--- tags: - sentence-transformers - feature-extraction - sentence-similarity - transformers - mteb model-index: - name: bge-large-en-v1.5 results: - task: type: Classification dataset: type: mteb/amazon_counterfactual name: MTEB AmazonCounterfactualClassification (en) config: en split: test revision: e8379541af4e31359cca9fbcf4b00f2671dba205 metrics: - type: accuracy value: 75.8507462686567 - type: ap value: 38.566457320228245 - type: f1 value: 69.69386648043475 - task: type: Classification dataset: type: mteb/amazon_polarity name: MTEB AmazonPolarityClassification config: default split: test revision: e2d317d38cd51312af73b3d32a06d1a08b442046 metrics: - type: accuracy value: 92.416675 - type: ap value: 89.1928861155922 - type: f1 value: 92.39477019574215 - task: type: Classification dataset: type: mteb/amazon_reviews_multi name: MTEB AmazonReviewsClassification (en) config: en split: test revision: 1399c76144fd37290681b995c656ef9b2e06e26d metrics: - type: accuracy value: 48.175999999999995 - type: f1 value: 47.80712792870253 - task: type: Retrieval dataset: type: arguana name: MTEB ArguAna config: default split: test revision: None metrics: - type: map_at_1 value: 40.184999999999995 - type: map_at_10 value: 55.654 - type: map_at_100 value: 56.25 - type: map_at_1000 value: 56.255 - type: map_at_3 value: 51.742999999999995 - type: map_at_5 value: 54.129000000000005 - type: mrr_at_1 value: 40.967 - type: mrr_at_10 value: 55.96 - type: mrr_at_100 value: 56.54900000000001 - type: mrr_at_1000 value: 56.554 - type: mrr_at_3 value: 51.980000000000004 - type: mrr_at_5 value: 54.44 - type: ndcg_at_1 value: 40.184999999999995 - type: ndcg_at_10 value: 63.542 - type: ndcg_at_100 value: 65.96499999999999 - type: ndcg_at_1000 value: 66.08699999999999 - type: ndcg_at_3 value: 55.582 - type: ndcg_at_5 value: 59.855000000000004 - type: precision_at_1 value: 40.184999999999995 - type: precision_at_10 value: 8.841000000000001 - type: precision_at_100 value: 0.987 - type: precision_at_1000 value: 0.1 - type: precision_at_3 value: 22.238 - type: precision_at_5 value: 15.405 - type: recall_at_1 value: 40.184999999999995 - type: recall_at_10 value: 88.407 - type: recall_at_100 value: 98.72 - type: recall_at_1000 value: 99.644 - type: recall_at_3 value: 66.714 - type: recall_at_5 value: 77.027 - task: type: Clustering dataset: type: mteb/arxiv-clustering-p2p name: MTEB ArxivClusteringP2P config: default split: test revision: a122ad7f3f0291bf49cc6f4d32aa80929df69d5d metrics: - type: v_measure value: 48.567077926750066 - task: type: Clustering dataset: type: mteb/arxiv-clustering-s2s name: MTEB ArxivClusteringS2S config: default split: test revision: f910caf1a6075f7329cdf8c1a6135696f37dbd53 metrics: - type: v_measure value: 43.19453389182364 - task: type: Reranking dataset: type: mteb/askubuntudupquestions-reranking name: MTEB AskUbuntuDupQuestions config: default split: test revision: 2000358ca161889fa9c082cb41daa8dcfb161a54 metrics: - type: map value: 64.46555939623092 - type: mrr value: 77.82361605768807 - task: type: STS dataset: type: mteb/biosses-sts name: MTEB BIOSSES config: default split: test revision: d3fb88f8f02e40887cd149695127462bbcf29b4a metrics: - type: cos_sim_pearson value: 84.9554128814735 - type: cos_sim_spearman value: 84.65373612172036 - type: euclidean_pearson value: 83.2905059954138 - type: euclidean_spearman value: 84.52240782811128 - type: manhattan_pearson value: 82.99533802997436 - type: manhattan_spearman value: 84.20673798475734 - task: type: Classification dataset: type: mteb/banking77 name: MTEB Banking77Classification config: default split: test revision: 0fd18e25b25c072e09e0d92ab615fda904d66300 metrics: - type: accuracy value: 87.78896103896103 - type: f1 value: 87.77189310964883 - task: type: Clustering dataset: type: mteb/biorxiv-clustering-p2p name: MTEB BiorxivClusteringP2P config: default split: test revision: 65b79d1d13f80053f67aca9498d9402c2d9f1f40 metrics: - type: v_measure value: 39.714538337650495 - task: type: Clustering dataset: type: mteb/biorxiv-clustering-s2s name: MTEB BiorxivClusteringS2S config: default split: test revision: 258694dd0231531bc1fd9de6ceb52a0853c6d908 metrics: - type: v_measure value: 36.90108349284447 - task: type: Retrieval dataset: type: BeIR/cqadupstack name: MTEB CQADupstackAndroidRetrieval config: default split: test revision: None metrics: - type: map_at_1 value: 32.795 - type: map_at_10 value: 43.669000000000004 - type: map_at_100 value: 45.151 - type: map_at_1000 value: 45.278 - type: map_at_3 value: 40.006 - type: map_at_5 value: 42.059999999999995 - type: mrr_at_1 value: 39.771 - type: mrr_at_10 value: 49.826 - type: mrr_at_100 value: 50.504000000000005 - type: mrr_at_1000 value: 50.549 - type: mrr_at_3 value: 47.115 - type: mrr_at_5 value: 48.832 - type: ndcg_at_1 value: 39.771 - type: ndcg_at_10 value: 50.217999999999996 - type: ndcg_at_100 value: 55.454 - type: ndcg_at_1000 value: 57.37 - type: ndcg_at_3 value: 44.885000000000005 - type: ndcg_at_5 value: 47.419 - type: precision_at_1 value: 39.771 - type: precision_at_10 value: 9.642000000000001 - type: precision_at_100 value: 1.538 - type: precision_at_1000 value: 0.198 - type: precision_at_3 value: 21.268 - type: precision_at_5 value: 15.536 - type: recall_at_1 value: 32.795 - type: recall_at_10 value: 62.580999999999996 - type: recall_at_100 value: 84.438 - type: recall_at_1000 value: 96.492 - type: recall_at_3 value: 47.071000000000005 - type: recall_at_5 value: 54.079 - task: type: Retrieval dataset: type: BeIR/cqadupstack name: MTEB CQADupstackEnglishRetrieval config: default split: test revision: None metrics: - type: map_at_1 value: 32.671 - type: map_at_10 value: 43.334 - type: map_at_100 value: 44.566 - type: map_at_1000 value: 44.702999999999996 - type: map_at_3 value: 40.343 - type: map_at_5 value: 41.983 - type: mrr_at_1 value: 40.764 - type: mrr_at_10 value: 49.382 - type: mrr_at_100 value: 49.988 - type: mrr_at_1000 value: 50.03300000000001 - type: mrr_at_3 value: 47.293 - type: mrr_at_5 value: 48.51 - type: ndcg_at_1 value: 40.764 - type: ndcg_at_10 value: 49.039 - type: ndcg_at_100 value: 53.259 - type: ndcg_at_1000 value: 55.253 - type: ndcg_at_3 value: 45.091 - type: ndcg_at_5 value: 46.839999999999996 - type: precision_at_1 value: 40.764 - type: precision_at_10 value: 9.191 - type: precision_at_100 value: 1.476 - type: precision_at_1000 value: 0.19499999999999998 - type: precision_at_3 value: 21.72 - type: precision_at_5 value: 15.299 - type: recall_at_1 value: 32.671 - type: recall_at_10 value: 58.816 - type: recall_at_100 value: 76.654 - type: recall_at_1000 value: 89.05999999999999 - type: recall_at_3 value: 46.743 - type: recall_at_5 value: 51.783 - task: type: Retrieval dataset: type: BeIR/cqadupstack name: MTEB CQADupstackGamingRetrieval config: default split: test revision: None metrics: - type: map_at_1 value: 40.328 - type: map_at_10 value: 53.32599999999999 - type: map_at_100 value: 54.37499999999999 - type: map_at_1000 value: 54.429 - type: map_at_3 value: 49.902 - type: map_at_5 value: 52.002 - type: mrr_at_1 value: 46.332 - type: mrr_at_10 value: 56.858 - type: mrr_at_100 value: 57.522 - type: mrr_at_1000 value: 57.54899999999999 - type: mrr_at_3 value: 54.472 - type: mrr_at_5 value: 55.996 - type: ndcg_at_1 value: 46.332 - type: ndcg_at_10 value: 59.313 - type: ndcg_at_100 value: 63.266999999999996 - type: ndcg_at_1000 value: 64.36 - type: ndcg_at_3 value: 53.815000000000005 - type: ndcg_at_5 value: 56.814 - type: precision_at_1 value: 46.332 - type: precision_at_10 value: 9.53 - type: precision_at_100 value: 1.238 - type: precision_at_1000 value: 0.13699999999999998 - type: precision_at_3 value: 24.054000000000002 - type: precision_at_5 value: 16.589000000000002 - type: recall_at_1 value: 40.328 - type: recall_at_10 value: 73.421 - type: recall_at_100 value: 90.059 - type: recall_at_1000 value: 97.81 - type: recall_at_3 value: 59.009 - type: recall_at_5 value: 66.352 - task: type: Retrieval dataset: type: BeIR/cqadupstack name: MTEB CQADupstackGisRetrieval config: default split: test revision: None metrics: - type: map_at_1 value: 27.424 - type: map_at_10 value: 36.332 - type: map_at_100 value: 37.347 - type: map_at_1000 value: 37.422 - type: map_at_3 value: 33.743 - type: map_at_5 value: 35.176 - type: mrr_at_1 value: 29.153000000000002 - type: mrr_at_10 value: 38.233 - type: mrr_at_100 value: 39.109 - type: mrr_at_1000 value: 39.164 - type: mrr_at_3 value: 35.876000000000005 - type: mrr_at_5 value: 37.169000000000004 - type: ndcg_at_1 value: 29.153000000000002 - type: ndcg_at_10 value: 41.439 - type: ndcg_at_100 value: 46.42 - type: ndcg_at_1000 value: 48.242000000000004 - type: ndcg_at_3 value: 36.362 - type: ndcg_at_5 value: 38.743 - type: precision_at_1 value: 29.153000000000002 - type: precision_at_10 value: 6.315999999999999 - type: precision_at_100 value: 0.927 - type: precision_at_1000 value: 0.11199999999999999 - type: precision_at_3 value: 15.443000000000001 - type: precision_at_5 value: 10.644 - type: recall_at_1 value: 27.424 - type: recall_at_10 value: 55.364000000000004 - type: recall_at_100 value: 78.211 - type: recall_at_1000 value: 91.74600000000001 - type: recall_at_3 value: 41.379 - type: recall_at_5 value: 47.14 - task: type: Retrieval dataset: type: BeIR/cqadupstack name: MTEB CQADupstackMathematicaRetrieval config: default split: test revision: None metrics: - type: map_at_1 value: 19.601 - type: map_at_10 value: 27.826 - type: map_at_100 value: 29.017 - type: map_at_1000 value: 29.137 - type: map_at_3 value: 25.125999999999998 - type: map_at_5 value: 26.765 - type: mrr_at_1 value: 24.005000000000003 - type: mrr_at_10 value: 32.716 - type: mrr_at_100 value: 33.631 - type: mrr_at_1000 value: 33.694 - type: mrr_at_3 value: 29.934 - type: mrr_at_5 value: 31.630999999999997 - type: ndcg_at_1 value: 24.005000000000003 - type: ndcg_at_10 value: 33.158 - type: ndcg_at_100 value: 38.739000000000004 - type: ndcg_at_1000 value: 41.495 - type: ndcg_at_3 value: 28.185 - type: ndcg_at_5 value: 30.796 - type: precision_at_1 value: 24.005000000000003 - type: precision_at_10 value: 5.908 - type: precision_at_100 value: 1.005 - type: precision_at_1000 value: 0.13899999999999998 - type: precision_at_3 value: 13.391 - type: precision_at_5 value: 9.876 - type: recall_at_1 value: 19.601 - type: recall_at_10 value: 44.746 - type: recall_at_100 value: 68.82300000000001 - type: recall_at_1000 value: 88.215 - type: recall_at_3 value: 31.239 - type: recall_at_5 value: 37.695 - task: type: Retrieval dataset: type: BeIR/cqadupstack name: MTEB CQADupstackPhysicsRetrieval config: default split: test revision: None metrics: - type: map_at_1 value: 30.130000000000003 - type: map_at_10 value: 40.96 - type: map_at_100 value: 42.282 - type: map_at_1000 value: 42.392 - type: map_at_3 value: 37.889 - type: map_at_5 value: 39.661 - type: mrr_at_1 value: 36.958999999999996 - type: mrr_at_10 value: 46.835 - type: mrr_at_100 value: 47.644 - type: mrr_at_1000 value: 47.688 - type: mrr_at_3 value: 44.562000000000005 - type: mrr_at_5 value: 45.938 - type: ndcg_at_1 value: 36.958999999999996 - type: ndcg_at_10 value: 47.06 - type: ndcg_at_100 value: 52.345 - type: ndcg_at_1000 value: 54.35 - type: ndcg_at_3 value: 42.301 - type: ndcg_at_5 value: 44.635999999999996 - type: precision_at_1 value: 36.958999999999996 - type: precision_at_10 value: 8.479000000000001 - type: precision_at_100 value: 1.284 - type: precision_at_1000 value: 0.163 - type: precision_at_3 value: 20.244 - type: precision_at_5 value: 14.224999999999998 - type: recall_at_1 value: 30.130000000000003 - type: recall_at_10 value: 59.27 - type: recall_at_100 value: 81.195 - type: recall_at_1000 value: 94.21199999999999 - type: recall_at_3 value: 45.885 - type: recall_at_5 value: 52.016 - task: type: Retrieval dataset: type: BeIR/cqadupstack name: MTEB CQADupstackProgrammersRetrieval config: default split: test revision: None metrics: - type: map_at_1 value: 26.169999999999998 - type: map_at_10 value: 36.451 - type: map_at_100 value: 37.791000000000004 - type: map_at_1000 value: 37.897 - type: map_at_3 value: 33.109 - type: map_at_5 value: 34.937000000000005 - type: mrr_at_1 value: 32.877 - type: mrr_at_10 value: 42.368 - type: mrr_at_100 value: 43.201 - type: mrr_at_1000 value: 43.259 - type: mrr_at_3 value: 39.763999999999996 - type: mrr_at_5 value: 41.260000000000005 - type: ndcg_at_1 value: 32.877 - type: ndcg_at_10 value: 42.659000000000006 - type: ndcg_at_100 value: 48.161 - type: ndcg_at_1000 value: 50.345 - type: ndcg_at_3 value: 37.302 - type: ndcg_at_5 value: 39.722 - type: precision_at_1 value: 32.877 - type: precision_at_10 value: 7.9 - type: precision_at_100 value: 1.236 - type: precision_at_1000 value: 0.158 - type: precision_at_3 value: 17.846 - type: precision_at_5 value: 12.9 - type: recall_at_1 value: 26.169999999999998 - type: recall_at_10 value: 55.35 - type: recall_at_100 value: 78.755 - type: recall_at_1000 value: 93.518 - type: recall_at_3 value: 40.176 - type: recall_at_5 value: 46.589000000000006 - task: type: Retrieval dataset: type: BeIR/cqadupstack name: MTEB CQADupstackRetrieval config: default split: test revision: None metrics: - type: map_at_1 value: 27.15516666666667 - type: map_at_10 value: 36.65741666666667 - type: map_at_100 value: 37.84991666666666 - type: map_at_1000 value: 37.96316666666667 - type: map_at_3 value: 33.74974999999999 - type: map_at_5 value: 35.3765 - type: mrr_at_1 value: 32.08233333333334 - type: mrr_at_10 value: 41.033833333333334 - type: mrr_at_100 value: 41.84524999999999 - type: mrr_at_1000 value: 41.89983333333333 - type: mrr_at_3 value: 38.62008333333333 - type: mrr_at_5 value: 40.03441666666666 - type: ndcg_at_1 value: 32.08233333333334 - type: ndcg_at_10 value: 42.229 - type: ndcg_at_100 value: 47.26716666666667 - type: ndcg_at_1000 value: 49.43466666666667 - type: ndcg_at_3 value: 37.36408333333333 - type: ndcg_at_5 value: 39.6715 - type: precision_at_1 value: 32.08233333333334 - type: precision_at_10 value: 7.382583333333334 - type: precision_at_100 value: 1.16625 - type: precision_at_1000 value: 0.15408333333333332 - type: precision_at_3 value: 17.218 - type: precision_at_5 value: 12.21875 - type: recall_at_1 value: 27.15516666666667 - type: recall_at_10 value: 54.36683333333333 - type: recall_at_100 value: 76.37183333333333 - type: recall_at_1000 value: 91.26183333333333 - type: recall_at_3 value: 40.769916666666674 - type: recall_at_5 value: 46.702333333333335 - task: type: Retrieval dataset: type: BeIR/cqadupstack name: MTEB CQADupstackStatsRetrieval config: default split: test revision: None metrics: - type: map_at_1 value: 25.749 - type: map_at_10 value: 33.001999999999995 - type: map_at_100 value: 33.891 - type: map_at_1000 value: 33.993 - type: map_at_3 value: 30.703999999999997 - type: map_at_5 value: 31.959 - type: mrr_at_1 value: 28.834 - type: mrr_at_10 value: 35.955 - type: mrr_at_100 value: 36.709 - type: mrr_at_1000 value: 36.779 - type: mrr_at_3 value: 33.947 - type: mrr_at_5 value: 35.089 - type: ndcg_at_1 value: 28.834 - type: ndcg_at_10 value: 37.329 - type: ndcg_at_100 value: 41.79 - type: ndcg_at_1000 value: 44.169000000000004 - type: ndcg_at_3 value: 33.184999999999995 - type: ndcg_at_5 value: 35.107 - type: precision_at_1 value: 28.834 - type: precision_at_10 value: 5.7669999999999995 - type: precision_at_100 value: 0.876 - type: precision_at_1000 value: 0.11399999999999999 - type: precision_at_3 value: 14.213000000000001 - type: precision_at_5 value: 9.754999999999999 - type: recall_at_1 value: 25.749 - type: recall_at_10 value: 47.791 - type: recall_at_100 value: 68.255 - type: recall_at_1000 value: 85.749 - type: recall_at_3 value: 36.199 - type: recall_at_5 value: 41.071999999999996 - task: type: Retrieval dataset: type: BeIR/cqadupstack name: MTEB CQADupstackTexRetrieval config: default split: test revision: None metrics: - type: map_at_1 value: 17.777 - type: map_at_10 value: 25.201 - type: map_at_100 value: 26.423999999999996 - type: map_at_1000 value: 26.544 - type: map_at_3 value: 22.869 - type: map_at_5 value: 24.023 - type: mrr_at_1 value: 21.473 - type: mrr_at_10 value: 29.12 - type: mrr_at_100 value: 30.144 - type: mrr_at_1000 value: 30.215999999999998 - type: mrr_at_3 value: 26.933 - type: mrr_at_5 value: 28.051 - type: ndcg_at_1 value: 21.473 - type: ndcg_at_10 value: 30.003 - type: ndcg_at_100 value: 35.766 - type: ndcg_at_1000 value: 38.501000000000005 - type: ndcg_at_3 value: 25.773000000000003 - type: ndcg_at_5 value: 27.462999999999997 - type: precision_at_1 value: 21.473 - type: precision_at_10 value: 5.482 - type: precision_at_100 value: 0.975 - type: precision_at_1000 value: 0.13799999999999998 - type: precision_at_3 value: 12.205 - type: precision_at_5 value: 8.692 - type: recall_at_1 value: 17.777 - type: recall_at_10 value: 40.582 - type: recall_at_100 value: 66.305 - type: recall_at_1000 value: 85.636 - type: recall_at_3 value: 28.687 - type: recall_at_5 value: 33.089 - task: type: Retrieval dataset: type: BeIR/cqadupstack name: MTEB CQADupstackUnixRetrieval config: default split: test revision: None metrics: - type: map_at_1 value: 26.677 - type: map_at_10 value: 36.309000000000005 - type: map_at_100 value: 37.403999999999996 - type: map_at_1000 value: 37.496 - type: map_at_3 value: 33.382 - type: map_at_5 value: 34.98 - type: mrr_at_1 value: 31.343 - type: mrr_at_10 value: 40.549 - type: mrr_at_100 value: 41.342 - type: mrr_at_1000 value: 41.397 - type: mrr_at_3 value: 38.029 - type: mrr_at_5 value: 39.451 - type: ndcg_at_1 value: 31.343 - type: ndcg_at_10 value: 42.1 - type: ndcg_at_100 value: 47.089999999999996 - type: ndcg_at_1000 value: 49.222 - type: ndcg_at_3 value: 36.836999999999996 - type: ndcg_at_5 value: 39.21 - type: precision_at_1 value: 31.343 - type: precision_at_10 value: 7.164 - type: precision_at_100 value: 1.0959999999999999 - type: precision_at_1000 value: 0.13899999999999998 - type: precision_at_3 value: 16.915 - type: precision_at_5 value: 11.940000000000001 - type: recall_at_1 value: 26.677 - type: recall_at_10 value: 55.54599999999999 - type: recall_at_100 value: 77.094 - type: recall_at_1000 value: 92.01 - type: recall_at_3 value: 41.191 - type: recall_at_5 value: 47.006 - task: type: Retrieval dataset: type: BeIR/cqadupstack name: MTEB CQADupstackWebmastersRetrieval config: default split: test revision: None metrics: - type: map_at_1 value: 24.501 - type: map_at_10 value: 33.102 - type: map_at_100 value: 34.676 - type: map_at_1000 value: 34.888000000000005 - type: map_at_3 value: 29.944 - type: map_at_5 value: 31.613999999999997 - type: mrr_at_1 value: 29.447000000000003 - type: mrr_at_10 value: 37.996 - type: mrr_at_100 value: 38.946 - type: mrr_at_1000 value: 38.995000000000005 - type: mrr_at_3 value: 35.079 - type: mrr_at_5 value: 36.69 - type: ndcg_at_1 value: 29.447000000000003 - type: ndcg_at_10 value: 39.232 - type: ndcg_at_100 value: 45.247 - type: ndcg_at_1000 value: 47.613 - type: ndcg_at_3 value: 33.922999999999995 - type: ndcg_at_5 value: 36.284 - type: precision_at_1 value: 29.447000000000003 - type: precision_at_10 value: 7.648000000000001 - type: precision_at_100 value: 1.516 - type: precision_at_1000 value: 0.23900000000000002 - type: precision_at_3 value: 16.008 - type: precision_at_5 value: 11.779 - type: recall_at_1 value: 24.501 - type: recall_at_10 value: 51.18899999999999 - type: recall_at_100 value: 78.437 - type: recall_at_1000 value: 92.842 - type: recall_at_3 value: 35.808 - type: recall_at_5 value: 42.197 - task: type: Retrieval dataset: type: BeIR/cqadupstack name: MTEB CQADupstackWordpressRetrieval config: default split: test revision: None metrics: - type: map_at_1 value: 22.039 - type: map_at_10 value: 30.377 - type: map_at_100 value: 31.275 - type: map_at_1000 value: 31.379 - type: map_at_3 value: 27.98 - type: map_at_5 value: 29.358 - type: mrr_at_1 value: 24.03 - type: mrr_at_10 value: 32.568000000000005 - type: mrr_at_100 value: 33.403 - type: mrr_at_1000 value: 33.475 - type: mrr_at_3 value: 30.436999999999998 - type: mrr_at_5 value: 31.796000000000003 - type: ndcg_at_1 value: 24.03 - type: ndcg_at_10 value: 35.198 - type: ndcg_at_100 value: 39.668 - type: ndcg_at_1000 value: 42.296 - type: ndcg_at_3 value: 30.709999999999997 - type: ndcg_at_5 value: 33.024 - type: precision_at_1 value: 24.03 - type: precision_at_10 value: 5.564 - type: precision_at_100 value: 0.828 - type: precision_at_1000 value: 0.117 - type: precision_at_3 value: 13.309000000000001 - type: precision_at_5 value: 9.39 - type: recall_at_1 value: 22.039 - type: recall_at_10 value: 47.746 - type: recall_at_100 value: 68.23599999999999 - type: recall_at_1000 value: 87.852 - type: recall_at_3 value: 35.852000000000004 - type: recall_at_5 value: 41.410000000000004 - task: type: Retrieval dataset: type: climate-fever name: MTEB ClimateFEVER config: default split: test revision: None metrics: - type: map_at_1 value: 15.692999999999998 - type: map_at_10 value: 26.903 - type: map_at_100 value: 28.987000000000002 - type: map_at_1000 value: 29.176999999999996 - type: map_at_3 value: 22.137 - type: map_at_5 value: 24.758 - type: mrr_at_1 value: 35.57 - type: mrr_at_10 value: 47.821999999999996 - type: mrr_at_100 value: 48.608000000000004 - type: mrr_at_1000 value: 48.638999999999996 - type: mrr_at_3 value: 44.452000000000005 - type: mrr_at_5 value: 46.546 - type: ndcg_at_1 value: 35.57 - type: ndcg_at_10 value: 36.567 - type: ndcg_at_100 value: 44.085 - type: ndcg_at_1000 value: 47.24 - type: ndcg_at_3 value: 29.964000000000002 - type: ndcg_at_5 value: 32.511 - type: precision_at_1 value: 35.57 - type: precision_at_10 value: 11.485 - type: precision_at_100 value: 1.9619999999999997 - type: precision_at_1000 value: 0.256 - type: precision_at_3 value: 22.237000000000002 - type: precision_at_5 value: 17.471999999999998 - type: recall_at_1 value: 15.692999999999998 - type: recall_at_10 value: 43.056 - type: recall_at_100 value: 68.628 - type: recall_at_1000 value: 86.075 - type: recall_at_3 value: 26.918999999999997 - type: recall_at_5 value: 34.14 - task: type: Retrieval dataset: type: dbpedia-entity name: MTEB DBPedia config: default split: test revision: None metrics: - type: map_at_1 value: 9.53 - type: map_at_10 value: 20.951 - type: map_at_100 value: 30.136000000000003 - type: map_at_1000 value: 31.801000000000002 - type: map_at_3 value: 15.021 - type: map_at_5 value: 17.471999999999998 - type: mrr_at_1 value: 71.0 - type: mrr_at_10 value: 79.176 - type: mrr_at_100 value: 79.418 - type: mrr_at_1000 value: 79.426 - type: mrr_at_3 value: 78.125 - type: mrr_at_5 value: 78.61200000000001 - type: ndcg_at_1 value: 58.5 - type: ndcg_at_10 value: 44.106 - type: ndcg_at_100 value: 49.268 - type: ndcg_at_1000 value: 56.711999999999996 - type: ndcg_at_3 value: 48.934 - type: ndcg_at_5 value: 45.826 - type: precision_at_1 value: 71.0 - type: precision_at_10 value: 35.0 - type: precision_at_100 value: 11.360000000000001 - type: precision_at_1000 value: 2.046 - type: precision_at_3 value: 52.833 - type: precision_at_5 value: 44.15 - type: recall_at_1 value: 9.53 - type: recall_at_10 value: 26.811 - type: recall_at_100 value: 55.916999999999994 - type: recall_at_1000 value: 79.973 - type: recall_at_3 value: 16.413 - type: recall_at_5 value: 19.980999999999998 - task: type: Classification dataset: type: mteb/emotion name: MTEB EmotionClassification config: default split: test revision: 4f58c6b202a23cf9a4da393831edf4f9183cad37 metrics: - type: accuracy value: 51.519999999999996 - type: f1 value: 46.36601294761231 - task: type: Retrieval dataset: type: fever name: MTEB FEVER config: default split: test revision: None metrics: - type: map_at_1 value: 74.413 - type: map_at_10 value: 83.414 - type: map_at_100 value: 83.621 - type: map_at_1000 value: 83.635 - type: map_at_3 value: 82.337 - type: map_at_5 value: 83.039 - type: mrr_at_1 value: 80.19800000000001 - type: mrr_at_10 value: 87.715 - type: mrr_at_100 value: 87.778 - type: mrr_at_1000 value: 87.779 - type: mrr_at_3 value: 87.106 - type: mrr_at_5 value: 87.555 - type: ndcg_at_1 value: 80.19800000000001 - type: ndcg_at_10 value: 87.182 - type: ndcg_at_100 value: 87.90299999999999 - type: ndcg_at_1000 value: 88.143 - type: ndcg_at_3 value: 85.60600000000001 - type: ndcg_at_5 value: 86.541 - type: precision_at_1 value: 80.19800000000001 - type: precision_at_10 value: 10.531 - type: precision_at_100 value: 1.113 - type: precision_at_1000 value: 0.11499999999999999 - type: precision_at_3 value: 32.933 - type: precision_at_5 value: 20.429 - type: recall_at_1 value: 74.413 - type: recall_at_10 value: 94.363 - type: recall_at_100 value: 97.165 - type: recall_at_1000 value: 98.668 - type: recall_at_3 value: 90.108 - type: recall_at_5 value: 92.52 - task: type: Retrieval dataset: type: fiqa name: MTEB FiQA2018 config: default split: test revision: None metrics: - type: map_at_1 value: 22.701 - type: map_at_10 value: 37.122 - type: map_at_100 value: 39.178000000000004 - type: map_at_1000 value: 39.326 - type: map_at_3 value: 32.971000000000004 - type: map_at_5 value: 35.332 - type: mrr_at_1 value: 44.753 - type: mrr_at_10 value: 53.452 - type: mrr_at_100 value: 54.198 - type: mrr_at_1000 value: 54.225 - type: mrr_at_3 value: 50.952 - type: mrr_at_5 value: 52.464 - type: ndcg_at_1 value: 44.753 - type: ndcg_at_10 value: 45.021 - type: ndcg_at_100 value: 52.028 - type: ndcg_at_1000 value: 54.596000000000004 - type: ndcg_at_3 value: 41.622 - type: ndcg_at_5 value: 42.736000000000004 - type: precision_at_1 value: 44.753 - type: precision_at_10 value: 12.284 - type: precision_at_100 value: 1.955 - type: precision_at_1000 value: 0.243 - type: precision_at_3 value: 27.828999999999997 - type: precision_at_5 value: 20.061999999999998 - type: recall_at_1 value: 22.701 - type: recall_at_10 value: 51.432 - type: recall_at_100 value: 77.009 - type: recall_at_1000 value: 92.511 - type: recall_at_3 value: 37.919000000000004 - type: recall_at_5 value: 44.131 - task: type: Retrieval dataset: type: hotpotqa name: MTEB HotpotQA config: default split: test revision: None metrics: - type: map_at_1 value: 40.189 - type: map_at_10 value: 66.24600000000001 - type: map_at_100 value: 67.098 - type: map_at_1000 value: 67.149 - type: map_at_3 value: 62.684 - type: map_at_5 value: 64.974 - type: mrr_at_1 value: 80.378 - type: mrr_at_10 value: 86.127 - type: mrr_at_100 value: 86.29299999999999 - type: mrr_at_1000 value: 86.297 - type: mrr_at_3 value: 85.31400000000001 - type: mrr_at_5 value: 85.858 - type: ndcg_at_1 value: 80.378 - type: ndcg_at_10 value: 74.101 - type: ndcg_at_100 value: 76.993 - type: ndcg_at_1000 value: 77.948 - type: ndcg_at_3 value: 69.232 - type: ndcg_at_5 value: 72.04599999999999 - type: precision_at_1 value: 80.378 - type: precision_at_10 value: 15.595999999999998 - type: precision_at_100 value: 1.7840000000000003 - type: precision_at_1000 value: 0.191 - type: precision_at_3 value: 44.884 - type: precision_at_5 value: 29.145 - type: recall_at_1 value: 40.189 - type: recall_at_10 value: 77.981 - type: recall_at_100 value: 89.21 - type: recall_at_1000 value: 95.48299999999999 - type: recall_at_3 value: 67.326 - type: recall_at_5 value: 72.863 - task: type: Classification dataset: type: mteb/imdb name: MTEB ImdbClassification config: default split: test revision: 3d86128a09e091d6018b6d26cad27f2739fc2db7 metrics: - type: accuracy value: 92.84599999999999 - type: ap value: 89.4710787567357 - type: f1 value: 92.83752676932258 - task: type: Retrieval dataset: type: msmarco name: MTEB MSMARCO config: default split: dev revision: None metrics: - type: map_at_1 value: 23.132 - type: map_at_10 value: 35.543 - type: map_at_100 value: 36.702 - type: map_at_1000 value: 36.748999999999995 - type: map_at_3 value: 31.737 - type: map_at_5 value: 33.927 - type: mrr_at_1 value: 23.782 - type: mrr_at_10 value: 36.204 - type: mrr_at_100 value: 37.29 - type: mrr_at_1000 value: 37.330999999999996 - type: mrr_at_3 value: 32.458999999999996 - type: mrr_at_5 value: 34.631 - type: ndcg_at_1 value: 23.782 - type: ndcg_at_10 value: 42.492999999999995 - type: ndcg_at_100 value: 47.985 - type: ndcg_at_1000 value: 49.141 - type: ndcg_at_3 value: 34.748000000000005 - type: ndcg_at_5 value: 38.651 - type: precision_at_1 value: 23.782 - type: precision_at_10 value: 6.665 - type: precision_at_100 value: 0.941 - type: precision_at_1000 value: 0.104 - type: precision_at_3 value: 14.776 - type: precision_at_5 value: 10.84 - type: recall_at_1 value: 23.132 - type: recall_at_10 value: 63.794 - type: recall_at_100 value: 89.027 - type: recall_at_1000 value: 97.807 - type: recall_at_3 value: 42.765 - type: recall_at_5 value: 52.11 - task: type: Classification dataset: type: mteb/mtop_domain name: MTEB MTOPDomainClassification (en) config: en split: test revision: d80d48c1eb48d3562165c59d59d0034df9fff0bf metrics: - type: accuracy value: 94.59188326493388 - type: f1 value: 94.3842594786827 - task: type: Classification dataset: type: mteb/mtop_intent name: MTEB MTOPIntentClassification (en) config: en split: test revision: ae001d0e6b1228650b7bd1c2c65fb50ad11a8aba metrics: - type: accuracy value: 79.49384404924761 - type: f1 value: 59.7580539534629 - task: type: Classification dataset: type: mteb/amazon_massive_intent name: MTEB MassiveIntentClassification (en) config: en split: test revision: 31efe3c427b0bae9c22cbb560b8f15491cc6bed7 metrics: - type: accuracy value: 77.56220578345663 - type: f1 value: 75.27228165561478 - task: type: Classification dataset: type: mteb/amazon_massive_scenario name: MTEB MassiveScenarioClassification (en) config: en split: test revision: 7d571f92784cd94a019292a1f45445077d0ef634 metrics: - type: accuracy value: 80.53463349024884 - type: f1 value: 80.4893958236536 - task: type: Clustering dataset: type: mteb/medrxiv-clustering-p2p name: MTEB MedrxivClusteringP2P config: default split: test revision: e7a26af6f3ae46b30dde8737f02c07b1505bcc73 metrics: - type: v_measure value: 32.56100273484962 - task: type: Clustering dataset: type: mteb/medrxiv-clustering-s2s name: MTEB MedrxivClusteringS2S config: default split: test revision: 35191c8c0dca72d8ff3efcd72aa802307d469663 metrics: - type: v_measure value: 31.470380028839607 - task: type: Reranking dataset: type: mteb/mind_small name: MTEB MindSmallReranking config: default split: test revision: 3bdac13927fdc888b903db93b2ffdbd90b295a69 metrics: - type: map value: 32.06102792457849 - type: mrr value: 33.30709199672238 - task: type: Retrieval dataset: type: nfcorpus name: MTEB NFCorpus config: default split: test revision: None metrics: - type: map_at_1 value: 6.776999999999999 - type: map_at_10 value: 14.924000000000001 - type: map_at_100 value: 18.955 - type: map_at_1000 value: 20.538999999999998 - type: map_at_3 value: 10.982 - type: map_at_5 value: 12.679000000000002 - type: mrr_at_1 value: 47.988 - type: mrr_at_10 value: 57.232000000000006 - type: mrr_at_100 value: 57.818999999999996 - type: mrr_at_1000 value: 57.847 - type: mrr_at_3 value: 54.901999999999994 - type: mrr_at_5 value: 56.481 - type: ndcg_at_1 value: 46.594 - type: ndcg_at_10 value: 38.129000000000005 - type: ndcg_at_100 value: 35.54 - type: ndcg_at_1000 value: 44.172 - type: ndcg_at_3 value: 43.025999999999996 - type: ndcg_at_5 value: 41.052 - type: precision_at_1 value: 47.988 - type: precision_at_10 value: 28.111000000000004 - type: precision_at_100 value: 8.929 - type: precision_at_1000 value: 2.185 - type: precision_at_3 value: 40.144000000000005 - type: precision_at_5 value: 35.232 - type: recall_at_1 value: 6.776999999999999 - type: recall_at_10 value: 19.289 - type: recall_at_100 value: 36.359 - type: recall_at_1000 value: 67.54 - type: recall_at_3 value: 11.869 - type: recall_at_5 value: 14.999 - task: type: Retrieval dataset: type: nq name: MTEB NQ config: default split: test revision: None metrics: - type: map_at_1 value: 31.108000000000004 - type: map_at_10 value: 47.126000000000005 - type: map_at_100 value: 48.171 - type: map_at_1000 value: 48.199 - type: map_at_3 value: 42.734 - type: map_at_5 value: 45.362 - type: mrr_at_1 value: 34.936 - type: mrr_at_10 value: 49.571 - type: mrr_at_100 value: 50.345 - type: mrr_at_1000 value: 50.363 - type: mrr_at_3 value: 45.959 - type: mrr_at_5 value: 48.165 - type: ndcg_at_1 value: 34.936 - type: ndcg_at_10 value: 55.028999999999996 - type: ndcg_at_100 value: 59.244 - type: ndcg_at_1000 value: 59.861 - type: ndcg_at_3 value: 46.872 - type: ndcg_at_5 value: 51.217999999999996 - type: precision_at_1 value: 34.936 - type: precision_at_10 value: 9.099 - type: precision_at_100 value: 1.145 - type: precision_at_1000 value: 0.12 - type: precision_at_3 value: 21.456 - type: precision_at_5 value: 15.411 - type: recall_at_1 value: 31.108000000000004 - type: recall_at_10 value: 76.53999999999999 - type: recall_at_100 value: 94.39 - type: recall_at_1000 value: 98.947 - type: recall_at_3 value: 55.572 - type: recall_at_5 value: 65.525 - task: type: Retrieval dataset: type: quora name: MTEB QuoraRetrieval config: default split: test revision: None metrics: - type: map_at_1 value: 71.56400000000001 - type: map_at_10 value: 85.482 - type: map_at_100 value: 86.114 - type: map_at_1000 value: 86.13 - type: map_at_3 value: 82.607 - type: map_at_5 value: 84.405 - type: mrr_at_1 value: 82.42 - type: mrr_at_10 value: 88.304 - type: mrr_at_100 value: 88.399 - type: mrr_at_1000 value: 88.399 - type: mrr_at_3 value: 87.37 - type: mrr_at_5 value: 88.024 - type: ndcg_at_1 value: 82.45 - type: ndcg_at_10 value: 89.06500000000001 - type: ndcg_at_100 value: 90.232 - type: ndcg_at_1000 value: 90.305 - type: ndcg_at_3 value: 86.375 - type: ndcg_at_5 value: 87.85300000000001 - type: precision_at_1 value: 82.45 - type: precision_at_10 value: 13.486999999999998 - type: precision_at_100 value: 1.534 - type: precision_at_1000 value: 0.157 - type: precision_at_3 value: 37.813 - type: precision_at_5 value: 24.773999999999997 - type: recall_at_1 value: 71.56400000000001 - type: recall_at_10 value: 95.812 - type: recall_at_100 value: 99.7 - type: recall_at_1000 value: 99.979 - type: recall_at_3 value: 87.966 - type: recall_at_5 value: 92.268 - task: type: Clustering dataset: type: mteb/reddit-clustering name: MTEB RedditClustering config: default split: test revision: 24640382cdbf8abc73003fb0fa6d111a705499eb metrics: - type: v_measure value: 57.241876648614145 - task: type: Clustering dataset: type: mteb/reddit-clustering-p2p name: MTEB RedditClusteringP2P config: default split: test revision: 282350215ef01743dc01b456c7f5241fa8937f16 metrics: - type: v_measure value: 64.66212576446223 - task: type: Retrieval dataset: type: scidocs name: MTEB SCIDOCS config: default split: test revision: None metrics: - type: map_at_1 value: 5.308 - type: map_at_10 value: 13.803 - type: map_at_100 value: 16.176 - type: map_at_1000 value: 16.561 - type: map_at_3 value: 9.761000000000001 - type: map_at_5 value: 11.802 - type: mrr_at_1 value: 26.200000000000003 - type: mrr_at_10 value: 37.621 - type: mrr_at_100 value: 38.767 - type: mrr_at_1000 value: 38.815 - type: mrr_at_3 value: 34.117 - type: mrr_at_5 value: 36.107 - type: ndcg_at_1 value: 26.200000000000003 - type: ndcg_at_10 value: 22.64 - type: ndcg_at_100 value: 31.567 - type: ndcg_at_1000 value: 37.623 - type: ndcg_at_3 value: 21.435000000000002 - type: ndcg_at_5 value: 18.87 - type: precision_at_1 value: 26.200000000000003 - type: precision_at_10 value: 11.74 - type: precision_at_100 value: 2.465 - type: precision_at_1000 value: 0.391 - type: precision_at_3 value: 20.033 - type: precision_at_5 value: 16.64 - type: recall_at_1 value: 5.308 - type: recall_at_10 value: 23.794999999999998 - type: recall_at_100 value: 50.015 - type: recall_at_1000 value: 79.283 - type: recall_at_3 value: 12.178 - type: recall_at_5 value: 16.882 - task: type: STS dataset: type: mteb/sickr-sts name: MTEB SICK-R config: default split: test revision: a6ea5a8cab320b040a23452cc28066d9beae2cee metrics: - type: cos_sim_pearson value: 84.93231134675553 - type: cos_sim_spearman value: 81.68319292603205 - type: euclidean_pearson value: 81.8396814380367 - type: euclidean_spearman value: 81.24641903349945 - type: manhattan_pearson value: 81.84698799204274 - type: manhattan_spearman value: 81.24269997904105 - task: type: STS dataset: type: mteb/sts12-sts name: MTEB STS12 config: default split: test revision: a0d554a64d88156834ff5ae9920b964011b16384 metrics: - type: cos_sim_pearson value: 86.73241671587446 - type: cos_sim_spearman value: 79.05091082971826 - type: euclidean_pearson value: 83.91146869578044 - type: euclidean_spearman value: 79.87978465370936 - type: manhattan_pearson value: 83.90888338917678 - type: manhattan_spearman value: 79.87482848584241 - task: type: STS dataset: type: mteb/sts13-sts name: MTEB STS13 config: default split: test revision: 7e90230a92c190f1bf69ae9002b8cea547a64cca metrics: - type: cos_sim_pearson value: 85.14970731146177 - type: cos_sim_spearman value: 86.37363490084627 - type: euclidean_pearson value: 83.02154218530433 - type: euclidean_spearman value: 83.80258761957367 - type: manhattan_pearson value: 83.01664495119347 - type: manhattan_spearman value: 83.77567458007952 - task: type: STS dataset: type: mteb/sts14-sts name: MTEB STS14 config: default split: test revision: 6031580fec1f6af667f0bd2da0a551cf4f0b2375 metrics: - type: cos_sim_pearson value: 83.40474139886784 - type: cos_sim_spearman value: 82.77768789165984 - type: euclidean_pearson value: 80.7065877443695 - type: euclidean_spearman value: 81.375940662505 - type: manhattan_pearson value: 80.6507552270278 - type: manhattan_spearman value: 81.32782179098741 - task: type: STS dataset: type: mteb/sts15-sts name: MTEB STS15 config: default split: test revision: ae752c7c21bf194d8b67fd573edf7ae58183cbe3 metrics: - type: cos_sim_pearson value: 87.08585968722274 - type: cos_sim_spearman value: 88.03110031451399 - type: euclidean_pearson value: 85.74012019602384 - type: euclidean_spearman value: 86.13592849438209 - type: manhattan_pearson value: 85.74404842369206 - type: manhattan_spearman value: 86.14492318960154 - task: type: STS dataset: type: mteb/sts16-sts name: MTEB STS16 config: default split: test revision: 4d8694f8f0e0100860b497b999b3dbed754a0513 metrics: - type: cos_sim_pearson value: 84.95069052788875 - type: cos_sim_spearman value: 86.4867991595147 - type: euclidean_pearson value: 84.31013325754635 - type: euclidean_spearman value: 85.01529258006482 - type: manhattan_pearson value: 84.26995570085374 - type: manhattan_spearman value: 84.96982104986162 - task: type: STS dataset: type: mteb/sts17-crosslingual-sts name: MTEB STS17 (en-en) config: en-en split: test revision: af5e6fb845001ecf41f4c1e033ce921939a2a68d metrics: - type: cos_sim_pearson value: 87.54617647971897 - type: cos_sim_spearman value: 87.49834181751034 - type: euclidean_pearson value: 86.01015322577122 - type: euclidean_spearman value: 84.63362652063199 - type: manhattan_pearson value: 86.13807574475706 - type: manhattan_spearman value: 84.7772370721132 - task: type: STS dataset: type: mteb/sts22-crosslingual-sts name: MTEB STS22 (en) config: en split: test revision: 6d1ba47164174a496b7fa5d3569dae26a6813b80 metrics: - type: cos_sim_pearson value: 67.20047755786615 - type: cos_sim_spearman value: 67.05324077987636 - type: euclidean_pearson value: 66.91930642976601 - type: euclidean_spearman value: 65.21491856099105 - type: manhattan_pearson value: 66.78756851976624 - type: manhattan_spearman value: 65.12356257740728 - task: type: STS dataset: type: mteb/stsbenchmark-sts name: MTEB STSBenchmark config: default split: test revision: b0fddb56ed78048fa8b90373c8a3cfc37b684831 metrics: - type: cos_sim_pearson value: 86.19852871539686 - type: cos_sim_spearman value: 87.5161895296395 - type: euclidean_pearson value: 84.59848645207485 - type: euclidean_spearman value: 85.26427328757919 - type: manhattan_pearson value: 84.59747366996524 - type: manhattan_spearman value: 85.24045855146915 - task: type: Reranking dataset: type: mteb/scidocs-reranking name: MTEB SciDocsRR config: default split: test revision: d3c5e1fc0b855ab6097bf1cda04dd73947d7caab metrics: - type: map value: 87.63320317811032 - type: mrr value: 96.26242947321379 - task: type: Retrieval dataset: type: scifact name: MTEB SciFact config: default split: test revision: None metrics: - type: map_at_1 value: 60.928000000000004 - type: map_at_10 value: 70.112 - type: map_at_100 value: 70.59299999999999 - type: map_at_1000 value: 70.623 - type: map_at_3 value: 66.846 - type: map_at_5 value: 68.447 - type: mrr_at_1 value: 64.0 - type: mrr_at_10 value: 71.212 - type: mrr_at_100 value: 71.616 - type: mrr_at_1000 value: 71.64500000000001 - type: mrr_at_3 value: 68.77799999999999 - type: mrr_at_5 value: 70.094 - type: ndcg_at_1 value: 64.0 - type: ndcg_at_10 value: 74.607 - type: ndcg_at_100 value: 76.416 - type: ndcg_at_1000 value: 77.102 - type: ndcg_at_3 value: 69.126 - type: ndcg_at_5 value: 71.41300000000001 - type: precision_at_1 value: 64.0 - type: precision_at_10 value: 9.933 - type: precision_at_100 value: 1.077 - type: precision_at_1000 value: 0.11299999999999999 - type: precision_at_3 value: 26.556 - type: precision_at_5 value: 17.467 - type: recall_at_1 value: 60.928000000000004 - type: recall_at_10 value: 87.322 - type: recall_at_100 value: 94.833 - type: recall_at_1000 value: 100.0 - type: recall_at_3 value: 72.628 - type: recall_at_5 value: 78.428 - task: type: PairClassification dataset: type: mteb/sprintduplicatequestions-pairclassification name: MTEB SprintDuplicateQuestions config: default split: test revision: d66bd1f72af766a5cc4b0ca5e00c162f89e8cc46 metrics: - type: cos_sim_accuracy value: 99.86237623762376 - type: cos_sim_ap value: 96.72586477206649 - type: cos_sim_f1 value: 93.01858362631845 - type: cos_sim_precision value: 93.4409687184662 - type: cos_sim_recall value: 92.60000000000001 - type: dot_accuracy value: 99.78019801980199 - type: dot_ap value: 93.72748205246228 - type: dot_f1 value: 89.04109589041096 - type: dot_precision value: 87.16475095785441 - type: dot_recall value: 91.0 - type: euclidean_accuracy value: 99.85445544554456 - type: euclidean_ap value: 96.6661459876145 - type: euclidean_f1 value: 92.58337481333997 - type: euclidean_precision value: 92.17046580773042 - type: euclidean_recall value: 93.0 - type: manhattan_accuracy value: 99.85445544554456 - type: manhattan_ap value: 96.6883549244056 - type: manhattan_f1 value: 92.57598405580468 - type: manhattan_precision value: 92.25422045680239 - type: manhattan_recall value: 92.9 - type: max_accuracy value: 99.86237623762376 - type: max_ap value: 96.72586477206649 - type: max_f1 value: 93.01858362631845 - task: type: Clustering dataset: type: mteb/stackexchange-clustering name: MTEB StackExchangeClustering config: default split: test revision: 6cbc1f7b2bc0622f2e39d2c77fa502909748c259 metrics: - type: v_measure value: 66.39930057069995 - task: type: Clustering dataset: type: mteb/stackexchange-clustering-p2p name: MTEB StackExchangeClusteringP2P config: default split: test revision: 815ca46b2622cec33ccafc3735d572c266efdb44 metrics: - type: v_measure value: 34.96398659903402 - task: type: Reranking dataset: type: mteb/stackoverflowdupquestions-reranking name: MTEB StackOverflowDupQuestions config: default split: test revision: e185fbe320c72810689fc5848eb6114e1ef5ec69 metrics: - type: map value: 55.946944700355395 - type: mrr value: 56.97151398438164 - task: type: Summarization dataset: type: mteb/summeval name: MTEB SummEval config: default split: test revision: cda12ad7615edc362dbf25a00fdd61d3b1eaf93c metrics: - type: cos_sim_pearson value: 31.541657650692905 - type: cos_sim_spearman value: 31.605804192286303 - type: dot_pearson value: 28.26905996736398 - type: dot_spearman value: 27.864801765851187 - task: type: Retrieval dataset: type: trec-covid name: MTEB TRECCOVID config: default split: test revision: None metrics: - type: map_at_1 value: 0.22599999999999998 - type: map_at_10 value: 1.8870000000000002 - type: map_at_100 value: 9.78 - type: map_at_1000 value: 22.514 - type: map_at_3 value: 0.6669999999999999 - type: map_at_5 value: 1.077 - type: mrr_at_1 value: 82.0 - type: mrr_at_10 value: 89.86699999999999 - type: mrr_at_100 value: 89.86699999999999 - type: mrr_at_1000 value: 89.86699999999999 - type: mrr_at_3 value: 89.667 - type: mrr_at_5 value: 89.667 - type: ndcg_at_1 value: 79.0 - type: ndcg_at_10 value: 74.818 - type: ndcg_at_100 value: 53.715999999999994 - type: ndcg_at_1000 value: 47.082 - type: ndcg_at_3 value: 82.134 - type: ndcg_at_5 value: 79.81899999999999 - type: precision_at_1 value: 82.0 - type: precision_at_10 value: 78.0 - type: precision_at_100 value: 54.48 - type: precision_at_1000 value: 20.518 - type: precision_at_3 value: 87.333 - type: precision_at_5 value: 85.2 - type: recall_at_1 value: 0.22599999999999998 - type: recall_at_10 value: 2.072 - type: recall_at_100 value: 13.013 - type: recall_at_1000 value: 43.462 - type: recall_at_3 value: 0.695 - type: recall_at_5 value: 1.139 - task: type: Retrieval dataset: type: webis-touche2020 name: MTEB Touche2020 config: default split: test revision: None metrics: - type: map_at_1 value: 2.328 - type: map_at_10 value: 9.795 - type: map_at_100 value: 15.801000000000002 - type: map_at_1000 value: 17.23 - type: map_at_3 value: 4.734 - type: map_at_5 value: 6.644 - type: mrr_at_1 value: 30.612000000000002 - type: mrr_at_10 value: 46.902 - type: mrr_at_100 value: 47.495 - type: mrr_at_1000 value: 47.495 - type: mrr_at_3 value: 41.156 - type: mrr_at_5 value: 44.218 - type: ndcg_at_1 value: 28.571 - type: ndcg_at_10 value: 24.806 - type: ndcg_at_100 value: 36.419000000000004 - type: ndcg_at_1000 value: 47.272999999999996 - type: ndcg_at_3 value: 25.666 - type: ndcg_at_5 value: 25.448999999999998 - type: precision_at_1 value: 30.612000000000002 - type: precision_at_10 value: 23.061 - type: precision_at_100 value: 7.714 - type: precision_at_1000 value: 1.484 - type: precision_at_3 value: 26.531 - type: precision_at_5 value: 26.122 - type: recall_at_1 value: 2.328 - type: recall_at_10 value: 16.524 - type: recall_at_100 value: 47.179 - type: recall_at_1000 value: 81.22200000000001 - type: recall_at_3 value: 5.745 - type: recall_at_5 value: 9.339 - task: type: Classification dataset: type: mteb/toxic_conversations_50k name: MTEB ToxicConversationsClassification config: default split: test revision: d7c0de2777da35d6aae2200a62c6e0e5af397c4c metrics: - type: accuracy value: 70.9142 - type: ap value: 14.335574772555415 - type: f1 value: 54.62839595194111 - task: type: Classification dataset: type: mteb/tweet_sentiment_extraction name: MTEB TweetSentimentExtractionClassification config: default split: test revision: d604517c81ca91fe16a244d1248fc021f9ecee7a metrics: - type: accuracy value: 59.94340690435768 - type: f1 value: 60.286487936731916 - task: type: Clustering dataset: type: mteb/twentynewsgroups-clustering name: MTEB TwentyNewsgroupsClustering config: default split: test revision: 6125ec4e24fa026cec8a478383ee943acfbd5449 metrics: - type: v_measure value: 51.26597708987974 - task: type: PairClassification dataset: type: mteb/twittersemeval2015-pairclassification name: MTEB TwitterSemEval2015 config: default split: test revision: 70970daeab8776df92f5ea462b6173c0b46fd2d1 metrics: - type: cos_sim_accuracy value: 87.48882398521786 - type: cos_sim_ap value: 79.04326607602204 - type: cos_sim_f1 value: 71.64566826860633 - type: cos_sim_precision value: 70.55512918905092 - type: cos_sim_recall value: 72.77044854881267 - type: dot_accuracy value: 84.19264469213805 - type: dot_ap value: 67.96360043562528 - type: dot_f1 value: 64.06418393006827 - type: dot_precision value: 58.64941898706424 - type: dot_recall value: 70.58047493403694 - type: euclidean_accuracy value: 87.45902127913214 - type: euclidean_ap value: 78.9742237648272 - type: euclidean_f1 value: 71.5553235908142 - type: euclidean_precision value: 70.77955601445535 - type: euclidean_recall value: 72.34828496042216 - type: manhattan_accuracy value: 87.41729749061214 - type: manhattan_ap value: 78.90073137580596 - type: manhattan_f1 value: 71.3942611553533 - type: manhattan_precision value: 68.52705653967483 - type: manhattan_recall value: 74.51187335092348 - type: max_accuracy value: 87.48882398521786 - type: max_ap value: 79.04326607602204 - type: max_f1 value: 71.64566826860633 - task: type: PairClassification dataset: type: mteb/twitterurlcorpus-pairclassification name: MTEB TwitterURLCorpus config: default split: test revision: 8b6510b0b1fa4e4c4f879467980e9be563ec1cdf metrics: - type: cos_sim_accuracy value: 88.68125897465751 - type: cos_sim_ap value: 85.6003454431979 - type: cos_sim_f1 value: 77.6957163958641 - type: cos_sim_precision value: 73.0110366307807 - type: cos_sim_recall value: 83.02279026793964 - type: dot_accuracy value: 87.7672992587418 - type: dot_ap value: 82.4971301112899 - type: dot_f1 value: 75.90528233151184 - type: dot_precision value: 72.0370626469368 - type: dot_recall value: 80.21250384970742 - type: euclidean_accuracy value: 88.4503434625684 - type: euclidean_ap value: 84.91949884748384 - type: euclidean_f1 value: 76.92365018444684 - type: euclidean_precision value: 74.53245721712759 - type: euclidean_recall value: 79.47336002463813 - type: manhattan_accuracy value: 88.47556952691427 - type: manhattan_ap value: 84.8963689101517 - type: manhattan_f1 value: 76.85901249256395 - type: manhattan_precision value: 74.31693989071039 - type: manhattan_recall value: 79.58115183246073 - type: max_accuracy value: 88.68125897465751 - type: max_ap value: 85.6003454431979 - type: max_f1 value: 77.6957163958641 license: mit language: - en --- <h1 align="center">FlagEmbedding</h1> <h4 align="center"> <p> <a href=#model-list>Model List</a> | <a href=#frequently-asked-questions>FAQ</a> | <a href=#usage>Usage</a> | <a href="#evaluation">Evaluation</a> | <a href="#train">Train</a> | <a href="#contact">Contact</a> | <a href="#citation">Citation</a> | <a href="#license">License</a> <p> </h4> More details please refer to our Github: [FlagEmbedding](https://github.com/FlagOpen/FlagEmbedding). [English](README.md) | [中文](https://github.com/FlagOpen/FlagEmbedding/blob/master/README_zh.md) FlagEmbedding can map any text to a low-dimensional dense vector which can be used for tasks like retrieval, classification, clustering, or semantic search. And it also can be used in vector databases for LLMs. ************* 🌟**Updates**🌟 ************* - 10/12/2023: Release [LLM-Embedder](./FlagEmbedding/llm_embedder/README.md), a unified embedding model to support diverse retrieval augmentation needs for LLMs. [Paper](https://arxiv.org/pdf/2310.07554.pdf) :fire: - 09/15/2023: The [technical report](https://arxiv.org/pdf/2309.07597.pdf) of BGE has been released - 09/15/2023: The [masive training data](https://data.baai.ac.cn/details/BAAI-MTP) of BGE has been released - 09/12/2023: New models: - **New reranker model**: release cross-encoder models `BAAI/bge-reranker-base` and `BAAI/bge-reranker-large`, which are more powerful than embedding model. We recommend to use/fine-tune them to re-rank top-k documents returned by embedding models. - **update embedding model**: release `bge-*-v1.5` embedding model to alleviate the issue of the similarity distribution, and enhance its retrieval ability without instruction. <details> <summary>More</summary> <!-- ### More --> - 09/07/2023: Update [fine-tune code](https://github.com/FlagOpen/FlagEmbedding/blob/master/FlagEmbedding/baai_general_embedding/README.md): Add script to mine hard negatives and support adding instruction during fine-tuning. - 08/09/2023: BGE Models are integrated into **Langchain**, you can use it like [this](#using-langchain); C-MTEB **leaderboard** is [available](https://huggingface.co/spaces/mteb/leaderboard). - 08/05/2023: Release base-scale and small-scale models, **best performance among the models of the same size 🤗** - 08/02/2023: Release `bge-large-*`(short for BAAI General Embedding) Models, **rank 1st on MTEB and C-MTEB benchmark!** :tada: :tada: - 08/01/2023: We release the [Chinese Massive Text Embedding Benchmark](https://github.com/FlagOpen/FlagEmbedding/blob/master/C_MTEB) (**C-MTEB**), consisting of 31 test dataset. </details> ## Model List `bge` is short for `BAAI general embedding`. | Model | Language | | Description | query instruction for retrieval [1] | |:-------------------------------|:--------:| :--------:| :--------:|:--------:| | [BAAI/llm-embedder](https://huggingface.co/BAAI/llm-embedder) | English | [Inference](./FlagEmbedding/llm_embedder/README.md) [Fine-tune](./FlagEmbedding/llm_embedder/README.md) | a unified embedding model to support diverse retrieval augmentation needs for LLMs | See [README](./FlagEmbedding/llm_embedder/README.md) | | [BAAI/bge-reranker-large](https://huggingface.co/BAAI/bge-reranker-large) | Chinese and English | [Inference](#usage-for-reranker) [Fine-tune](https://github.com/FlagOpen/FlagEmbedding/tree/master/examples/reranker) | a cross-encoder model which is more accurate but less efficient [2] | | | [BAAI/bge-reranker-base](https://huggingface.co/BAAI/bge-reranker-base) | Chinese and English | [Inference](#usage-for-reranker) [Fine-tune](https://github.com/FlagOpen/FlagEmbedding/tree/master/examples/reranker) | a cross-encoder model which is more accurate but less efficient [2] | | | [BAAI/bge-large-en-v1.5](https://huggingface.co/BAAI/bge-large-en-v1.5) | English | [Inference](#usage-for-embedding-model) [Fine-tune](https://github.com/FlagOpen/FlagEmbedding/tree/master/examples/finetune) | version 1.5 with more reasonable similarity distribution | `Represent this sentence for searching relevant passages: ` | | [BAAI/bge-base-en-v1.5](https://huggingface.co/BAAI/bge-base-en-v1.5) | English | [Inference](#usage-for-embedding-model) [Fine-tune](https://github.com/FlagOpen/FlagEmbedding/tree/master/examples/finetune) | version 1.5 with more reasonable similarity distribution | `Represent this sentence for searching relevant passages: ` | | [BAAI/bge-small-en-v1.5](https://huggingface.co/BAAI/bge-small-en-v1.5) | English | [Inference](#usage-for-embedding-model) [Fine-tune](https://github.com/FlagOpen/FlagEmbedding/tree/master/examples/finetune) | version 1.5 with more reasonable similarity distribution | `Represent this sentence for searching relevant passages: ` | | [BAAI/bge-large-zh-v1.5](https://huggingface.co/BAAI/bge-large-zh-v1.5) | Chinese | [Inference](#usage-for-embedding-model) [Fine-tune](https://github.com/FlagOpen/FlagEmbedding/tree/master/examples/finetune) | version 1.5 with more reasonable similarity distribution | `为这个句子生成表示以用于检索相关文章:` | | [BAAI/bge-base-zh-v1.5](https://huggingface.co/BAAI/bge-base-zh-v1.5) | Chinese | [Inference](#usage-for-embedding-model) [Fine-tune](https://github.com/FlagOpen/FlagEmbedding/tree/master/examples/finetune) | version 1.5 with more reasonable similarity distribution | `为这个句子生成表示以用于检索相关文章:` | | [BAAI/bge-small-zh-v1.5](https://huggingface.co/BAAI/bge-small-zh-v1.5) | Chinese | [Inference](#usage-for-embedding-model) [Fine-tune](https://github.com/FlagOpen/FlagEmbedding/tree/master/examples/finetune) | version 1.5 with more reasonable similarity distribution | `为这个句子生成表示以用于检索相关文章:` | | [BAAI/bge-large-en](https://huggingface.co/BAAI/bge-large-en) | English | [Inference](#usage-for-embedding-model) [Fine-tune](https://github.com/FlagOpen/FlagEmbedding/tree/master/examples/finetune) | :trophy: rank **1st** in [MTEB](https://huggingface.co/spaces/mteb/leaderboard) leaderboard | `Represent this sentence for searching relevant passages: ` | | [BAAI/bge-base-en](https://huggingface.co/BAAI/bge-base-en) | English | [Inference](#usage-for-embedding-model) [Fine-tune](https://github.com/FlagOpen/FlagEmbedding/tree/master/examples/finetune) | a base-scale model but with similar ability to `bge-large-en` | `Represent this sentence for searching relevant passages: ` | | [BAAI/bge-small-en](https://huggingface.co/BAAI/bge-small-en) | English | [Inference](#usage-for-embedding-model) [Fine-tune](https://github.com/FlagOpen/FlagEmbedding/tree/master/examples/finetune) |a small-scale model but with competitive performance | `Represent this sentence for searching relevant passages: ` | | [BAAI/bge-large-zh](https://huggingface.co/BAAI/bge-large-zh) | Chinese | [Inference](#usage-for-embedding-model) [Fine-tune](https://github.com/FlagOpen/FlagEmbedding/tree/master/examples/finetune) | :trophy: rank **1st** in [C-MTEB](https://github.com/FlagOpen/FlagEmbedding/tree/master/C_MTEB) benchmark | `为这个句子生成表示以用于检索相关文章:` | | [BAAI/bge-base-zh](https://huggingface.co/BAAI/bge-base-zh) | Chinese | [Inference](#usage-for-embedding-model) [Fine-tune](https://github.com/FlagOpen/FlagEmbedding/tree/master/examples/finetune) | a base-scale model but with similar ability to `bge-large-zh` | `为这个句子生成表示以用于检索相关文章:` | | [BAAI/bge-small-zh](https://huggingface.co/BAAI/bge-small-zh) | Chinese | [Inference](#usage-for-embedding-model) [Fine-tune](https://github.com/FlagOpen/FlagEmbedding/tree/master/examples/finetune) | a small-scale model but with competitive performance | `为这个句子生成表示以用于检索相关文章:` | [1\]: If you need to search the relevant passages to a query, we suggest to add the instruction to the query; in other cases, no instruction is needed, just use the original query directly. In all cases, **no instruction** needs to be added to passages. [2\]: Different from embedding model, reranker uses question and document as input and directly output similarity instead of embedding. To balance the accuracy and time cost, cross-encoder is widely used to re-rank top-k documents retrieved by other simple models. For examples, use bge embedding model to retrieve top 100 relevant documents, and then use bge reranker to re-rank the top 100 document to get the final top-3 results. All models have been uploaded to Huggingface Hub, and you can see them at https://huggingface.co/BAAI. If you cannot open the Huggingface Hub, you also can download the models at https://model.baai.ac.cn/models . ## Frequently asked questions <details> <summary>1. How to fine-tune bge embedding model?</summary> <!-- ### How to fine-tune bge embedding model? --> Following this [example](https://github.com/FlagOpen/FlagEmbedding/tree/master/examples/finetune) to prepare data and fine-tune your model. Some suggestions: - Mine hard negatives following this [example](https://github.com/FlagOpen/FlagEmbedding/tree/master/examples/finetune#hard-negatives), which can improve the retrieval performance. - If you pre-train bge on your data, the pre-trained model cannot be directly used to calculate similarity, and it must be fine-tuned with contrastive learning before computing similarity. - If the accuracy of the fine-tuned model is still not high, it is recommended to use/fine-tune the cross-encoder model (bge-reranker) to re-rank top-k results. Hard negatives also are needed to fine-tune reranker. </details> <details> <summary>2. The similarity score between two dissimilar sentences is higher than 0.5</summary> <!-- ### The similarity score between two dissimilar sentences is higher than 0.5 --> **Suggest to use bge v1.5, which alleviates the issue of the similarity distribution.** Since we finetune the models by contrastive learning with a temperature of 0.01, the similarity distribution of the current BGE model is about in the interval \[0.6, 1\]. So a similarity score greater than 0.5 does not indicate that the two sentences are similar. For downstream tasks, such as passage retrieval or semantic similarity, **what matters is the relative order of the scores, not the absolute value.** If you need to filter similar sentences based on a similarity threshold, please select an appropriate similarity threshold based on the similarity distribution on your data (such as 0.8, 0.85, or even 0.9). </details> <details> <summary>3. When does the query instruction need to be used</summary> <!-- ### When does the query instruction need to be used --> For the `bge-*-v1.5`, we improve its retrieval ability when not using instruction. No instruction only has a slight degradation in retrieval performance compared with using instruction. So you can generate embedding without instruction in all cases for convenience. For a retrieval task that uses short queries to find long related documents, it is recommended to add instructions for these short queries. **The best method to decide whether to add instructions for queries is choosing the setting that achieves better performance on your task.** In all cases, the documents/passages do not need to add the instruction. </details> ## Usage ### Usage for Embedding Model Here are some examples for using `bge` models with [FlagEmbedding](#using-flagembedding), [Sentence-Transformers](#using-sentence-transformers), [Langchain](#using-langchain), or [Huggingface Transformers](#using-huggingface-transformers). #### Using FlagEmbedding ``` pip install -U FlagEmbedding ``` If it doesn't work for you, you can see [FlagEmbedding](https://github.com/FlagOpen/FlagEmbedding/blob/master/FlagEmbedding/baai_general_embedding/README.md) for more methods to install FlagEmbedding. ```python from FlagEmbedding import FlagModel sentences_1 = ["样例数据-1", "样例数据-2"] sentences_2 = ["样例数据-3", "样例数据-4"] model = FlagModel('BAAI/bge-large-zh-v1.5', query_instruction_for_retrieval="为这个句子生成表示以用于检索相关文章:", use_fp16=True) # Setting use_fp16 to True speeds up computation with a slight performance degradation embeddings_1 = model.encode(sentences_1) embeddings_2 = model.encode(sentences_2) similarity = embeddings_1 @ embeddings_2.T print(similarity) # for s2p(short query to long passage) retrieval task, suggest to use encode_queries() which will automatically add the instruction to each query # corpus in retrieval task can still use encode() or encode_corpus(), since they don't need instruction queries = ['query_1', 'query_2'] passages = ["样例文档-1", "样例文档-2"] q_embeddings = model.encode_queries(queries) p_embeddings = model.encode(passages) scores = q_embeddings @ p_embeddings.T ``` For the value of the argument `query_instruction_for_retrieval`, see [Model List](https://github.com/FlagOpen/FlagEmbedding/tree/master#model-list). By default, FlagModel will use all available GPUs when encoding. Please set `os.environ["CUDA_VISIBLE_DEVICES"]` to select specific GPUs. You also can set `os.environ["CUDA_VISIBLE_DEVICES"]=""` to make all GPUs unavailable. #### Using Sentence-Transformers You can also use the `bge` models with [sentence-transformers](https://www.SBERT.net): ``` pip install -U sentence-transformers ``` ```python from sentence_transformers import SentenceTransformer sentences_1 = ["样例数据-1", "样例数据-2"] sentences_2 = ["样例数据-3", "样例数据-4"] model = SentenceTransformer('BAAI/bge-large-zh-v1.5') embeddings_1 = model.encode(sentences_1, normalize_embeddings=True) embeddings_2 = model.encode(sentences_2, normalize_embeddings=True) similarity = embeddings_1 @ embeddings_2.T print(similarity) ``` For s2p(short query to long passage) retrieval task, each short query should start with an instruction (instructions see [Model List](https://github.com/FlagOpen/FlagEmbedding/tree/master#model-list)). But the instruction is not needed for passages. ```python from sentence_transformers import SentenceTransformer queries = ['query_1', 'query_2'] passages = ["样例文档-1", "样例文档-2"] instruction = "为这个句子生成表示以用于检索相关文章:" model = SentenceTransformer('BAAI/bge-large-zh-v1.5') q_embeddings = model.encode([instruction+q for q in queries], normalize_embeddings=True) p_embeddings = model.encode(passages, normalize_embeddings=True) scores = q_embeddings @ p_embeddings.T ``` #### Using Langchain You can use `bge` in langchain like this: ```python from langchain.embeddings import HuggingFaceBgeEmbeddings model_name = "BAAI/bge-large-en-v1.5" model_kwargs = {'device': 'cuda'} encode_kwargs = {'normalize_embeddings': True} # set True to compute cosine similarity model = HuggingFaceBgeEmbeddings( model_name=model_name, model_kwargs=model_kwargs, encode_kwargs=encode_kwargs, query_instruction="为这个句子生成表示以用于检索相关文章:" ) model.query_instruction = "为这个句子生成表示以用于检索相关文章:" ``` #### Using HuggingFace Transformers With the transformers package, you can use the model like this: First, you pass your input through the transformer model, then you select the last hidden state of the first token (i.e., [CLS]) as the sentence embedding. ```python from transformers import AutoTokenizer, AutoModel import torch # Sentences we want sentence embeddings for sentences = ["样例数据-1", "样例数据-2"] # Load model from HuggingFace Hub tokenizer = AutoTokenizer.from_pretrained('BAAI/bge-large-zh-v1.5') model = AutoModel.from_pretrained('BAAI/bge-large-zh-v1.5') model.eval() # Tokenize sentences encoded_input = tokenizer(sentences, padding=True, truncation=True, return_tensors='pt') # for s2p(short query to long passage) retrieval task, add an instruction to query (not add instruction for passages) # encoded_input = tokenizer([instruction + q for q in queries], padding=True, truncation=True, return_tensors='pt') # Compute token embeddings with torch.no_grad(): model_output = model(**encoded_input) # Perform pooling. In this case, cls pooling. sentence_embeddings = model_output[0][:, 0] # normalize embeddings sentence_embeddings = torch.nn.functional.normalize(sentence_embeddings, p=2, dim=1) print("Sentence embeddings:", sentence_embeddings) ``` ### Usage for Reranker Different from embedding model, reranker uses question and document as input and directly output similarity instead of embedding. You can get a relevance score by inputting query and passage to the reranker. The reranker is optimized based cross-entropy loss, so the relevance score is not bounded to a specific range. #### Using FlagEmbedding ``` pip install -U FlagEmbedding ``` Get relevance scores (higher scores indicate more relevance): ```python from FlagEmbedding import FlagReranker reranker = FlagReranker('BAAI/bge-reranker-large', use_fp16=True) # Setting use_fp16 to True speeds up computation with a slight performance degradation score = reranker.compute_score(['query', 'passage']) print(score) scores = reranker.compute_score([['what is panda?', 'hi'], ['what is panda?', 'The giant panda (Ailuropoda melanoleuca), sometimes called a panda bear or simply panda, is a bear species endemic to China.']]) print(scores) ``` #### Using Huggingface transformers ```python import torch from transformers import AutoModelForSequenceClassification, AutoTokenizer tokenizer = AutoTokenizer.from_pretrained('BAAI/bge-reranker-large') model = AutoModelForSequenceClassification.from_pretrained('BAAI/bge-reranker-large') model.eval() pairs = [['what is panda?', 'hi'], ['what is panda?', 'The giant panda (Ailuropoda melanoleuca), sometimes called a panda bear or simply panda, is a bear species endemic to China.']] with torch.no_grad(): inputs = tokenizer(pairs, padding=True, truncation=True, return_tensors='pt', max_length=512) scores = model(**inputs, return_dict=True).logits.view(-1, ).float() print(scores) ``` ## Evaluation `baai-general-embedding` models achieve **state-of-the-art performance on both MTEB and C-MTEB leaderboard!** For more details and evaluation tools see our [scripts](https://github.com/FlagOpen/FlagEmbedding/blob/master/C_MTEB/README.md). - **MTEB**: | Model Name | Dimension | Sequence Length | Average (56) | Retrieval (15) |Clustering (11) | Pair Classification (3) | Reranking (4) | STS (10) | Summarization (1) | Classification (12) | |:----:|:---:|:---:|:---:|:---:|:---:|:---:|:---:|:---:|:---:|:---:| | [BAAI/bge-large-en-v1.5](https://huggingface.co/BAAI/bge-large-en-v1.5) | 1024 | 512 | **64.23** | **54.29** | 46.08 | 87.12 | 60.03 | 83.11 | 31.61 | 75.97 | | [BAAI/bge-base-en-v1.5](https://huggingface.co/BAAI/bge-base-en-v1.5) | 768 | 512 | 63.55 | 53.25 | 45.77 | 86.55 | 58.86 | 82.4 | 31.07 | 75.53 | | [BAAI/bge-small-en-v1.5](https://huggingface.co/BAAI/bge-small-en-v1.5) | 384 | 512 | 62.17 |51.68 | 43.82 | 84.92 | 58.36 | 81.59 | 30.12 | 74.14 | | [bge-large-en](https://huggingface.co/BAAI/bge-large-en) | 1024 | 512 | 63.98 | 53.9 | 46.98 | 85.8 | 59.48 | 81.56 | 32.06 | 76.21 | | [bge-base-en](https://huggingface.co/BAAI/bge-base-en) | 768 | 512 | 63.36 | 53.0 | 46.32 | 85.86 | 58.7 | 81.84 | 29.27 | 75.27 | | [gte-large](https://huggingface.co/thenlper/gte-large) | 1024 | 512 | 63.13 | 52.22 | 46.84 | 85.00 | 59.13 | 83.35 | 31.66 | 73.33 | | [gte-base](https://huggingface.co/thenlper/gte-base) | 768 | 512 | 62.39 | 51.14 | 46.2 | 84.57 | 58.61 | 82.3 | 31.17 | 73.01 | | [e5-large-v2](https://huggingface.co/intfloat/e5-large-v2) | 1024| 512 | 62.25 | 50.56 | 44.49 | 86.03 | 56.61 | 82.05 | 30.19 | 75.24 | | [bge-small-en](https://huggingface.co/BAAI/bge-small-en) | 384 | 512 | 62.11 | 51.82 | 44.31 | 83.78 | 57.97 | 80.72 | 30.53 | 74.37 | | [instructor-xl](https://huggingface.co/hkunlp/instructor-xl) | 768 | 512 | 61.79 | 49.26 | 44.74 | 86.62 | 57.29 | 83.06 | 32.32 | 61.79 | | [e5-base-v2](https://huggingface.co/intfloat/e5-base-v2) | 768 | 512 | 61.5 | 50.29 | 43.80 | 85.73 | 55.91 | 81.05 | 30.28 | 73.84 | | [gte-small](https://huggingface.co/thenlper/gte-small) | 384 | 512 | 61.36 | 49.46 | 44.89 | 83.54 | 57.7 | 82.07 | 30.42 | 72.31 | | [text-embedding-ada-002](https://platform.openai.com/docs/guides/embeddings) | 1536 | 8192 | 60.99 | 49.25 | 45.9 | 84.89 | 56.32 | 80.97 | 30.8 | 70.93 | | [e5-small-v2](https://huggingface.co/intfloat/e5-base-v2) | 384 | 512 | 59.93 | 49.04 | 39.92 | 84.67 | 54.32 | 80.39 | 31.16 | 72.94 | | [sentence-t5-xxl](https://huggingface.co/sentence-transformers/sentence-t5-xxl) | 768 | 512 | 59.51 | 42.24 | 43.72 | 85.06 | 56.42 | 82.63 | 30.08 | 73.42 | | [all-mpnet-base-v2](https://huggingface.co/sentence-transformers/all-mpnet-base-v2) | 768 | 514 | 57.78 | 43.81 | 43.69 | 83.04 | 59.36 | 80.28 | 27.49 | 65.07 | | [sgpt-bloom-7b1-msmarco](https://huggingface.co/bigscience/sgpt-bloom-7b1-msmarco) | 4096 | 2048 | 57.59 | 48.22 | 38.93 | 81.9 | 55.65 | 77.74 | 33.6 | 66.19 | - **C-MTEB**: We create the benchmark C-MTEB for Chinese text embedding which consists of 31 datasets from 6 tasks. Please refer to [C_MTEB](https://github.com/FlagOpen/FlagEmbedding/blob/master/C_MTEB/README.md) for a detailed introduction. | Model | Embedding dimension | Avg | Retrieval | STS | PairClassification | Classification | Reranking | Clustering | |:-------------------------------|:--------:|:--------:|:--------:|:--------:|:--------:|:--------:|:--------:|:--------:| | [**BAAI/bge-large-zh-v1.5**](https://huggingface.co/BAAI/bge-large-zh-v1.5) | 1024 | **64.53** | 70.46 | 56.25 | 81.6 | 69.13 | 65.84 | 48.99 | | [BAAI/bge-base-zh-v1.5](https://huggingface.co/BAAI/bge-base-zh-v1.5) | 768 | 63.13 | 69.49 | 53.72 | 79.75 | 68.07 | 65.39 | 47.53 | | [BAAI/bge-small-zh-v1.5](https://huggingface.co/BAAI/bge-small-zh-v1.5) | 512 | 57.82 | 61.77 | 49.11 | 70.41 | 63.96 | 60.92 | 44.18 | | [BAAI/bge-large-zh](https://huggingface.co/BAAI/bge-large-zh) | 1024 | 64.20 | 71.53 | 54.98 | 78.94 | 68.32 | 65.11 | 48.39 | | [bge-large-zh-noinstruct](https://huggingface.co/BAAI/bge-large-zh-noinstruct) | 1024 | 63.53 | 70.55 | 53 | 76.77 | 68.58 | 64.91 | 50.01 | | [BAAI/bge-base-zh](https://huggingface.co/BAAI/bge-base-zh) | 768 | 62.96 | 69.53 | 54.12 | 77.5 | 67.07 | 64.91 | 47.63 | | [multilingual-e5-large](https://huggingface.co/intfloat/multilingual-e5-large) | 1024 | 58.79 | 63.66 | 48.44 | 69.89 | 67.34 | 56.00 | 48.23 | | [BAAI/bge-small-zh](https://huggingface.co/BAAI/bge-small-zh) | 512 | 58.27 | 63.07 | 49.45 | 70.35 | 63.64 | 61.48 | 45.09 | | [m3e-base](https://huggingface.co/moka-ai/m3e-base) | 768 | 57.10 | 56.91 | 50.47 | 63.99 | 67.52 | 59.34 | 47.68 | | [m3e-large](https://huggingface.co/moka-ai/m3e-large) | 1024 | 57.05 | 54.75 | 50.42 | 64.3 | 68.2 | 59.66 | 48.88 | | [multilingual-e5-base](https://huggingface.co/intfloat/multilingual-e5-base) | 768 | 55.48 | 61.63 | 46.49 | 67.07 | 65.35 | 54.35 | 40.68 | | [multilingual-e5-small](https://huggingface.co/intfloat/multilingual-e5-small) | 384 | 55.38 | 59.95 | 45.27 | 66.45 | 65.85 | 53.86 | 45.26 | | [text-embedding-ada-002(OpenAI)](https://platform.openai.com/docs/guides/embeddings/what-are-embeddings) | 1536 | 53.02 | 52.0 | 43.35 | 69.56 | 64.31 | 54.28 | 45.68 | | [luotuo](https://huggingface.co/silk-road/luotuo-bert-medium) | 1024 | 49.37 | 44.4 | 42.78 | 66.62 | 61 | 49.25 | 44.39 | | [text2vec-base](https://huggingface.co/shibing624/text2vec-base-chinese) | 768 | 47.63 | 38.79 | 43.41 | 67.41 | 62.19 | 49.45 | 37.66 | | [text2vec-large](https://huggingface.co/GanymedeNil/text2vec-large-chinese) | 1024 | 47.36 | 41.94 | 44.97 | 70.86 | 60.66 | 49.16 | 30.02 | - **Reranking**: See [C_MTEB](https://github.com/FlagOpen/FlagEmbedding/blob/master/C_MTEB/) for evaluation script. | Model | T2Reranking | T2RerankingZh2En\* | T2RerankingEn2Zh\* | MMarcoReranking | CMedQAv1 | CMedQAv2 | Avg | |:-------------------------------|:--------:|:--------:|:--------:|:--------:|:--------:|:--------:|:--------:| | text2vec-base-multilingual | 64.66 | 62.94 | 62.51 | 14.37 | 48.46 | 48.6 | 50.26 | | multilingual-e5-small | 65.62 | 60.94 | 56.41 | 29.91 | 67.26 | 66.54 | 57.78 | | multilingual-e5-large | 64.55 | 61.61 | 54.28 | 28.6 | 67.42 | 67.92 | 57.4 | | multilingual-e5-base | 64.21 | 62.13 | 54.68 | 29.5 | 66.23 | 66.98 | 57.29 | | m3e-base | 66.03 | 62.74 | 56.07 | 17.51 | 77.05 | 76.76 | 59.36 | | m3e-large | 66.13 | 62.72 | 56.1 | 16.46 | 77.76 | 78.27 | 59.57 | | bge-base-zh-v1.5 | 66.49 | 63.25 | 57.02 | 29.74 | 80.47 | 84.88 | 63.64 | | bge-large-zh-v1.5 | 65.74 | 63.39 | 57.03 | 28.74 | 83.45 | 85.44 | 63.97 | | [BAAI/bge-reranker-base](https://huggingface.co/BAAI/bge-reranker-base) | 67.28 | 63.95 | 60.45 | 35.46 | 81.26 | 84.1 | 65.42 | | [BAAI/bge-reranker-large](https://huggingface.co/BAAI/bge-reranker-large) | 67.6 | 64.03 | 61.44 | 37.16 | 82.15 | 84.18 | 66.09 | \* : T2RerankingZh2En and T2RerankingEn2Zh are cross-language retrieval tasks ## Train ### BAAI Embedding We pre-train the models using [retromae](https://github.com/staoxiao/RetroMAE) and train them on large-scale pairs data using contrastive learning. **You can fine-tune the embedding model on your data following our [examples](https://github.com/FlagOpen/FlagEmbedding/tree/master/examples/finetune).** We also provide a [pre-train example](https://github.com/FlagOpen/FlagEmbedding/tree/master/examples/pretrain). Note that the goal of pre-training is to reconstruct the text, and the pre-trained model cannot be used for similarity calculation directly, it needs to be fine-tuned. More training details for bge see [baai_general_embedding](https://github.com/FlagOpen/FlagEmbedding/blob/master/FlagEmbedding/baai_general_embedding/README.md). ### BGE Reranker Cross-encoder will perform full-attention over the input pair, which is more accurate than embedding model (i.e., bi-encoder) but more time-consuming than embedding model. Therefore, it can be used to re-rank the top-k documents returned by embedding model. We train the cross-encoder on a multilingual pair data, The data format is the same as embedding model, so you can fine-tune it easily following our [example](https://github.com/FlagOpen/FlagEmbedding/tree/master/examples/reranker). More details please refer to [./FlagEmbedding/reranker/README.md](https://github.com/FlagOpen/FlagEmbedding/tree/master/FlagEmbedding/reranker) ## Contact If you have any question or suggestion related to this project, feel free to open an issue or pull request. You also can email Shitao Xiao(stxiao@baai.ac.cn) and Zheng Liu(liuzheng@baai.ac.cn). ## Citation If you find this repository useful, please consider giving a star :star: and citation ``` @misc{bge_embedding, title={C-Pack: Packaged Resources To Advance General Chinese Embedding}, author={Shitao Xiao and Zheng Liu and Peitian Zhang and Niklas Muennighoff}, year={2023}, eprint={2309.07597}, archivePrefix={arXiv}, primaryClass={cs.CL} } ``` ## License FlagEmbedding is licensed under the [MIT License](https://github.com/FlagOpen/FlagEmbedding/blob/master/LICENSE). The released models can be used for commercial purposes free of charge.