mteb 2.1.7__py3-none-any.whl → 2.1.8__py3-none-any.whl

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
Files changed (239) hide show
  1. mteb/_create_dataloaders.py +6 -3
  2. mteb/_evaluators/any_sts_evaluator.py +14 -12
  3. mteb/_evaluators/clustering_evaluator.py +1 -1
  4. mteb/_evaluators/image/imagetext_pairclassification_evaluator.py +2 -2
  5. mteb/_evaluators/pair_classification_evaluator.py +3 -1
  6. mteb/_evaluators/sklearn_evaluator.py +15 -28
  7. mteb/_evaluators/text/bitext_mining_evaluator.py +4 -1
  8. mteb/_evaluators/text/summarization_evaluator.py +4 -2
  9. mteb/_evaluators/zeroshot_classification_evaluator.py +2 -2
  10. mteb/abstasks/clustering.py +1 -1
  11. mteb/abstasks/multilabel_classification.py +2 -2
  12. mteb/abstasks/task_metadata.py +1 -0
  13. mteb/benchmarks/benchmark.py +9 -0
  14. mteb/benchmarks/benchmarks/__init__.py +2 -0
  15. mteb/benchmarks/benchmarks/benchmarks.py +40 -1
  16. mteb/descriptive_stats/Image/DocumentUnderstanding/Vidore3ComputerScienceRetrieval.json +214 -0
  17. mteb/descriptive_stats/Image/DocumentUnderstanding/Vidore3EnergyRetrieval.json +214 -0
  18. mteb/descriptive_stats/Image/DocumentUnderstanding/Vidore3FinanceEnRetrieval.json +214 -0
  19. mteb/descriptive_stats/Image/DocumentUnderstanding/Vidore3FinanceFrRetrieval.json +214 -0
  20. mteb/descriptive_stats/Image/DocumentUnderstanding/Vidore3HrRetrieval.json +214 -0
  21. mteb/descriptive_stats/Image/DocumentUnderstanding/Vidore3IndustrialRetrieval.json +214 -0
  22. mteb/descriptive_stats/Image/DocumentUnderstanding/Vidore3NuclearRetrieval.json +214 -0
  23. mteb/descriptive_stats/Image/DocumentUnderstanding/Vidore3PharmaceuticalsRetrieval.json +214 -0
  24. mteb/descriptive_stats/Image/DocumentUnderstanding/Vidore3PhysicsRetrieval.json +214 -0
  25. mteb/descriptive_stats/Image/DocumentUnderstanding/Vidore3TelecomRetrieval.json +214 -0
  26. mteb/models/cache_wrappers/cache_wrapper.py +1 -1
  27. mteb/models/model_implementations/align_models.py +6 -0
  28. mteb/models/model_implementations/ara_models.py +7 -0
  29. mteb/models/model_implementations/blip2_models.py +9 -0
  30. mteb/models/model_implementations/blip_models.py +19 -0
  31. mteb/models/model_implementations/cadet_models.py +8 -0
  32. mteb/models/model_implementations/cde_models.py +12 -0
  33. mteb/models/model_implementations/codefuse_models.py +15 -0
  34. mteb/models/model_implementations/codesage_models.py +12 -0
  35. mteb/models/model_implementations/misc_models.py +6 -0
  36. mteb/models/model_implementations/moco_models.py +9 -0
  37. mteb/models/model_implementations/openclip_models.py +16 -0
  38. mteb/models/model_implementations/piccolo_models.py +6 -0
  39. mteb/models/model_implementations/rasgaard_models.py +7 -1
  40. mteb/models/model_implementations/tarka_models.py +317 -0
  41. mteb/models/search_wrappers.py +5 -5
  42. mteb/tasks/bitext_mining/multilingual/ru_sci_bench_bitext_mining.py +1 -5
  43. mteb/tasks/bitext_mining/multilingual/web_faq_bitext_mining.py +2 -6
  44. mteb/tasks/classification/ara/ajgt.py +1 -2
  45. mteb/tasks/classification/ara/hotel_review_sentiment_classification.py +1 -2
  46. mteb/tasks/classification/ara/online_store_review_sentiment_classification.py +1 -2
  47. mteb/tasks/classification/ara/restaurant_review_sentiment_classification.py +1 -2
  48. mteb/tasks/classification/ara/tweet_emotion_classification.py +1 -2
  49. mteb/tasks/classification/ara/tweet_sarcasm_classification.py +1 -2
  50. mteb/tasks/classification/ben/bengali_document_classification.py +1 -2
  51. mteb/tasks/classification/ben/bengali_hate_speech_classification.py +1 -2
  52. mteb/tasks/classification/ben/bengali_sentiment_analysis.py +1 -2
  53. mteb/tasks/classification/ces/csfdcz_movie_review_sentiment_classification.py +1 -2
  54. mteb/tasks/classification/ces/czech_product_review_sentiment_classification.py +1 -2
  55. mteb/tasks/classification/ces/czech_so_me_sentiment_classification.py +1 -2
  56. mteb/tasks/classification/dan/angry_tweets_classification.py +1 -2
  57. mteb/tasks/classification/dan/danish_political_comments_classification.py +1 -2
  58. mteb/tasks/classification/dan/ddisco_cohesion_classification.py +1 -2
  59. mteb/tasks/classification/dan/dk_hate_classification.py +1 -2
  60. mteb/tasks/classification/deu/german_politicians_twitter_sentiment_classification.py +1 -2
  61. mteb/tasks/classification/deu/ten_k_gnad_classification.py +1 -2
  62. mteb/tasks/classification/eng/amazon_polarity_classification.py +1 -2
  63. mteb/tasks/classification/eng/arxiv_classification.py +1 -2
  64. mteb/tasks/classification/eng/banking77_classification.py +1 -2
  65. mteb/tasks/classification/eng/dbpedia_classification.py +1 -2
  66. mteb/tasks/classification/eng/emotion_classification.py +1 -2
  67. mteb/tasks/classification/eng/financial_phrasebank_classification.py +1 -2
  68. mteb/tasks/classification/eng/frenk_en_classification.py +1 -2
  69. mteb/tasks/classification/eng/gtsrb_classification.py +1 -1
  70. mteb/tasks/classification/eng/imdb_classification.py +1 -2
  71. mteb/tasks/classification/eng/legal_bench_classification.py +14 -120
  72. mteb/tasks/classification/eng/news_classification.py +1 -2
  73. mteb/tasks/classification/eng/patch_camelyon_classification.py +1 -1
  74. mteb/tasks/classification/eng/patent_classification.py +1 -2
  75. mteb/tasks/classification/eng/poem_sentiment_classification.py +1 -2
  76. mteb/tasks/classification/eng/sds_eye_protection_classification.py +1 -2
  77. mteb/tasks/classification/eng/sds_gloves_classification.py +1 -2
  78. mteb/tasks/classification/eng/toxic_chat_classification.py +2 -19
  79. mteb/tasks/classification/eng/toxic_conversations_classification.py +1 -2
  80. mteb/tasks/classification/eng/tweet_sentiment_extraction_classification.py +1 -2
  81. mteb/tasks/classification/eng/tweet_topic_single_classification.py +2 -13
  82. mteb/tasks/classification/eng/ucf101_classification.py +1 -5
  83. mteb/tasks/classification/eng/wikipedia_bio_met_chem_classification.py +1 -2
  84. mteb/tasks/classification/eng/wikipedia_chem_fields_classification.py +1 -2
  85. mteb/tasks/classification/eng/wikipedia_comp_chem_spectroscopy_classification.py +1 -2
  86. mteb/tasks/classification/eng/wikipedia_crystallography_analytical_classification.py +1 -2
  87. mteb/tasks/classification/eng/wikipedia_theoretical_applied_classification.py +1 -2
  88. mteb/tasks/classification/eng/yahoo_answers_topics_classification.py +1 -2
  89. mteb/tasks/classification/eng/yelp_review_full_classification.py +1 -2
  90. mteb/tasks/classification/est/estonian_valence.py +1 -2
  91. mteb/tasks/classification/fas/fa_mteb_classification.py +7 -14
  92. mteb/tasks/classification/fil/filipino_hate_speech_classification.py +1 -2
  93. mteb/tasks/classification/fin/fin_toxicity_classification.py +2 -11
  94. mteb/tasks/classification/fra/french_book_reviews.py +1 -2
  95. mteb/tasks/classification/fra/movie_review_sentiment_classification.py +1 -2
  96. mteb/tasks/classification/guj/gujarati_news_classification.py +1 -2
  97. mteb/tasks/classification/heb/hebrew_sentiment_analysis.py +1 -2
  98. mteb/tasks/classification/hin/hindi_discourse_classification.py +1 -2
  99. mteb/tasks/classification/hin/sentiment_analysis_hindi.py +1 -2
  100. mteb/tasks/classification/hrv/frenk_hr_classification.py +1 -2
  101. mteb/tasks/classification/ind/indonesian_id_clickbait_classification.py +1 -2
  102. mteb/tasks/classification/ind/indonesian_mongabay_conservation_classification.py +1 -2
  103. mteb/tasks/classification/ita/italian_linguist_acceptability_classification.py +1 -2
  104. mteb/tasks/classification/jav/javanese_imdb_classification.py +1 -2
  105. mteb/tasks/classification/jpn/wrime_classification.py +1 -2
  106. mteb/tasks/classification/kan/kannada_news_classification.py +1 -2
  107. mteb/tasks/classification/kor/klue_tc.py +1 -2
  108. mteb/tasks/classification/kor/kor_hate_classification.py +2 -17
  109. mteb/tasks/classification/kor/kor_sarcasm_classification.py +2 -19
  110. mteb/tasks/classification/kur/kurdish_sentiment_classification.py +1 -2
  111. mteb/tasks/classification/mal/malayalam_news_classification.py +1 -2
  112. mteb/tasks/classification/mar/marathi_news_classification.py +1 -2
  113. mteb/tasks/classification/mkd/macedonian_tweet_sentiment_classification.py +1 -2
  114. mteb/tasks/classification/multilingual/catalonia_tweet_classification.py +1 -6
  115. mteb/tasks/classification/multilingual/multi_hate_classification.py +1 -4
  116. mteb/tasks/classification/multilingual/ru_sci_bench_classification.py +4 -23
  117. mteb/tasks/classification/multilingual/scala_classification.py +1 -2
  118. mteb/tasks/classification/multilingual/sib200_classification.py +1 -6
  119. mteb/tasks/classification/mya/myanmar_news.py +1 -2
  120. mteb/tasks/classification/nep/nepali_news_classification.py +1 -2
  121. mteb/tasks/classification/nld/dutch_book_review_sentiment_classification.py +1 -3
  122. mteb/tasks/classification/nob/no_rec_classification.py +1 -2
  123. mteb/tasks/classification/nob/norwegian_parliament_classification.py +1 -2
  124. mteb/tasks/classification/ory/odia_news_classification.py +1 -2
  125. mteb/tasks/classification/pol/polish_classification.py +3 -6
  126. mteb/tasks/classification/ron/moroco.py +1 -2
  127. mteb/tasks/classification/ron/romanian_reviews_sentiment.py +1 -2
  128. mteb/tasks/classification/ron/romanian_sentiment_classification.py +1 -2
  129. mteb/tasks/classification/rus/georeview_classification.py +1 -2
  130. mteb/tasks/classification/rus/headline_classification.py +1 -2
  131. mteb/tasks/classification/rus/inappropriateness_classification.py +1 -2
  132. mteb/tasks/classification/rus/ru_reviews_classification.py +1 -2
  133. mteb/tasks/classification/rus/ru_toixic_classification_okmlcup.py +1 -2
  134. mteb/tasks/classification/rus/senti_ru_eval.py +1 -2
  135. mteb/tasks/classification/sin/sinhala_news_classification.py +1 -2
  136. mteb/tasks/classification/sin/sinhala_news_source_classification.py +1 -2
  137. mteb/tasks/classification/slk/csfdsk_movie_review_sentiment_classification.py +1 -2
  138. mteb/tasks/classification/slk/slovak_hate_speech_classification.py +1 -2
  139. mteb/tasks/classification/slk/slovak_movie_review_sentiment_classification.py +1 -2
  140. mteb/tasks/classification/slv/frenk_sl_classification.py +1 -2
  141. mteb/tasks/classification/spa/spanish_news_classification.py +1 -2
  142. mteb/tasks/classification/spa/spanish_sentiment_classification.py +1 -2
  143. mteb/tasks/classification/ssw/siswati_news_classification.py +1 -2
  144. mteb/tasks/classification/swa/swahili_news_classification.py +1 -2
  145. mteb/tasks/classification/swe/dalaj_classification.py +1 -2
  146. mteb/tasks/classification/swe/swe_rec_classification.py +1 -2
  147. mteb/tasks/classification/swe/swedish_sentiment_classification.py +1 -2
  148. mteb/tasks/classification/tam/tamil_news_classification.py +1 -2
  149. mteb/tasks/classification/tel/telugu_andhra_jyoti_news_classification.py +1 -2
  150. mteb/tasks/classification/tha/wisesight_sentiment_classification.py +1 -2
  151. mteb/tasks/classification/tsn/tswana_news_classification.py +1 -2
  152. mteb/tasks/classification/tur/turkish_movie_sentiment_classification.py +1 -2
  153. mteb/tasks/classification/tur/turkish_product_sentiment_classification.py +1 -2
  154. mteb/tasks/classification/ukr/ukr_formality_classification.py +2 -15
  155. mteb/tasks/classification/urd/urdu_roman_sentiment_classification.py +1 -2
  156. mteb/tasks/classification/vie/amazon_counterfactual_vn_classification.py +1 -6
  157. mteb/tasks/classification/vie/amazon_polarity_vn_classification.py +1 -6
  158. mteb/tasks/classification/vie/amazon_reviews_vn_classification.py +1 -5
  159. mteb/tasks/classification/vie/banking77_vn_classification.py +1 -5
  160. mteb/tasks/classification/vie/emotion_vn_classification.py +1 -5
  161. mteb/tasks/classification/vie/imdb_vn_classification.py +1 -5
  162. mteb/tasks/classification/vie/massive_intent_vn_classification.py +1 -5
  163. mteb/tasks/classification/vie/massive_scenario_vn_classification.py +1 -5
  164. mteb/tasks/classification/vie/mtop_domain_vn_classification.py +1 -5
  165. mteb/tasks/classification/vie/mtop_intent_vn_classification.py +1 -5
  166. mteb/tasks/classification/vie/toxic_conversations_vn_classification.py +1 -5
  167. mteb/tasks/classification/vie/tweet_sentiment_extraction_vn_classification.py +1 -5
  168. mteb/tasks/classification/vie/vie_student_feedback_classification.py +1 -2
  169. mteb/tasks/classification/zho/cmteb_classification.py +5 -10
  170. mteb/tasks/classification/zho/yue_openrice_review_classification.py +1 -2
  171. mteb/tasks/classification/zul/isi_zulu_news_classification.py +1 -2
  172. mteb/tasks/clustering/jpn/mews_c16_ja_clustering.py +1 -3
  173. mteb/tasks/clustering/multilingual/sib200_clustering_s2s.py +1 -6
  174. mteb/tasks/clustering/vie/reddit_clustering_p2p_vn.py +1 -5
  175. mteb/tasks/clustering/vie/reddit_clustering_vn.py +1 -5
  176. mteb/tasks/clustering/vie/stack_exchange_clustering_p2p_vn.py +1 -5
  177. mteb/tasks/clustering/vie/stack_exchange_clustering_vn.py +1 -5
  178. mteb/tasks/clustering/vie/twenty_newsgroups_clustering_vn.py +1 -5
  179. mteb/tasks/multilabel_classification/ita/emit_classification.py +1 -5
  180. mteb/tasks/multilabel_classification/kor/kor_hate_speech_ml_classification.py +1 -9
  181. mteb/tasks/multilabel_classification/mlt/maltese_news_classification.py +1 -6
  182. mteb/tasks/multilabel_classification/por/brazilian_toxic_tweets_classification.py +1 -6
  183. mteb/tasks/multilabel_classification/swe/swedish_patent_cpc_group_classification.py +1 -1
  184. mteb/tasks/multilabel_classification/swe/swedish_patent_cpc_subclass_classification.py +1 -2
  185. mteb/tasks/pair_classification/dan/talemaader_pc.py +1 -6
  186. mteb/tasks/pair_classification/eng/legal_bench_pc.py +1 -9
  187. mteb/tasks/pair_classification/vie/sprint_duplicate_questions_pcvn.py +1 -5
  188. mteb/tasks/pair_classification/vie/twitter_sem_eval2015_pcvn.py +1 -5
  189. mteb/tasks/pair_classification/vie/twitter_url_corpus_pcvn.py +1 -5
  190. mteb/tasks/regression/multilingual/ru_sci_bench_regression.py +2 -6
  191. mteb/tasks/reranking/multilingual/x_glue_wpr_reranking.py +1 -2
  192. mteb/tasks/reranking/vie/ask_ubuntu_dup_questions_vn.py +1 -5
  193. mteb/tasks/reranking/vie/sci_docs_reranking_vn.py +1 -5
  194. mteb/tasks/reranking/vie/stack_overflow_dup_questions_vn.py +1 -5
  195. mteb/tasks/retrieval/eng/lit_search_retrieval.py +1 -8
  196. mteb/tasks/retrieval/jpn/ja_cwir_retrieval.py +1 -4
  197. mteb/tasks/retrieval/multilingual/__init__.py +22 -0
  198. mteb/tasks/retrieval/multilingual/mkqa_retrieval.py +1 -2
  199. mteb/tasks/retrieval/multilingual/mlqa_retrieval.py +1 -4
  200. mteb/tasks/retrieval/multilingual/multi_long_doc_retrieval.py +1 -2
  201. mteb/tasks/retrieval/multilingual/ru_sci_bench_retrieval.py +2 -12
  202. mteb/tasks/retrieval/multilingual/vidore3_bench_retrieval.py +399 -0
  203. mteb/tasks/retrieval/slk/slovak_sum_retrieval.py +1 -7
  204. mteb/tasks/retrieval/vie/argu_ana_vn_retrieval.py +1 -5
  205. mteb/tasks/retrieval/vie/climate_fevervn_retrieval.py +1 -5
  206. mteb/tasks/retrieval/vie/cqa_dupstack_android_vn_retrieval.py +1 -5
  207. mteb/tasks/retrieval/vie/cqa_dupstack_gis_vn_retrieval.py +1 -5
  208. mteb/tasks/retrieval/vie/cqa_dupstack_mathematica_vn_retrieval.py +1 -5
  209. mteb/tasks/retrieval/vie/cqa_dupstack_physics_vn_retrieval.py +1 -5
  210. mteb/tasks/retrieval/vie/cqa_dupstack_programmers_vn_retrieval.py +1 -5
  211. mteb/tasks/retrieval/vie/cqa_dupstack_stats_vn_retrieval.py +1 -5
  212. mteb/tasks/retrieval/vie/cqa_dupstack_tex_vn_retrieval.py +1 -5
  213. mteb/tasks/retrieval/vie/cqa_dupstack_unix_vn_retrieval.py +1 -5
  214. mteb/tasks/retrieval/vie/cqa_dupstack_webmasters_vn_retrieval.py +1 -5
  215. mteb/tasks/retrieval/vie/cqa_dupstack_wordpress_vn_retrieval.py +1 -5
  216. mteb/tasks/retrieval/vie/db_pedia_vn_retrieval.py +1 -5
  217. mteb/tasks/retrieval/vie/fevervn_retrieval.py +1 -7
  218. mteb/tasks/retrieval/vie/fi_qa2018_vn_retrieval.py +1 -5
  219. mteb/tasks/retrieval/vie/hotpot_qavn_retrieval.py +1 -6
  220. mteb/tasks/retrieval/vie/msmarcovn_retrieval.py +1 -5
  221. mteb/tasks/retrieval/vie/nf_corpus_vn_retrieval.py +1 -5
  222. mteb/tasks/retrieval/vie/nqvn_retrieval.py +1 -5
  223. mteb/tasks/retrieval/vie/quora_vn_retrieval.py +1 -6
  224. mteb/tasks/retrieval/vie/sci_fact_vn_retrieval.py +1 -5
  225. mteb/tasks/retrieval/vie/scidocsvn_retrieval.py +1 -6
  226. mteb/tasks/retrieval/vie/touche2020_vn_retrieval.py +1 -5
  227. mteb/tasks/retrieval/vie/treccovidvn_retrieval.py +1 -5
  228. mteb/tasks/sts/vie/biosses_stsvn.py +1 -5
  229. mteb/tasks/sts/vie/sickr_stsvn.py +1 -5
  230. mteb/tasks/sts/vie/sts_benchmark_stsvn.py +1 -5
  231. mteb/tasks/zeroshot_classification/eng/gtsrb.py +1 -1
  232. mteb/tasks/zeroshot_classification/eng/patch_camelyon.py +1 -1
  233. mteb/tasks/zeroshot_classification/eng/ucf101.py +1 -5
  234. {mteb-2.1.7.dist-info → mteb-2.1.8.dist-info}/METADATA +1 -1
  235. {mteb-2.1.7.dist-info → mteb-2.1.8.dist-info}/RECORD +239 -228
  236. {mteb-2.1.7.dist-info → mteb-2.1.8.dist-info}/WHEEL +0 -0
  237. {mteb-2.1.7.dist-info → mteb-2.1.8.dist-info}/entry_points.txt +0 -0
  238. {mteb-2.1.7.dist-info → mteb-2.1.8.dist-info}/licenses/LICENSE +0 -0
  239. {mteb-2.1.7.dist-info → mteb-2.1.8.dist-info}/top_level.txt +0 -0
mteb/descriptive_stats/Image/DocumentUnderstanding/Vidore3PharmaceuticalsRetrieval.json ADDED
@@ -0,0 +1,214 @@
1
+ {
2
+ "test": {
3
+ "num_samples": 16062,
4
+ "number_of_characters": 273516,
5
+ "documents_text_statistics": null,
6
+ "documents_image_statistics": {
7
+ "min_image_width": 1221,
8
+ "average_image_width": 1857.1033290099438,
9
+ "max_image_width": 4000,
10
+ "min_image_height": 1125,
11
+ "average_image_height": 1578.8672719412018,
12
+ "max_image_height": 2250,
13
+ "unique_images": 2298
14
+ },
15
+ "queries_text_statistics": {
16
+ "total_text_length": 273516,
17
+ "min_text_length": 25,
18
+ "average_text_length": 125.23626373626374,
19
+ "max_text_length": 295,
20
+ "unique_texts": 2184
21
+ },
22
+ "queries_image_statistics": null,
23
+ "relevant_docs_statistics": {
24
+ "num_relevant_docs": 62352,
25
+ "min_relevant_docs_per_query": 1,
26
+ "average_relevant_docs_per_query": 4.758241758241758,
27
+ "max_relevant_docs_per_query": 26,
28
+ "unique_relevant_docs": 5676
29
+ },
30
+ "top_ranked_statistics": null,
31
+ "hf_subset_descriptive_stats": {
32
+ "french": {
33
+ "num_samples": 2677,
34
+ "number_of_characters": 49632,
35
+ "documents_text_statistics": null,
36
+ "documents_image_statistics": {
37
+ "min_image_width": 1221,
38
+ "average_image_width": 1857.1033290099438,
39
+ "max_image_width": 4000,
40
+ "min_image_height": 1125,
41
+ "average_image_height": 1578.8672719412018,
42
+ "max_image_height": 2250,
43
+ "unique_images": 2298
44
+ },
45
+ "queries_text_statistics": {
46
+ "total_text_length": 49632,
47
+ "min_text_length": 37,
48
+ "average_text_length": 136.35164835164835,
49
+ "max_text_length": 273,
50
+ "unique_texts": 364
51
+ },
52
+ "queries_image_statistics": null,
53
+ "relevant_docs_statistics": {
54
+ "num_relevant_docs": 10392,
55
+ "min_relevant_docs_per_query": 1,
56
+ "average_relevant_docs_per_query": 4.758241758241758,
57
+ "max_relevant_docs_per_query": 26,
58
+ "unique_relevant_docs": 946
59
+ },
60
+ "top_ranked_statistics": null
61
+ },
62
+ "spanish": {
63
+ "num_samples": 2677,
64
+ "number_of_characters": 46902,
65
+ "documents_text_statistics": null,
66
+ "documents_image_statistics": {
67
+ "min_image_width": 1221,
68
+ "average_image_width": 1857.1033290099438,
69
+ "max_image_width": 4000,
70
+ "min_image_height": 1125,
71
+ "average_image_height": 1578.8672719412018,
72
+ "max_image_height": 2250,
73
+ "unique_images": 2298
74
+ },
75
+ "queries_text_statistics": {
76
+ "total_text_length": 46902,
77
+ "min_text_length": 34,
78
+ "average_text_length": 128.85164835164835,
79
+ "max_text_length": 256,
80
+ "unique_texts": 364
81
+ },
82
+ "queries_image_statistics": null,
83
+ "relevant_docs_statistics": {
84
+ "num_relevant_docs": 10392,
85
+ "min_relevant_docs_per_query": 1,
86
+ "average_relevant_docs_per_query": 4.758241758241758,
87
+ "max_relevant_docs_per_query": 26,
88
+ "unique_relevant_docs": 946
89
+ },
90
+ "top_ranked_statistics": null
91
+ },
92
+ "english": {
93
+ "num_samples": 2677,
94
+ "number_of_characters": 39067,
95
+ "documents_text_statistics": null,
96
+ "documents_image_statistics": {
97
+ "min_image_width": 1221,
98
+ "average_image_width": 1857.1033290099438,
99
+ "max_image_width": 4000,
100
+ "min_image_height": 1125,
101
+ "average_image_height": 1578.8672719412018,
102
+ "max_image_height": 2250,
103
+ "unique_images": 2298
104
+ },
105
+ "queries_text_statistics": {
106
+ "total_text_length": 39067,
107
+ "min_text_length": 25,
108
+ "average_text_length": 107.32692307692308,
109
+ "max_text_length": 213,
110
+ "unique_texts": 364
111
+ },
112
+ "queries_image_statistics": null,
113
+ "relevant_docs_statistics": {
114
+ "num_relevant_docs": 10392,
115
+ "min_relevant_docs_per_query": 1,
116
+ "average_relevant_docs_per_query": 4.758241758241758,
117
+ "max_relevant_docs_per_query": 26,
118
+ "unique_relevant_docs": 946
119
+ },
120
+ "top_ranked_statistics": null
121
+ },
122
+ "german": {
123
+ "num_samples": 2677,
124
+ "number_of_characters": 47757,
125
+ "documents_text_statistics": null,
126
+ "documents_image_statistics": {
127
+ "min_image_width": 1221,
128
+ "average_image_width": 1857.1033290099438,
129
+ "max_image_width": 4000,
130
+ "min_image_height": 1125,
131
+ "average_image_height": 1578.8672719412018,
132
+ "max_image_height": 2250,
133
+ "unique_images": 2298
134
+ },
135
+ "queries_text_statistics": {
136
+ "total_text_length": 47757,
137
+ "min_text_length": 36,
138
+ "average_text_length": 131.20054945054946,
139
+ "max_text_length": 295,
140
+ "unique_texts": 364
141
+ },
142
+ "queries_image_statistics": null,
143
+ "relevant_docs_statistics": {
144
+ "num_relevant_docs": 10392,
145
+ "min_relevant_docs_per_query": 1,
146
+ "average_relevant_docs_per_query": 4.758241758241758,
147
+ "max_relevant_docs_per_query": 26,
148
+ "unique_relevant_docs": 946
149
+ },
150
+ "top_ranked_statistics": null
151
+ },
152
+ "italian": {
153
+ "num_samples": 2677,
154
+ "number_of_characters": 46246,
155
+ "documents_text_statistics": null,
156
+ "documents_image_statistics": {
157
+ "min_image_width": 1221,
158
+ "average_image_width": 1857.1033290099438,
159
+ "max_image_width": 4000,
160
+ "min_image_height": 1125,
161
+ "average_image_height": 1578.8672719412018,
162
+ "max_image_height": 2250,
163
+ "unique_images": 2298
164
+ },
165
+ "queries_text_statistics": {
166
+ "total_text_length": 46246,
167
+ "min_text_length": 31,
168
+ "average_text_length": 127.04945054945055,
169
+ "max_text_length": 280,
170
+ "unique_texts": 364
171
+ },
172
+ "queries_image_statistics": null,
173
+ "relevant_docs_statistics": {
174
+ "num_relevant_docs": 10392,
175
+ "min_relevant_docs_per_query": 1,
176
+ "average_relevant_docs_per_query": 4.758241758241758,
177
+ "max_relevant_docs_per_query": 26,
178
+ "unique_relevant_docs": 946
179
+ },
180
+ "top_ranked_statistics": null
181
+ },
182
+ "portuguese": {
183
+ "num_samples": 2677,
184
+ "number_of_characters": 43912,
185
+ "documents_text_statistics": null,
186
+ "documents_image_statistics": {
187
+ "min_image_width": 1221,
188
+ "average_image_width": 1857.1033290099438,
189
+ "max_image_width": 4000,
190
+ "min_image_height": 1125,
191
+ "average_image_height": 1578.8672719412018,
192
+ "max_image_height": 2250,
193
+ "unique_images": 2298
194
+ },
195
+ "queries_text_statistics": {
196
+ "total_text_length": 43912,
197
+ "min_text_length": 36,
198
+ "average_text_length": 120.63736263736264,
199
+ "max_text_length": 252,
200
+ "unique_texts": 364
201
+ },
202
+ "queries_image_statistics": null,
203
+ "relevant_docs_statistics": {
204
+ "num_relevant_docs": 10392,
205
+ "min_relevant_docs_per_query": 1,
206
+ "average_relevant_docs_per_query": 4.758241758241758,
207
+ "max_relevant_docs_per_query": 26,
208
+ "unique_relevant_docs": 946
209
+ },
210
+ "top_ranked_statistics": null
211
+ }
212
+ }
213
+ }
214
+ }
mteb/descriptive_stats/Image/DocumentUnderstanding/Vidore3PhysicsRetrieval.json ADDED
@@ -0,0 +1,214 @@
1
+ {
2
+ "test": {
3
+ "num_samples": 11856,
4
+ "number_of_characters": 220508,
5
+ "documents_text_statistics": null,
6
+ "documents_image_statistics": {
7
+ "min_image_width": 2667,
8
+ "average_image_width": 2667.0,
9
+ "max_image_width": 2667,
10
+ "min_image_height": 1500,
11
+ "average_image_height": 1500.0,
12
+ "max_image_height": 1500,
13
+ "unique_images": 1674
14
+ },
15
+ "queries_text_statistics": {
16
+ "total_text_length": 220508,
17
+ "min_text_length": 37,
18
+ "average_text_length": 121.69315673289184,
19
+ "max_text_length": 286,
20
+ "unique_texts": 1812
21
+ },
22
+ "queries_image_statistics": null,
23
+ "relevant_docs_statistics": {
24
+ "num_relevant_docs": 78408,
25
+ "min_relevant_docs_per_query": 1,
26
+ "average_relevant_docs_per_query": 7.211920529801325,
27
+ "max_relevant_docs_per_query": 28,
28
+ "unique_relevant_docs": 5772
29
+ },
30
+ "top_ranked_statistics": null,
31
+ "hf_subset_descriptive_stats": {
32
+ "french": {
33
+ "num_samples": 1976,
34
+ "number_of_characters": 38216,
35
+ "documents_text_statistics": null,
36
+ "documents_image_statistics": {
37
+ "min_image_width": 2667,
38
+ "average_image_width": 2667.0,
39
+ "max_image_width": 2667,
40
+ "min_image_height": 1500,
41
+ "average_image_height": 1500.0,
42
+ "max_image_height": 1500,
43
+ "unique_images": 1674
44
+ },
45
+ "queries_text_statistics": {
46
+ "total_text_length": 38216,
47
+ "min_text_length": 40,
48
+ "average_text_length": 126.54304635761589,
49
+ "max_text_length": 267,
50
+ "unique_texts": 302
51
+ },
52
+ "queries_image_statistics": null,
53
+ "relevant_docs_statistics": {
54
+ "num_relevant_docs": 13068,
55
+ "min_relevant_docs_per_query": 1,
56
+ "average_relevant_docs_per_query": 7.211920529801325,
57
+ "max_relevant_docs_per_query": 28,
58
+ "unique_relevant_docs": 962
59
+ },
60
+ "top_ranked_statistics": null
61
+ },
62
+ "spanish": {
63
+ "num_samples": 1976,
64
+ "number_of_characters": 37169,
65
+ "documents_text_statistics": null,
66
+ "documents_image_statistics": {
67
+ "min_image_width": 2667,
68
+ "average_image_width": 2667.0,
69
+ "max_image_width": 2667,
70
+ "min_image_height": 1500,
71
+ "average_image_height": 1500.0,
72
+ "max_image_height": 1500,
73
+ "unique_images": 1674
74
+ },
75
+ "queries_text_statistics": {
76
+ "total_text_length": 37169,
77
+ "min_text_length": 39,
78
+ "average_text_length": 123.07615894039735,
79
+ "max_text_length": 263,
80
+ "unique_texts": 302
81
+ },
82
+ "queries_image_statistics": null,
83
+ "relevant_docs_statistics": {
84
+ "num_relevant_docs": 13068,
85
+ "min_relevant_docs_per_query": 1,
86
+ "average_relevant_docs_per_query": 7.211920529801325,
87
+ "max_relevant_docs_per_query": 28,
88
+ "unique_relevant_docs": 962
89
+ },
90
+ "top_ranked_statistics": null
91
+ },
92
+ "english": {
93
+ "num_samples": 1976,
94
+ "number_of_characters": 34092,
95
+ "documents_text_statistics": null,
96
+ "documents_image_statistics": {
97
+ "min_image_width": 2667,
98
+ "average_image_width": 2667.0,
99
+ "max_image_width": 2667,
100
+ "min_image_height": 1500,
101
+ "average_image_height": 1500.0,
102
+ "max_image_height": 1500,
103
+ "unique_images": 1674
104
+ },
105
+ "queries_text_statistics": {
106
+ "total_text_length": 34092,
107
+ "min_text_length": 38,
108
+ "average_text_length": 112.88741721854305,
109
+ "max_text_length": 244,
110
+ "unique_texts": 302
111
+ },
112
+ "queries_image_statistics": null,
113
+ "relevant_docs_statistics": {
114
+ "num_relevant_docs": 13068,
115
+ "min_relevant_docs_per_query": 1,
116
+ "average_relevant_docs_per_query": 7.211920529801325,
117
+ "max_relevant_docs_per_query": 28,
118
+ "unique_relevant_docs": 962
119
+ },
120
+ "top_ranked_statistics": null
121
+ },
122
+ "german": {
123
+ "num_samples": 1976,
124
+ "number_of_characters": 39603,
125
+ "documents_text_statistics": null,
126
+ "documents_image_statistics": {
127
+ "min_image_width": 2667,
128
+ "average_image_width": 2667.0,
129
+ "max_image_width": 2667,
130
+ "min_image_height": 1500,
131
+ "average_image_height": 1500.0,
132
+ "max_image_height": 1500,
133
+ "unique_images": 1674
134
+ },
135
+ "queries_text_statistics": {
136
+ "total_text_length": 39603,
137
+ "min_text_length": 37,
138
+ "average_text_length": 131.13576158940398,
139
+ "max_text_length": 286,
140
+ "unique_texts": 302
141
+ },
142
+ "queries_image_statistics": null,
143
+ "relevant_docs_statistics": {
144
+ "num_relevant_docs": 13068,
145
+ "min_relevant_docs_per_query": 1,
146
+ "average_relevant_docs_per_query": 7.211920529801325,
147
+ "max_relevant_docs_per_query": 28,
148
+ "unique_relevant_docs": 962
149
+ },
150
+ "top_ranked_statistics": null
151
+ },
152
+ "italian": {
153
+ "num_samples": 1976,
154
+ "number_of_characters": 36485,
155
+ "documents_text_statistics": null,
156
+ "documents_image_statistics": {
157
+ "min_image_width": 2667,
158
+ "average_image_width": 2667.0,
159
+ "max_image_width": 2667,
160
+ "min_image_height": 1500,
161
+ "average_image_height": 1500.0,
162
+ "max_image_height": 1500,
163
+ "unique_images": 1674
164
+ },
165
+ "queries_text_statistics": {
166
+ "total_text_length": 36485,
167
+ "min_text_length": 39,
168
+ "average_text_length": 120.8112582781457,
169
+ "max_text_length": 253,
170
+ "unique_texts": 302
171
+ },
172
+ "queries_image_statistics": null,
173
+ "relevant_docs_statistics": {
174
+ "num_relevant_docs": 13068,
175
+ "min_relevant_docs_per_query": 1,
176
+ "average_relevant_docs_per_query": 7.211920529801325,
177
+ "max_relevant_docs_per_query": 28,
178
+ "unique_relevant_docs": 962
179
+ },
180
+ "top_ranked_statistics": null
181
+ },
182
+ "portuguese": {
183
+ "num_samples": 1976,
184
+ "number_of_characters": 34943,
185
+ "documents_text_statistics": null,
186
+ "documents_image_statistics": {
187
+ "min_image_width": 2667,
188
+ "average_image_width": 2667.0,
189
+ "max_image_width": 2667,
190
+ "min_image_height": 1500,
191
+ "average_image_height": 1500.0,
192
+ "max_image_height": 1500,
193
+ "unique_images": 1674
194
+ },
195
+ "queries_text_statistics": {
196
+ "total_text_length": 34943,
197
+ "min_text_length": 38,
198
+ "average_text_length": 115.70529801324503,
199
+ "max_text_length": 240,
200
+ "unique_texts": 302
201
+ },
202
+ "queries_image_statistics": null,
203
+ "relevant_docs_statistics": {
204
+ "num_relevant_docs": 13068,
205
+ "min_relevant_docs_per_query": 1,
206
+ "average_relevant_docs_per_query": 7.211920529801325,
207
+ "max_relevant_docs_per_query": 28,
208
+ "unique_relevant_docs": 962
209
+ },
210
+ "top_ranked_statistics": null
211
+ }
212
+ }
213
+ }
214
+ }
mteb/descriptive_stats/Image/DocumentUnderstanding/Vidore3TelecomRetrieval.json ADDED
@@ -0,0 +1,214 @@
1
+ {
2
+ "test": {
3
+ "num_samples": 29850,
4
+ "number_of_characters": 259999,
5
+ "documents_text_statistics": null,
6
+ "documents_image_statistics": {
7
+ "min_image_width": 1653,
8
+ "average_image_width": 1692.308839420791,
9
+ "max_image_width": 1700,
10
+ "min_image_height": 2197,
11
+ "average_image_height": 2222.7112599956777,
12
+ "max_image_height": 2339,
13
+ "unique_images": 4624
14
+ },
15
+ "queries_text_statistics": {
16
+ "total_text_length": 259999,
17
+ "min_text_length": 17,
18
+ "average_text_length": 124.5205938697318,
19
+ "max_text_length": 326,
20
+ "unique_texts": 2088
21
+ },
22
+ "queries_image_statistics": null,
23
+ "relevant_docs_statistics": {
24
+ "num_relevant_docs": 56196,
25
+ "min_relevant_docs_per_query": 1,
26
+ "average_relevant_docs_per_query": 4.485632183908046,
27
+ "max_relevant_docs_per_query": 24,
28
+ "unique_relevant_docs": 6510
29
+ },
30
+ "top_ranked_statistics": null,
31
+ "hf_subset_descriptive_stats": {
32
+ "french": {
33
+ "num_samples": 4975,
34
+ "number_of_characters": 46656,
35
+ "documents_text_statistics": null,
36
+ "documents_image_statistics": {
37
+ "min_image_width": 1653,
38
+ "average_image_width": 1692.308839420791,
39
+ "max_image_width": 1700,
40
+ "min_image_height": 2197,
41
+ "average_image_height": 2222.7112599956777,
42
+ "max_image_height": 2339,
43
+ "unique_images": 4624
44
+ },
45
+ "queries_text_statistics": {
46
+ "total_text_length": 46656,
47
+ "min_text_length": 20,
48
+ "average_text_length": 134.06896551724137,
49
+ "max_text_length": 323,
50
+ "unique_texts": 348
51
+ },
52
+ "queries_image_statistics": null,
53
+ "relevant_docs_statistics": {
54
+ "num_relevant_docs": 9366,
55
+ "min_relevant_docs_per_query": 1,
56
+ "average_relevant_docs_per_query": 4.485632183908046,
57
+ "max_relevant_docs_per_query": 24,
58
+ "unique_relevant_docs": 1085
59
+ },
60
+ "top_ranked_statistics": null
61
+ },
62
+ "spanish": {
63
+ "num_samples": 4975,
64
+ "number_of_characters": 44339,
65
+ "documents_text_statistics": null,
66
+ "documents_image_statistics": {
67
+ "min_image_width": 1653,
68
+ "average_image_width": 1692.308839420791,
69
+ "max_image_width": 1700,
70
+ "min_image_height": 2197,
71
+ "average_image_height": 2222.7112599956777,
72
+ "max_image_height": 2339,
73
+ "unique_images": 4624
74
+ },
75
+ "queries_text_statistics": {
76
+ "total_text_length": 44339,
77
+ "min_text_length": 21,
78
+ "average_text_length": 127.41091954022988,
79
+ "max_text_length": 301,
80
+ "unique_texts": 348
81
+ },
82
+ "queries_image_statistics": null,
83
+ "relevant_docs_statistics": {
84
+ "num_relevant_docs": 9366,
85
+ "min_relevant_docs_per_query": 1,
86
+ "average_relevant_docs_per_query": 4.485632183908046,
87
+ "max_relevant_docs_per_query": 24,
88
+ "unique_relevant_docs": 1085
89
+ },
90
+ "top_ranked_statistics": null
91
+ },
92
+ "english": {
93
+ "num_samples": 4975,
94
+ "number_of_characters": 38411,
95
+ "documents_text_statistics": null,
96
+ "documents_image_statistics": {
97
+ "min_image_width": 1653,
98
+ "average_image_width": 1692.308839420791,
99
+ "max_image_width": 1700,
100
+ "min_image_height": 2197,
101
+ "average_image_height": 2222.7112599956777,
102
+ "max_image_height": 2339,
103
+ "unique_images": 4624
104
+ },
105
+ "queries_text_statistics": {
106
+ "total_text_length": 38411,
107
+ "min_text_length": 18,
108
+ "average_text_length": 110.3764367816092,
109
+ "max_text_length": 252,
110
+ "unique_texts": 348
111
+ },
112
+ "queries_image_statistics": null,
113
+ "relevant_docs_statistics": {
114
+ "num_relevant_docs": 9366,
115
+ "min_relevant_docs_per_query": 1,
116
+ "average_relevant_docs_per_query": 4.485632183908046,
117
+ "max_relevant_docs_per_query": 24,
118
+ "unique_relevant_docs": 1085
119
+ },
120
+ "top_ranked_statistics": null
121
+ },
122
+ "german": {
123
+ "num_samples": 4975,
124
+ "number_of_characters": 44640,
125
+ "documents_text_statistics": null,
126
+ "documents_image_statistics": {
127
+ "min_image_width": 1653,
128
+ "average_image_width": 1692.308839420791,
129
+ "max_image_width": 1700,
130
+ "min_image_height": 2197,
131
+ "average_image_height": 2222.7112599956777,
132
+ "max_image_height": 2339,
133
+ "unique_images": 4624
134
+ },
135
+ "queries_text_statistics": {
136
+ "total_text_length": 44640,
137
+ "min_text_length": 17,
138
+ "average_text_length": 128.27586206896552,
139
+ "max_text_length": 326,
140
+ "unique_texts": 348
141
+ },
142
+ "queries_image_statistics": null,
143
+ "relevant_docs_statistics": {
144
+ "num_relevant_docs": 9366,
145
+ "min_relevant_docs_per_query": 1,
146
+ "average_relevant_docs_per_query": 4.485632183908046,
147
+ "max_relevant_docs_per_query": 24,
148
+ "unique_relevant_docs": 1085
149
+ },
150
+ "top_ranked_statistics": null
151
+ },
152
+ "italian": {
153
+ "num_samples": 4975,
154
+ "number_of_characters": 44058,
155
+ "documents_text_statistics": null,
156
+ "documents_image_statistics": {
157
+ "min_image_width": 1653,
158
+ "average_image_width": 1692.308839420791,
159
+ "max_image_width": 1700,
160
+ "min_image_height": 2197,
161
+ "average_image_height": 2222.7112599956777,
162
+ "max_image_height": 2339,
163
+ "unique_images": 4624
164
+ },
165
+ "queries_text_statistics": {
166
+ "total_text_length": 44058,
167
+ "min_text_length": 17,
168
+ "average_text_length": 126.60344827586206,
169
+ "max_text_length": 290,
170
+ "unique_texts": 348
171
+ },
172
+ "queries_image_statistics": null,
173
+ "relevant_docs_statistics": {
174
+ "num_relevant_docs": 9366,
175
+ "min_relevant_docs_per_query": 1,
176
+ "average_relevant_docs_per_query": 4.485632183908046,
177
+ "max_relevant_docs_per_query": 24,
178
+ "unique_relevant_docs": 1085
179
+ },
180
+ "top_ranked_statistics": null
181
+ },
182
+ "portuguese": {
183
+ "num_samples": 4975,
184
+ "number_of_characters": 41895,
185
+ "documents_text_statistics": null,
186
+ "documents_image_statistics": {
187
+ "min_image_width": 1653,
188
+ "average_image_width": 1692.308839420791,
189
+ "max_image_width": 1700,
190
+ "min_image_height": 2197,
191
+ "average_image_height": 2222.7112599956777,
192
+ "max_image_height": 2339,
193
+ "unique_images": 4624
194
+ },
195
+ "queries_text_statistics": {
196
+ "total_text_length": 41895,
197
+ "min_text_length": 21,
198
+ "average_text_length": 120.38793103448276,
199
+ "max_text_length": 283,
200
+ "unique_texts": 348
201
+ },
202
+ "queries_image_statistics": null,
203
+ "relevant_docs_statistics": {
204
+ "num_relevant_docs": 9366,
205
+ "min_relevant_docs_per_query": 1,
206
+ "average_relevant_docs_per_query": 4.485632183908046,
207
+ "max_relevant_docs_per_query": 24,
208
+ "unique_relevant_docs": 1085
209
+ },
210
+ "top_ranked_statistics": null
211
+ }
212
+ }
213
+ }
214
+ }
mteb/models/cache_wrappers/cache_wrapper.py CHANGED
@@ -112,7 +112,7 @@ class CachedEmbeddingWrapper:
112
112
  dataset,
113
113
  task_metadata=task_metadata,
114
114
  prompt_type=prompt_type,
115
- batch_size=batch_size,
115
+ **kwargs,
116
116
  )
117
117
  new_vectors = self._model.encode(
118
118
  dl,
mteb/models/model_implementations/align_models.py CHANGED
@@ -124,4 +124,10 @@ align_base = ModelMeta(
124
124
  training_datasets=set(
125
125
  # COYO-700M
126
126
  ),
127
+ citation="""@misc{kakaobrain2022coyo-align,
128
+ title = {COYO-ALIGN},
129
+ author = {Yoon, Boogeo and Lee, Youhan and Baek, Woonhyuk},
130
+ year = {2022},
131
+ howpublished = {https://github.com/kakaobrain/coyo-align},
132
+ }""",
127
133
  )
mteb/models/model_implementations/ara_models.py CHANGED
@@ -23,4 +23,11 @@ arabic_triplet_matryoshka = ModelMeta(
23
23
  training_datasets=set(
24
24
  # "akhooli/arabic-triplets-1m-curated-sims-len"
25
25
  ),
26
+ citation="""
27
+ @article{nacar2025gate,
28
+ title={GATE: General Arabic Text Embedding for Enhanced Semantic Textual Similarity with Matryoshka Representation Learning and Hybrid Loss Training},
29
+ author={Nacar, Omer and Koubaa, Anis and Sibaee, Serry and Al-Habashi, Yasser and Ammar, Adel and Boulila, Wadii},
30
+ journal={arXiv preprint arXiv:2505.24581},
31
+ year={2025}
32
+ }""",
26
33
  )