644 --- a/src/main/resources/docgen/templates/dl19-passage.unicoil.cached.template +++ b/src/main/resources/docgen/templates/dl19-passage.unicoil.cached.template @@ -1,6 +1,6 @@ # Anserini Regressions: TREC 2019 Deep Learning Track (Passage) -**Model**: uniCOIL (with doc2query-T5 expansions) +**Model**: uniCOIL with doc2query-T5 expansions (using cached queries) This page describes regression experiments, integrated into Anserini's regression testing framework, using uniCOIL (with doc2query-T5 expansions) on the [TREC 2019 Deep Learning Track passage ranking task](https://trec.nist.gov/data/deep2019.html). The uniCOIL model is described in the following paper: @@ -10,6 +10,8 @@ The uniCOIL model is described in the following paper: The experiments on this page are not actually reported in the paper. However, the model is the same. +In these experiments, we are using cached queries (i.e., cached results of query encoding). + Note that the NIST relevance judgments provide far more relevant passages per topic, unlike the "sparse" judgments provided by Microsoft (these are sometimes called "dense" judgments to emphasize this contrast). For additional instructions on working with MS MARCO passage collection, refer to [this page](${root_path}/docs/experiments-msmarco-passage.md). diff --git a/src/main/resources/docgen/templates/dl20-passage.bge-base-en-v1.5.hnsw-int8.cached.template b/src/main/resources/docgen/templates/dl20-passage.bge-base-en-v1.5.hnsw-int8.cached.template index 3e92b869a3..ec86f6078b 100644 --- a/src/main/resources/docgen/templates/dl20-passage.bge-base-en-v1.5.hnsw-int8.cached.template +++ b/src/main/resources/docgen/templates/dl20-passage.bge-base-en-v1.5.hnsw-int8.cached.template @@ -1,12 +1,12 @@ # Anserini Regressions: TREC 2020 Deep Learning Track (Passage) -**Model**: [BGE-base-en-v1.5](https://huggingface.co/BAAI/bge-base-en-v1.5) with HNSW quantized indexes (using pre-encoded queries) +**Model**: [BGE-base-en-v1.5](https://huggingface.co/BAAI/bge-base-en-v1.5) with quantized HNSW indexes (using cached queries) This page describes regression experiments, integrated into Anserini's regression testing framework, using the [BGE-base-en-v1.5](https://huggingface.co/BAAI/bge-base-en-v1.5) model on the [TREC 2020 Deep Learning Track passage ranking task](https://trec.nist.gov/data/deep2019.html), as described in the following paper: > Shitao Xiao, Zheng Liu, Peitian Zhang, and Niklas Muennighoff. [C-Pack: Packaged Resources To Advance General Chinese Embedding.](https://arxiv.org/abs/2309.07597) _arXiv:2309.07597_, 2023. -In these experiments, we are using pre-encoded queries (i.e., cached results of query encoding). +In these experiments, we are using cached queries (i.e., cached results of query encoding). Note that the NIST relevance judgments provide far more relevant passages per topic, unlike the "sparse" judgments provided by Microsoft (these are sometimes called "dense" judgments to emphasize this contrast). For additional instructions on working with MS MARCO passage collection, refer to [this page](experiments-msmarco-passage.md). @@ -49,7 +49,7 @@ python src/main/python/run_regression.py --index --verify --search --regression ## Indexing -Sample indexing command, building HNSW indexes: +Sample indexing command, building quantized HNSW indexes: ```bash ${index_cmds} diff --git a/src/main/resources/docgen/templates/dl20-passage.bge-base-en-v1.5.hnsw-int8.onnx.template b/src/main/resources/docgen/templates/dl20-passage.bge-base-en-v1.5.hnsw-int8.onnx.template index cf974d5160..5500041509 100644 --- a/src/main/resources/docgen/templates/dl20-passage.bge-base-en-v1.5.hnsw-int8.onnx.template +++ b/src/main/resources/docgen/templates/dl20-passage.bge-base-en-v1.5.hnsw-int8.onnx.template @@ -1,6 +1,6 @@ # Anserini Regressions: TREC 2020 Deep Learning Track (Passage) -**Model**: [BGE-base-en-v1.5](https://huggingface.co/BAAI/bge-base-en-v1.5) with HNSW quantized indexes (using ONNX for on-the-fly query encoding) +**Model**: [BGE-base-en-v1.5](https://huggingface.co/BAAI/bge-base-en-v1.5) with quantized HNSW indexes (using ONNX for on-the-fly query encoding) This page describes regression experiments, integrated into Anserini's regression testing framework, using the [BGE-base-en-v1.5](https://huggingface.co/BAAI/bge-base-en-v1.5) model on the [TREC 2020 Deep Learning Track passage ranking task](https://trec.nist.gov/data/deep2019.html), as described in the following paper: @@ -49,7 +49,7 @@ python src/main/python/run_regression.py --index --verify --search --regression ## Indexing -Sample indexing command, building HNSW indexes: +Sample indexing command, building quantized HNSW indexes: ```bash ${index_cmds} diff --git a/src/main/resources/docgen/templates/dl20-passage.bge-base-en-v1.5.hnsw.cached.template b/src/main/resources/docgen/templates/dl20-passage.bge-base-en-v1.5.hnsw.cached.template index 6b2520e432..324a9c2a42 100644 --- a/src/main/resources/docgen/templates/dl20-passage.bge-base-en-v1.5.hnsw.cached.template +++ b/src/main/resources/docgen/templates/dl20-passage.bge-base-en-v1.5.hnsw.cached.template @@ -1,12 +1,12 @@ # Anserini Regressions: TREC 2020 Deep Learning Track (Passage) -**Model**: [BGE-base-en-v1.5](https://huggingface.co/BAAI/bge-base-en-v1.5) with HNSW indexes (using pre-encoded queries) +**Model**: [BGE-base-en-v1.5](https://huggingface.co/BAAI/bge-base-en-v1.5) with HNSW indexes (using cached queries) This page describes regression experiments, integrated into Anserini's regression testing framework, using the [BGE-base-en-v1.5](https://huggingface.co/BAAI/bge-base-en-v1.5) model on the [TREC 2020 Deep Learning Track passage ranking task](https://trec.nist.gov/data/deep2019.html), as described in the following paper: > Shitao Xiao, Zheng Liu, Peitian Zhang, and Niklas Muennighoff. [C-Pack: Packaged Resources To Advance General Chinese Embedding.](https://arxiv.org/abs/2309.07597) _arXiv:2309.07597_, 2023. -In these experiments, we are using pre-encoded queries (i.e., cached results of query encoding). +In these experiments, we are using cached queries (i.e., cached results of query encoding). Note that the NIST relevance judgments provide far more relevant passages per topic, unlike the "sparse" judgments provided by Microsoft (these are sometimes called "dense" judgments to emphasize this contrast). For additional instructions on working with MS MARCO passage collection, refer to [this page](experiments-msmarco-passage.md). diff --git a/src/main/resources/docgen/templates/dl20-passage.cohere-embed-english-v3.0.hnsw-int8.cached.template b/src/main/resources/docgen/templates/dl20-passage.cohere-embed-english-v3.0.hnsw-int8.cached.template index 1290fd9d0a..6afb3a9c6a 100644 --- a/src/main/resources/docgen/templates/dl20-passage.cohere-embed-english-v3.0.hnsw-int8.cached.template +++ b/src/main/resources/docgen/templates/dl20-passage.cohere-embed-english-v3.0.hnsw-int8.cached.template @@ -1,10 +1,10 @@ # Anserini Regressions: TREC 2020 Deep Learning Track (Passage) -**Model**: [Cohere embed-english-v3.0](https://docs.cohere.com/reference/embed) with HNSW quantized indexes (using pre-encoded queries) +**Model**: [Cohere embed-english-v3.0](https://docs.cohere.com/reference/embed) with quantized HNSW indexes (using cached queries) This page describes regression experiments, integrated into Anserini's regression testing framework, using the [Cohere embed-english-v3.0](https://docs.cohere.com/reference/embed) model on the [TREC 2020 Deep Learning Track passage ranking task](https://trec.nist.gov/data/deep2020.html). -In these experiments, we are using pre-encoded queries (i.e., cached results of query encoding). +In these experiments, we are using cached queries (i.e., cached results of query encoding). The exact configurations for these regressions are stored in [this YAML file](${yaml}). Note that this page is automatically generated from [this template](${template}) as part of Anserini's regression pipeline, so do not modify this page directly; modify the template instead and then run `bin/build.sh` to rebuild the documentation. @@ -44,7 +44,7 @@ python src/main/python/run_regression.py --index --verify --search --regression ## Indexing -Sample indexing command, building HNSW indexes: +Sample indexing command, building quantized HNSW indexes: ```bash ${index_cmds} diff --git a/src/main/resources/docgen/templates/dl20-passage.cohere-embed-english-v3.0.hnsw.cached.template b/src/main/resources/docgen/templates/dl20-passage.cohere-embed-english-v3.0.hnsw.cached.template index 3073bd30f4..ac85f4fae7 100644 --- a/src/main/resources/docgen/templates/dl20-passage.cohere-embed-english-v3.0.hnsw.cached.template +++ b/src/main/resources/docgen/templates/dl20-passage.cohere-embed-english-v3.0.hnsw.cached.template @@ -1,10 +1,10 @@ # Anserini Regressions: TREC 2020 Deep Learning Track (Passage) -**Model**: [Cohere embed-english-v3.0](https://docs.cohere.com/reference/embed) with HNSW indexes (using pre-encoded queries) +**Model**: [Cohere embed-english-v3.0](https://docs.cohere.com/reference/embed) with HNSW indexes (using cached queries) This page describes regression experiments, integrated into Anserini's regression testing framework, using the [Cohere embed-english-v3.0](https://docs.cohere.com/reference/embed) model on the [TREC 2020 Deep Learning Track passage ranking task](https://trec.nist.gov/data/deep2020.html). -In these experiments, we are using pre-encoded queries (i.e., cached results of query encoding). +In these experiments, we are using cached queries (i.e., cached results of query encoding). The exact configurations for these regressions are stored in [this YAML file](${yaml}). Note that this page is automatically generated from [this template](${template}) as part of Anserini's regression pipeline, so do not modify this page directly; modify the template instead and then run `bin/build.sh` to rebuild the documentation. diff --git a/src/main/resources/docgen/templates/dl20-passage.cos-dpr-distil.fw.template b/src/main/resources/docgen/templates/dl20-passage.cos-dpr-distil.fw.template index f687b5d7a7..1b493f9a9a 100644 --- a/src/main/resources/docgen/templates/dl20-passage.cos-dpr-distil.fw.template +++ b/src/main/resources/docgen/templates/dl20-passage.cos-dpr-distil.fw.template @@ -1,9 +1,9 @@ # Anserini Regressions: TREC 2020 Deep Learning Track (Passage) -**Model**: cosDPR-distil with inverted indexes using the "fake-words" technique (q=40); pre-encoded queries +**Model**: cosDPR-distil with inverted indexes using the "fake-words" technique (q=40) using cached queries This page describes regression experiments, integrated into Anserini's regression testing framework, using the cosDPR-distil model on the [TREC 2020 Deep Learning Track passage ranking task](https://trec.nist.gov/data/deep2019.html). -In these experiments, we are using pre-encoded queries (i.e., cached results of query encoding). +In these experiments, we are using cached queries (i.e., cached results of query encoding). Note that the NIST relevance judgments provide far more relevant passages per topic, unlike the "sparse" judgments provided by Microsoft (these are sometimes called "dense" judgments to emphasize this contrast). For additional instructions on working with MS MARCO passage collection, refer to [this page](experiments-msmarco-passage.md). diff --git a/src/main/resources/docgen/templates/dl20-passage.cos-dpr-distil.hnsw-int8.cached.template b/src/main/resources/docgen/templates/dl20-passage.cos-dpr-distil.hnsw-int8.cached.template index 7c4dc1ab0a..b9622009f8 100644 --- a/src/main/resources/docgen/templates/dl20-passage.cos-dpr-distil.hnsw-int8.cached.template +++ b/src/main/resources/docgen/templates/dl20-passage.cos-dpr-distil.hnsw-int8.cached.template @@ -1,12 +1,12 @@ # Anserini Regressions: TREC 2020 Deep Learning Track (Passage) -**Model**: cosDPR-distil with HNSW quantized indexes (using pre-encoded queries) +**Model**: cosDPR-distil with quantized HNSW indexes (using cached queries) This page describes regression experiments, integrated into Anserini's regression testing framework, using the cosDPR-distil model on the [TREC 2020 Deep Learning Track passage ranking task](https://trec.nist.gov/data/deep2019.html), as described in the following paper: > Xueguang Ma, Tommaso Teofili, and Jimmy Lin. [Anserini Gets Dense Retrieval: Integration of Lucene's HNSW Indexes.](https://dl.acm.org/doi/10.1145/3583780.3615112) _Proceedings of the 32nd International Conference on Information and Knowledge Management (CIKM 2023)_, October 2023, pages 5366–5370, Birmingham, the United Kingdom. -In these experiments, we are using pre-encoded queries (i.e., cached results of query encoding). +In these experiments, we are using cached queries (i.e., cached results of query encoding). Note that the NIST relevance judgments provide far more relevant passages per topic, unlike the "sparse" judgments provided by Microsoft (these are sometimes called "dense" judgments to emphasize this contrast). For additional instructions on working with MS MARCO passage collection, refer to [this page](experiments-msmarco-passage.md). @@ -49,7 +49,7 @@ python src/main/python/run_regression.py --index --verify --search --regression ## Indexing -Sample indexing command, building HNSW indexes: +Sample indexing command, building quantized HNSW indexes: ```bash ${index_cmds} diff --git a/src/main/resources/docgen/templates/dl20-passage.cos-dpr-distil.hnsw-int8.onnx.template b/src/main/resources/docgen/templates/dl20-passage.cos-dpr-distil.hnsw-int8.onnx.template index 82a6c8fcf9..55826ab0be 100644 --- a/src/main/resources/docgen/templates/dl20-passage.cos-dpr-distil.hnsw-int8.onnx.template +++ b/src/main/resources/docgen/templates/dl20-passage.cos-dpr-distil.hnsw-int8.onnx.template @@ -1,6 +1,6 @@ # Anserini Regressions: TREC 2020 Deep Learning Track (Passage) -**Model**: cosDPR-distil with HNSW quantized indexes (using ONNX for on-the-fly query encoding) +**Model**: cosDPR-distil with quantized HNSW indexes (using ONNX for on-the-fly query encoding) This page describes regression experiments, integrated into Anserini's regression testing framework, using the cosDPR-distil model on the [TREC 2020 Deep Learning Track passage ranking task](https://trec.nist.gov/data/deep2019.html), as described in the following paper: @@ -49,7 +49,7 @@ python src/main/python/run_regression.py --index --verify --search --regression ## Indexing -Sample indexing command, building HNSW indexes: +Sample indexing command, building quantized HNSW indexes: ```bash ${index_cmds} diff --git a/src/main/resources/docgen/templates/dl20-passage.cos-dpr-distil.hnsw.cached.template b/src/main/resources/docgen/templates/dl20-passage.cos-dpr-distil.hnsw.cached.template index 0da92e354a..fafa6ebc2d 100644 --- a/src/main/resources/docgen/templates/dl20-passage.cos-dpr-distil.hnsw.cached.template +++ b/src/main/resources/docgen/templates/dl20-passage.cos-dpr-distil.hnsw.cached.template @@ -1,12 +1,12 @@ # Anserini Regressions: TREC 2020 Deep Learning Track (Passage) -**Model**: cosDPR-distil with HNSW indexes (using pre-encoded queries) +**Model**: cosDPR-distil with HNSW indexes (using cached queries) This page describes regression experiments, integrated into Anserini's regression testing framework, using the cosDPR-distil model on the [TREC 2020 Deep Learning Track passage ranking task](https://trec.nist.gov/data/deep2019.html), as described in the following paper: > Xueguang Ma, Tommaso Teofili, and Jimmy Lin. [Anserini Gets Dense Retrieval: Integration of Lucene's HNSW Indexes.](https://dl.acm.org/doi/10.1145/3583780.3615112) _Proceedings of the 32nd International Conference on Information and Knowledge Management (CIKM 2023)_, October 2023, pages 5366–5370, Birmingham, the United Kingdom. -In these experiments, we are using pre-encoded queries (i.e., cached results of query encoding). +In these experiments, we are using cached queries (i.e., cached results of query encoding). Note that the NIST relevance judgments provide far more relevant passages per topic, unlike the "sparse" judgments provided by Microsoft (these are sometimes called "dense" judgments to emphasize this contrast). For additional instructions on working with MS MARCO passage collection, refer to [this page](experiments-msmarco-passage.md). diff --git a/src/main/resources/docgen/templates/dl20-passage.cos-dpr-distil.lexlsh.template b/src/main/resources/docgen/templates/dl20-passage.cos-dpr-distil.lexlsh.template index a5c06fd8e5..099fa17d28 100644 --- a/src/main/resources/docgen/templates/dl20-passage.cos-dpr-distil.lexlsh.template +++ b/src/main/resources/docgen/templates/dl20-passage.cos-dpr-distil.lexlsh.template @@ -1,9 +1,9 @@ # Anserini Regressions: TREC 2020 Deep Learning Track (Passage) -**Model**: cosDPR-distil with inverted indexes using the "LexLSH" technique (b=600); pre-encoded queries +**Model**: cosDPR-distil with inverted indexes using the "LexLSH" technique (b=600) using cached queries This page describes regression experiments, integrated into Anserini's regression testing framework, using the cosDPR-distil model on the [TREC 2020 Deep Learning Track passage ranking task](https://trec.nist.gov/data/deep2019.html). -In these experiments, we are using pre-encoded queries (i.e., cached results of query encoding). +In these experiments, we are using cached queries (i.e., cached results of query encoding). Note that the NIST relevance judgments provide far more relevant passages per topic, unlike the "sparse" judgments provided by Microsoft (these are sometimes called "dense" judgments to emphasize this contrast). For additional instructions on working with MS MARCO passage collection, refer to [this page](experiments-msmarco-passage.md). diff --git a/src/main/resources/docgen/templates/dl20-passage.openai-ada2.hnsw-int8.cached.template b/src/main/resources/docgen/templates/dl20-passage.openai-ada2.hnsw-int8.cached.template index 352106c366..3633e0aa96 100644 --- a/src/main/resources/docgen/templates/dl20-passage.openai-ada2.hnsw-int8.cached.template +++ b/src/main/resources/docgen/templates/dl20-passage.openai-ada2.hnsw-int8.cached.template @@ -1,12 +1,12 @@ # Anserini Regressions: TREC 2020 Deep Learning Track (Passage) -**Model**: OpenAI-ada2 embeddings (using pre-encoded queries) with HNSW quantized indexes +**Model**: OpenAI-ada2 embeddings with quantized HNSW indexes (using cached queries) This page describes regression experiments, integrated into Anserini's regression testing framework, using OpenAI-ada2 embeddings on the [TREC 2020 Deep Learning Track passage ranking task](https://trec.nist.gov/data/deep2019.html), as described in the following paper: > Jimmy Lin, Ronak Pradeep, Tommaso Teofili, and Jasper Xian. [Vector Search with OpenAI Embeddings: Lucene Is All You Need.](https://arxiv.org/abs/2308.14963) _arXiv:2308.14963_, 2023. -In these experiments, we are using pre-encoded queries (i.e., cached results of query encoding). +In these experiments, we are using cached queries (i.e., cached results of query encoding). Note that the NIST relevance judgments provide far more relevant passages per topic, unlike the "sparse" judgments provided by Microsoft (these are sometimes called "dense" judgments to emphasize this contrast). For additional instructions on working with MS MARCO passage collection, refer to [this page](experiments-msmarco-passage.md). @@ -49,7 +49,7 @@ python src/main/python/run_regression.py --index --verify --search --regression ## Indexing -Sample indexing command, building HNSW indexes: +Sample indexing command, building quantized HNSW indexes: ```bash ${index_cmds} diff --git a/src/main/resources/docgen/templates/dl20-passage.openai-ada2.hnsw.cached.template b/src/main/resources/docgen/templates/dl20-passage.openai-ada2.hnsw.cached.template index 02547dd7e1..c7ee57e2ff 100644 --- a/src/main/resources/docgen/templates/dl20-passage.openai-ada2.hnsw.cached.template +++ b/src/main/resources/docgen/templates/dl20-passage.openai-ada2.hnsw.cached.template @@ -1,12 +1,12 @@ # Anserini Regressions: TREC 2020 Deep Learning Track (Passage) -**Model**: OpenAI-ada2 embeddings (using pre-encoded queries) with HNSW indexes +**Model**: OpenAI-ada2 embeddings with HNSW indexes (using cached queries) This page describes regression experiments, integrated into Anserini's regression testing framework, using OpenAI-ada2 embeddings on the [TREC 2020 Deep Learning Track passage ranking task](https://trec.nist.gov/data/deep2019.html), as described in the following paper: > Jimmy Lin, Ronak Pradeep, Tommaso Teofili, and Jasper Xian. [Vector Search with OpenAI Embeddings: Lucene Is All You Need.](https://arxiv.org/abs/2308.14963) _arXiv:2308.14963_, 2023. -In these experiments, we are using pre-encoded queries (i.e., cached results of query encoding). +In these experiments, we are using cached queries (i.e., cached results of query encoding). Note that the NIST relevance judgments provide far more relevant passages per topic, unlike the "sparse" judgments provided by Microsoft (these are sometimes called "dense" judgments to emphasize this contrast). For additional instructions on working with MS MARCO passage collection, refer to [this page](experiments-msmarco-passage.md). diff --git a/src/main/resources/docgen/templates/dl20-passage.splade-pp-ed.cached.template b/src/main/resources/docgen/templates/dl20-passage.splade-pp-ed.cached.template index 205484a472..92df296b23 100644 --- a/src/main/resources/docgen/templates/dl20-passage.splade-pp-ed.cached.template +++ b/src/main/resources/docgen/templates/dl20-passage.splade-pp-ed.cached.template @@ -1,12 +1,12 @@ # Anserini Regressions: TREC 2020 Deep Learning Track (Passage) -**Model**: SPLADE++ CoCondenser-EnsembleDistil (using pre-encoded queries) +**Model**: SPLADE++ CoCondenser-EnsembleDistil (using cached queries) This page describes regression experiments, integrated into Anserini's regression testing framework, using the [SPLADE++ CoCondenser-EnsembleDistil](https://huggingface.co/naver/splade-cocondenser-ensembledistil) model on the [TREC 2020 Deep Learning Track passage ranking task](https://trec.nist.gov/data/deep2019.html), as described in the following paper: > Thibault Formal, Carlos Lassance, Benjamin Piwowarski, and Stéphane Clinchant. [From Distillation to Hard Negative Sampling: Making Sparse Neural IR Models More Effective.](https://dl.acm.org/doi/10.1145/3477495.3531857) _Proceedings of the 45th International ACM SIGIR Conference on Research and Development in Information Retrieval_, pages 2353–2359. -In these experiments, we are using pre-encoded queries (i.e., cached results of query encoding). +In these experiments, we are using cached queries (i.e., cached results of query encoding). Note that the NIST relevance judgments provide far more relevant passages per topic, unlike the "sparse" judgments provided by Microsoft (these are sometimes called "dense" judgments to emphasize this contrast). For additional instructions on working with MS MARCO passage collection, refer to [this page](${root_path}/docs/experiments-msmarco-passage.md). diff --git a/src/main/resources/docgen/templates/dl20-passage.splade-pp-sd.cached.template b/src/main/resources/docgen/templates/dl20-passage.splade-pp-sd.cached.template index abe581df33..8286c4c6f0 100644 --- a/src/main/resources/docgen/templates/dl20-passage.splade-pp-sd.cached.template +++ b/src/main/resources/docgen/templates/dl20-passage.splade-pp-sd.cached.template @@ -1,12 +1,12 @@ # Anserini Regressions: TREC 2020 Deep Learning Track (Passage) -**Model**: SPLADE++ CoCondenser-SelfDistil (using pre-encoded queries) +**Model**: SPLADE++ CoCondenser-SelfDistil (using cached queries) This page describes regression experiments, integrated into Anserini's regression testing framework, using the [SPLADE++ CoCondenser-SelfDistil](https://huggingface.co/naver/splade-cocondenser-selfdistil) model on the [TREC 2020 Deep Learning Track passage ranking task](https://trec.nist.gov/data/deep2019.html), as described in the following paper: > Thibault Formal, Carlos Lassance, Benjamin Piwowarski, and Stéphane Clinchant. [From Distillation to Hard Negative Sampling: Making Sparse Neural IR Models More Effective.](https://dl.acm.org/doi/10.1145/3477495.3531857) _Proceedings of the 45th International ACM SIGIR Conference on Research and Development in Information Retrieval_, pages 2353–2359. -In these experiments, we are using pre-encoded queries (i.e., cached results of query encoding). +In these experiments, we are using cached queries (i.e., cached results of query encoding). Note that the NIST relevance judgments provide far more relevant passages per topic, unlike the "sparse" judgments provided by Microsoft (these are sometimes called "dense" judgments to emphasize this contrast). For additional instructions on working with MS MARCO passage collection, refer to [this page](${root_path}/docs/experiments-msmarco-passage.md). diff --git a/src/main/resources/docgen/templates/dl20-passage.unicoil-noexp.cached.template b/src/main/resources/docgen/templates/dl20-passage.unicoil-noexp.cached.template index 6a5ae3a303..60b2acd90f 100644 --- a/src/main/resources/docgen/templates/dl20-passage.unicoil-noexp.cached.template +++ b/src/main/resources/docgen/templates/dl20-passage.unicoil-noexp.cached.template @@ -1,6 +1,6 @@ # Anserini Regressions: TREC 2020 Deep Learning Track (Passage) -**Model**: uniCOIL (without any expansions) +**Model**: uniCOIL without any expansions (using cached queries) This page describes regression experiments, integrated into Anserini's regression testing framework, using uniCOIL (without any expansions) on the [TREC 2020 Deep Learning Track passage ranking task](https://trec.nist.gov/data/deep2020.html). The uniCOIL model is described in the following paper: @@ -10,6 +10,8 @@ The uniCOIL model is described in the following paper: The experiments on this page are not actually reported in the paper. Here, a variant model without expansion is used. +In these experiments, we are using cached queries (i.e., cached results of query encoding). + Note that the NIST relevance judgments provide far more relevant passages per topic, unlike the "sparse" judgments provided by Microsoft (these are sometimes called "dense" judgments to emphasize this contrast). For additional instructions on working with MS MARCO passage collection, refer to [this page](${root_path}/docs/experiments-msmarco-passage.md). diff --git a/src/main/resources/docgen/templates/dl20-passage.unicoil.cached.template b/src/main/resources/docgen/templates/dl20-passage.unicoil.cached.template index 9c4b5f73f6..7f22233f35 100644 --- a/src/main/resources/docgen/templates/dl20-passage.unicoil.cached.template +++ b/src/main/resources/docgen/templates/dl20-passage.unicoil.cached.template @@ -1,6 +1,6 @@ # Anserini Regressions: TREC 2020 Deep Learning Track (Passage) -**Model**: uniCOIL (with doc2query-T5 expansions) +**Model**: uniCOIL with doc2query-T5 expansions (using cached queries) This page describes regression experiments, integrated into Anserini's regression testing framework, using uniCOIL (with doc2query-T5 expansions) on the [TREC 2020 Deep Learning Track passage ranking task](https://trec.nist.gov/data/deep2020.html). The uniCOIL model is described in the following paper: @@ -10,6 +10,8 @@ The uniCOIL model is described in the following paper: The experiments on this page are not actually reported in the paper. However, the model is the same. +In these experiments, we are using cached queries (i.e., cached results of query encoding). + Note that the NIST relevance judgments provide far more relevant passages per topic, unlike the "sparse" judgments provided by Microsoft (these are sometimes called "dense" judgments to emphasize this contrast). For additional instructions on working with MS MARCO passage collection, refer to [this page](${root_path}/docs/experiments-msmarco-passage.md). diff --git a/src/main/resources/docgen/templates/dl21-passage.splade-pp-ed.cached.template b/src/main/resources/docgen/templates/dl21-passage.splade-pp-ed.cached.template index 2355f6a937..a8a99dc233 100644 --- a/src/main/resources/docgen/templates/dl21-passage.splade-pp-ed.cached.template +++ b/src/main/resources/docgen/templates/dl21-passage.splade-pp-ed.cached.template @@ -1,9 +1,9 @@ # Anserini Regressions: TREC 2021 Deep Learning Track (Passage) -**Model**: SPLADE++ CoCondenser-EnsembleDistil (using pre-encoded queries) +**Model**: SPLADE++ CoCondenser-EnsembleDistil (using cached queries) This page describes regression experiments, integrated into Anserini's regression testing framework, applying the [SPLADE++ CoCondenser-EnsembleDistil](https://huggingface.co/naver/splade-cocondenser-ensembledistil) model to the MS MARCO V2 passage corpus. -Here, we evaluate on the [TREC 2021 Deep Learning Track passage ranking task](https://trec.nist.gov/data/deep2021.html), using pre-encoded queries (i.e., cached results of query encoding). +Here, we evaluate on the [TREC 2021 Deep Learning Track passage ranking task](https://trec.nist.gov/data/deep2021.html), using cached queries (i.e., cached results of query encoding). The model is described in the following paper: diff --git a/src/main/resources/docgen/templates/dl21-passage.splade-pp-sd.cached.template b/src/main/resources/docgen/templates/dl21-passage.splade-pp-sd.cached.template index 6340c37d3c..ef0779e63b 100644 --- a/src/main/resources/docgen/templates/dl21-passage.splade-pp-sd.cached.template +++ b/src/main/resources/docgen/templates/dl21-passage.splade-pp-sd.cached.template @@ -1,9 +1,9 @@ # Anserini Regressions: TREC 2021 Deep Learning Track (Passage) -**Model**: SPLADE++ CoCondenser-SelfDistil (using pre-encoded queries) +**Model**: SPLADE++ CoCondenser-SelfDistil (using cached queries) This page describes regression experiments, integrated into Anserini's regression testing framework, applying the [SPLADE++ CoCondenser-SelfDistil](https://huggingface.co/naver/splade-cocondenser-selfdistil) model to the MS MARCO V2 passage corpus. -Here, we evaluate on the [TREC 2021 Deep Learning Track passage ranking task](https://trec.nist.gov/data/deep2021.html), using pre-encoded queries (i.e., cached results of query encoding). +Here, we evaluate on the [TREC 2021 Deep Learning Track passage ranking task](https://trec.nist.gov/data/deep2021.html), using cached queries (i.e., cached results of query encoding). The model is described in the following paper: diff --git a/src/main/resources/docgen/templates/dl21-passage.unicoil-0shot.cached.template b/src/main/resources/docgen/templates/dl21-passage.unicoil-0shot.cached.template index 937683eca3..9cd6328753 100644 --- a/src/main/resources/docgen/templates/dl21-passage.unicoil-0shot.cached.template +++ b/src/main/resources/docgen/templates/dl21-passage.unicoil-0shot.cached.template @@ -95,4 +95,4 @@ With the above commands, you should be able to reproduce the following results: ${effectiveness} This run roughly corresponds to run `d_unicoil0` submitted to the TREC 2021 Deep Learning Track under the "baseline" group. -The difference is that here we are using pre-encoded queries, whereas the official submission performed query encoding on the fly. +The difference is that here we are using cached queries (i.e., cached results of query encoding), whereas the official submission performed query encoding on the fly. diff --git a/src/main/resources/docgen/templates/dl21-passage.unicoil-noexp-0shot.cached.template b/src/main/resources/docgen/templates/dl21-passage.unicoil-noexp-0shot.cached.template index b643e2f0be..4e83eb36a4 100644 --- a/src/main/resources/docgen/templates/dl21-passage.unicoil-noexp-0shot.cached.template +++ b/src/main/resources/docgen/templates/dl21-passage.unicoil-noexp-0shot.cached.template @@ -95,4 +95,4 @@ With the above commands, you should be able to reproduce the following results: ${effectiveness} This run roughly corresponds to run `d_unicoil0` submitted to the TREC 2021 Deep Learning Track under the "baseline" group. -The difference is that here we are using pre-encoded queries, whereas the official submission performed query encoding on the fly. +The difference is that here we are using cached queries (i.e., cached results of query encoding), whereas the official submission performed query encoding on the fly. diff --git a/src/main/resources/docgen/templates/dl22-passage.splade-pp-ed.cached.template b/src/main/resources/docgen/templates/dl22-passage.splade-pp-ed.cached.template index 3eefc52170..1e3be6daf7 100644 --- a/src/main/resources/docgen/templates/dl22-passage.splade-pp-ed.cached.template +++ b/src/main/resources/docgen/templates/dl22-passage.splade-pp-ed.cached.template @@ -1,9 +1,9 @@ # Anserini Regressions: TREC 2022 Deep Learning Track (Passage) -**Model**: SPLADE++ CoCondenser-EnsembleDistil (using pre-encoded queries) +**Model**: SPLADE++ CoCondenser-EnsembleDistil (using cached queries) This page describes regression experiments, integrated into Anserini's regression testing framework, applying the [SPLADE++ CoCondenser-EnsembleDistil](https://huggingface.co/naver/splade-cocondenser-ensembledistil) model to the MS MARCO V2 passage corpus. -Here, we evaluate on the [TREC 2022 Deep Learning Track passage ranking task](https://trec.nist.gov/data/deep2022.html), using pre-encoded queries (i.e., cached results of query encoding). +Here, we evaluate on the [TREC 2022 Deep Learning Track passage ranking task](https://trec.nist.gov/data/deep2022.html), using cached queries (i.e., cached results of query encoding). The model is described in the following paper: diff --git a/src/main/resources/docgen/templates/dl22-passage.splade-pp-sd.cached.template b/src/main/resources/docgen/templates/dl22-passage.splade-pp-sd.cached.template index 8b8863b599..7987116174 100644 --- a/src/main/resources/docgen/templates/dl22-passage.splade-pp-sd.cached.template +++ b/src/main/resources/docgen/templates/dl22-passage.splade-pp-sd.cached.template @@ -1,9 +1,9 @@ # Anserini Regressions: TREC 2022 Deep Learning Track (Passage) -**Model**: SPLADE++ CoCondenser-SelfDistil (using pre-encoded queries) +**Model**: SPLADE++ CoCondenser-SelfDistil (using cached queries) This page describes regression experiments, integrated into Anserini's regression testing framework, applying the [SPLADE++ CoCondenser-SelfDistil](https://huggingface.co/naver/splade-cocondenser-selfdistil) model to the MS MARCO V2 passage corpus. -Here, we evaluate on the [TREC 2022 Deep Learning Track passage ranking task](https://trec.nist.gov/data/deep2022.html), using pre-encoded queries (i.e., cached results of query encoding). +Here, we evaluate on the [TREC 2022 Deep Learning Track passage ranking task](https://trec.nist.gov/data/deep2022.html), using cached queries (i.e., cached results of query encoding). The model is described in the following paper: diff --git a/src/main/resources/docgen/templates/dl23-passage.splade-pp-ed.cached.template b/src/main/resources/docgen/templates/dl23-passage.splade-pp-ed.cached.template index 3598f822f6..f7c0e34f7e 100644 --- a/src/main/resources/docgen/templates/dl23-passage.splade-pp-ed.cached.template +++ b/src/main/resources/docgen/templates/dl23-passage.splade-pp-ed.cached.template @@ -1,9 +1,9 @@ # Anserini Regressions: TREC 2023 Deep Learning Track (Passage) -**Model**: SPLADE++ CoCondenser-EnsembleDistil (using pre-encoded queries) +**Model**: SPLADE++ CoCondenser-EnsembleDistil (using cached queries) This page describes regression experiments, integrated into Anserini's regression testing framework, applying the [SPLADE++ CoCondenser-EnsembleDistil](https://huggingface.co/naver/splade-cocondenser-ensembledistil) model to the MS MARCO V2 passage corpus. -Here, we evaluate on the [TREC 2023 Deep Learning Track passage ranking task](https://trec.nist.gov/data/deep2023.html), using pre-encoded queries (i.e., cached results of query encoding). +Here, we evaluate on the [TREC 2023 Deep Learning Track passage ranking task](https://trec.nist.gov/data/deep2023.html), using cached queries (i.e., cached results of query encoding). The model is described in the following paper: diff --git a/src/main/resources/docgen/templates/dl23-passage.splade-pp-sd.cached.template b/src/main/resources/docgen/templates/dl23-passage.splade-pp-sd.cached.template index b9eb1ab1cb..e88f2c2b60 100644 --- a/src/main/resources/docgen/templates/dl23-passage.splade-pp-sd.cached.template +++ b/src/main/resources/docgen/templates/dl23-passage.splade-pp-sd.cached.template @@ -1,9 +1,9 @@ # Anserini Regressions: TREC 2023 Deep Learning Track (Passage) -**Model**: SPLADE++ CoCondenser-SelfDistil (using pre-encoded queries) +**Model**: SPLADE++ CoCondenser-SelfDistil (using cached queries) This page describes regression experiments, integrated into Anserini's regression testing framework, applying the [SPLADE++ CoCondenser-SelfDistil](https://huggingface.co/naver/splade-cocondenser-selfdistil) model to the MS MARCO V2 passage corpus. -Here, we evaluate on the [TREC 2023 Deep Learning Track passage ranking task](https://trec.nist.gov/data/deep2023.html), using pre-encoded queries (i.e., cached results of query encoding). +Here, we evaluate on the [TREC 2023 Deep Learning Track passage ranking task](https://trec.nist.gov/data/deep2023.html), using cached queries (i.e., cached results of query encoding). The model is described in the following paper: diff --git a/src/main/resources/docgen/templates/msmarco-v1-passage.bge-base-en-v1.5.hnsw-int8.cached.template b/src/main/resources/docgen/templates/msmarco-v1-passage.bge-base-en-v1.5.hnsw-int8.cached.template index 59436c7610..af69251204 100644 --- a/src/main/resources/docgen/templates/msmarco-v1-passage.bge-base-en-v1.5.hnsw-int8.cached.template +++ b/src/main/resources/docgen/templates/msmarco-v1-passage.bge-base-en-v1.5.hnsw-int8.cached.template @@ -1,12 +1,12 @@ # Anserini Regressions: MS MARCO Passage Ranking -**Model**: [BGE-base-en-v1.5](https://huggingface.co/BAAI/bge-base-en-v1.5) with HNSW quantized indexes (using pre-encoded queries) +**Model**: [BGE-base-en-v1.5](https://huggingface.co/BAAI/bge-base-en-v1.5) with quantized HNSW indexes (using cached queries) This page describes regression experiments, integrated into Anserini's regression testing framework, using the [BGE-base-en-v1.5](https://huggingface.co/BAAI/bge-base-en-v1.5) model on the [MS MARCO passage ranking task](/~https://github.com/microsoft/MSMARCO-Passage-Ranking), as described in the following paper: > Shitao Xiao, Zheng Liu, Peitian Zhang, and Niklas Muennighoff. [C-Pack: Packaged Resources To Advance General Chinese Embedding.](https://arxiv.org/abs/2309.07597) _arXiv:2309.07597_, 2023. -In these experiments, we are using pre-encoded queries (i.e., cached results of query encoding). +In these experiments, we are using cached queries (i.e., cached results of query encoding). The exact configurations for these regressions are stored in [this YAML file](${yaml}). Note that this page is automatically generated from [this template](${template}) as part of Anserini's regression pipeline, so do not modify this page directly; modify the template instead and then run `bin/build.sh` to rebuild the documentation. @@ -46,7 +46,7 @@ python src/main/python/run_regression.py --index --verify --search --regression ## Indexing -Sample indexing command, building HNSW indexes: +Sample indexing command, building quantized HNSW indexes: ```bash ${index_cmds} diff --git a/src/main/resources/docgen/templates/msmarco-v1-passage.bge-base-en-v1.5.hnsw-int8.onnx.template b/src/main/resources/docgen/templates/msmarco-v1-passage.bge-base-en-v1.5.hnsw-int8.onnx.template index 83df36bfc5..94075818a8 100644 --- a/src/main/resources/docgen/templates/msmarco-v1-passage.bge-base-en-v1.5.hnsw-int8.onnx.template +++ b/src/main/resources/docgen/templates/msmarco-v1-passage.bge-base-en-v1.5.hnsw-int8.onnx.template @@ -1,6 +1,6 @@ # Anserini Regressions: MS MARCO Passage Ranking -**Model**: [BGE-base-en-v1.5](https://huggingface.co/BAAI/bge-base-en-v1.5) with HNSW quantized indexes (using ONNX for on-the-fly query encoding) +**Model**: [BGE-base-en-v1.5](https://huggingface.co/BAAI/bge-base-en-v1.5) with quantized HNSW indexes (using ONNX for on-the-fly query encoding) This page describes regression experiments, integrated into Anserini's regression testing framework, using the [BGE-base-en-v1.5](https://huggingface.co/BAAI/bge-base-en-v1.5) model on the [MS MARCO passage ranking task](/~https://github.com/microsoft/MSMARCO-Passage-Ranking), as described in the following paper: @@ -46,7 +46,7 @@ python src/main/python/run_regression.py --index --verify --search --regression ## Indexing -Sample indexing command, building HNSW indexes: +Sample indexing command, building quantized HNSW indexes: ```bash ${index_cmds} diff --git a/src/main/resources/docgen/templates/msmarco-v1-passage.bge-base-en-v1.5.hnsw.cached.template b/src/main/resources/docgen/templates/msmarco-v1-passage.bge-base-en-v1.5.hnsw.cached.template index d481fc5c68..e75f7be10e 100644 --- a/src/main/resources/docgen/templates/msmarco-v1-passage.bge-base-en-v1.5.hnsw.cached.template +++ b/src/main/resources/docgen/templates/msmarco-v1-passage.bge-base-en-v1.5.hnsw.cached.template @@ -1,12 +1,12 @@ # Anserini Regressions: MS MARCO Passage Ranking -**Model**: [BGE-base-en-v1.5](https://huggingface.co/BAAI/bge-base-en-v1.5) with HNSW indexes (using pre-encoded queries) +**Model**: [BGE-base-en-v1.5](https://huggingface.co/BAAI/bge-base-en-v1.5) with HNSW indexes (using cached queries) This page describes regression experiments, integrated into Anserini's regression testing framework, using the [BGE-base-en-v1.5](https://huggingface.co/BAAI/bge-base-en-v1.5) model on the [MS MARCO passage ranking task](/~https://github.com/microsoft/MSMARCO-Passage-Ranking), as described in the following paper: > Shitao Xiao, Zheng Liu, Peitian Zhang, and Niklas Muennighoff. [C-Pack: Packaged Resources To Advance General Chinese Embedding.](https://arxiv.org/abs/2309.07597) _arXiv:2309.07597_, 2023. -In these experiments, we are using pre-encoded queries (i.e., cached results of query encoding). +In these experiments, we are using cached queries (i.e., cached results of query encoding). The exact configurations for these regressions are stored in [this YAML file](${yaml}). Note that this page is automatically generated from [this template](${template}) as part of Anserini's regression pipeline, so do not modify this page directly; modify the template instead and then run `bin/build.sh` to rebuild the documentation. diff --git a/src/main/resources/docgen/templates/msmarco-v1-passage.cohere-embed-english-v3.0.hnsw-int8.cached.template b/src/main/resources/docgen/templates/msmarco-v1-passage.cohere-embed-english-v3.0.hnsw-int8.cached.template index 1ff40fa029..c809996e13 100644 --- a/src/main/resources/docgen/templates/msmarco-v1-passage.cohere-embed-english-v3.0.hnsw-int8.cached.template +++ b/src/main/resources/docgen/templates/msmarco-v1-passage.cohere-embed-english-v3.0.hnsw-int8.cached.template @@ -1,10 +1,10 @@ # Anserini Regressions: MS MARCO Passage Ranking -**Model**: [Cohere embed-english-v3.0](https://docs.cohere.com/reference/embed) with HNSW quantized indexes (using pre-encoded queries) +**Model**: [Cohere embed-english-v3.0](https://docs.cohere.com/reference/embed) with quantized HNSW indexes (using cached queries) This page describes regression experiments, integrated into Anserini's regression testing framework, using the [Cohere embed-english-v3.0](https://docs.cohere.com/reference/embed) model on the [MS MARCO passage ranking task](/~https://github.com/microsoft/MSMARCO-Passage-Ranking). -In these experiments, we are using pre-encoded queries (i.e., cached results of query encoding). +In these experiments, we are using cached queries (i.e., cached results of query encoding). The exact configurations for these regressions are stored in [this YAML file](${yaml}). Note that this page is automatically generated from [this template](${template}) as part of Anserini's regression pipeline, so do not modify this page directly; modify the template instead and then run `bin/build.sh` to rebuild the documentation. @@ -44,7 +44,7 @@ python src/main/python/run_regression.py --index --verify --search --regression ## Indexing -Sample indexing command, building HNSW indexes: +Sample indexing command, building quantized HNSW indexes: ```bash ${index_cmds} diff --git a/src/main/resources/docgen/templates/msmarco-v1-passage.cohere-embed-english-v3.0.hnsw.cached.template b/src/main/resources/docgen/templates/msmarco-v1-passage.cohere-embed-english-v3.0.hnsw.cached.template index 7370c4fc7b..f159ced864 100644 --- a/src/main/resources/docgen/templates/msmarco-v1-passage.cohere-embed-english-v3.0.hnsw.cached.template +++ b/src/main/resources/docgen/templates/msmarco-v1-passage.cohere-embed-english-v3.0.hnsw.cached.template @@ -1,10 +1,10 @@ # Anserini Regressions: MS MARCO Passage Ranking -**Model**: [Cohere embed-english-v3.0](https://docs.cohere.com/reference/embed) with HNSW indexes (using pre-encoded queries) +**Model**: [Cohere embed-english-v3.0](https://docs.cohere.com/reference/embed) with HNSW indexes (using cached queries) This page describes regression experiments, integrated into Anserini's regression testing framework, using the [Cohere embed-english-v3.0](https://docs.cohere.com/reference/embed) model on the [MS MARCO passage ranking task](/~https://github.com/microsoft/MSMARCO-Passage-Ranking). -In these experiments, we are using pre-encoded queries (i.e., cached results of query encoding). +In these experiments, we are using cached queries (i.e., cached results of query encoding). The exact configurations for these regressions are stored in [this YAML file](${yaml}). Note that this page is automatically generated from [this template](${template}) as part of Anserini's regression pipeline, so do not modify this page directly; modify the template instead and then run `bin/build.sh` to rebuild the documentation. diff --git a/src/main/resources/docgen/templates/msmarco-v1-passage.cos-dpr-distil.fw.template b/src/main/resources/docgen/templates/msmarco-v1-passage.cos-dpr-distil.fw.template index f52efc57ce..314f13194f 100644 --- a/src/main/resources/docgen/templates/msmarco-v1-passage.cos-dpr-distil.fw.template +++ b/src/main/resources/docgen/templates/msmarco-v1-passage.cos-dpr-distil.fw.template @@ -1,9 +1,9 @@ # Anserini Regressions: MS MARCO Passage Ranking -**Model**: cosDPR-distil with inverted indexes using the "fake-words" technique (q=40); pre-encoded queries +**Model**: cosDPR-distil with inverted indexes using the "fake-words" technique (q=40) using cached queries This page describes regression experiments, integrated into Anserini's regression testing framework, using the cosDPR-distil model on the [MS MARCO passage ranking task](/~https://github.com/microsoft/MSMARCO-Passage-Ranking). -In these experiments, we are using pre-encoded queries (i.e., cached results of query encoding). +In these experiments, we are using cached queries (i.e., cached results of query encoding). The exact configurations for these regressions are stored in [this YAML file](${yaml}). Note that this page is automatically generated from [this template](${template}) as part of Anserini's regression pipeline, so do not modify this page directly; modify the template instead and then run `bin/build.sh` to rebuild the documentation. diff --git a/src/main/resources/docgen/templates/msmarco-v1-passage.cos-dpr-distil.hnsw-int8.cached.template b/src/main/resources/docgen/templates/msmarco-v1-passage.cos-dpr-distil.hnsw-int8.cached.template index 8489b89171..5f85acc05b 100644 --- a/src/main/resources/docgen/templates/msmarco-v1-passage.cos-dpr-distil.hnsw-int8.cached.template +++ b/src/main/resources/docgen/templates/msmarco-v1-passage.cos-dpr-distil.hnsw-int8.cached.template @@ -1,12 +1,12 @@ # Anserini Regressions: MS MARCO Passage Ranking -**Model**: cosDPR-distil with HNSW quantized indexes (using pre-encoded queries) +**Model**: cosDPR-distil with quantized HNSW indexes (using cached queries) This page describes regression experiments, integrated into Anserini's regression testing framework, using the cosDPR-distil model on the [MS MARCO passage ranking task](/~https://github.com/microsoft/MSMARCO-Passage-Ranking), as described in the following paper: > Xueguang Ma, Tommaso Teofili, and Jimmy Lin. [Anserini Gets Dense Retrieval: Integration of Lucene's HNSW Indexes.](https://dl.acm.org/doi/10.1145/3583780.3615112) _Proceedings of the 32nd International Conference on Information and Knowledge Management (CIKM 2023)_, October 2023, pages 5366–5370, Birmingham, the United Kingdom. -In these experiments, we are using pre-encoded queries (i.e., cached results of query encoding). +In these experiments, we are using cached queries (i.e., cached results of query encoding). The exact configurations for these regressions are stored in [this YAML file](${yaml}). Note that this page is automatically generated from [this template](${template}) as part of Anserini's regression pipeline, so do not modify this page directly; modify the template instead and then run `bin/build.sh` to rebuild the documentation. @@ -46,7 +46,7 @@ python src/main/python/run_regression.py --index --verify --search --regression ## Indexing -Sample indexing command, building HNSW indexes: +Sample indexing command, building quantized HNSW indexes: ```bash ${index_cmds} diff --git a/src/main/resources/docgen/templates/msmarco-v1-passage.cos-dpr-distil.hnsw-int8.onnx.template b/src/main/resources/docgen/templates/msmarco-v1-passage.cos-dpr-distil.hnsw-int8.onnx.template index de54bd521e..aa88ab7d67 100644 --- a/src/main/resources/docgen/templates/msmarco-v1-passage.cos-dpr-distil.hnsw-int8.onnx.template +++ b/src/main/resources/docgen/templates/msmarco-v1-passage.cos-dpr-distil.hnsw-int8.onnx.template @@ -1,6 +1,6 @@ # Anserini Regressions: MS MARCO Passage Ranking -**Model**: cosDPR-distil with HNSW quantized indexes (using ONNX for on-the-fly query encoding) +**Model**: cosDPR-distil with quantized HNSW indexes (using ONNX for on-the-fly query encoding) This page describes regression experiments, integrated into Anserini's regression testing framework, using the cosDPR-distil model on the [MS MARCO passage ranking task](/~https://github.com/microsoft/MSMARCO-Passage-Ranking), as described in the following paper: @@ -46,7 +46,7 @@ python src/main/python/run_regression.py --index --verify --search --regression ## Indexing -Sample indexing command, building HNSW indexes: +Sample indexing command, building quantized HNSW indexes: ```bash ${index_cmds} diff --git a/src/main/resources/docgen/templates/msmarco-v1-passage.cos-dpr-distil.hnsw.cached.template b/src/main/resources/docgen/templates/msmarco-v1-passage.cos-dpr-distil.hnsw.cached.template index f5d7400267..affa039161 100644 --- a/src/main/resources/docgen/templates/msmarco-v1-passage.cos-dpr-distil.hnsw.cached.template +++ b/src/main/resources/docgen/templates/msmarco-v1-passage.cos-dpr-distil.hnsw.cached.template @@ -1,12 +1,12 @@ # Anserini Regressions: MS MARCO Passage Ranking -**Model**: cosDPR-distil with HNSW indexes (using pre-encoded queries) +**Model**: cosDPR-distil with HNSW indexes (using cached queries) This page describes regression experiments, integrated into Anserini's regression testing framework, using the cosDPR-distil model on the [MS MARCO passage ranking task](/~https://github.com/microsoft/MSMARCO-Passage-Ranking), as described in the following paper: > Xueguang Ma, Tommaso Teofili, and Jimmy Lin. [Anserini Gets Dense Retrieval: Integration of Lucene's HNSW Indexes.](https://dl.acm.org/doi/10.1145/3583780.3615112) _Proceedings of the 32nd International Conference on Information and Knowledge Management (CIKM 2023)_, October 2023, pages 5366–5370, Birmingham, the United Kingdom. -In these experiments, we are using pre-encoded queries (i.e., cached results of query encoding). +In these experiments, we are using cached queries (i.e., cached results of query encoding). The exact configurations for these regressions are stored in [this YAML file](${yaml}). Note that this page is automatically generated from [this template](${template}) as part of Anserini's regression pipeline, so do not modify this page directly; modify the template instead and then run `bin/build.sh` to rebuild the documentation. diff --git a/src/main/resources/docgen/templates/msmarco-v1-passage.cos-dpr-distil.lexlsh.template b/src/main/resources/docgen/templates/msmarco-v1-passage.cos-dpr-distil.lexlsh.template index 26d61c1cc7..9902ff3401 100644 --- a/src/main/resources/docgen/templates/msmarco-v1-passage.cos-dpr-distil.lexlsh.template +++ b/src/main/resources/docgen/templates/msmarco-v1-passage.cos-dpr-distil.lexlsh.template @@ -1,9 +1,9 @@ # Anserini Regressions: MS MARCO Passage Ranking -**Model**: cosDPR-distil with inverted indexes using the "LexLSH" technique (b=600); pre-encoded queries +**Model**: cosDPR-distil with inverted indexes using the "LexLSH" technique (b=600) using cached queries This page describes regression experiments, integrated into Anserini's regression testing framework, using the cosDPR-distil model on the [MS MARCO passage ranking task](/~https://github.com/microsoft/MSMARCO-Passage-Ranking). -In these experiments, we are using pre-encoded queries (i.e., cached results of query encoding). +In these experiments, we are using cached queries (i.e., cached results of query encoding). The exact configurations for these regressions are stored in [this YAML file](${yaml}). Note that this page is automatically generated from [this template](${template}) as part of Anserini's regression pipeline, so do not modify this page directly; modify the template instead and then run `bin/build.sh` to rebuild the documentation. diff --git a/src/main/resources/docgen/templates/msmarco-v1-passage.deepimpact.cached.template b/src/main/resources/docgen/templates/msmarco-v1-passage.deepimpact.cached.template index 0ca788e934..731230d53b 100644 --- a/src/main/resources/docgen/templates/msmarco-v1-passage.deepimpact.cached.template +++ b/src/main/resources/docgen/templates/msmarco-v1-passage.deepimpact.cached.template @@ -1,12 +1,14 @@ # Anserini Regressions: MS MARCO Passage Ranking -**Model**: DeepImpact +**Model**: DeepImpact (using cached queries) This page describes regression experiments, integrated into Anserini's regression testing framework, using DeepImpact on the [MS MARCO passage ranking task](/~https://github.com/microsoft/MSMARCO-Passage-Ranking). The DeepImpact model is described in the following paper: > Antonio Mallia, Omar Khattab, Nicola Tonellotto, and Torsten Suel. [Learning Passage Impacts for Inverted Indexes.](https://dl.acm.org/doi/10.1145/3404835.3463030) _SIGIR 2021_. +In these experiments, we are using cached queries (i.e., cached results of query encoding). + The exact configurations for these regressions are stored in [this YAML file](${yaml}). Note that this page is automatically generated from [this template](${template}) as part of Anserini's regression pipeline, so do not modify this page directly; modify the template instead. diff --git a/src/main/resources/docgen/templates/msmarco-v1-passage.distill-splade-max.cached.template b/src/main/resources/docgen/templates/msmarco-v1-passage.distill-splade-max.cached.template index 3328f9413a..d19f8043dc 100644 --- a/src/main/resources/docgen/templates/msmarco-v1-passage.distill-splade-max.cached.template +++ b/src/main/resources/docgen/templates/msmarco-v1-passage.distill-splade-max.cached.template @@ -1,12 +1,14 @@ # Anserini Regressions: MS MARCO Passage Ranking -**Model**: DistilSPLADE-max +**Model**: DistilSPLADE-max (using cached queries) This page describes regression experiments, integrated into Anserini's regression testing framework, using the DistilSPLADE-max model from SPLADEv2 on the [MS MARCO passage ranking task](/~https://github.com/microsoft/MSMARCO-Passage-Ranking). The DistilSPLADE-max model is described in the following paper: > Thibault Formal, Carlos Lassance, Benjamin Piwowarski, Stéphane Clinchant. [SPLADE v2: Sparse Lexical and Expansion Model for Information Retrieval.](https://arxiv.org/abs/2109.10086) _arXiv:2109.10086_. +In these experiments, we are using cached queries (i.e., cached results of query encoding). + The exact configurations for these regressions are stored in [this YAML file](${yaml}). Note that this page is automatically generated from [this template](${template}) as part of Anserini's regression pipeline, so do not modify this page directly; modify the template instead. diff --git a/src/main/resources/docgen/templates/msmarco-v1-passage.openai-ada2.hnsw-int8.cached.template b/src/main/resources/docgen/templates/msmarco-v1-passage.openai-ada2.hnsw-int8.cached.template index d1773246fa..cdf4ff2c98 100644 --- a/src/main/resources/docgen/templates/msmarco-v1-passage.openai-ada2.hnsw-int8.cached.template +++ b/src/main/resources/docgen/templates/msmarco-v1-passage.openai-ada2.hnsw-int8.cached.template @@ -1,12 +1,12 @@ # Anserini Regressions: MS MARCO Passage Ranking -**Model**: OpenAI-ada2 embeddings (using pre-encoded queries) with HNSW quantized indexes +**Model**: OpenAI-ada2 embeddings with quantized HNSW indexes (using cached queries) This page describes regression experiments, integrated into Anserini's regression testing framework, using OpenAI-ada2 embeddings on the [MS MARCO passage ranking task](/~https://github.com/microsoft/MSMARCO-Passage-Ranking), as described in the following paper: > Jimmy Lin, Ronak Pradeep, Tommaso Teofili, and Jasper Xian. [Vector Search with OpenAI Embeddings: Lucene Is All You Need.](https://arxiv.org/abs/2308.14963) _arXiv:2308.14963_, 2023. -In these experiments, we are using pre-encoded queries (i.e., cached results of query encoding). +In these experiments, we are using cached queries (i.e., cached results of query encoding). The exact configurations for these regressions are stored in [this YAML file](${yaml}). Note that this page is automatically generated from [this template](${template}) as part of Anserini's regression pipeline, so do not modify this page directly; modify the template instead and then run `bin/build.sh` to rebuild the documentation. @@ -46,7 +46,7 @@ python src/main/python/run_regression.py --index --verify --search --regression ## Indexing -Sample indexing command, building HNSW indexes: +Sample indexing command, building quantized HNSW indexes: ```bash ${index_cmds} diff --git a/src/main/resources/docgen/templates/msmarco-v1-passage.openai-ada2.hnsw.cached.template b/src/main/resources/docgen/templates/msmarco-v1-passage.openai-ada2.hnsw.cached.template index 7bf567dda7..6b705c906e 100644 --- a/src/main/resources/docgen/templates/msmarco-v1-passage.openai-ada2.hnsw.cached.template +++ b/src/main/resources/docgen/templates/msmarco-v1-passage.openai-ada2.hnsw.cached.template @@ -1,12 +1,12 @@ # Anserini Regressions: MS MARCO Passage Ranking -**Model**: OpenAI-ada2 embeddings (using pre-encoded queries) with HNSW indexes +**Model**: OpenAI-ada2 embeddings with HNSW indexes (using cached queries) This page describes regression experiments, integrated into Anserini's regression testing framework, using OpenAI-ada2 embeddings on the [MS MARCO passage ranking task](/~https://github.com/microsoft/MSMARCO-Passage-Ranking), as described in the following paper: > Jimmy Lin, Ronak Pradeep, Tommaso Teofili, and Jasper Xian. [Vector Search with OpenAI Embeddings: Lucene Is All You Need.](https://arxiv.org/abs/2308.14963) _arXiv:2308.14963_, 2023. -In these experiments, we are using pre-encoded queries (i.e., cached results of query encoding). +In these experiments, we are using cached queries (i.e., cached results of query encoding). The exact configurations for these regressions are stored in [this YAML file](${yaml}). Note that this page is automatically generated from [this template](${template}) as part of Anserini's regression pipeline, so do not modify this page directly; modify the template instead and then run `bin/build.sh` to rebuild the documentation. diff --git a/src/main/resources/docgen/templates/msmarco-v1-passage.splade-pp-ed.cached.template b/src/main/resources/docgen/templates/msmarco-v1-passage.splade-pp-ed.cached.template index c4ddea3366..7985f66782 100644 --- a/src/main/resources/docgen/templates/msmarco-v1-passage.splade-pp-ed.cached.template +++ b/src/main/resources/docgen/templates/msmarco-v1-passage.splade-pp-ed.cached.template @@ -1,12 +1,12 @@ # Anserini Regressions: MS MARCO Passage Ranking -**Model**: SPLADE++ CoCondenser-EnsembleDistil (using pre-encoded queries) +**Model**: SPLADE++ CoCondenser-EnsembleDistil (using cached queries) This page describes regression experiments, integrated into Anserini's regression testing framework, using the [SPLADE++ CoCondenser-EnsembleDistil](https://huggingface.co/naver/splade-cocondenser-ensembledistil) model on the [MS MARCO passage ranking task](/~https://github.com/microsoft/MSMARCO-Passage-Ranking), as described in the following paper: > Thibault Formal, Carlos Lassance, Benjamin Piwowarski, and Stéphane Clinchant. [From Distillation to Hard Negative Sampling: Making Sparse Neural IR Models More Effective.](https://dl.acm.org/doi/10.1145/3477495.3531857) _Proceedings of the 45th International ACM SIGIR Conference on Research and Development in Information Retrieval_, pages 2353–2359. -In these experiments, we are using pre-encoded queries (i.e., cached results of query encoding). +In these experiments, we are using cached queries (i.e., cached results of query encoding). The exact configurations for these regressions are stored in [this YAML file](${yaml}). Note that this page is automatically generated from [this template](${template}) as part of Anserini's regression pipeline, so do not modify this page directly; modify the template instead and then run `bin/build.sh` to rebuild the documentation. diff --git a/src/main/resources/docgen/templates/msmarco-v1-passage.splade-pp-sd.cached.template b/src/main/resources/docgen/templates/msmarco-v1-passage.splade-pp-sd.cached.template index 171a7318dc..ad59ce1668 100644 --- a/src/main/resources/docgen/templates/msmarco-v1-passage.splade-pp-sd.cached.template +++ b/src/main/resources/docgen/templates/msmarco-v1-passage.splade-pp-sd.cached.template @@ -1,12 +1,12 @@ # Anserini Regressions: MS MARCO Passage Ranking -**Model**: SPLADE++ CoCondenser-SelfDistil (using pre-encoded queries) +**Model**: SPLADE++ CoCondenser-SelfDistil (using cached queries) This page describes regression experiments, integrated into Anserini's regression testing framework, using the [SPLADE++ CoCondenser-SelfDistil](https://huggingface.co/naver/splade-cocondenser-selfdistil) model on the [MS MARCO passage ranking task](/~https://github.com/microsoft/MSMARCO-Passage-Ranking), as described in the following paper: > Thibault Formal, Carlos Lassance, Benjamin Piwowarski, and Stéphane Clinchant. [From Distillation to Hard Negative Sampling: Making Sparse Neural IR Models More Effective.](https://dl.acm.org/doi/10.1145/3477495.3531857) _Proceedings of the 45th International ACM SIGIR Conference on Research and Development in Information Retrieval_, pages 2353–2359. -In these experiments, we are using pre-encoded queries (i.e., cached results of query encoding). +In these experiments, we are using cached queries (i.e., cached results of query encoding). The exact configurations for these regressions are stored in [this YAML file](${yaml}). Note that this page is automatically generated from [this template](${template}) as part of Anserini's regression pipeline, so do not modify this page directly; modify the template instead and then run `bin/build.sh` to rebuild the documentation. diff --git a/src/main/resources/docgen/templates/msmarco-v1-passage.unicoil-noexp.cached.template b/src/main/resources/docgen/templates/msmarco-v1-passage.unicoil-noexp.cached.template index 9d82d8faa3..7536d6785c 100644 --- a/src/main/resources/docgen/templates/msmarco-v1-passage.unicoil-noexp.cached.template +++ b/src/main/resources/docgen/templates/msmarco-v1-passage.unicoil-noexp.cached.template @@ -1,6 +1,6 @@ # Anserini Regressions: MS MARCO Passage Ranking -**Model**: uniCOIL (without any expansions) +**Model**: uniCOIL without any expansions (using cached queries) This page describes regression experiments, integrated into Anserini's regression testing framework, using uniCOIL (without any expansions) on the [MS MARCO passage ranking task](/~https://github.com/microsoft/MSMARCO-Passage-Ranking). The uniCOIL model is described in the following paper: @@ -10,6 +10,8 @@ The uniCOIL model is described in the following paper: The experiments on this page are not actually reported in the paper. Here, a variant model without expansion is used. +In these experiments, we are using cached queries (i.e., cached results of query encoding). + The exact configurations for these regressions are stored in [this YAML file](${yaml}). Note that this page is automatically generated from [this template](${template}) as part of Anserini's regression pipeline, so do not modify this page directly; modify the template instead and then run `bin/build.sh` to rebuild the documentation. diff --git a/src/main/resources/docgen/templates/msmarco-v1-passage.unicoil-tilde-expansion.cached.template b/src/main/resources/docgen/templates/msmarco-v1-passage.unicoil-tilde-expansion.cached.template index 4f47cd9926..7ebe8f20df 100644 --- a/src/main/resources/docgen/templates/msmarco-v1-passage.unicoil-tilde-expansion.cached.template +++ b/src/main/resources/docgen/templates/msmarco-v1-passage.unicoil-tilde-expansion.cached.template @@ -1,12 +1,14 @@ # Anserini Regressions: MS MARCO Passage Ranking -**Model**: uniCOIL (with TILDE expansions) +**Model**: uniCOIL with TILDE expansions (using cached queries) This page describes regression experiments, integrated into Anserini's regression testing framework, using uniCOIL (with TILDE expansions) on the [MS MARCO passage ranking task](/~https://github.com/microsoft/MSMARCO-Passage-Ranking). The uniCOIL+TILDE model is described in the following paper: > Shengyao Zhuang and Guido Zuccon. [Fast Passage Re-ranking with Contextualized Exact Term Matching and Efficient Passage Expansion.](https://arxiv.org/pdf/2108.08513) _arXiv:2108.08513_. +In these experiments, we are using cached queries (i.e., cached results of query encoding). + The exact configurations for these regressions are stored in [this YAML file](${yaml}). Note that this page is automatically generated from [this template](${template}) as part of Anserini's regression pipeline, so do not modify this page directly; modify the template instead. diff --git a/src/main/resources/docgen/templates/msmarco-v1-passage.unicoil.cached.template b/src/main/resources/docgen/templates/msmarco-v1-passage.unicoil.cached.template index 5026217ba6..4476d7e4b4 100644 --- a/src/main/resources/docgen/templates/msmarco-v1-passage.unicoil.cached.template +++ b/src/main/resources/docgen/templates/msmarco-v1-passage.unicoil.cached.template @@ -1,12 +1,14 @@ # Anserini Regressions: MS MARCO Passage Ranking -**Model**: uniCOIL (with doc2query-T5 expansions) +**Model**: uniCOIL with doc2query-T5 expansions (using cached queries) This page describes regression experiments, integrated into Anserini's regression testing framework, using uniCOIL (with doc2query-T5 expansions) on the [MS MARCO passage ranking task](/~https://github.com/microsoft/MSMARCO-Passage-Ranking). The uniCOIL model is described in the following paper: > Jimmy Lin and Xueguang Ma. [A Few Brief Notes on DeepImpact, COIL, and a Conceptual Framework for Information Retrieval Techniques.](https://arxiv.org/abs/2106.14807) _arXiv:2106.14807_. +In these experiments, we are using cached queries (i.e., cached results of query encoding). + The exact configurations for these regressions are stored in [this YAML file](${yaml}). Note that this page is automatically generated from [this template](${template}) as part of Anserini's regression pipeline, so do not modify this page directly; modify the template instead and then run `bin/build.sh` to rebuild the documentation. diff --git a/src/main/resources/docgen/templates/msmarco-v2-passage.splade-pp-ed.cached.template b/src/main/resources/docgen/templates/msmarco-v2-passage.splade-pp-ed.cached.template index 3cb54a910d..775bef83e1 100644 --- a/src/main/resources/docgen/templates/msmarco-v2-passage.splade-pp-ed.cached.template +++ b/src/main/resources/docgen/templates/msmarco-v2-passage.splade-pp-ed.cached.template @@ -1,9 +1,9 @@ # Anserini Regressions: MS MARCO (V2) Passage Ranking -**Model**: SPLADE++ CoCondenser-EnsembleDistil (using pre-encoded queries) +**Model**: SPLADE++ CoCondenser-EnsembleDistil (using cached queries) This page describes regression experiments, integrated into Anserini's regression testing framework, applying the [SPLADE++ CoCondenser-EnsembleDistil](https://huggingface.co/naver/splade-cocondenser-ensembledistil) model to the MS MARCO V2 passage corpus. -Here, we evaluate on the dev queries, using pre-encoded queries (i.e., cached results of query encoding). +Here, we evaluate on the dev queries, using cached queries (i.e., cached results of query encoding). The model is described in the following paper: diff --git a/src/main/resources/docgen/templates/msmarco-v2-passage.splade-pp-sd.cached.template b/src/main/resources/docgen/templates/msmarco-v2-passage.splade-pp-sd.cached.template index 591b193e23..1cf6f312b9 100644 --- a/src/main/resources/docgen/templates/msmarco-v2-passage.splade-pp-sd.cached.template +++ b/src/main/resources/docgen/templates/msmarco-v2-passage.splade-pp-sd.cached.template @@ -1,9 +1,9 @@ # Anserini Regressions: MS MARCO (V2) Passage Ranking -**Model**: SPLADE++ CoCondenser-SelfDistil (using pre-encoded queries) +**Model**: SPLADE++ CoCondenser-SelfDistil (using cached queries) This page describes regression experiments, integrated into Anserini's regression testing framework, applying the [SPLADE++ CoCondenser-SelfDistil](https://huggingface.co/naver/splade-cocondenser-selfdistil) model to the MS MARCO V2 passage corpus. -Here, we evaluate on the dev queries, using pre-encoded queries (i.e., cached results of query encoding). +Here, we evaluate on the dev queries, using cached queries (i.e., cached results of query encoding). The model is described in the following paper: