Merge pull request #7 from embeddings-benchmark/fix_pawsx

boards_data/en/data_overall/default.jsonl

@@ -1,12 +1,12 @@
 {"index":91,"Rank":1,"Model":"<a target=\"_blank\" style=\"text-decoration: underline\" href=\"https:\/\/huggingface.co\/Salesforce\/SFR-Embedding-2_R\">SFR-Embedding-2_R<\/a>","Model Size (Million Parameters)":7111,"Memory Usage (GB, fp32)":26.49,"Embedding Dimensions":4096,"Max Tokens":32768,"Average (56 datasets)":70.31,"Classification Average (12 datasets)":89.05,"Clustering Average (11 datasets)":56.17,"PairClassification Average (3 datasets)":88.07,"Reranking Average (4 datasets)":60.14,"Retrieval Average (15 datasets)":60.18,"STS Average (10 datasets)":81.26,"Summarization Average (1 datasets)":30.71}
-{"index":16,"Rank":2,"Model":"<a target=\"_blank\" style=\"text-decoration: underline\" href=\"https:\/\/huggingface.co\/Alibaba-NLP\/gte-Qwen2-7B-instruct\">gte-Qwen2-7B-instruct<\/a>","Model Size (Million Parameters)":7613,"Memory Usage (GB, fp32)":28.36,"Embedding Dimensions":3584,"Max Tokens":131072,"Average (56 datasets)":70.24,"Classification Average (12 datasets)":86.58,"Clustering Average (11 datasets)":56.92,"PairClassification Average (3 datasets)":85.79,"Reranking Average (4 datasets)":61.42,"Retrieval Average (15 datasets)":60.25,"STS Average (10 datasets)":83.04,"Summarization Average (1 datasets)":31.35}
+{"index":16,"Rank":2,"Model":"<a target=\"_blank\" style=\"text-decoration: underline\" href=\"https:\/\/huggingface.co\/Alibaba-NLP\/gte-Qwen2-7B-instruct\">gte-Qwen2-7B-instruct<\/a>","Model Size (Million Parameters)":7613,"Memory Usage (GB, fp32)":28.36,"Embedding Dimensions":3584,"Max Tokens":32768,"Average (56 datasets)":70.24,"Classification Average (12 datasets)":86.58,"Clustering Average (11 datasets)":56.92,"PairClassification Average (3 datasets)":85.79,"Reranking Average (4 datasets)":61.42,"Retrieval Average (15 datasets)":60.25,"STS Average (10 datasets)":83.04,"Summarization Average (1 datasets)":31.35}
 {"index":48,"Rank":3,"Model":"<a target=\"_blank\" style=\"text-decoration: underline\" href=\"https:\/\/huggingface.co\/Intel\/neural-embedding-v1\">neural-embedding-v1<\/a>","Model Size (Million Parameters)":"","Memory Usage (GB, fp32)":"","Embedding Dimensions":"","Max Tokens":"","Average (56 datasets)":69.94,"Classification Average (12 datasets)":87.91,"Clustering Average (11 datasets)":54.32,"PairClassification Average (3 datasets)":87.68,"Reranking Average (4 datasets)":61.49,"Retrieval Average (15 datasets)":58.12,"STS Average (10 datasets)":85.24,"Summarization Average (1 datasets)":30.87}
 {"index":197,"Rank":4,"Model":"<a target=\"_blank\" style=\"text-decoration: underline\" href=\"https:\/\/huggingface.co\/nvidia\/NV-Embed-v1\">NV-Embed-v1<\/a>","Model Size (Million Parameters)":7851,"Memory Usage (GB, fp32)":29.25,"Embedding Dimensions":4096,"Max Tokens":32768,"Average (56 datasets)":69.32,"Classification Average (12 datasets)":87.35,"Clustering Average (11 datasets)":52.8,"PairClassification Average (3 datasets)":86.91,"Reranking Average (4 datasets)":60.54,"Retrieval Average (15 datasets)":59.36,"STS Average (10 datasets)":82.84,"Summarization Average (1 datasets)":31.2}
 {"index":6,"Rank":5,"Model":"<a target=\"_blank\" style=\"text-decoration: underline\" href=\"https:\/\/docs.voyageai.com\/embeddings\/\">voyage-large-2-instruct<\/a>","Model Size (Million Parameters)":"","Memory Usage (GB, fp32)":"","Embedding Dimensions":1024,"Max Tokens":16000,"Average (56 datasets)":68.28,"Classification Average (12 datasets)":81.49,"Clustering Average (11 datasets)":53.35,"PairClassification Average (3 datasets)":89.24,"Reranking Average (4 datasets)":60.09,"Retrieval Average (15 datasets)":58.28,"STS Average (10 datasets)":84.58,"Summarization Average (1 datasets)":30.84}
 {"index":54,"Rank":6,"Model":"<a target=\"_blank\" style=\"text-decoration: underline\" href=\"https:\/\/huggingface.co\/Linq-AI-Research\/Linq-Embed-Mistral\">Linq-Embed-Mistral<\/a>","Model Size (Million Parameters)":7111,"Memory Usage (GB, fp32)":26.49,"Embedding Dimensions":4096,"Max Tokens":32768,"Average (56 datasets)":68.17,"Classification Average (12 datasets)":80.2,"Clustering Average (11 datasets)":51.42,"PairClassification Average (3 datasets)":88.35,"Reranking Average (4 datasets)":60.29,"Retrieval Average (15 datasets)":60.19,"STS Average (10 datasets)":84.97,"Summarization Average (1 datasets)":30.98}
 {"index":92,"Rank":7,"Model":"<a target=\"_blank\" style=\"text-decoration: underline\" href=\"https:\/\/huggingface.co\/Salesforce\/SFR-Embedding-Mistral\">SFR-Embedding-Mistral<\/a>","Model Size (Million Parameters)":7111,"Memory Usage (GB, fp32)":26.49,"Embedding Dimensions":4096,"Max Tokens":32768,"Average (56 datasets)":67.56,"Classification Average (12 datasets)":78.33,"Clustering Average (11 datasets)":51.67,"PairClassification Average (3 datasets)":88.54,"Reranking Average (4 datasets)":60.64,"Retrieval Average (15 datasets)":59.0,"STS Average (10 datasets)":85.05,"Summarization Average (1 datasets)":31.16}
 {"index":14,"Rank":8,"Model":"<a target=\"_blank\" style=\"text-decoration: underline\" href=\"https:\/\/huggingface.co\/Alibaba-NLP\/gte-Qwen1.5-7B-instruct\">gte-Qwen1.5-7B-instruct<\/a>","Model Size (Million Parameters)":7099,"Memory Usage (GB, fp32)":26.45,"Embedding Dimensions":4096,"Max Tokens":32768,"Average (56 datasets)":67.34,"Classification Average (12 datasets)":79.6,"Clustering Average (11 datasets)":55.83,"PairClassification Average (3 datasets)":87.38,"Reranking Average (4 datasets)":60.13,"Retrieval Average (15 datasets)":56.24,"STS Average (10 datasets)":82.42,"Summarization Average (1 datasets)":31.46}
-{"index":15,"Rank":9,"Model":"<a target=\"_blank\" style=\"text-decoration: underline\" href=\"https:\/\/huggingface.co\/Alibaba-NLP\/gte-Qwen2-1.5B-instruct\">gte-Qwen2-1.5B-instruct<\/a>","Model Size (Million Parameters)":1776,"Memory Usage (GB, fp32)":6.62,"Embedding Dimensions":4096,"Max Tokens":131072,"Average (56 datasets)":67.16,"Classification Average (12 datasets)":82.47,"Clustering Average (11 datasets)":48.75,"PairClassification Average (3 datasets)":87.51,"Reranking Average (4 datasets)":59.98,"Retrieval Average (15 datasets)":58.29,"STS Average (10 datasets)":82.73,"Summarization Average (1 datasets)":31.17}
+{"index":15,"Rank":9,"Model":"<a target=\"_blank\" style=\"text-decoration: underline\" href=\"https:\/\/huggingface.co\/Alibaba-NLP\/gte-Qwen2-1.5B-instruct\">gte-Qwen2-1.5B-instruct<\/a>","Model Size (Million Parameters)":1776,"Memory Usage (GB, fp32)":6.62,"Embedding Dimensions":4096,"Max Tokens":32768,"Average (56 datasets)":67.16,"Classification Average (12 datasets)":82.47,"Clustering Average (11 datasets)":48.75,"PairClassification Average (3 datasets)":87.51,"Reranking Average (4 datasets)":59.98,"Retrieval Average (15 datasets)":58.29,"STS Average (10 datasets)":82.73,"Summarization Average (1 datasets)":31.17}
 {"index":9,"Rank":10,"Model":"<a target=\"_blank\" style=\"text-decoration: underline\" href=\"https:\/\/docs.voyageai.com\/embeddings\/\">voyage-lite-02-instruct<\/a>","Model Size (Million Parameters)":1220,"Memory Usage (GB, fp32)":4.54,"Embedding Dimensions":1024,"Max Tokens":4000,"Average (56 datasets)":67.13,"Classification Average (12 datasets)":79.25,"Clustering Average (11 datasets)":52.42,"PairClassification Average (3 datasets)":86.87,"Reranking Average (4 datasets)":58.24,"Retrieval Average (15 datasets)":56.6,"STS Average (10 datasets)":85.79,"Summarization Average (1 datasets)":31.01}
 {"index":39,"Rank":11,"Model":"<a target=\"_blank\" style=\"text-decoration: underline\" href=\"https:\/\/huggingface.co\/GritLM\/GritLM-7B\">GritLM-7B<\/a>","Model Size (Million Parameters)":7242,"Memory Usage (GB, fp32)":26.98,"Embedding Dimensions":4096,"Max Tokens":32768,"Average (56 datasets)":66.76,"Classification Average (12 datasets)":79.46,"Clustering Average (11 datasets)":50.61,"PairClassification Average (3 datasets)":87.16,"Reranking Average (4 datasets)":60.49,"Retrieval Average (15 datasets)":57.41,"STS Average (10 datasets)":83.35,"Summarization Average (1 datasets)":30.37}
 {"index":142,"Rank":12,"Model":"<a target=\"_blank\" style=\"text-decoration: underline\" href=\"https:\/\/huggingface.co\/intfloat\/e5-mistral-7b-instruct\">e5-mistral-7b-instruct<\/a>","Model Size (Million Parameters)":7111,"Memory Usage (GB, fp32)":26.49,"Embedding Dimensions":4096,"Max Tokens":32768,"Average (56 datasets)":66.63,"Classification Average (12 datasets)":78.47,"Clustering Average (11 datasets)":50.26,"PairClassification Average (3 datasets)":88.34,"Reranking Average (4 datasets)":60.21,"Retrieval Average (15 datasets)":56.89,"STS Average (10 datasets)":84.63,"Summarization Average (1 datasets)":31.4}

boards_data/fr/data_overall/default.jsonl

@@ -1,5 +1,5 @@
-{"index":9,"Rank":1,"Model":"<a target=\"_blank\" style=\"text-decoration: underline\" href=\"https:\/\/huggingface.co\/Alibaba-NLP\/gte-Qwen2-7B-instruct\">gte-Qwen2-7B-instruct<\/a>","Model Size (Million Parameters)":7613,"Memory Usage (GB, fp32)":28.36,"Embedding Dimensions":3584,"Max Tokens":131072,"Average (26 datasets)":68.25,"Classification Average (6 datasets)":81.76,"Clustering Average (7 datasets)":55.56,"PairClassification Average (2 datasets)":90.43,"Reranking Average (2 datasets)":78.7,"Retrieval Average (5 datasets)":55.65,"STS Average (3 datasets)":82.31,"Summarization Average (1 datasets)":31.45}
-{"index":8,"Rank":2,"Model":"<a target=\"_blank\" style=\"text-decoration: underline\" href=\"https:\/\/huggingface.co\/Alibaba-NLP\/gte-Qwen2-1.5B-instruct\">gte-Qwen2-1.5B-instruct<\/a>","Model Size (Million Parameters)":1776,"Memory Usage (GB, fp32)":6.62,"Embedding Dimensions":4096,"Max Tokens":131072,"Average (26 datasets)":66.6,"Classification Average (6 datasets)":78.02,"Clustering Average (7 datasets)":55.01,"PairClassification Average (2 datasets)":86.88,"Reranking Average (2 datasets)":83.76,"Retrieval Average (5 datasets)":52.56,"STS Average (3 datasets)":81.26,"Summarization Average (1 datasets)":30.5}
+{"index":9,"Rank":1,"Model":"<a target=\"_blank\" style=\"text-decoration: underline\" href=\"https:\/\/huggingface.co\/Alibaba-NLP\/gte-Qwen2-7B-instruct\">gte-Qwen2-7B-instruct<\/a>","Model Size (Million Parameters)":7613,"Memory Usage (GB, fp32)":28.36,"Embedding Dimensions":3584,"Max Tokens":32768,"Average (26 datasets)":68.25,"Classification Average (6 datasets)":81.76,"Clustering Average (7 datasets)":55.56,"PairClassification Average (2 datasets)":90.43,"Reranking Average (2 datasets)":78.7,"Retrieval Average (5 datasets)":55.65,"STS Average (3 datasets)":82.31,"Summarization Average (1 datasets)":31.45}
+{"index":8,"Rank":2,"Model":"<a target=\"_blank\" style=\"text-decoration: underline\" href=\"https:\/\/huggingface.co\/Alibaba-NLP\/gte-Qwen2-1.5B-instruct\">gte-Qwen2-1.5B-instruct<\/a>","Model Size (Million Parameters)":1776,"Memory Usage (GB, fp32)":6.62,"Embedding Dimensions":4096,"Max Tokens":32768,"Average (26 datasets)":66.6,"Classification Average (6 datasets)":78.02,"Clustering Average (7 datasets)":55.01,"PairClassification Average (2 datasets)":86.88,"Reranking Average (2 datasets)":83.76,"Retrieval Average (5 datasets)":52.56,"STS Average (3 datasets)":81.26,"Summarization Average (1 datasets)":30.5}
 {"index":4,"Rank":3,"Model":"<a target=\"_blank\" style=\"text-decoration: underline\" href=\"https:\/\/docs.voyageai.com\/embeddings\/\">voyage-multilingual-2<\/a>","Model Size (Million Parameters)":"","Memory Usage (GB, fp32)":"","Embedding Dimensions":1024,"Max Tokens":32000,"Average (26 datasets)":61.65,"Classification Average (6 datasets)":68.56,"Clustering Average (7 datasets)":46.57,"PairClassification Average (2 datasets)":78.66,"Reranking Average (2 datasets)":82.59,"Retrieval Average (5 datasets)":54.56,"STS Average (3 datasets)":80.13,"Summarization Average (1 datasets)":29.96}
 {"index":3,"Rank":4,"Model":"<a target=\"_blank\" style=\"text-decoration: underline\" href=\"https:\/\/docs.voyageai.com\/embeddings\/\">voyage-law-2<\/a>","Model Size (Million Parameters)":"","Memory Usage (GB, fp32)":"","Embedding Dimensions":1024,"Max Tokens":16000,"Average (26 datasets)":60.58,"Classification Average (6 datasets)":68.45,"Clustering Average (7 datasets)":44.23,"PairClassification Average (2 datasets)":77.3,"Reranking Average (2 datasets)":82.06,"Retrieval Average (5 datasets)":52.98,"STS Average (3 datasets)":80.29,"Summarization Average (1 datasets)":30.34}
 {"index":0,"Rank":5,"Model":"<a target=\"_blank\" style=\"text-decoration: underline\" href=\"https:\/\/docs.mistral.ai\/guides\/embeddings\">mistral-embed<\/a>","Model Size (Million Parameters)":"","Memory Usage (GB, fp32)":"","Embedding Dimensions":1024,"Max Tokens":"","Average (26 datasets)":59.41,"Classification Average (6 datasets)":68.61,"Clustering Average (7 datasets)":44.74,"PairClassification Average (2 datasets)":77.32,"Reranking Average (2 datasets)":80.46,"Retrieval Average (5 datasets)":46.81,"STS Average (3 datasets)":79.56,"Summarization Average (1 datasets)":31.47}

boards_data/pl/data_overall/default.jsonl

@@ -1,5 +1,5 @@
-{"index":2,"Rank":1,"Model":"<a target=\"_blank\" style=\"text-decoration: underline\" href=\"https:\/\/huggingface.co\/Alibaba-NLP\/gte-Qwen2-7B-instruct\">gte-Qwen2-7B-instruct<\/a>","Model Size (Million Parameters)":7613,"Memory Usage (GB, fp32)":28.36,"Embedding Dimensions":3584,"Max Tokens":131072,"Average (26 datasets)":67.86,"Classification Average (7 datasets)":77.84,"Clustering Average (1 datasets)":51.36,"PairClassification Average (4 datasets)":88.48,"Retrieval Average (11 datasets)":54.69,"STS Average (3 datasets)":70.86}
-{"index":1,"Rank":2,"Model":"<a target=\"_blank\" style=\"text-decoration: underline\" href=\"https:\/\/huggingface.co\/Alibaba-NLP\/gte-Qwen2-1.5B-instruct\">gte-Qwen2-1.5B-instruct<\/a>","Model Size (Million Parameters)":1776,"Memory Usage (GB, fp32)":6.62,"Embedding Dimensions":4096,"Max Tokens":131072,"Average (26 datasets)":64.04,"Classification Average (7 datasets)":72.29,"Clustering Average (1 datasets)":44.59,"PairClassification Average (4 datasets)":84.87,"Retrieval Average (11 datasets)":51.88,"STS Average (3 datasets)":68.12}
+{"index":2,"Rank":1,"Model":"<a target=\"_blank\" style=\"text-decoration: underline\" href=\"https:\/\/huggingface.co\/Alibaba-NLP\/gte-Qwen2-7B-instruct\">gte-Qwen2-7B-instruct<\/a>","Model Size (Million Parameters)":7613,"Memory Usage (GB, fp32)":28.36,"Embedding Dimensions":3584,"Max Tokens":32768,"Average (26 datasets)":67.86,"Classification Average (7 datasets)":77.84,"Clustering Average (1 datasets)":51.36,"PairClassification Average (4 datasets)":88.48,"Retrieval Average (11 datasets)":54.69,"STS Average (3 datasets)":70.86}
+{"index":1,"Rank":2,"Model":"<a target=\"_blank\" style=\"text-decoration: underline\" href=\"https:\/\/huggingface.co\/Alibaba-NLP\/gte-Qwen2-1.5B-instruct\">gte-Qwen2-1.5B-instruct<\/a>","Model Size (Million Parameters)":1776,"Memory Usage (GB, fp32)":6.62,"Embedding Dimensions":4096,"Max Tokens":32768,"Average (26 datasets)":64.04,"Classification Average (7 datasets)":72.29,"Clustering Average (1 datasets)":44.59,"PairClassification Average (4 datasets)":84.87,"Retrieval Average (11 datasets)":51.88,"STS Average (3 datasets)":68.12}
 {"index":34,"Rank":3,"Model":"<a target=\"_blank\" style=\"text-decoration: underline\" href=\"https:\/\/huggingface.co\/sdadas\/mmlw-roberta-large\">mmlw-roberta-large<\/a>","Model Size (Million Parameters)":435,"Memory Usage (GB, fp32)":1.62,"Embedding Dimensions":1024,"Max Tokens":514,"Average (26 datasets)":63.23,"Classification Average (7 datasets)":66.39,"Clustering Average (1 datasets)":31.16,"PairClassification Average (4 datasets)":89.13,"Retrieval Average (11 datasets)":52.71,"STS Average (3 datasets)":70.59}
 {"index":31,"Rank":4,"Model":"<a target=\"_blank\" style=\"text-decoration: underline\" href=\"https:\/\/huggingface.co\/sdadas\/mmlw-e5-large\">mmlw-e5-large<\/a>","Model Size (Million Parameters)":560,"Memory Usage (GB, fp32)":2.09,"Embedding Dimensions":1024,"Max Tokens":514,"Average (26 datasets)":61.17,"Classification Average (7 datasets)":61.07,"Clustering Average (1 datasets)":30.62,"PairClassification Average (4 datasets)":85.9,"Retrieval Average (11 datasets)":52.63,"STS Average (3 datasets)":69.98}
 {"index":33,"Rank":5,"Model":"<a target=\"_blank\" style=\"text-decoration: underline\" href=\"https:\/\/huggingface.co\/sdadas\/mmlw-roberta-base\">mmlw-roberta-base<\/a>","Model Size (Million Parameters)":124,"Memory Usage (GB, fp32)":0.46,"Embedding Dimensions":768,"Max Tokens":514,"Average (26 datasets)":61.05,"Classification Average (7 datasets)":62.92,"Clustering Average (1 datasets)":33.08,"PairClassification Average (4 datasets)":88.14,"Retrieval Average (11 datasets)":49.92,"STS Average (3 datasets)":70.7}

boards_data/zh/data_overall/default.jsonl

@@ -1,5 +1,5 @@
 {"index":194,"Rank":1,"Model":"<a target=\"_blank\" style=\"text-decoration: underline\" href=\"https:\/\/huggingface.co\/lier007\/xiaobu-embedding-v2\">xiaobu-embedding-v2<\/a>","Model Size (Million Parameters)":326,"Memory Usage (GB, fp32)":1.21,"Embedding Dimensions":1792,"Max Tokens":512,"Average (35 datasets)":72.43,"Classification Average (9 datasets)":74.67,"Clustering Average (4 datasets)":65.17,"PairClassification Average (2 datasets)":91.87,"Reranking Average (4 datasets)":72.58,"Retrieval Average (8 datasets)":76.5,"STS Average (8 datasets)":64.53}
-{"index":17,"Rank":2,"Model":"<a target=\"_blank\" style=\"text-decoration: underline\" href=\"https:\/\/huggingface.co\/Alibaba-NLP\/gte-Qwen2-7B-instruct\">gte-Qwen2-7B-instruct<\/a>","Model Size (Million Parameters)":7613,"Memory Usage (GB, fp32)":28.36,"Embedding Dimensions":3584,"Max Tokens":131072,"Average (35 datasets)":72.05,"Classification Average (9 datasets)":75.09,"Clustering Average (4 datasets)":66.06,"PairClassification Average (2 datasets)":87.48,"Reranking Average (4 datasets)":68.92,"Retrieval Average (8 datasets)":76.03,"STS Average (8 datasets)":65.33}
+{"index":17,"Rank":2,"Model":"<a target=\"_blank\" style=\"text-decoration: underline\" href=\"https:\/\/huggingface.co\/Alibaba-NLP\/gte-Qwen2-7B-instruct\">gte-Qwen2-7B-instruct<\/a>","Model Size (Million Parameters)":7613,"Memory Usage (GB, fp32)":28.36,"Embedding Dimensions":3584,"Max Tokens":32768,"Average (35 datasets)":72.05,"Classification Average (9 datasets)":75.09,"Clustering Average (4 datasets)":66.06,"PairClassification Average (2 datasets)":87.48,"Reranking Average (4 datasets)":68.92,"Retrieval Average (8 datasets)":76.03,"STS Average (8 datasets)":65.33}
 {"index":154,"Rank":3,"Model":"<a target=\"_blank\" style=\"text-decoration: underline\" href=\"https:\/\/huggingface.co\/iampanda\/zpoint_large_embedding_zh\">zpoint_large_embedding_zh<\/a>","Model Size (Million Parameters)":326,"Memory Usage (GB, fp32)":1.21,"Embedding Dimensions":1792,"Max Tokens":512,"Average (35 datasets)":71.88,"Classification Average (9 datasets)":74.43,"Clustering Average (4 datasets)":62.23,"PairClassification Average (2 datasets)":91.55,"Reranking Average (4 datasets)":72.34,"Retrieval Average (8 datasets)":76.36,"STS Average (8 datasets)":64.22}
 {"index":48,"Rank":4,"Model":"<a target=\"_blank\" style=\"text-decoration: underline\" href=\"https:\/\/huggingface.co\/Erin\/IYun-large-zh\">IYun-large-zh<\/a>","Model Size (Million Parameters)":"","Memory Usage (GB, fp32)":"","Embedding Dimensions":"","Max Tokens":"","Average (35 datasets)":71.04,"Classification Average (9 datasets)":74.18,"Clustering Average (4 datasets)":66.35,"PairClassification Average (2 datasets)":90.87,"Reranking Average (4 datasets)":69.3,"Retrieval Average (8 datasets)":73.56,"STS Average (8 datasets)":63.23}
 {"index":237,"Rank":5,"Model":"<a target=\"_blank\" style=\"text-decoration: underline\" href=\"https:\/\/huggingface.co\/sensenova\/piccolo-large-zh-v2\">piccolo-large-zh-v2<\/a>","Model Size (Million Parameters)":"","Memory Usage (GB, fp32)":"","Embedding Dimensions":"","Max Tokens":"","Average (35 datasets)":70.95,"Classification Average (9 datasets)":74.59,"Clustering Average (4 datasets)":62.17,"PairClassification Average (2 datasets)":90.24,"Reranking Average (4 datasets)":70.0,"Retrieval Average (8 datasets)":74.36,"STS Average (8 datasets)":63.5}
@@ -12,7 +12,7 @@
 {"index":161,"Rank":12,"Model":"<a target=\"_blank\" style=\"text-decoration: underline\" href=\"https:\/\/huggingface.co\/infgrad\/stella-large-zh-v3-1792d\">stella-large-zh-v3-1792d<\/a>","Model Size (Million Parameters)":325,"Memory Usage (GB, fp32)":1.21,"Embedding Dimensions":1792,"Max Tokens":512,"Average (35 datasets)":68.48,"Classification Average (9 datasets)":71.5,"Clustering Average (4 datasets)":53.9,"PairClassification Average (2 datasets)":88.1,"Reranking Average (4 datasets)":68.26,"Retrieval Average (8 datasets)":73.6,"STS Average (8 datasets)":62.46}
 {"index":298,"Rank":13,"Model":"<a target=\"_blank\" style=\"text-decoration: underline\" href=\"https:\/\/platform.baichuan-ai.com\/docs\/text-Embedding\">Baichuan-text-embedding<\/a>","Model Size (Million Parameters)":"","Memory Usage (GB, fp32)":"","Embedding Dimensions":1024,"Max Tokens":512,"Average (35 datasets)":68.34,"Classification Average (9 datasets)":72.84,"Clustering Average (4 datasets)":56.88,"PairClassification Average (2 datasets)":82.32,"Reranking Average (4 datasets)":69.67,"Retrieval Average (8 datasets)":73.12,"STS Average (8 datasets)":60.07}
 {"index":158,"Rank":14,"Model":"<a target=\"_blank\" style=\"text-decoration: underline\" href=\"https:\/\/huggingface.co\/infgrad\/stella-base-zh-v3-1792d\">stella-base-zh-v3-1792d<\/a>","Model Size (Million Parameters)":102,"Memory Usage (GB, fp32)":0.38,"Embedding Dimensions":1792,"Max Tokens":1024,"Average (35 datasets)":67.96,"Classification Average (9 datasets)":71.12,"Clustering Average (4 datasets)":53.3,"PairClassification Average (2 datasets)":87.93,"Reranking Average (4 datasets)":67.84,"Retrieval Average (8 datasets)":72.28,"STS Average (8 datasets)":62.49}
-{"index":16,"Rank":15,"Model":"<a target=\"_blank\" style=\"text-decoration: underline\" href=\"https:\/\/huggingface.co\/Alibaba-NLP\/gte-Qwen2-1.5B-instruct\">gte-Qwen2-1.5B-instruct<\/a>","Model Size (Million Parameters)":1776,"Memory Usage (GB, fp32)":6.62,"Embedding Dimensions":4096,"Max Tokens":131072,"Average (35 datasets)":67.65,"Classification Average (9 datasets)":71.12,"Clustering Average (4 datasets)":54.61,"PairClassification Average (2 datasets)":86.91,"Reranking Average (4 datasets)":68.21,"Retrieval Average (8 datasets)":71.86,"STS Average (8 datasets)":60.96}
+{"index":16,"Rank":15,"Model":"<a target=\"_blank\" style=\"text-decoration: underline\" href=\"https:\/\/huggingface.co\/Alibaba-NLP\/gte-Qwen2-1.5B-instruct\">gte-Qwen2-1.5B-instruct<\/a>","Model Size (Million Parameters)":1776,"Memory Usage (GB, fp32)":6.62,"Embedding Dimensions":4096,"Max Tokens":32768,"Average (35 datasets)":67.65,"Classification Average (9 datasets)":71.12,"Clustering Average (4 datasets)":54.61,"PairClassification Average (2 datasets)":86.91,"Reranking Average (4 datasets)":68.21,"Retrieval Average (8 datasets)":71.86,"STS Average (8 datasets)":60.96}
 {"index":44,"Rank":16,"Model":"<a target=\"_blank\" style=\"text-decoration: underline\" href=\"https:\/\/huggingface.co\/DMetaSoul\/Dmeta-embedding-zh\">Dmeta-embedding-zh<\/a>","Model Size (Million Parameters)":103,"Memory Usage (GB, fp32)":0.38,"Embedding Dimensions":768,"Max Tokens":1024,"Average (35 datasets)":67.51,"Classification Average (9 datasets)":70.0,"Clustering Average (4 datasets)":50.96,"PairClassification Average (2 datasets)":88.92,"Reranking Average (4 datasets)":67.17,"Retrieval Average (8 datasets)":70.41,"STS Average (8 datasets)":64.89}
 {"index":193,"Rank":17,"Model":"<a target=\"_blank\" style=\"text-decoration: underline\" href=\"https:\/\/huggingface.co\/lier007\/xiaobu-embedding\">xiaobu-embedding<\/a>","Model Size (Million Parameters)":326,"Memory Usage (GB, fp32)":1.21,"Embedding Dimensions":1024,"Max Tokens":512,"Average (35 datasets)":67.28,"Classification Average (9 datasets)":71.2,"Clustering Average (4 datasets)":54.62,"PairClassification Average (2 datasets)":85.3,"Reranking Average (4 datasets)":67.34,"Retrieval Average (8 datasets)":73.41,"STS Average (8 datasets)":58.52}
 {"index":102,"Rank":18,"Model":"<a target=\"_blank\" style=\"text-decoration: underline\" href=\"https:\/\/huggingface.co\/Pristinenlp\/alime-embedding-large-zh\">alime-embedding-large-zh<\/a>","Model Size (Million Parameters)":326,"Memory Usage (GB, fp32)":1.21,"Embedding Dimensions":1024,"Max Tokens":512,"Average (35 datasets)":67.17,"Classification Average (9 datasets)":71.35,"Clustering Average (4 datasets)":54.0,"PairClassification Average (2 datasets)":84.34,"Reranking Average (4 datasets)":67.61,"Retrieval Average (8 datasets)":73.3,"STS Average (8 datasets)":58.41}
@@ -94,7 +94,7 @@
 {"index":66,"Rank":104,"Model":"<a target=\"_blank\" style=\"text-decoration: underline\" href=\"https:\/\/huggingface.co\/Lajavaness\/bilingual-embedding-base\">bilingual-embedding-base<\/a>","Model Size (Million Parameters)":278,"Memory Usage (GB, fp32)":1.04,"Embedding Dimensions":768,"Max Tokens":514,"Average (35 datasets)":"","Classification Average (9 datasets)":"","Clustering Average (4 datasets)":"","PairClassification Average (2 datasets)":"","Reranking Average (4 datasets)":"","Retrieval Average (8 datasets)":"","STS Average (8 datasets)":""}
 {"index":67,"Rank":105,"Model":"<a target=\"_blank\" style=\"text-decoration: underline\" href=\"https:\/\/huggingface.co\/Lajavaness\/bilingual-embedding-large\">bilingual-embedding-large<\/a>","Model Size (Million Parameters)":560,"Memory Usage (GB, fp32)":2.09,"Embedding Dimensions":1024,"Max Tokens":514,"Average (35 datasets)":"","Classification Average (9 datasets)":"","Clustering Average (4 datasets)":"","PairClassification Average (2 datasets)":"","Reranking Average (4 datasets)":"","Retrieval Average (8 datasets)":"","STS Average (8 datasets)":""}
 {"index":68,"Rank":106,"Model":"<a target=\"_blank\" style=\"text-decoration: underline\" href=\"https:\/\/huggingface.co\/Lajavaness\/bilingual-embedding-large-8k\">bilingual-embedding-large-8k<\/a>","Model Size (Million Parameters)":568,"Memory Usage (GB, fp32)":2.12,"Embedding Dimensions":1024,"Max Tokens":8194,"Average (35 datasets)":"","Classification Average (9 datasets)":"","Clustering Average (4 datasets)":"","PairClassification Average (2 datasets)":"","Reranking Average (4 datasets)":"","Retrieval Average (8 datasets)":"","STS Average (8 datasets)":""}
-{"index":69,"Rank":107,"Model":"<a target=\"_blank\" style=\"text-decoration: underline\" href=\"https:\/\/huggingface.co\/Lenovo-Zhihui\/Zhihui_LLM_Embedding\">Zhihui_LLM_Embedding<\/a>","Model Size (Million Parameters)":7069,"Memory Usage (GB, fp32)":26.33,"Embedding Dimensions":3584,"Max Tokens":131072,"Average (35 datasets)":"","Classification Average (9 datasets)":"","Clustering Average (4 datasets)":"","PairClassification Average (2 datasets)":"","Reranking Average (4 datasets)":"","Retrieval Average (8 datasets)":76.74,"STS Average (8 datasets)":""}
+{"index":69,"Rank":107,"Model":"<a target=\"_blank\" style=\"text-decoration: underline\" href=\"https:\/\/huggingface.co\/Lenovo-Zhihui\/Zhihui_LLM_Embedding\">Zhihui_LLM_Embedding<\/a>","Model Size (Million Parameters)":7069,"Memory Usage (GB, fp32)":26.33,"Embedding Dimensions":3584,"Max Tokens":32768,"Average (35 datasets)":"","Classification Average (9 datasets)":"","Clustering Average (4 datasets)":"","PairClassification Average (2 datasets)":"","Reranking Average (4 datasets)":"","Retrieval Average (8 datasets)":76.74,"STS Average (8 datasets)":""}
 {"index":70,"Rank":108,"Model":"<a target=\"_blank\" style=\"text-decoration: underline\" href=\"https:\/\/huggingface.co\/Linq-AI-Research\/Linq-Embed-Mistral\">Linq-Embed-Mistral<\/a>","Model Size (Million Parameters)":7111,"Memory Usage (GB, fp32)":26.49,"Embedding Dimensions":4096,"Max Tokens":32768,"Average (35 datasets)":"","Classification Average (9 datasets)":"","Clustering Average (4 datasets)":"","PairClassification Average (2 datasets)":"","Reranking Average (4 datasets)":"","Retrieval Average (8 datasets)":"","STS Average (8 datasets)":""}
 {"index":72,"Rank":110,"Model":"<a target=\"_blank\" style=\"text-decoration: underline\" href=\"https:\/\/huggingface.co\/McGill-NLP\/LLM2Vec-Llama-2-7b-chat-hf-mntp-supervised\">LLM2Vec-Llama-2-supervised<\/a>","Model Size (Million Parameters)":6607,"Memory Usage (GB, fp32)":24.61,"Embedding Dimensions":4096,"Max Tokens":4096,"Average (35 datasets)":"","Classification Average (9 datasets)":"","Clustering Average (4 datasets)":"","PairClassification Average (2 datasets)":"","Reranking Average (4 datasets)":"","Retrieval Average (8 datasets)":"","STS Average (8 datasets)":""}
 {"index":73,"Rank":111,"Model":"<a target=\"_blank\" style=\"text-decoration: underline\" href=\"https:\/\/huggingface.co\/McGill-NLP\/LLM2Vec-Llama-2-7b-chat-hf-mntp-unsup-simcse\">LLM2Vec-Llama-2-unsupervised<\/a>","Model Size (Million Parameters)":6607,"Memory Usage (GB, fp32)":24.61,"Embedding Dimensions":4096,"Max Tokens":4096,"Average (35 datasets)":"","Classification Average (9 datasets)":"","Clustering Average (4 datasets)":"","PairClassification Average (2 datasets)":"","Reranking Average (4 datasets)":"","Retrieval Average (8 datasets)":"","STS Average (8 datasets)":""}

refresh.py

@@ -323,8 +323,17 @@ def get_mteb_data(tasks=["Clustering"], langs=[], datasets=[], fillna=True, add_
             df['MLSUMClusteringS2S (fr)'] = df['MLSUMClusteringS2S (fr)'].fillna(df['MLSUMClusteringS2S'])
             datasets.remove('MLSUMClusteringS2S')
         if ('PawsXPairClassification (fr)' in datasets) and ('PawsX (fr)' in cols):
-            df['PawsXPairClassification (fr)'] = df['PawsXPairClassification (fr)'].fillna(df['PawsX (fr)'])
+             # for the first bit no model has it, hence no column for it. We can remove this in a month or so
+            if "PawsXPairClassification (fr)" not in cols:
+                df['PawsXPairClassification (fr)'] = df['PawsX (fr)']
+            else:
+                df['PawsXPairClassification (fr)'] = df['PawsXPairClassification (fr)'].fillna(df['PawsX (fr)'])
+            # make all the columns the same
             datasets.remove('PawsX (fr)')
+            cols.remove('PawsX (fr)')
+            df.drop(columns=['PawsX (fr)'], inplace=True)
+            cols.append('PawsXPairClassification (fr)')
+            
         # Filter invalid columns
         cols = [col for col in cols if col in base_columns + datasets]
     i = 0
@@ -353,7 +362,6 @@ def get_mteb_average(task_dict: dict):
     )
     # Debugging:
     # DATA_OVERALL.to_csv("overall.csv")
-    
     DATA_OVERALL.insert(1, f"Average ({len(all_tasks)} datasets)", DATA_OVERALL[all_tasks].mean(axis=1, skipna=False))
     for i, (task_category, task_category_list) in enumerate(task_dict.items()):
         DATA_OVERALL.insert(i+2, f"{task_category} Average ({len(task_category_list)} datasets)", DATA_OVERALL[task_category_list].mean(axis=1, skipna=False))
@@ -420,7 +428,7 @@ def refresh_leaderboard():
 
 
 
-def write_out_results(item: dict, item_name: str):
+def write_out_results(item, item_name: str):
     """
     Due to their complex structure, let's recursively create subfolders until we reach the end
         of the item and then save the DFs as jsonl files