Datasets:
mteb
/

Samoed commited on
Commit
11a9906
β€’
1 Parent(s): 1b85142

fix: Standartize results folders (#34)

Browse files

* before add instruct

* udpate paths

* fix test

This view is limited to 50 files because it contains too many changes. Β  See raw diff
Files changed (50) hide show
  1. paths.json +0 -0
  2. results.py +504 -251
  3. results/{gte-Qwen1.5-7B-instruct β†’ Alibaba-NLP__gte-Qwen1.5-7B-instruct}/no_revision_available/AFQMC.json +0 -0
  4. results/{gte-Qwen1.5-7B-instruct β†’ Alibaba-NLP__gte-Qwen1.5-7B-instruct}/no_revision_available/ATEC.json +0 -0
  5. results/{gte-Qwen1.5-7B-instruct β†’ Alibaba-NLP__gte-Qwen1.5-7B-instruct}/no_revision_available/AmazonCounterfactualClassification.json +0 -0
  6. results/{gte-Qwen1.5-7B-instruct β†’ Alibaba-NLP__gte-Qwen1.5-7B-instruct}/no_revision_available/AmazonPolarityClassification.json +0 -0
  7. results/{gte-Qwen1.5-7B-instruct β†’ Alibaba-NLP__gte-Qwen1.5-7B-instruct}/no_revision_available/AmazonReviewsClassification.json +0 -0
  8. results/{gte-Qwen1.5-7B-instruct β†’ Alibaba-NLP__gte-Qwen1.5-7B-instruct}/no_revision_available/ArguAna.json +0 -0
  9. results/{gte-Qwen1.5-7B-instruct β†’ Alibaba-NLP__gte-Qwen1.5-7B-instruct}/no_revision_available/ArxivClusteringP2P.json +0 -0
  10. results/{gte-Qwen1.5-7B-instruct β†’ Alibaba-NLP__gte-Qwen1.5-7B-instruct}/no_revision_available/ArxivClusteringS2S.json +0 -0
  11. results/{gte-Qwen1.5-7B-instruct β†’ Alibaba-NLP__gte-Qwen1.5-7B-instruct}/no_revision_available/AskUbuntuDupQuestions.json +0 -0
  12. results/{gte-Qwen1.5-7B-instruct β†’ Alibaba-NLP__gte-Qwen1.5-7B-instruct}/no_revision_available/BIOSSES.json +0 -0
  13. results/{gte-Qwen1.5-7B-instruct β†’ Alibaba-NLP__gte-Qwen1.5-7B-instruct}/no_revision_available/BQ.json +0 -0
  14. results/{gte-Qwen1.5-7B-instruct β†’ Alibaba-NLP__gte-Qwen1.5-7B-instruct}/no_revision_available/Banking77Classification.json +0 -0
  15. results/{gte-Qwen1.5-7B-instruct β†’ Alibaba-NLP__gte-Qwen1.5-7B-instruct}/no_revision_available/BiorxivClusteringP2P.json +0 -0
  16. results/{gte-Qwen1.5-7B-instruct β†’ Alibaba-NLP__gte-Qwen1.5-7B-instruct}/no_revision_available/BiorxivClusteringS2S.json +0 -0
  17. results/{gte-Qwen1.5-7B-instruct β†’ Alibaba-NLP__gte-Qwen1.5-7B-instruct}/no_revision_available/BrightRetrieval.json +0 -0
  18. results/{gte-Qwen1.5-7B-instruct β†’ Alibaba-NLP__gte-Qwen1.5-7B-instruct}/no_revision_available/CLSClusteringP2P.json +0 -0
  19. results/{gte-Qwen1.5-7B-instruct β†’ Alibaba-NLP__gte-Qwen1.5-7B-instruct}/no_revision_available/CLSClusteringS2S.json +0 -0
  20. results/{gte-Qwen1.5-7B-instruct β†’ Alibaba-NLP__gte-Qwen1.5-7B-instruct}/no_revision_available/CMedQAv1.json +0 -0
  21. results/{gte-Qwen1.5-7B-instruct β†’ Alibaba-NLP__gte-Qwen1.5-7B-instruct}/no_revision_available/CMedQAv2.json +0 -0
  22. results/{gte-Qwen1.5-7B-instruct β†’ Alibaba-NLP__gte-Qwen1.5-7B-instruct}/no_revision_available/CQADupstackAndroidRetrieval.json +0 -0
  23. results/{gte-Qwen1.5-7B-instruct β†’ Alibaba-NLP__gte-Qwen1.5-7B-instruct}/no_revision_available/CQADupstackEnglishRetrieval.json +0 -0
  24. results/{gte-Qwen1.5-7B-instruct β†’ Alibaba-NLP__gte-Qwen1.5-7B-instruct}/no_revision_available/CQADupstackGamingRetrieval.json +0 -0
  25. results/{gte-Qwen1.5-7B-instruct β†’ Alibaba-NLP__gte-Qwen1.5-7B-instruct}/no_revision_available/CQADupstackGisRetrieval.json +0 -0
  26. results/{gte-Qwen1.5-7B-instruct β†’ Alibaba-NLP__gte-Qwen1.5-7B-instruct}/no_revision_available/CQADupstackMathematicaRetrieval.json +0 -0
  27. results/{gte-Qwen1.5-7B-instruct β†’ Alibaba-NLP__gte-Qwen1.5-7B-instruct}/no_revision_available/CQADupstackPhysicsRetrieval.json +0 -0
  28. results/{gte-Qwen1.5-7B-instruct β†’ Alibaba-NLP__gte-Qwen1.5-7B-instruct}/no_revision_available/CQADupstackProgrammersRetrieval.json +0 -0
  29. results/{gte-Qwen1.5-7B-instruct β†’ Alibaba-NLP__gte-Qwen1.5-7B-instruct}/no_revision_available/CQADupstackRetrieval.json +0 -0
  30. results/{gte-Qwen1.5-7B-instruct β†’ Alibaba-NLP__gte-Qwen1.5-7B-instruct}/no_revision_available/CQADupstackStatsRetrieval.json +0 -0
  31. results/{gte-Qwen1.5-7B-instruct β†’ Alibaba-NLP__gte-Qwen1.5-7B-instruct}/no_revision_available/CQADupstackTexRetrieval.json +0 -0
  32. results/{gte-Qwen1.5-7B-instruct β†’ Alibaba-NLP__gte-Qwen1.5-7B-instruct}/no_revision_available/CQADupstackUnixRetrieval.json +0 -0
  33. results/{gte-Qwen1.5-7B-instruct β†’ Alibaba-NLP__gte-Qwen1.5-7B-instruct}/no_revision_available/CQADupstackWebmastersRetrieval.json +0 -0
  34. results/{gte-Qwen1.5-7B-instruct β†’ Alibaba-NLP__gte-Qwen1.5-7B-instruct}/no_revision_available/CQADupstackWordpressRetrieval.json +0 -0
  35. results/{gte-Qwen1.5-7B-instruct β†’ Alibaba-NLP__gte-Qwen1.5-7B-instruct}/no_revision_available/ClimateFEVER.json +0 -0
  36. results/{gte-Qwen1.5-7B-instruct β†’ Alibaba-NLP__gte-Qwen1.5-7B-instruct}/no_revision_available/CmedqaRetrieval.json +0 -0
  37. results/{gte-Qwen1.5-7B-instruct β†’ Alibaba-NLP__gte-Qwen1.5-7B-instruct}/no_revision_available/Cmnli.json +0 -0
  38. results/{gte-Qwen1.5-7B-instruct β†’ Alibaba-NLP__gte-Qwen1.5-7B-instruct}/no_revision_available/CovidRetrieval.json +0 -0
  39. results/{gte-Qwen1.5-7B-instruct β†’ Alibaba-NLP__gte-Qwen1.5-7B-instruct}/no_revision_available/DBPedia.json +0 -0
  40. results/{gte-Qwen1.5-7B-instruct β†’ Alibaba-NLP__gte-Qwen1.5-7B-instruct}/no_revision_available/DuRetrieval.json +0 -0
  41. results/{gte-Qwen1.5-7B-instruct β†’ Alibaba-NLP__gte-Qwen1.5-7B-instruct}/no_revision_available/EcomRetrieval.json +0 -0
  42. results/{gte-Qwen1.5-7B-instruct β†’ Alibaba-NLP__gte-Qwen1.5-7B-instruct}/no_revision_available/EmotionClassification.json +0 -0
  43. results/{gte-Qwen1.5-7B-instruct β†’ Alibaba-NLP__gte-Qwen1.5-7B-instruct}/no_revision_available/FEVER.json +0 -0
  44. results/{gte-Qwen1.5-7B-instruct β†’ Alibaba-NLP__gte-Qwen1.5-7B-instruct}/no_revision_available/FiQA2018.json +0 -0
  45. results/{gte-Qwen1.5-7B-instruct β†’ Alibaba-NLP__gte-Qwen1.5-7B-instruct}/no_revision_available/HotpotQA.json +0 -0
  46. results/{gte-Qwen1.5-7B-instruct β†’ Alibaba-NLP__gte-Qwen1.5-7B-instruct}/no_revision_available/IFlyTek.json +0 -0
  47. results/{gte-Qwen1.5-7B-instruct β†’ Alibaba-NLP__gte-Qwen1.5-7B-instruct}/no_revision_available/ImdbClassification.json +0 -0
  48. results/{gte-Qwen1.5-7B-instruct β†’ Alibaba-NLP__gte-Qwen1.5-7B-instruct}/no_revision_available/JDReview.json +0 -0
  49. results/{gte-Qwen1.5-7B-instruct β†’ Alibaba-NLP__gte-Qwen1.5-7B-instruct}/no_revision_available/LCQMC.json +0 -0
  50. results/{gte-Qwen1.5-7B-instruct β†’ Alibaba-NLP__gte-Qwen1.5-7B-instruct}/no_revision_available/MMarcoReranking.json +0 -0
paths.json CHANGED
The diff for this file is too large to render. See raw diff
 
results.py CHANGED
@@ -1,4 +1,5 @@
1
  """MTEB Results"""
 
2
  from __future__ import annotations
3
 
4
  import json
@@ -24,7 +25,205 @@ _DESCRIPTION = """Results on MTEB"""
24
 
25
  URL = "https://huggingface.co/datasets/mteb/results/resolve/main/paths.json"
26
  VERSION = datasets.Version("1.0.1")
27
- EVAL_LANGS = ['af', 'afr-eng', 'am', "amh", 'amh-eng', 'ang-eng', 'ar', 'ar-ar', 'ara-eng', 'arq-eng', 'arz-eng', 'ast-eng', 'awa-eng', 'az', 'aze-eng', 'bel-eng', 'ben-eng', 'ber-eng', 'bn', 'bos-eng', 'bre-eng', 'bul-eng', 'cat-eng', 'cbk-eng', 'ceb-eng', 'ces-eng', 'cha-eng', 'cmn-eng', 'cor-eng', 'csb-eng', 'cy', 'cym-eng', 'da', 'dan-eng', 'de', 'de-fr', 'de-pl', 'deu-eng', 'dsb-eng', 'dtp-eng', 'el', 'ell-eng', 'en', 'en-ar', 'en-de', 'en-en', 'en-tr', 'eng', 'epo-eng', 'es', 'es-en', 'es-es', 'es-it', 'est-eng', 'eus-eng', 'fa', 'fao-eng', 'fi', 'fin-eng', 'fr', 'fr-en', 'fr-pl', 'fra', 'fra-eng', 'fry-eng', 'gla-eng', 'gle-eng', 'glg-eng', 'gsw-eng', 'hau', 'he', 'heb-eng', 'hi', 'hin-eng', 'hrv-eng', 'hsb-eng', 'hu', 'hun-eng', 'hy', 'hye-eng', 'ibo', 'id', 'ido-eng', 'ile-eng', 'ina-eng', 'ind-eng', 'is', 'isl-eng', 'it', 'it-en', 'ita-eng', 'ja', 'jav-eng', 'jpn-eng', 'jv', 'ka', 'kab-eng', 'kat-eng', 'kaz-eng', 'khm-eng', 'km', 'kn', 'ko', 'ko-ko', 'kor-eng', 'kur-eng', 'kzj-eng', 'lat-eng', 'lfn-eng', 'lit-eng', 'lin', 'lug', 'lv', 'lvs-eng', 'mal-eng', 'mar-eng', 'max-eng', 'mhr-eng', 'mkd-eng', 'ml', 'mn', 'mon-eng', 'ms', 'my', 'nb', 'nds-eng', 'nl', 'nl-ende-en', 'nld-eng', 'nno-eng', 'nob-eng', 'nov-eng', 'oci-eng', 'orm', 'orv-eng', 'pam-eng', 'pcm', 'pes-eng', 'pl', 'pl-en', 'pms-eng', 'pol-eng', 'por-eng', 'pt', 'ro', 'ron-eng', 'ru', 'run', 'rus-eng', 'sl', 'slk-eng', 'slv-eng', 'spa-eng', 'sna', 'som', 'sq', 'sqi-eng', 'srp-eng', 'sv', 'sw', 'swa', 'swe-eng', 'swg-eng', 'swh-eng', 'ta', 'tam-eng', 'tat-eng', 'te', 'tel-eng', 'tgl-eng', 'th', 'tha-eng', 'tir', 'tl', 'tr', 'tuk-eng', 'tur-eng', 'tzl-eng', 'uig-eng', 'ukr-eng', 'ur', 'urd-eng', 'uzb-eng', 'vi', 'vie-eng', 'war-eng', 'wuu-eng', 'xho', 'xho-eng', 'yid-eng', 'yor', 'yue-eng', 'zh', 'zh-CN', 'zh-TW', 'zh-en', 'zsm-eng']
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
28
 
29
  # v_measures key is somehow present in voyage-2-law results and is a list
30
  SKIP_KEYS = ["std", "evaluation_time", "main_score", "threshold", "v_measures", "scores_per_experiment"]
@@ -32,9 +231,35 @@ SKIP_KEYS = ["std", "evaluation_time", "main_score", "threshold", "v_measures",
32
  # Use "train" split instead
33
  TRAIN_SPLIT = ["DanishPoliticalCommentsClassification"]
34
  # Use "validation" split instead
35
- VALIDATION_SPLIT = ["AFQMC", "Cmnli", "IFlyTek", "LEMBSummScreenFDRetrieval", "MSMARCO", "MSMARCO-PL", "MultilingualSentiment", "Ocnli", "TNews"]
 
 
 
 
 
 
 
 
 
 
36
  # Use "dev" split instead
37
- DEV_SPLIT = ["CmedqaRetrieval", "CovidRetrieval", "DuRetrieval", "EcomRetrieval", "MedicalRetrieval", "MMarcoReranking", "MMarcoRetrieval", "MSMARCO", "MSMARCO-PL", "T2Reranking", "T2Retrieval", "VideoRetrieval", "TERRa", "MIRACLReranking", "MIRACLRetrieval"]
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
38
  # Use "test.full" split
39
  TESTFULL_SPLIT = ["OpusparcusPC"]
40
  # Use "standard" split
@@ -43,219 +268,237 @@ STANDARD_SPLIT = ["BrightRetrieval"]
43
  DEVTEST_SPLIT = ["FloresBitextMining"]
44
 
45
  TEST_AVG_SPLIT = {
46
- "LEMBNeedleRetrieval": ["test_256", "test_512", "test_1024", "test_2048", "test_4096", "test_8192", "test_16384", "test_32768"],
47
- "LEMBPasskeyRetrieval": ["test_256", "test_512", "test_1024", "test_2048", "test_4096", "test_8192", "test_16384", "test_32768"],
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
48
  }
49
 
50
  MODELS = [
51
- "Baichuan-text-embedding",
52
- "Cohere-embed-english-v3.0",
53
- "Cohere-embed-english-v3.0-instruct",
54
- "Cohere-embed-multilingual-light-v3.0",
55
- "Cohere-embed-multilingual-v3.0",
56
- "DanskBERT",
57
- "FollowIR-7B",
58
- "GritLM-7B",
59
- "GritLM-7B-noinstruct",
60
- "LASER2",
61
- "LLM2Vec-Llama-2-supervised",
62
- "LLM2Vec-Llama-2-unsupervised",
63
- "LLM2Vec-Meta-Llama-3-supervised",
64
- "LLM2Vec-Meta-Llama-3-unsupervised",
65
- "LLM2Vec-Mistral-supervised",
66
- "LLM2Vec-Mistral-unsupervised",
67
- "LLM2Vec-Sheared-Llama-supervised",
68
- "LLM2Vec-Sheared-Llama-unsupervised",
69
- "LaBSE",
70
- "OpenSearch-text-hybrid",
71
- "SFR-Embedding-Mistral",
72
- "all-MiniLM-L6-v2",
73
- "all-MiniLM-L6-v2-instruct",
74
- "all-mpnet-base-v2",
75
- "all-mpnet-base-v2-instruct",
76
- "allenai-specter",
77
- "bert-base-10lang-cased",
78
- "bert-base-15lang-cased",
79
- "bert-base-25lang-cased",
80
- "bert-base-multilingual-cased",
81
- "bert-base-multilingual-uncased",
82
- "bert-base-swedish-cased",
83
- "bert-base-uncased",
84
- "bge-base-en-v1.5",
85
- "bge-base-en-v1.5-instruct",
86
- "bge-base-en",
87
- "bge-base-zh",
88
- "bge-base-zh-v1.5",
89
- "bge-large-en",
90
- "bge-large-en-v1.5",
91
- "bge-large-en-v1.5-instruct",
92
- "bge-large-zh",
93
- "bge-large-zh-noinstruct",
94
- "bge-large-zh-v1.5",
95
- "bge-m3",
96
- "bge-m3-instruct",
97
- "bge-small-en-v1.5",
98
- "bge-small-en-v1.5-instruct",
99
- "bge-small-zh",
100
- "bge-small-zh-v1.5",
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
101
  "bm25",
102
  "bm25s",
103
- "camembert-base",
104
- "camembert-large",
105
- "contriever",
106
- "contriever-instruct",
107
- "contriever-base-msmarco",
108
- "cross-en-de-roberta-sentence-transformer",
109
- "dfm-encoder-large-v1",
110
- "distilbert-base-25lang-cased",
111
- "distilbert-base-en-fr-cased",
112
- "distilbert-base-en-fr-es-pt-it-cased",
113
- "distilbert-base-fr-cased",
114
- "distilbert-base-uncased",
115
- "distiluse-base-multilingual-cased-v2",
116
- "dragon-plus",
117
- "dragon-plus-instruct",
118
- "e5-base",
119
- "e5-base-4k",
120
- "e5-base-v2",
121
- "e5-large",
122
- "e5-large-v2",
123
- "e5-mistral-7b-instruct",
124
- "e5-mistral-7b-instruct-noinstruct",
125
- "e5-small",
126
- "e5-small-v2",
127
- "electra-small-nordic",
128
- "electra-small-swedish-cased-discriminator",
129
- "elser-v2",
130
- "embedder-100p",
131
- "facebook-dpr-ctx_encoder-multiset-base",
132
- "facebookdragon-plus-context-encoder",
133
- "flan-t5-base",
134
- "flan-t5-large",
135
- "flaubert_base_cased",
136
- "flaubert_base_uncased",
137
- "flaubert_large_cased",
138
- "gbert-base",
139
- "gbert-large",
140
- "gelectra-base",
141
- "gelectra-large",
142
- "glove.6B.300d",
143
- "google-gecko-256.text-embedding-preview-0409",
144
- "google-gecko.text-embedding-preview-0409",
145
- "gottbert-base",
146
- "gte-Qwen1.5-7B-instruct",
147
- "gte-Qwen2-7B-instruct",
148
- "gtr-t5-base",
149
- "gtr-t5-large",
150
- "gtr-t5-xl",
151
- "gtr-t5-xxl",
152
- "herbert-base-retrieval-v2",
153
- "instructor-base",
154
- "instructor-large",
155
- "instructor-xl",
156
- "jina-embeddings-v2-base-en",
157
- "komninos",
158
- "llama-2-7b-chat",
159
- "luotuo-bert-medium",
160
- "m3e-base",
161
- "m3e-large",
162
- "mistral-7b-instruct-v0.2",
163
- "mistral-embed",
164
- "monobert-large-msmarco",
165
- "monot5-3b-msmarco-10k",
166
- "monot5-base-msmarco-10k",
167
- "msmarco-bert-co-condensor",
168
- "multi-qa-MiniLM-L6-cos-v1",
169
- "multilingual-e5-base",
170
- "multilingual-e5-large",
171
- "multilingual-e5-large-instruct",
172
- "multilingual-e5-small",
173
- "mxbai-embed-large-v1",
174
- "nb-bert-base",
175
- "nb-bert-large",
176
- "nomic-embed-text-v1",
177
- "nomic-embed-text-v1.5-128",
178
- "nomic-embed-text-v1.5-256",
179
- "nomic-embed-text-v1.5-512",
180
- "nomic-embed-text-v1.5-64",
181
- "norbert3-base",
182
- "norbert3-large",
183
- "paraphrase-multilingual-MiniLM-L12-v2",
184
- "paraphrase-multilingual-mpnet-base-v2",
185
- "rubert-tiny",
186
- "rubert-tiny2",
187
- "sbert_large_mt_nlu_ru",
188
- "sbert_large_nlu_ru",
189
- "sentence-bert-swedish-cased",
190
- "sentence-camembert-base",
191
- "sentence-camembert-large",
192
- "sentence-croissant-llm-base",
193
- "sentence-t5-base",
194
- "sentence-t5-large",
195
- "sentence-t5-xl",
196
- "sentence-t5-xxl",
197
- "all-MiniLM-L12-v2",
198
- "sgpt-bloom-1b7-nli",
199
- "sgpt-bloom-7b1-msmarco",
200
- "SGPT-125M-weightedmean-nli-bitfit",
201
- "SGPT-1.3B-weightedmean-msmarco-specb-bitfit",
202
- "SGPT-5.8B-weightedmean-msmarco-specb-bitfit-que",
203
- "SGPT-5.8B-weightedmean-msmarco-specb-bitfit",
204
- "SGPT-5.8B-weightedmean-nli-bitfit",
205
- "SGPT-2.7B-weightedmean-msmarco-specb-bitfit",
206
- "SGPT-125M-weightedmean-msmarco-specb-bitfit-que",
207
- "SGPT-125M-weightedmean-msmarco-specb-bitfit-doc",
208
- "SGPT-125M-weightedmean-msmarco-specb-bitfit",
209
- "silver-retriever-base-v1",
210
- "st-polish-paraphrase-from-distilroberta",
211
- "st-polish-paraphrase-from-mpnet",
212
- "sup-simcse-bert-base-uncased",
213
- "tart-dual-contriever-msmarco",
214
- "tart-full-flan-t5-xl",
215
- "text-embedding-3-large",
216
- "text-embedding-3-large-instruct",
217
- "text-embedding-3-large-256",
218
- "text-embedding-3-small",
219
- "text-embedding-3-small-instruct",
220
- "text-embedding-ada-002",
221
- "text-embedding-ada-002-instruct",
222
- "text-search-ada-001",
223
- "text-search-ada-doc-001",
224
- "text-search-babbage-001",
225
- "text-search-curie-001",
226
- "text-search-davinci-001",
227
- "text-similarity-ada-001",
228
- "text-similarity-babbage-001",
229
- "text-similarity-curie-001",
230
- "text-similarity-davinci-001",
231
- "text2vec-base-chinese",
232
- "text2vec-base-multilingual",
233
- "text2vec-large-chinese",
234
- "titan-embed-text-v1",
235
- "udever-bloom-1b1",
236
- "udever-bloom-560m",
237
- "universal-sentence-encoder-multilingual-3",
238
- "universal-sentence-encoder-multilingual-large-3",
239
- "unsup-simcse-bert-base-uncased",
240
- "use-cmlm-multilingual",
241
- "voyage-2",
242
- "voyage-code-2",
243
- "voyage-large-2-instruct",
244
- "voyage-law-2",
245
- "voyage-lite-01-instruct",
246
- "voyage-lite-02-instruct",
247
- "voyage-multilingual-2",
248
- "xlm-roberta-base",
249
- "xlm-roberta-large",
250
- "deberta-v1-base",
251
- "USER-bge-m3",
252
- "USER-base",
253
- "rubert-tiny-turbo",
254
- "LaBSE-ru-turbo",
255
- "distilrubert-small-cased-conversational",
256
- "rubert-base-cased",
257
- "rubert-base-cased-sentence",
258
- "LaBSE-en-ru",
259
  ]
260
 
261
 
@@ -269,6 +512,7 @@ def get_model_for_current_dir(dir_name: str) -> str | None:
269
  # Needs to be run whenever new files are added
270
  def get_paths():
271
  import collections, json, os
 
272
  files = collections.defaultdict(list)
273
  for model_dir in os.listdir("results"):
274
  results_model_dir = os.path.join("results", model_dir)
@@ -283,7 +527,9 @@ def get_paths():
283
  if not os.path.isdir(os.path.join(results_model_dir, revision_folder)):
284
  continue
285
  for res_file in os.listdir(os.path.join(results_model_dir, revision_folder)):
286
- if (res_file.endswith(".json")) and not(res_file.endswith(("overall_results.json", "model_meta.json"))):
 
 
287
  results_model_file = os.path.join(results_model_dir, revision_folder, res_file)
288
  files[model_name].append(results_model_file)
289
  with open("paths.json", "w") as f:
@@ -327,12 +573,7 @@ class MTEBResults(datasets.GeneratorBasedBuilder):
327
  with open(path_file) as f:
328
  files = json.load(f)
329
  downloaded_files = dl_manager.download_and_extract(files[self.config.name])
330
- return [
331
- datasets.SplitGenerator(
332
- name=datasets.Split.TEST,
333
- gen_kwargs={'filepath': downloaded_files}
334
- )
335
- ]
336
 
337
  def _generate_examples(self, filepath):
338
  """This function returns the examples in the raw (text) form."""
@@ -356,7 +597,7 @@ class MTEBResults(datasets.GeneratorBasedBuilder):
356
  split = "dev"
357
  elif (ds_name in TESTFULL_SPLIT) and ("test.full" in res_dict):
358
  split = "test.full"
359
- elif (ds_name in STANDARD_SPLIT):
360
  split = []
361
  if "standard" in res_dict:
362
  split += ["standard"]
@@ -364,7 +605,7 @@ class MTEBResults(datasets.GeneratorBasedBuilder):
364
  split += ["long"]
365
  elif (ds_name in DEVTEST_SPLIT) and ("devtest" in res_dict):
366
  split = "devtest"
367
- elif (ds_name in TEST_AVG_SPLIT):
368
  # Average splits
369
  res_dict = {}
370
  for split in TEST_AVG_SPLIT[ds_name]:
@@ -385,7 +626,8 @@ class MTEBResults(datasets.GeneratorBasedBuilder):
385
  for k, v in res_dict[split][0].items():
386
  if k in ["hf_subset", "languages"]:
387
  res_dict[k] = v
388
- if not isinstance(v, float): continue
 
389
  v /= len(TEST_AVG_SPLIT[ds_name])
390
  if k not in res_dict:
391
  res_dict[k] = v
@@ -414,41 +656,48 @@ class MTEBResults(datasets.GeneratorBasedBuilder):
414
  if not lang:
415
  lang = subset
416
  for metric, score in res.items():
417
- if metric in SKIP_KEYS: continue
 
418
  if isinstance(score, dict):
419
  # Legacy format with e.g. {cosine: {spearman: ...}}
420
  # Now it is {cosine_spearman: ...}
421
  for k, v in score.items():
422
  if not isinstance(v, float):
423
- print(f'WARNING: Expected float, got {v} for {ds_name} {lang} {metric} {k}')
424
  continue
425
- if metric in SKIP_KEYS: continue
426
- out.append({
427
- "mteb_dataset_name": ds_name,
428
- "eval_language": lang,
429
- "metric": metric + "_" + k,
430
- "score": v * 100,
431
- "hf_subset": subset,
432
- })
 
 
 
433
  else:
434
  if not isinstance(score, float):
435
- print(f'WARNING: Expected float, got {score} for {ds_name} {lang} {metric}')
436
  continue
437
- out.append({
438
- "mteb_dataset_name": ds_name,
439
- "eval_language": lang,
440
- "metric": metric,
441
- "score": score * 100,
442
- "split": split,
443
- "hf_subset": subset,
444
- })
 
 
445
 
446
  ### Old MTEB format ###
447
  else:
448
  is_multilingual = any(x in res_dict for x in EVAL_LANGS)
449
  langs = res_dict.keys() if is_multilingual else ["en"]
450
  for lang in langs:
451
- if lang in SKIP_KEYS: continue
 
452
  test_result_lang = res_dict.get(lang) if is_multilingual else res_dict
453
  subset = test_result_lang.pop("hf_subset", "")
454
  if subset == "" and is_multilingual:
@@ -457,16 +706,20 @@ class MTEBResults(datasets.GeneratorBasedBuilder):
457
  if not isinstance(score, dict):
458
  score = {metric: score}
459
  for sub_metric, sub_score in score.items():
460
- if any(x in sub_metric for x in SKIP_KEYS): continue
461
- if isinstance(sub_score, dict): continue
462
- out.append({
463
- "mteb_dataset_name": ds_name,
464
- "eval_language": lang if is_multilingual else "",
465
- "metric": f"{metric}_{sub_metric}" if metric != sub_metric else metric,
466
- "score": sub_score * 100,
467
- "split": split,
468
- "hf_subset": subset,
469
- })
 
 
 
 
470
  for idx, row in enumerate(sorted(out, key=lambda x: x["mteb_dataset_name"])):
471
  yield idx, row
472
 
 
1
  """MTEB Results"""
2
+
3
  from __future__ import annotations
4
 
5
  import json
 
25
 
26
  URL = "https://huggingface.co/datasets/mteb/results/resolve/main/paths.json"
27
  VERSION = datasets.Version("1.0.1")
28
+ EVAL_LANGS = [
29
+ "af",
30
+ "afr-eng",
31
+ "am",
32
+ "amh",
33
+ "amh-eng",
34
+ "ang-eng",
35
+ "ar",
36
+ "ar-ar",
37
+ "ara-eng",
38
+ "arq-eng",
39
+ "arz-eng",
40
+ "ast-eng",
41
+ "awa-eng",
42
+ "az",
43
+ "aze-eng",
44
+ "bel-eng",
45
+ "ben-eng",
46
+ "ber-eng",
47
+ "bn",
48
+ "bos-eng",
49
+ "bre-eng",
50
+ "bul-eng",
51
+ "cat-eng",
52
+ "cbk-eng",
53
+ "ceb-eng",
54
+ "ces-eng",
55
+ "cha-eng",
56
+ "cmn-eng",
57
+ "cor-eng",
58
+ "csb-eng",
59
+ "cy",
60
+ "cym-eng",
61
+ "da",
62
+ "dan-eng",
63
+ "de",
64
+ "de-fr",
65
+ "de-pl",
66
+ "deu-eng",
67
+ "dsb-eng",
68
+ "dtp-eng",
69
+ "el",
70
+ "ell-eng",
71
+ "en",
72
+ "en-ar",
73
+ "en-de",
74
+ "en-en",
75
+ "en-tr",
76
+ "eng",
77
+ "epo-eng",
78
+ "es",
79
+ "es-en",
80
+ "es-es",
81
+ "es-it",
82
+ "est-eng",
83
+ "eus-eng",
84
+ "fa",
85
+ "fao-eng",
86
+ "fi",
87
+ "fin-eng",
88
+ "fr",
89
+ "fr-en",
90
+ "fr-pl",
91
+ "fra",
92
+ "fra-eng",
93
+ "fry-eng",
94
+ "gla-eng",
95
+ "gle-eng",
96
+ "glg-eng",
97
+ "gsw-eng",
98
+ "hau",
99
+ "he",
100
+ "heb-eng",
101
+ "hi",
102
+ "hin-eng",
103
+ "hrv-eng",
104
+ "hsb-eng",
105
+ "hu",
106
+ "hun-eng",
107
+ "hy",
108
+ "hye-eng",
109
+ "ibo",
110
+ "id",
111
+ "ido-eng",
112
+ "ile-eng",
113
+ "ina-eng",
114
+ "ind-eng",
115
+ "is",
116
+ "isl-eng",
117
+ "it",
118
+ "it-en",
119
+ "ita-eng",
120
+ "ja",
121
+ "jav-eng",
122
+ "jpn-eng",
123
+ "jv",
124
+ "ka",
125
+ "kab-eng",
126
+ "kat-eng",
127
+ "kaz-eng",
128
+ "khm-eng",
129
+ "km",
130
+ "kn",
131
+ "ko",
132
+ "ko-ko",
133
+ "kor-eng",
134
+ "kur-eng",
135
+ "kzj-eng",
136
+ "lat-eng",
137
+ "lfn-eng",
138
+ "lit-eng",
139
+ "lin",
140
+ "lug",
141
+ "lv",
142
+ "lvs-eng",
143
+ "mal-eng",
144
+ "mar-eng",
145
+ "max-eng",
146
+ "mhr-eng",
147
+ "mkd-eng",
148
+ "ml",
149
+ "mn",
150
+ "mon-eng",
151
+ "ms",
152
+ "my",
153
+ "nb",
154
+ "nds-eng",
155
+ "nl",
156
+ "nl-ende-en",
157
+ "nld-eng",
158
+ "nno-eng",
159
+ "nob-eng",
160
+ "nov-eng",
161
+ "oci-eng",
162
+ "orm",
163
+ "orv-eng",
164
+ "pam-eng",
165
+ "pcm",
166
+ "pes-eng",
167
+ "pl",
168
+ "pl-en",
169
+ "pms-eng",
170
+ "pol-eng",
171
+ "por-eng",
172
+ "pt",
173
+ "ro",
174
+ "ron-eng",
175
+ "ru",
176
+ "run",
177
+ "rus-eng",
178
+ "sl",
179
+ "slk-eng",
180
+ "slv-eng",
181
+ "spa-eng",
182
+ "sna",
183
+ "som",
184
+ "sq",
185
+ "sqi-eng",
186
+ "srp-eng",
187
+ "sv",
188
+ "sw",
189
+ "swa",
190
+ "swe-eng",
191
+ "swg-eng",
192
+ "swh-eng",
193
+ "ta",
194
+ "tam-eng",
195
+ "tat-eng",
196
+ "te",
197
+ "tel-eng",
198
+ "tgl-eng",
199
+ "th",
200
+ "tha-eng",
201
+ "tir",
202
+ "tl",
203
+ "tr",
204
+ "tuk-eng",
205
+ "tur-eng",
206
+ "tzl-eng",
207
+ "uig-eng",
208
+ "ukr-eng",
209
+ "ur",
210
+ "urd-eng",
211
+ "uzb-eng",
212
+ "vi",
213
+ "vie-eng",
214
+ "war-eng",
215
+ "wuu-eng",
216
+ "xho",
217
+ "xho-eng",
218
+ "yid-eng",
219
+ "yor",
220
+ "yue-eng",
221
+ "zh",
222
+ "zh-CN",
223
+ "zh-TW",
224
+ "zh-en",
225
+ "zsm-eng",
226
+ ]
227
 
228
  # v_measures key is somehow present in voyage-2-law results and is a list
229
  SKIP_KEYS = ["std", "evaluation_time", "main_score", "threshold", "v_measures", "scores_per_experiment"]
 
231
  # Use "train" split instead
232
  TRAIN_SPLIT = ["DanishPoliticalCommentsClassification"]
233
  # Use "validation" split instead
234
+ VALIDATION_SPLIT = [
235
+ "AFQMC",
236
+ "Cmnli",
237
+ "IFlyTek",
238
+ "LEMBSummScreenFDRetrieval",
239
+ "MSMARCO",
240
+ "MSMARCO-PL",
241
+ "MultilingualSentiment",
242
+ "Ocnli",
243
+ "TNews",
244
+ ]
245
  # Use "dev" split instead
246
+ DEV_SPLIT = [
247
+ "CmedqaRetrieval",
248
+ "CovidRetrieval",
249
+ "DuRetrieval",
250
+ "EcomRetrieval",
251
+ "MedicalRetrieval",
252
+ "MMarcoReranking",
253
+ "MMarcoRetrieval",
254
+ "MSMARCO",
255
+ "MSMARCO-PL",
256
+ "T2Reranking",
257
+ "T2Retrieval",
258
+ "VideoRetrieval",
259
+ "TERRa",
260
+ "MIRACLReranking",
261
+ "MIRACLRetrieval",
262
+ ]
263
  # Use "test.full" split
264
  TESTFULL_SPLIT = ["OpusparcusPC"]
265
  # Use "standard" split
 
268
  DEVTEST_SPLIT = ["FloresBitextMining"]
269
 
270
  TEST_AVG_SPLIT = {
271
+ "LEMBNeedleRetrieval": [
272
+ "test_256",
273
+ "test_512",
274
+ "test_1024",
275
+ "test_2048",
276
+ "test_4096",
277
+ "test_8192",
278
+ "test_16384",
279
+ "test_32768",
280
+ ],
281
+ "LEMBPasskeyRetrieval": [
282
+ "test_256",
283
+ "test_512",
284
+ "test_1024",
285
+ "test_2048",
286
+ "test_4096",
287
+ "test_8192",
288
+ "test_16384",
289
+ "test_32768",
290
+ ],
291
  }
292
 
293
  MODELS = [
294
+ "Alibaba-NLP__gte-Qwen1.5-7B-instruct",
295
+ "Alibaba-NLP__gte-Qwen2-7B-instruct",
296
+ "BAAI__bge-base-en",
297
+ "BAAI__bge-base-en-v1.5",
298
+ "BAAI__bge-base-en-v1.5-instruct",
299
+ "BAAI__bge-base-zh",
300
+ "BAAI__bge-base-zh-v1.5",
301
+ "BAAI__bge-large-en",
302
+ "BAAI__bge-large-en-v1.5",
303
+ "BAAI__bge-large-en-v1.5-instruct",
304
+ "BAAI__bge-large-zh",
305
+ "BAAI__bge-large-zh-noinstruct",
306
+ "BAAI__bge-large-zh-v1.5",
307
+ "BAAI__bge-m3",
308
+ "BAAI__bge-m3-instruct",
309
+ "BAAI__bge-small-en-v1.5",
310
+ "BAAI__bge-small-en-v1.5-instruct",
311
+ "BAAI__bge-small-zh",
312
+ "BAAI__bge-small-zh-v1.5",
313
+ "Cohere__Cohere-embed-english-v3.0",
314
+ "Cohere__Cohere-embed-english-v3.0-instruct",
315
+ "Cohere__Cohere-embed-multilingual-light-v3.0",
316
+ "Cohere__Cohere-embed-multilingual-v3.0",
317
+ "DeepPavlov__distilrubert-small-cased-conversational",
318
+ "DeepPavlov__rubert-base-cased",
319
+ "DeepPavlov__rubert-base-cased-sentence",
320
+ "FacebookAI__xlm-roberta-base",
321
+ "FacebookAI__xlm-roberta-large",
322
+ "Geotrend__bert-base-10lang-cased",
323
+ "Geotrend__bert-base-15lang-cased",
324
+ "Geotrend__bert-base-25lang-cased",
325
+ "Geotrend__distilbert-base-25lang-cased",
326
+ "Geotrend__distilbert-base-en-fr-cased",
327
+ "Geotrend__distilbert-base-en-fr-es-pt-it-cased",
328
+ "Geotrend__distilbert-base-fr-cased",
329
+ "GritLM__GritLM-7B",
330
+ "GritLM__GritLM-7B-noinstruct",
331
+ "KBLab__electra-small-swedish-cased-discriminator",
332
+ "KBLab__sentence-bert-swedish-cased",
333
+ "KB__bert-base-swedish-cased",
334
+ "McGill-NLP__LLM2Vec-Llama-2-7b-chat-hf-mntp-supervised",
335
+ "McGill-NLP__LLM2Vec-Llama-2-unsupervised",
336
+ "McGill-NLP__LLM2Vec-Meta-Llama-3-supervised",
337
+ "McGill-NLP__LLM2Vec-Meta-Llama-3-unsupervised",
338
+ "McGill-NLP__LLM2Vec-Mistral-supervised",
339
+ "McGill-NLP__LLM2Vec-Mistral-unsupervised",
340
+ "McGill-NLP__LLM2Vec-Sheared-Llama-supervised",
341
+ "McGill-NLP__LLM2Vec-Sheared-Llama-unsupervised",
342
+ "Muennighoff__SGPT-1.3B-weightedmean-msmarco-specb-bitfit",
343
+ "Muennighoff__SGPT-125M-weightedmean-msmarco-specb-bitfit",
344
+ "Muennighoff__SGPT-125M-weightedmean-msmarco-specb-bitfit-doc",
345
+ "Muennighoff__SGPT-125M-weightedmean-msmarco-specb-bitfit-que",
346
+ "Muennighoff__SGPT-125M-weightedmean-nli-bitfit",
347
+ "Muennighoff__SGPT-2.7B-weightedmean-msmarco-specb-bitfit",
348
+ "Muennighoff__SGPT-5.8B-weightedmean-msmarco-specb-bitfit",
349
+ "Muennighoff__SGPT-5.8B-weightedmean-msmarco-specb-bitfit-que",
350
+ "Muennighoff__SGPT-5.8B-weightedmean-nli-bitfit",
351
+ "NbAiLab__nb-bert-base",
352
+ "NbAiLab__nb-bert-large",
353
+ "Salesforce__SFR-Embedding-Mistral",
354
+ "T-Systems-onsite__cross-en-de-roberta-sentence-transformer",
355
+ "Wissam42__sentence-croissant-llm-base",
356
+ "ai-forever__sbert_large_mt_nlu_ru",
357
+ "ai-forever__sbert_large_nlu_ru",
358
+ "aliyun__OpenSearch-text-hybrid",
359
+ "almanach__camembert-base",
360
+ "almanach__camembert-large",
361
+ "amazon__titan-embed-text-v1",
362
+ "baichuan-ai__text-embedding",
363
+ "bigscience-data__sgpt-bloom-1b7-nli",
364
+ "bigscience-data__sgpt-bloom-7b1-msmarco",
365
  "bm25",
366
  "bm25s",
367
+ "castorini__monobert-large-msmarco",
368
+ "castorini__monot5-3b-msmarco-10k",
369
+ "castorini__monot5-base-msmarco-10k",
370
+ "chcaa__dfm-encoder-large-v1",
371
+ "cointegrated__LaBSE-en-ru",
372
+ "cointegrated__rubert-tiny",
373
+ "cointegrated__rubert-tiny2",
374
+ "dangvantuan__sentence-camembert-base",
375
+ "dangvantuan__sentence-camembert-large",
376
+ "deepfile__embedder-100p",
377
+ "deepset__gbert-base",
378
+ "deepset__gbert-large",
379
+ "deepset__gelectra-base",
380
+ "deepset__gelectra-large",
381
+ "deepvk__USER-base",
382
+ "deepvk__USER-bge-m3",
383
+ "deepvk__deberta-v1-base",
384
+ "distilbert__distilbert-base-uncased",
385
+ "dwzhu__e5-base-4k",
386
+ "elastic__elser-v2",
387
+ "facebook__contriever",
388
+ "facebook__contriever-instruct",
389
+ "facebook__dpr-ctx_encoder-multiset-base",
390
+ "facebook__dragon-plus-context-encoder",
391
+ "facebook__tart-full-flan-t5-xl",
392
+ "facebookresearch__LASER2",
393
+ "facebookresearch__dragon-plus",
394
+ "facebookresearch__dragon-plus-instruct",
395
+ "flaubert__flaubert_base_cased",
396
+ "flaubert__flaubert_base_uncased",
397
+ "flaubert__flaubert_large_cased",
398
+ "google-bert__bert-base-multilingual-cased",
399
+ "google-bert__bert-base-multilingual-uncased",
400
+ "google-bert__bert-base-uncased",
401
+ "google-gecko__text-embedding-preview-0409",
402
+ "google-gecko__text-embedding-preview-0409-256",
403
+ "google__flan-t5-base",
404
+ "google__flan-t5-large",
405
+ "hkunlp__instructor-base",
406
+ "hkunlp__instructor-large",
407
+ "hkunlp__instructor-xl",
408
+ "intfloat__e5-base",
409
+ "intfloat__e5-base-v2",
410
+ "intfloat__e5-large",
411
+ "intfloat__e5-large-v2",
412
+ "intfloat__e5-mistral-7b-instruct",
413
+ "intfloat__e5-mistral-7b-instruct-noinstruct",
414
+ "intfloat__e5-small",
415
+ "intfloat__e5-small-v2",
416
+ "intfloat__multilingual-e5-base",
417
+ "intfloat__multilingual-e5-large",
418
+ "intfloat__multilingual-e5-large-instruct",
419
+ "intfloat__multilingual-e5-small",
420
+ "ipipan__herbert-base-retrieval-v2",
421
+ "ipipan__silver-retriever-base-v1",
422
+ "izhx__udever-bloom-1b1",
423
+ "izhx__udever-bloom-560m",
424
+ "jhu-clsp__FollowIR-7B",
425
+ "jinaai__jina-embeddings-v2-base-en",
426
+ "jonfd__electra-small-nordic",
427
+ "ltg__norbert3-base",
428
+ "ltg__norbert3-large",
429
+ "meta-llama__llama-2-7b-chat",
430
+ "mistral__mistral-embed",
431
+ "mistralai__mistral-7b-instruct-v0.2",
432
+ "mixedbread-ai__mxbai-embed-large-v1",
433
+ "moka-ai__m3e-base",
434
+ "moka-ai__m3e-large",
435
+ "nomic-ai__nomic-embed-text-v1",
436
+ "nomic-ai__nomic-embed-text-v1.5-128",
437
+ "nomic-ai__nomic-embed-text-v1.5-256",
438
+ "nomic-ai__nomic-embed-text-v1.5-512",
439
+ "nomic-ai__nomic-embed-text-v1.5-64",
440
+ "nthakur__contriever-base-msmarco",
441
+ "openai__text-embedding-3-large",
442
+ "openai__text-embedding-3-large-256",
443
+ "openai__text-embedding-3-large-instruct",
444
+ "openai__text-embedding-3-small-instruct",
445
+ "openai__text-embedding-ada-002",
446
+ "openai__text-embedding-ada-002-instruct",
447
+ "openai__text-search-ada-001",
448
+ "openai__text-search-ada-doc-001",
449
+ "openai__text-search-babbage-001",
450
+ "openai__text-search-curie-001",
451
+ "openai__text-search-davinci-001",
452
+ "openai__text-similarity-ada-001",
453
+ "openai__text-similarity-babbage-001",
454
+ "openai__text-similarity-curie-001",
455
+ "openai__text-similarity-davinci-001",
456
+ "openai__text-embedding-3-small",
457
+ "orionweller__tart-dual-contriever-msmarco",
458
+ "princeton-nlp__sup-simcse-bert-base-uncased",
459
+ "princeton-nlp__unsup-simcse-bert-base-uncased",
460
+ "sdadas__st-polish-paraphrase-from-distilroberta",
461
+ "sdadas__st-polish-paraphrase-from-mpnet",
462
+ "sentence-transformers__LaBSE",
463
+ "sentence-transformers__all-MiniLM-L12-v2",
464
+ "sentence-transformers__all-MiniLM-L6-v2",
465
+ "sentence-transformers__all-MiniLM-L6-v2-instruct",
466
+ "sentence-transformers__all-mpnet-base-v2",
467
+ "sentence-transformers__all-mpnet-base-v2-instruct",
468
+ "sentence-transformers__allenai-specter",
469
+ "sentence-transformers__average_word_embeddings_glove.6B.300d",
470
+ "sentence-transformers__average_word_embeddings_komninos",
471
+ "sentence-transformers__distiluse-base-multilingual-cased-v2",
472
+ "sentence-transformers__gtr-t5-base",
473
+ "sentence-transformers__gtr-t5-large",
474
+ "sentence-transformers__gtr-t5-xl",
475
+ "sentence-transformers__gtr-t5-xxl",
476
+ "sentence-transformers__msmarco-bert-co-condensor",
477
+ "sentence-transformers__multi-qa-MiniLM-L6-cos-v1",
478
+ "sentence-transformers__paraphrase-multilingual-MiniLM-L12-v2",
479
+ "sentence-transformers__paraphrase-multilingual-mpnet-base-v2",
480
+ "sentence-transformers__sentence-t5-base",
481
+ "sentence-transformers__sentence-t5-large",
482
+ "sentence-transformers__sentence-t5-xl",
483
+ "sentence-transformers__sentence-t5-xxl",
484
+ "sentence-transformers__use-cmlm-multilingual",
485
+ "sergeyzh__LaBSE-ru-turbo",
486
+ "sergeyzh__rubert-tiny-turbo",
487
+ "shibing624__text2vec-base-chinese",
488
+ "shibing624__text2vec-base-multilingual",
489
+ "shibing624__text2vec-large-chinese",
490
+ "silk-road__luotuo-bert-medium",
491
+ "uklfr__gottbert-base",
492
+ "vesteinn__DanskBERT",
493
+ "voyageai__voyage-2",
494
+ "voyageai__voyage-code-2",
495
+ "voyageai__voyage-large-2-instruct",
496
+ "voyageai__voyage-law-2",
497
+ "voyageai__voyage-lite-01-instruct",
498
+ "voyageai__voyage-lite-02-instruct",
499
+ "voyageai__voyage-multilingual-2",
500
+ "vprelovac__universal-sentence-encoder-multilingual-3",
501
+ "vprelovac__universal-sentence-encoder-multilingual-large-3",
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
502
  ]
503
 
504
 
 
512
  # Needs to be run whenever new files are added
513
  def get_paths():
514
  import collections, json, os
515
+
516
  files = collections.defaultdict(list)
517
  for model_dir in os.listdir("results"):
518
  results_model_dir = os.path.join("results", model_dir)
 
527
  if not os.path.isdir(os.path.join(results_model_dir, revision_folder)):
528
  continue
529
  for res_file in os.listdir(os.path.join(results_model_dir, revision_folder)):
530
+ if (res_file.endswith(".json")) and not (
531
+ res_file.endswith(("overall_results.json", "model_meta.json"))
532
+ ):
533
  results_model_file = os.path.join(results_model_dir, revision_folder, res_file)
534
  files[model_name].append(results_model_file)
535
  with open("paths.json", "w") as f:
 
573
  with open(path_file) as f:
574
  files = json.load(f)
575
  downloaded_files = dl_manager.download_and_extract(files[self.config.name])
576
+ return [datasets.SplitGenerator(name=datasets.Split.TEST, gen_kwargs={"filepath": downloaded_files})]
 
 
 
 
 
577
 
578
  def _generate_examples(self, filepath):
579
  """This function returns the examples in the raw (text) form."""
 
597
  split = "dev"
598
  elif (ds_name in TESTFULL_SPLIT) and ("test.full" in res_dict):
599
  split = "test.full"
600
+ elif ds_name in STANDARD_SPLIT:
601
  split = []
602
  if "standard" in res_dict:
603
  split += ["standard"]
 
605
  split += ["long"]
606
  elif (ds_name in DEVTEST_SPLIT) and ("devtest" in res_dict):
607
  split = "devtest"
608
+ elif ds_name in TEST_AVG_SPLIT:
609
  # Average splits
610
  res_dict = {}
611
  for split in TEST_AVG_SPLIT[ds_name]:
 
626
  for k, v in res_dict[split][0].items():
627
  if k in ["hf_subset", "languages"]:
628
  res_dict[k] = v
629
+ if not isinstance(v, float):
630
+ continue
631
  v /= len(TEST_AVG_SPLIT[ds_name])
632
  if k not in res_dict:
633
  res_dict[k] = v
 
656
  if not lang:
657
  lang = subset
658
  for metric, score in res.items():
659
+ if metric in SKIP_KEYS:
660
+ continue
661
  if isinstance(score, dict):
662
  # Legacy format with e.g. {cosine: {spearman: ...}}
663
  # Now it is {cosine_spearman: ...}
664
  for k, v in score.items():
665
  if not isinstance(v, float):
666
+ print(f"WARNING: Expected float, got {v} for {ds_name} {lang} {metric} {k}")
667
  continue
668
+ if metric in SKIP_KEYS:
669
+ continue
670
+ out.append(
671
+ {
672
+ "mteb_dataset_name": ds_name,
673
+ "eval_language": lang,
674
+ "metric": metric + "_" + k,
675
+ "score": v * 100,
676
+ "hf_subset": subset,
677
+ }
678
+ )
679
  else:
680
  if not isinstance(score, float):
681
+ print(f"WARNING: Expected float, got {score} for {ds_name} {lang} {metric}")
682
  continue
683
+ out.append(
684
+ {
685
+ "mteb_dataset_name": ds_name,
686
+ "eval_language": lang,
687
+ "metric": metric,
688
+ "score": score * 100,
689
+ "split": split,
690
+ "hf_subset": subset,
691
+ }
692
+ )
693
 
694
  ### Old MTEB format ###
695
  else:
696
  is_multilingual = any(x in res_dict for x in EVAL_LANGS)
697
  langs = res_dict.keys() if is_multilingual else ["en"]
698
  for lang in langs:
699
+ if lang in SKIP_KEYS:
700
+ continue
701
  test_result_lang = res_dict.get(lang) if is_multilingual else res_dict
702
  subset = test_result_lang.pop("hf_subset", "")
703
  if subset == "" and is_multilingual:
 
706
  if not isinstance(score, dict):
707
  score = {metric: score}
708
  for sub_metric, sub_score in score.items():
709
+ if any(x in sub_metric for x in SKIP_KEYS):
710
+ continue
711
+ if isinstance(sub_score, dict):
712
+ continue
713
+ out.append(
714
+ {
715
+ "mteb_dataset_name": ds_name,
716
+ "eval_language": lang if is_multilingual else "",
717
+ "metric": f"{metric}_{sub_metric}" if metric != sub_metric else metric,
718
+ "score": sub_score * 100,
719
+ "split": split,
720
+ "hf_subset": subset,
721
+ }
722
+ )
723
  for idx, row in enumerate(sorted(out, key=lambda x: x["mteb_dataset_name"])):
724
  yield idx, row
725
 
results/{gte-Qwen1.5-7B-instruct β†’ Alibaba-NLP__gte-Qwen1.5-7B-instruct}/no_revision_available/AFQMC.json RENAMED
File without changes
results/{gte-Qwen1.5-7B-instruct β†’ Alibaba-NLP__gte-Qwen1.5-7B-instruct}/no_revision_available/ATEC.json RENAMED
File without changes
results/{gte-Qwen1.5-7B-instruct β†’ Alibaba-NLP__gte-Qwen1.5-7B-instruct}/no_revision_available/AmazonCounterfactualClassification.json RENAMED
File without changes
results/{gte-Qwen1.5-7B-instruct β†’ Alibaba-NLP__gte-Qwen1.5-7B-instruct}/no_revision_available/AmazonPolarityClassification.json RENAMED
File without changes
results/{gte-Qwen1.5-7B-instruct β†’ Alibaba-NLP__gte-Qwen1.5-7B-instruct}/no_revision_available/AmazonReviewsClassification.json RENAMED
File without changes
results/{gte-Qwen1.5-7B-instruct β†’ Alibaba-NLP__gte-Qwen1.5-7B-instruct}/no_revision_available/ArguAna.json RENAMED
File without changes
results/{gte-Qwen1.5-7B-instruct β†’ Alibaba-NLP__gte-Qwen1.5-7B-instruct}/no_revision_available/ArxivClusteringP2P.json RENAMED
File without changes
results/{gte-Qwen1.5-7B-instruct β†’ Alibaba-NLP__gte-Qwen1.5-7B-instruct}/no_revision_available/ArxivClusteringS2S.json RENAMED
File without changes
results/{gte-Qwen1.5-7B-instruct β†’ Alibaba-NLP__gte-Qwen1.5-7B-instruct}/no_revision_available/AskUbuntuDupQuestions.json RENAMED
File without changes
results/{gte-Qwen1.5-7B-instruct β†’ Alibaba-NLP__gte-Qwen1.5-7B-instruct}/no_revision_available/BIOSSES.json RENAMED
File without changes
results/{gte-Qwen1.5-7B-instruct β†’ Alibaba-NLP__gte-Qwen1.5-7B-instruct}/no_revision_available/BQ.json RENAMED
File without changes
results/{gte-Qwen1.5-7B-instruct β†’ Alibaba-NLP__gte-Qwen1.5-7B-instruct}/no_revision_available/Banking77Classification.json RENAMED
File without changes
results/{gte-Qwen1.5-7B-instruct β†’ Alibaba-NLP__gte-Qwen1.5-7B-instruct}/no_revision_available/BiorxivClusteringP2P.json RENAMED
File without changes
results/{gte-Qwen1.5-7B-instruct β†’ Alibaba-NLP__gte-Qwen1.5-7B-instruct}/no_revision_available/BiorxivClusteringS2S.json RENAMED
File without changes
results/{gte-Qwen1.5-7B-instruct β†’ Alibaba-NLP__gte-Qwen1.5-7B-instruct}/no_revision_available/BrightRetrieval.json RENAMED
File without changes
results/{gte-Qwen1.5-7B-instruct β†’ Alibaba-NLP__gte-Qwen1.5-7B-instruct}/no_revision_available/CLSClusteringP2P.json RENAMED
File without changes
results/{gte-Qwen1.5-7B-instruct β†’ Alibaba-NLP__gte-Qwen1.5-7B-instruct}/no_revision_available/CLSClusteringS2S.json RENAMED
File without changes
results/{gte-Qwen1.5-7B-instruct β†’ Alibaba-NLP__gte-Qwen1.5-7B-instruct}/no_revision_available/CMedQAv1.json RENAMED
File without changes
results/{gte-Qwen1.5-7B-instruct β†’ Alibaba-NLP__gte-Qwen1.5-7B-instruct}/no_revision_available/CMedQAv2.json RENAMED
File without changes
results/{gte-Qwen1.5-7B-instruct β†’ Alibaba-NLP__gte-Qwen1.5-7B-instruct}/no_revision_available/CQADupstackAndroidRetrieval.json RENAMED
File without changes
results/{gte-Qwen1.5-7B-instruct β†’ Alibaba-NLP__gte-Qwen1.5-7B-instruct}/no_revision_available/CQADupstackEnglishRetrieval.json RENAMED
File without changes
results/{gte-Qwen1.5-7B-instruct β†’ Alibaba-NLP__gte-Qwen1.5-7B-instruct}/no_revision_available/CQADupstackGamingRetrieval.json RENAMED
File without changes
results/{gte-Qwen1.5-7B-instruct β†’ Alibaba-NLP__gte-Qwen1.5-7B-instruct}/no_revision_available/CQADupstackGisRetrieval.json RENAMED
File without changes
results/{gte-Qwen1.5-7B-instruct β†’ Alibaba-NLP__gte-Qwen1.5-7B-instruct}/no_revision_available/CQADupstackMathematicaRetrieval.json RENAMED
File without changes
results/{gte-Qwen1.5-7B-instruct β†’ Alibaba-NLP__gte-Qwen1.5-7B-instruct}/no_revision_available/CQADupstackPhysicsRetrieval.json RENAMED
File without changes
results/{gte-Qwen1.5-7B-instruct β†’ Alibaba-NLP__gte-Qwen1.5-7B-instruct}/no_revision_available/CQADupstackProgrammersRetrieval.json RENAMED
File without changes
results/{gte-Qwen1.5-7B-instruct β†’ Alibaba-NLP__gte-Qwen1.5-7B-instruct}/no_revision_available/CQADupstackRetrieval.json RENAMED
File without changes
results/{gte-Qwen1.5-7B-instruct β†’ Alibaba-NLP__gte-Qwen1.5-7B-instruct}/no_revision_available/CQADupstackStatsRetrieval.json RENAMED
File without changes
results/{gte-Qwen1.5-7B-instruct β†’ Alibaba-NLP__gte-Qwen1.5-7B-instruct}/no_revision_available/CQADupstackTexRetrieval.json RENAMED
File without changes
results/{gte-Qwen1.5-7B-instruct β†’ Alibaba-NLP__gte-Qwen1.5-7B-instruct}/no_revision_available/CQADupstackUnixRetrieval.json RENAMED
File without changes
results/{gte-Qwen1.5-7B-instruct β†’ Alibaba-NLP__gte-Qwen1.5-7B-instruct}/no_revision_available/CQADupstackWebmastersRetrieval.json RENAMED
File without changes
results/{gte-Qwen1.5-7B-instruct β†’ Alibaba-NLP__gte-Qwen1.5-7B-instruct}/no_revision_available/CQADupstackWordpressRetrieval.json RENAMED
File without changes
results/{gte-Qwen1.5-7B-instruct β†’ Alibaba-NLP__gte-Qwen1.5-7B-instruct}/no_revision_available/ClimateFEVER.json RENAMED
File without changes
results/{gte-Qwen1.5-7B-instruct β†’ Alibaba-NLP__gte-Qwen1.5-7B-instruct}/no_revision_available/CmedqaRetrieval.json RENAMED
File without changes
results/{gte-Qwen1.5-7B-instruct β†’ Alibaba-NLP__gte-Qwen1.5-7B-instruct}/no_revision_available/Cmnli.json RENAMED
File without changes
results/{gte-Qwen1.5-7B-instruct β†’ Alibaba-NLP__gte-Qwen1.5-7B-instruct}/no_revision_available/CovidRetrieval.json RENAMED
File without changes
results/{gte-Qwen1.5-7B-instruct β†’ Alibaba-NLP__gte-Qwen1.5-7B-instruct}/no_revision_available/DBPedia.json RENAMED
File without changes
results/{gte-Qwen1.5-7B-instruct β†’ Alibaba-NLP__gte-Qwen1.5-7B-instruct}/no_revision_available/DuRetrieval.json RENAMED
File without changes
results/{gte-Qwen1.5-7B-instruct β†’ Alibaba-NLP__gte-Qwen1.5-7B-instruct}/no_revision_available/EcomRetrieval.json RENAMED
File without changes
results/{gte-Qwen1.5-7B-instruct β†’ Alibaba-NLP__gte-Qwen1.5-7B-instruct}/no_revision_available/EmotionClassification.json RENAMED
File without changes
results/{gte-Qwen1.5-7B-instruct β†’ Alibaba-NLP__gte-Qwen1.5-7B-instruct}/no_revision_available/FEVER.json RENAMED
File without changes
results/{gte-Qwen1.5-7B-instruct β†’ Alibaba-NLP__gte-Qwen1.5-7B-instruct}/no_revision_available/FiQA2018.json RENAMED
File without changes
results/{gte-Qwen1.5-7B-instruct β†’ Alibaba-NLP__gte-Qwen1.5-7B-instruct}/no_revision_available/HotpotQA.json RENAMED
File without changes
results/{gte-Qwen1.5-7B-instruct β†’ Alibaba-NLP__gte-Qwen1.5-7B-instruct}/no_revision_available/IFlyTek.json RENAMED
File without changes
results/{gte-Qwen1.5-7B-instruct β†’ Alibaba-NLP__gte-Qwen1.5-7B-instruct}/no_revision_available/ImdbClassification.json RENAMED
File without changes
results/{gte-Qwen1.5-7B-instruct β†’ Alibaba-NLP__gte-Qwen1.5-7B-instruct}/no_revision_available/JDReview.json RENAMED
File without changes
results/{gte-Qwen1.5-7B-instruct β†’ Alibaba-NLP__gte-Qwen1.5-7B-instruct}/no_revision_available/LCQMC.json RENAMED
File without changes
results/{gte-Qwen1.5-7B-instruct β†’ Alibaba-NLP__gte-Qwen1.5-7B-instruct}/no_revision_available/MMarcoReranking.json RENAMED
File without changes