minor fixes

app.py

@@ -28,13 +28,13 @@ def make_clickable_model(model_name, link=None):
         f'<a target="_blank" style="text-decoration: underline" href="{link}">{model_name.split("/")[-1]}</a>'
     )
 
-EXTERNAL_MODELS = {k for k,v in MODEL_META["models_meta"].items() if v.get("is_external", False)}
-EXTERNAL_MODEL_TO_LINK = {k: v["link"] for k,v in MODEL_META["models_meta"].items() if v.get("link", False)}
-EXTERNAL_MODEL_TO_DIM = {k: v["dim"] for k,v in MODEL_META["models_meta"].items() if v.get("dim", False)}
-EXTERNAL_MODEL_TO_SEQLEN = {k: v["seq_len"] for k,v in MODEL_META["models_meta"].items() if v.get("seq_len", False)}
-EXTERNAL_MODEL_TO_SIZE = {k: v["size"] for k,v in MODEL_META["models_meta"].items() if v.get("size", False)}
-PROPRIETARY_MODELS = {k for k,v in MODEL_META["models_meta"].items() if v.get("is_proprietary", False)}
-SENTENCE_TRANSFORMERS_COMPATIBLE_MODELS = {k for k,v in MODEL_META["models_meta"].items() if v.get("is_sentence_transformers_compatible", False)}
+EXTERNAL_MODELS = {k for k,v in MODEL_META["model_meta"].items() if v.get("is_external", False)}
+EXTERNAL_MODEL_TO_LINK = {k: v["link"] for k,v in MODEL_META["model_meta"].items() if v.get("link", False)}
+EXTERNAL_MODEL_TO_DIM = {k: v["dim"] for k,v in MODEL_META["model_meta"].items() if v.get("dim", False)}
+EXTERNAL_MODEL_TO_SEQLEN = {k: v["seq_len"] for k,v in MODEL_META["model_meta"].items() if v.get("seq_len", False)}
+EXTERNAL_MODEL_TO_SIZE = {k: v["size"] for k,v in MODEL_META["model_meta"].items() if v.get("size", False)}
+PROPRIETARY_MODELS = {k for k,v in MODEL_META["model_meta"].items() if v.get("is_proprietary", False)}
+SENTENCE_TRANSFORMERS_COMPATIBLE_MODELS = {k for k,v in MODEL_META["model_meta"].items() if v.get("is_sentence_transformers_compatible", False)}
 MODELS_TO_SKIP = MODEL_META["models_to_skip"]
 
 PROPRIETARY_MODELS = {
@@ -48,8 +48,8 @@ SENTENCE_TRANSFORMERS_COMPATIBLE_MODELS = {
 }
 
 TASK_TO_TASK_TYPE = {task_category: [] for task_category in TASKS}
-for board in BOARDS_CONFIG.values():
-    for task_category, task_list in board["tasks"].items():
+for board_config in BOARDS_CONFIG.values():
+    for task_category, task_list in board_config["tasks"].items():
         TASK_TO_TASK_TYPE[task_category].extend(task_list)
 
 def add_lang(examples):
@@ -365,7 +365,7 @@ for board, board_config in BOARDS_CONFIG.items():
             "language_long": board_config["language_long"],
             "description": f"**Overall MTEB {overall_pretty_name}** 🔮{board_icon}",
             "data": boards_data[board]["data_overall"],
-            "refresh": partial(get_mteb_average, board_config["tasks"]),
+            "refresh": lambda: get_mteb_average(board_config["tasks"])[0],#partial(get_mteb_average, board_config["tasks"]),
             "credits": credits,
         })
     for task_category, task_category_list in board_config["tasks"].items():

config.yaml

@@ -1,4 +1,3 @@
-version: 0.0.1
 config:
   REPO_ID: "mteb/leaderboard"
   RESULTS_REPO: mteb/results
@@ -111,7 +110,7 @@ boards:
         - SummEval
   en-x:
     title: "English-X"
-    language_long: " 117 (Pairs of: English & other language)"
+    language_long: "117 (Pairs of: English & other language)"
     has_overall: false
     acronym: null
     icon: null

envs.py

@@ -37,6 +37,7 @@ RESULTS_REPO = get_config("RESULTS_REPO", "mteb/results")
 CACHE_PATH=get_config("HF_HOME", ".")
 os.environ["HF_HOME"] = CACHE_PATH
 
+# Check if it is using persistent storage
 if not os.access(CACHE_PATH, os.W_OK):
     print(f"No write access to HF_HOME: {CACHE_PATH}. Resetting to current directory.")
     CACHE_PATH = "."

model_meta.yaml

@@ -1,1028 +1,1028 @@
-models_meta:
+model_meta:
   Baichuan-text-embedding:
-    dim: 1024
-    is_external: true
-    is_proprietary: true
-    is_sentence_transformers_compatible: false
     link: https://platform.baichuan-ai.com/docs/text-Embedding
     seq_len: 512
     size: null
-  Cohere-embed-english-v3.0:
     dim: 1024
     is_external: true
     is_proprietary: true
     is_sentence_transformers_compatible: false
+  Cohere-embed-english-v3.0:
     link: https://huggingface.co/Cohere/Cohere-embed-english-v3.0
     seq_len: 512
     size: null
-  Cohere-embed-multilingual-light-v3.0:
-    dim: 384
+    dim: 1024
     is_external: true
     is_proprietary: true
     is_sentence_transformers_compatible: false
+  Cohere-embed-multilingual-light-v3.0:
     link: https://huggingface.co/Cohere/Cohere-embed-multilingual-light-v3.0
     seq_len: 512
     size: null
-  Cohere-embed-multilingual-v3.0:
-    dim: 1024
+    dim: 384
     is_external: true
     is_proprietary: true
     is_sentence_transformers_compatible: false
+  Cohere-embed-multilingual-v3.0:
     link: https://huggingface.co/Cohere/Cohere-embed-multilingual-v3.0
     seq_len: 512
     size: null
-  DanskBERT:
-    dim: 768
+    dim: 1024
     is_external: true
-    is_proprietary: false
-    is_sentence_transformers_compatible: true
+    is_proprietary: true
+    is_sentence_transformers_compatible: false
+  DanskBERT:
     link: https://huggingface.co/vesteinn/DanskBERT
     seq_len: 514
     size: 125
-  LASER2:
-    dim: 1024
+    dim: 768
     is_external: true
     is_proprietary: false
-    is_sentence_transformers_compatible: false
+    is_sentence_transformers_compatible: true
+  LASER2:
     link: https://github.com/facebookresearch/LASER
     seq_len: N/A
     size: 43
-  LLM2Vec-Llama-supervised:
-    dim: 4096
+    dim: 1024
     is_external: true
     is_proprietary: false
     is_sentence_transformers_compatible: false
+  LLM2Vec-Llama-supervised:
     link: https://huggingface.co/McGill-NLP/LLM2Vec-Llama-2-7b-chat-hf-mntp-supervised
     seq_len: 4096
     size: 6607
-  LLM2Vec-Llama-unsupervised:
     dim: 4096
     is_external: true
     is_proprietary: false
     is_sentence_transformers_compatible: false
+  LLM2Vec-Llama-unsupervised:
     link: https://huggingface.co/McGill-NLP/LLM2Vec-Llama-2-7b-chat-hf-mntp
     seq_len: 4096
     size: 6607
-  LLM2Vec-Mistral-supervised:
     dim: 4096
     is_external: true
     is_proprietary: false
     is_sentence_transformers_compatible: false
+  LLM2Vec-Mistral-supervised:
     link: https://huggingface.co/McGill-NLP/LLM2Vec-Mistral-7B-Instruct-v2-mntp-supervised
     seq_len: 32768
     size: 7111
-  LLM2Vec-Mistral-unsupervised:
     dim: 4096
     is_external: true
     is_proprietary: false
     is_sentence_transformers_compatible: false
+  LLM2Vec-Mistral-unsupervised:
     link: https://huggingface.co/McGill-NLP/LLM2Vec-Mistral-7B-Instruct-v2-mntp
     seq_len: 32768
     size: 7111
-  LLM2Vec-Sheared-Llama-supervised:
-    dim: 2048
+    dim: 4096
     is_external: true
     is_proprietary: false
     is_sentence_transformers_compatible: false
+  LLM2Vec-Sheared-Llama-supervised:
     link: https://huggingface.co/McGill-NLP/LLM2Vec-Sheared-LLaMA-mntp-supervised
     seq_len: 4096
     size: 1280
-  LLM2Vec-Sheared-Llama-unsupervised:
     dim: 2048
     is_external: true
     is_proprietary: false
     is_sentence_transformers_compatible: false
+  LLM2Vec-Sheared-Llama-unsupervised:
     link: https://huggingface.co/McGill-NLP/LLM2Vec-Sheared-LLaMA-mntp
     seq_len: 4096
     size: 1280
-  LaBSE:
-    dim: 768
+    dim: 2048
     is_external: true
     is_proprietary: false
-    is_sentence_transformers_compatible: true
+    is_sentence_transformers_compatible: false
+  LaBSE:
     link: https://huggingface.co/sentence-transformers/LaBSE
     seq_len: 512
     size: 471
+    dim: 768
+    is_external: true
+    is_proprietary: false
+    is_sentence_transformers_compatible: true
   OpenSearch-text-hybrid:
+    link: https://help.aliyun.com/zh/open-search/vector-search-edition/hybrid-retrieval
+    seq_len: 512
+    size: null
     dim: 1792
     is_external: true
     is_proprietary: true
     is_sentence_transformers_compatible: false
-    link: https://help.aliyun.com/zh/open-search/vector-search-edition/hybrid-retrieval
-    seq_len: 512
-    size: null
   all-MiniLM-L12-v2:
-    dim: 384
-    is_external: true
-    is_proprietary: false
-    is_sentence_transformers_compatible: true
     link: https://huggingface.co/sentence-transformers/all-MiniLM-L12-v2
     seq_len: 512
     size: 33
-  all-MiniLM-L6-v2:
     dim: 384
     is_external: true
     is_proprietary: false
     is_sentence_transformers_compatible: true
+  all-MiniLM-L6-v2:
     link: https://huggingface.co/sentence-transformers/all-MiniLM-L6-v2
     seq_len: 512
     size: 23
-  all-mpnet-base-v2:
-    dim: 768
+    dim: 384
     is_external: true
     is_proprietary: false
     is_sentence_transformers_compatible: true
+  all-mpnet-base-v2:
     link: https://huggingface.co/sentence-transformers/all-mpnet-base-v2
     seq_len: 514
     size: 110
-  allenai-specter:
     dim: 768
     is_external: true
     is_proprietary: false
     is_sentence_transformers_compatible: true
+  allenai-specter:
     link: https://huggingface.co/sentence-transformers/allenai-specter
     seq_len: 512
     size: 110
-  bert-base-10lang-cased:
     dim: 768
     is_external: true
     is_proprietary: false
     is_sentence_transformers_compatible: true
+  bert-base-10lang-cased:
     link: https://huggingface.co/Geotrend/bert-base-10lang-cased
     seq_len: 512
     size: 138
-  bert-base-15lang-cased:
     dim: 768
     is_external: true
     is_proprietary: false
     is_sentence_transformers_compatible: true
+  bert-base-15lang-cased:
     link: https://huggingface.co/Geotrend/bert-base-15lang-cased
     seq_len: 512
     size: 138
-  bert-base-25lang-cased:
     dim: 768
     is_external: true
     is_proprietary: false
     is_sentence_transformers_compatible: true
+  bert-base-25lang-cased:
     link: https://huggingface.co/Geotrend/bert-base-25lang-cased
     seq_len: 512
     size: 138
-  bert-base-multilingual-cased:
     dim: 768
     is_external: true
     is_proprietary: false
     is_sentence_transformers_compatible: true
+  bert-base-multilingual-cased:
     link: https://huggingface.co/google-bert/bert-base-multilingual-cased
     seq_len: 512
     size: 179
-  bert-base-multilingual-uncased:
     dim: 768
     is_external: true
     is_proprietary: false
     is_sentence_transformers_compatible: true
+  bert-base-multilingual-uncased:
     link: https://huggingface.co/google-bert/bert-base-multilingual-uncased
     seq_len: 512
     size: 168
-  bert-base-swedish-cased:
     dim: 768
     is_external: true
     is_proprietary: false
     is_sentence_transformers_compatible: true
+  bert-base-swedish-cased:
     link: https://huggingface.co/KB/bert-base-swedish-cased
     seq_len: 512
     size: 125
-  bert-base-uncased:
     dim: 768
     is_external: true
     is_proprietary: false
     is_sentence_transformers_compatible: true
+  bert-base-uncased:
     link: https://huggingface.co/bert-base-uncased
     seq_len: 512
     size: 110
-  bge-base-zh-v1.5:
     dim: 768
     is_external: true
     is_proprietary: false
     is_sentence_transformers_compatible: true
+  bge-base-zh-v1.5:
     link: https://huggingface.co/BAAI/bge-base-zh-v1.5
     seq_len: 512
     size: 102
-  bge-large-en-v1.5:
-    dim: 1024
+    dim: 768
     is_external: true
     is_proprietary: false
-    is_sentence_transformers_compatible: false
+    is_sentence_transformers_compatible: true
+  bge-large-en-v1.5:
     link: https://huggingface.co/BAAI/bge-large-en-v1.5
     seq_len: 512
     size: null
-  bge-large-zh-noinstruct:
     dim: 1024
     is_external: true
     is_proprietary: false
-    is_sentence_transformers_compatible: true
+    is_sentence_transformers_compatible: false
+  bge-large-zh-noinstruct:
     link: https://huggingface.co/BAAI/bge-large-zh-noinstruct
     seq_len: 512
     size: 326
-  bge-large-zh-v1.5:
     dim: 1024
     is_external: true
     is_proprietary: false
     is_sentence_transformers_compatible: true
+  bge-large-zh-v1.5:
     link: https://huggingface.co/BAAI/bge-large-zh-v1.5
     seq_len: 512
     size: 326
-  bge-small-zh-v1.5:
-    dim: 512
+    dim: 1024
     is_external: true
     is_proprietary: false
     is_sentence_transformers_compatible: true
+  bge-small-zh-v1.5:
     link: https://huggingface.co/BAAI/bge-small-zh-v1.5
     seq_len: 512
     size: 24
-  camembert-base:
     dim: 512
-    is_external: false
+    is_external: true
     is_proprietary: false
     is_sentence_transformers_compatible: true
+  camembert-base:
     link: https://huggingface.co/almanach/camembert-base
     seq_len: 512
     size: 111
-  camembert-large:
-    dim: 768
+    dim: 512
     is_external: false
     is_proprietary: false
     is_sentence_transformers_compatible: true
+  camembert-large:
     link: https://huggingface.co/almanach/camembert-large
     seq_len: 512
     size: 338
-  contriever-base-msmarco:
     dim: 768
-    is_external: true
+    is_external: false
     is_proprietary: false
     is_sentence_transformers_compatible: true
+  contriever-base-msmarco:
     link: https://huggingface.co/nthakur/contriever-base-msmarco
     seq_len: 512
     size: 110
-  cross-en-de-roberta-sentence-transformer:
     dim: 768
     is_external: true
     is_proprietary: false
     is_sentence_transformers_compatible: true
+  cross-en-de-roberta-sentence-transformer:
     link: https://huggingface.co/T-Systems-onsite/cross-en-de-roberta-sentence-transformer
     seq_len: 514
     size: 278
-  dfm-encoder-large-v1:
-    dim: 1024
+    dim: 768
     is_external: true
     is_proprietary: false
     is_sentence_transformers_compatible: true
+  dfm-encoder-large-v1:
     link: https://huggingface.co/chcaa/dfm-encoder-large-v1
     seq_len: 512
     size: 355
-  dfm-sentence-encoder-large-1:
     dim: 1024
     is_external: true
     is_proprietary: false
     is_sentence_transformers_compatible: true
+  dfm-sentence-encoder-large-1:
     link: https://huggingface.co/chcaa/dfm-encoder-large-v1
     seq_len: 512
     size: 355
-  distilbert-base-25lang-cased:
-    dim: 768
-    is_external: false
+    dim: 1024
+    is_external: true
     is_proprietary: false
     is_sentence_transformers_compatible: true
+  distilbert-base-25lang-cased:
     link: https://huggingface.co/Geotrend/distilbert-base-25lang-cased
     seq_len: 512
     size: 110
-  distilbert-base-en-fr-cased:
     dim: 768
     is_external: false
     is_proprietary: false
     is_sentence_transformers_compatible: true
+  distilbert-base-en-fr-cased:
     link: https://huggingface.co/Geotrend/distilbert-base-en-fr-cased
     seq_len: 512
     size: 110
-  distilbert-base-en-fr-es-pt-it-cased:
     dim: 768
     is_external: false
     is_proprietary: false
     is_sentence_transformers_compatible: true
+  distilbert-base-en-fr-es-pt-it-cased:
     link: https://huggingface.co/Geotrend/distilbert-base-en-fr-es-pt-it-cased
     seq_len: 512
     size: 110
-  distilbert-base-fr-cased:
     dim: 768
     is_external: false
     is_proprietary: false
     is_sentence_transformers_compatible: true
+  distilbert-base-fr-cased:
     link: https://huggingface.co/Geotrend/distilbert-base-fr-cased
     seq_len: 512
     size: 110
-  distilbert-base-uncased:
     dim: 768
     is_external: false
     is_proprietary: false
     is_sentence_transformers_compatible: true
+  distilbert-base-uncased:
     link: https://huggingface.co/distilbert-base-uncased
     seq_len: 512
     size: 110
-  distiluse-base-multilingual-cased-v2:
-    dim: 512
-    is_external: true
+    dim: 768
+    is_external: false
     is_proprietary: false
     is_sentence_transformers_compatible: true
+  distiluse-base-multilingual-cased-v2:
     link: https://huggingface.co/sentence-transformers/distiluse-base-multilingual-cased-v2
     seq_len: 512
     size: 135
-  e5-base:
-    dim: 768
+    dim: 512
     is_external: true
     is_proprietary: false
     is_sentence_transformers_compatible: true
+  e5-base:
     link: https://huggingface.co/intfloat/e5-base
     seq_len: 512
     size: 110
-  e5-large:
-    dim: 1024
+    dim: 768
     is_external: true
     is_proprietary: false
     is_sentence_transformers_compatible: true
+  e5-large:
     link: https://huggingface.co/intfloat/e5-large
     seq_len: 512
     size: 335
-  e5-mistral-7b-instruct:
-    dim: 4096
+    dim: 1024
     is_external: true
     is_proprietary: false
     is_sentence_transformers_compatible: true
+  e5-mistral-7b-instruct:
     link: https://huggingface.co/intfloat/e5-mistral-7b-instruct
     seq_len: 32768
     size: 7111
-  e5-small:
-    dim: 384
+    dim: 4096
     is_external: true
     is_proprietary: false
     is_sentence_transformers_compatible: true
+  e5-small:
     link: https://huggingface.co/intfloat/e5-small
     seq_len: 512
     size: 33
-  electra-small-nordic:
-    dim: 256
+    dim: 384
     is_external: true
     is_proprietary: false
     is_sentence_transformers_compatible: true
+  electra-small-nordic:
     link: https://huggingface.co/jonfd/electra-small-nordic
     seq_len: 512
     size: 23
-  electra-small-swedish-cased-discriminator:
     dim: 256
     is_external: true
     is_proprietary: false
     is_sentence_transformers_compatible: true
+  electra-small-swedish-cased-discriminator:
     link: https://huggingface.co/KBLab/electra-small-swedish-cased-discriminator
     seq_len: 512
     size: 16
-  flaubert_base_cased:
-    dim: 768
+    dim: 256
     is_external: true
     is_proprietary: false
     is_sentence_transformers_compatible: true
+  flaubert_base_cased:
     link: https://huggingface.co/flaubert/flaubert_base_cased
     seq_len: 512
     size: 138
-  flaubert_base_uncased:
     dim: 768
     is_external: true
     is_proprietary: false
     is_sentence_transformers_compatible: true
+  flaubert_base_uncased:
     link: https://huggingface.co/flaubert/flaubert_base_uncased
     seq_len: 512
     size: 138
-  flaubert_large_cased:
-    dim: 1024
+    dim: 768
     is_external: true
     is_proprietary: false
     is_sentence_transformers_compatible: true
+  flaubert_large_cased:
     link: https://huggingface.co/flaubert/flaubert_large_cased
     seq_len: 512
     size: 372
-  gbert-base:
-    dim: 768
+    dim: 1024
     is_external: true
     is_proprietary: false
     is_sentence_transformers_compatible: true
+  gbert-base:
     link: https://huggingface.co/deepset/gbert-base
     seq_len: 512
     size: 110
-  gbert-large:
-    dim: 1024
+    dim: 768
     is_external: true
     is_proprietary: false
     is_sentence_transformers_compatible: true
+  gbert-large:
     link: https://huggingface.co/deepset/gbert-large
     seq_len: 512
     size: 337
-  gelectra-base:
-    dim: 768
+    dim: 1024
     is_external: true
     is_proprietary: false
     is_sentence_transformers_compatible: true
+  gelectra-base:
     link: https://huggingface.co/deepset/gelectra-base
     seq_len: 512
     size: 110
-  gelectra-large:
-    dim: 1024
+    dim: 768
     is_external: true
     is_proprietary: false
     is_sentence_transformers_compatible: true
+  gelectra-large:
     link: https://huggingface.co/deepset/gelectra-large
     seq_len: 512
     size: 335
-  glove.6B.300d:
-    dim: 300
+    dim: 1024
     is_external: true
     is_proprietary: false
     is_sentence_transformers_compatible: true
+  glove.6B.300d:
     link: https://huggingface.co/sentence-transformers/average_word_embeddings_glove.6B.300d
     seq_len: N/A
     size: 120
+    dim: 300
+    is_external: true
+    is_proprietary: false
+    is_sentence_transformers_compatible: true
   google-gecko-256.text-embedding-preview-0409:
+    link: https://cloud.google.com/vertex-ai/generative-ai/docs/embeddings/get-text-embeddings#latest_models
+    seq_len: 2048
+    size: 1200
     dim: 256
     is_external: true
     is_proprietary: true
     is_sentence_transformers_compatible: false
+  google-gecko.text-embedding-preview-0409:
     link: https://cloud.google.com/vertex-ai/generative-ai/docs/embeddings/get-text-embeddings#latest_models
     seq_len: 2048
     size: 1200
-  google-gecko.text-embedding-preview-0409:
     dim: 768
     is_external: true
     is_proprietary: true
     is_sentence_transformers_compatible: false
-    link: https://cloud.google.com/vertex-ai/generative-ai/docs/embeddings/get-text-embeddings#latest_models
-    seq_len: 2048
-    size: 1200
   gottbert-base:
-    dim: 768
-    is_external: true
-    is_proprietary: false
-    is_sentence_transformers_compatible: true
     link: https://huggingface.co/uklfr/gottbert-base
     seq_len: 512
     size: 127
-  gtr-t5-base:
     dim: 768
     is_external: true
     is_proprietary: false
     is_sentence_transformers_compatible: true
+  gtr-t5-base:
     link: https://huggingface.co/sentence-transformers/gtr-t5-base
     seq_len: 512
     size: 110
-  gtr-t5-large:
     dim: 768
     is_external: true
     is_proprietary: false
     is_sentence_transformers_compatible: true
+  gtr-t5-large:
     link: https://huggingface.co/sentence-transformers/gtr-t5-large
     seq_len: 512
     size: 168
-  gtr-t5-xl:
     dim: 768
     is_external: true
     is_proprietary: false
     is_sentence_transformers_compatible: true
+  gtr-t5-xl:
     link: https://huggingface.co/sentence-transformers/gtr-t5-xl
     seq_len: 512
     size: 1240
-  gtr-t5-xxl:
     dim: 768
     is_external: true
     is_proprietary: false
     is_sentence_transformers_compatible: true
+  gtr-t5-xxl:
     link: https://huggingface.co/sentence-transformers/gtr-t5-xxl
     seq_len: 512
     size: 4865
-  herbert-base-retrieval-v2:
     dim: 768
     is_external: true
     is_proprietary: false
     is_sentence_transformers_compatible: true
+  herbert-base-retrieval-v2:
     link: https://huggingface.co/ipipan/herbert-base-retrieval-v2
     seq_len: 514
     size: 125
-  komninos:
-    dim: 300
+    dim: 768
     is_external: true
     is_proprietary: false
     is_sentence_transformers_compatible: true
+  komninos:
     link: https://huggingface.co/sentence-transformers/average_word_embeddings_komninos
     seq_len: N/A
     size: 134
-  luotuo-bert-medium:
-    dim: 768
+    dim: 300
     is_external: true
     is_proprietary: false
     is_sentence_transformers_compatible: true
+  luotuo-bert-medium:
     link: https://huggingface.co/silk-road/luotuo-bert-medium
     seq_len: 512
     size: 328
-  m3e-base:
     dim: 768
     is_external: true
     is_proprietary: false
     is_sentence_transformers_compatible: true
+  m3e-base:
     link: https://huggingface.co/moka-ai/m3e-base
     seq_len: 512
     size: 102
-  m3e-large:
     dim: 768
     is_external: true
     is_proprietary: false
     is_sentence_transformers_compatible: true
+  m3e-large:
     link: https://huggingface.co/moka-ai/m3e-large
     seq_len: 512
     size: 102
+    dim: 768
+    is_external: true
+    is_proprietary: false
+    is_sentence_transformers_compatible: true
   mistral-embed:
+    link: https://docs.mistral.ai/guides/embeddings
+    seq_len: null
+    size: null
     dim: 1024
     is_external: true
     is_proprietary: true
     is_sentence_transformers_compatible: false
-    link: https://docs.mistral.ai/guides/embeddings
-    seq_len: null
-    size: null
   msmarco-bert-co-condensor:
-    dim: 768
-    is_external: true
-    is_proprietary: false
-    is_sentence_transformers_compatible: true
     link: https://huggingface.co/sentence-transformers/msmarco-bert-co-condensor
     seq_len: 512
     size: 110
-  multi-qa-MiniLM-L6-cos-v1:
-    dim: 384
+    dim: 768
     is_external: true
     is_proprietary: false
     is_sentence_transformers_compatible: true
+  multi-qa-MiniLM-L6-cos-v1:
     link: https://huggingface.co/sentence-transformers/multi-qa-MiniLM-L6-cos-v1
     seq_len: 512
     size: 23
-  multilingual-e5-base:
-    dim: 768
+    dim: 384
     is_external: true
     is_proprietary: false
     is_sentence_transformers_compatible: true
+  multilingual-e5-base:
     link: https://huggingface.co/intfloat/multilingual-e5-base
     seq_len: 514
     size: 278
-  multilingual-e5-large:
-    dim: 1024
+    dim: 768
     is_external: true
     is_proprietary: false
     is_sentence_transformers_compatible: true
+  multilingual-e5-large:
     link: https://huggingface.co/intfloat/multilingual-e5-large
     seq_len: 514
     size: 560
-  multilingual-e5-small:
-    dim: 384
+    dim: 1024
     is_external: true
     is_proprietary: false
     is_sentence_transformers_compatible: true
+  multilingual-e5-small:
     link: https://huggingface.co/intfloat/multilingual-e5-small
     seq_len: 512
     size: 118
-  nb-bert-base:
-    dim: 768
+    dim: 384
     is_external: true
     is_proprietary: false
     is_sentence_transformers_compatible: true
+  nb-bert-base:
     link: https://huggingface.co/NbAiLab/nb-bert-base
     seq_len: 512
     size: 179
-  nb-bert-large:
-    dim: 1024
+    dim: 768
     is_external: true
     is_proprietary: false
     is_sentence_transformers_compatible: true
+  nb-bert-large:
     link: https://huggingface.co/NbAiLab/nb-bert-large
     seq_len: 512
     size: 355
-  nomic-embed-text-v1.5-128:
-    dim: 128
+    dim: 1024
     is_external: true
     is_proprietary: false
     is_sentence_transformers_compatible: true
+  nomic-embed-text-v1.5-128:
     link: https://huggingface.co/nomic-ai/nomic-embed-text-v1.5
     seq_len: 8192
     size: 138
-  nomic-embed-text-v1.5-256:
-    dim: 256
+    dim: 128
     is_external: true
     is_proprietary: false
     is_sentence_transformers_compatible: true
+  nomic-embed-text-v1.5-256:
     link: https://huggingface.co/nomic-ai/nomic-embed-text-v1.5
     seq_len: 8192
     size: 138
-  nomic-embed-text-v1.5-512:
-    dim: 512
+    dim: 256
     is_external: true
     is_proprietary: false
     is_sentence_transformers_compatible: true
+  nomic-embed-text-v1.5-512:
     link: https://huggingface.co/nomic-ai/nomic-embed-text-v1.5
     seq_len: 8192
     size: 138
-  nomic-embed-text-v1.5-64:
-    dim: 64
+    dim: 512
     is_external: true
     is_proprietary: false
     is_sentence_transformers_compatible: true
+  nomic-embed-text-v1.5-64:
     link: https://huggingface.co/nomic-ai/nomic-embed-text-v1.5
     seq_len: 8192
     size: 138
-  norbert3-base:
-    dim: 768
+    dim: 64
     is_external: true
     is_proprietary: false
     is_sentence_transformers_compatible: true
+  norbert3-base:
     link: https://huggingface.co/ltg/norbert3-base
     seq_len: 512
     size: 131
-  norbert3-large:
-    dim: 1024
+    dim: 768
     is_external: true
     is_proprietary: false
     is_sentence_transformers_compatible: true
+  norbert3-large:
     link: https://huggingface.co/ltg/norbert3-large
     seq_len: 512
     size: 368
-  paraphrase-multilingual-MiniLM-L12-v2:
-    dim: 384
+    dim: 1024
     is_external: true
     is_proprietary: false
     is_sentence_transformers_compatible: true
+  paraphrase-multilingual-MiniLM-L12-v2:
     link: https://huggingface.co/sentence-transformers/paraphrase-multilingual-MiniLM-L12-v2
     seq_len: 512
     size: 118
-  paraphrase-multilingual-mpnet-base-v2:
-    dim: 768
+    dim: 384
     is_external: true
     is_proprietary: false
     is_sentence_transformers_compatible: true
+  paraphrase-multilingual-mpnet-base-v2:
     link: https://huggingface.co/sentence-transformers/paraphrase-multilingual-mpnet-base-v2
     seq_len: 514
     size: 278
-  sentence-bert-swedish-cased:
     dim: 768
     is_external: true
     is_proprietary: false
     is_sentence_transformers_compatible: true
+  sentence-bert-swedish-cased:
     link: https://huggingface.co/KBLab/sentence-bert-swedish-cased
     seq_len: 512
     size: 125
-  sentence-camembert-base:
     dim: 768
     is_external: true
     is_proprietary: false
     is_sentence_transformers_compatible: true
+  sentence-camembert-base:
     link: https://huggingface.co/dangvantuan/sentence-camembert-base
     seq_len: 512
     size: 110
-  sentence-camembert-large:
-    dim: 1024
+    dim: 768
     is_external: true
     is_proprietary: false
     is_sentence_transformers_compatible: true
+  sentence-camembert-large:
     link: https://huggingface.co/dangvantuan/sentence-camembert-large
     seq_len: 512
     size: 337
-  sentence-croissant-llm-base:
-    dim: 2048
+    dim: 1024
     is_external: true
     is_proprietary: false
     is_sentence_transformers_compatible: true
+  sentence-croissant-llm-base:
     link: https://huggingface.co/Wissam42/sentence-croissant-llm-base
     seq_len: 2048
     size: 1280
-  sentence-t5-base:
-    dim: 768
+    dim: 2048
     is_external: true
     is_proprietary: false
     is_sentence_transformers_compatible: true
+  sentence-t5-base:
     link: https://huggingface.co/sentence-transformers/sentence-t5-base
     seq_len: 512
     size: 110
-  sentence-t5-large:
     dim: 768
     is_external: true
     is_proprietary: false
     is_sentence_transformers_compatible: true
+  sentence-t5-large:
     link: https://huggingface.co/sentence-transformers/sentence-t5-large
     seq_len: 512
     size: 168
-  sentence-t5-xl:
     dim: 768
     is_external: true
     is_proprietary: false
     is_sentence_transformers_compatible: true
+  sentence-t5-xl:
     link: https://huggingface.co/sentence-transformers/sentence-t5-xl
     seq_len: 512
     size: 1240
-  sentence-t5-xxl:
     dim: 768
     is_external: true
     is_proprietary: false
     is_sentence_transformers_compatible: true
+  sentence-t5-xxl:
     link: https://huggingface.co/sentence-transformers/sentence-t5-xxl
     seq_len: 512
     size: 4865
-  silver-retriever-base-v1:
     dim: 768
     is_external: true
     is_proprietary: false
     is_sentence_transformers_compatible: true
+  silver-retriever-base-v1:
     link: https://huggingface.co/ipipan/silver-retriever-base-v1
     seq_len: 514
     size: 125
-  st-polish-paraphrase-from-distilroberta:
     dim: 768
     is_external: true
     is_proprietary: false
     is_sentence_transformers_compatible: true
+  st-polish-paraphrase-from-distilroberta:
     link: https://huggingface.co/sdadas/st-polish-paraphrase-from-distilroberta
     seq_len: 514
     size: 125
-  st-polish-paraphrase-from-mpnet:
     dim: 768
     is_external: true
     is_proprietary: false
     is_sentence_transformers_compatible: true
+  st-polish-paraphrase-from-mpnet:
     link: https://huggingface.co/sdadas/st-polish-paraphrase-from-mpnet
     seq_len: 514
     size: 125
-  sup-simcse-bert-base-uncased:
     dim: 768
     is_external: true
     is_proprietary: false
     is_sentence_transformers_compatible: true
+  sup-simcse-bert-base-uncased:
     link: https://huggingface.co/princeton-nlp/sup-simcse-bert-base-uncased
     seq_len: 512
     size: 110
+    dim: 768
+    is_external: true
+    is_proprietary: false
+    is_sentence_transformers_compatible: true
   text-embedding-3-large:
+    link: https://openai.com/blog/new-embedding-models-and-api-updates
+    seq_len: 8191
+    size: null
     dim: 3072
     is_external: true
     is_proprietary: true
     is_sentence_transformers_compatible: false
+  text-embedding-3-large-256:
     link: https://openai.com/blog/new-embedding-models-and-api-updates
     seq_len: 8191
     size: null
-  text-embedding-3-large-256:
     dim: 256
     is_external: true
     is_proprietary: true
     is_sentence_transformers_compatible: false
+  text-embedding-3-small:
     link: https://openai.com/blog/new-embedding-models-and-api-updates
     seq_len: 8191
     size: null
-  text-embedding-3-small:
     dim: 1536
     is_external: true
     is_proprietary: true
     is_sentence_transformers_compatible: false
-    link: https://openai.com/blog/new-embedding-models-and-api-updates
+  text-embedding-ada-002:
+    link: https://openai.com/blog/new-and-improved-embedding-model
     seq_len: 8191
     size: null
-  text-embedding-ada-002:
     dim: 1536
     is_external: true
     is_proprietary: true
     is_sentence_transformers_compatible: false
-    link: https://openai.com/blog/new-and-improved-embedding-model
-    seq_len: 8191
-    size: null
   text-search-ada-001:
+    link: https://openai.com/blog/introducing-text-and-code-embeddings
+    seq_len: 2046
+    size: null
     dim: 1024
     is_external: true
     is_proprietary: true
     is_sentence_transformers_compatible: false
+  text-search-ada-doc-001:
     link: https://openai.com/blog/introducing-text-and-code-embeddings
     seq_len: 2046
     size: null
-  text-search-ada-doc-001:
     dim: 1024
     is_external: true
     is_proprietary: true
     is_sentence_transformers_compatible: false
+  text-search-ada-query-001:
     link: https://openai.com/blog/introducing-text-and-code-embeddings
     seq_len: 2046
     size: null
-  text-search-ada-query-001:
     dim: 1024
     is_external: false
     is_proprietary: true
     is_sentence_transformers_compatible: false
+  text-search-babbage-001:
     link: https://openai.com/blog/introducing-text-and-code-embeddings
     seq_len: 2046
     size: null
-  text-search-babbage-001:
     dim: 2048
     is_external: true
     is_proprietary: true
     is_sentence_transformers_compatible: false
+  text-search-curie-001:
     link: https://openai.com/blog/introducing-text-and-code-embeddings
     seq_len: 2046
     size: null
-  text-search-curie-001:
     dim: 4096
     is_external: true
     is_proprietary: true
     is_sentence_transformers_compatible: false
+  text-search-davinci-001:
     link: https://openai.com/blog/introducing-text-and-code-embeddings
     seq_len: 2046
     size: null
-  text-search-davinci-001:
     dim: 12288
     is_external: true
     is_proprietary: true
     is_sentence_transformers_compatible: false
+  text-similarity-ada-001:
     link: https://openai.com/blog/introducing-text-and-code-embeddings
     seq_len: 2046
     size: null
-  text-similarity-ada-001:
     dim: 1024
     is_external: true
     is_proprietary: true
     is_sentence_transformers_compatible: false
+  text-similarity-babbage-001:
     link: https://openai.com/blog/introducing-text-and-code-embeddings
     seq_len: 2046
     size: null
-  text-similarity-babbage-001:
     dim: 2048
     is_external: true
     is_proprietary: true
     is_sentence_transformers_compatible: false
+  text-similarity-curie-001:
     link: https://openai.com/blog/introducing-text-and-code-embeddings
     seq_len: 2046
     size: null
-  text-similarity-curie-001:
     dim: 4096
     is_external: true
     is_proprietary: true
     is_sentence_transformers_compatible: false
+  text-similarity-davinci-001:
     link: https://openai.com/blog/introducing-text-and-code-embeddings
     seq_len: 2046
     size: null
-  text-similarity-davinci-001:
     dim: 12288
     is_external: true
     is_proprietary: true
     is_sentence_transformers_compatible: false
-    link: https://openai.com/blog/introducing-text-and-code-embeddings
-    seq_len: 2046
-    size: null
   text2vec-base-chinese:
+    link: https://huggingface.co/shibing624/text2vec-base-chinese
+    seq_len: 512
+    size: 102
     dim: 768
     is_external: true
     is_proprietary: false
     is_sentence_transformers_compatible: true
-    link: https://huggingface.co/shibing624/text2vec-base-chinese
-    seq_len: 512
-    size: 102
   text2vec-base-multilingual:
+    link: null
+    seq_len: null
+    size: null
     dim: null
     is_external: true
     is_proprietary: false
     is_sentence_transformers_compatible: false
-    link: null
-    seq_len: null
-    size: null
   text2vec-large-chinese:
+    link: https://huggingface.co/GanymedeNil/text2vec-large-chinese
+    seq_len: 512
+    size: 326
     dim: 1024
     is_external: true
     is_proprietary: false
     is_sentence_transformers_compatible: true
-    link: https://huggingface.co/GanymedeNil/text2vec-large-chinese
-    seq_len: 512
-    size: 326
   titan-embed-text-v1:
-    dim: 1536
-    is_external: true
-    is_proprietary: true
-    is_sentence_transformers_compatible: false
     link: https://docs.aws.amazon.com/bedrock/latest/userguide/embeddings.html
     seq_len: 8000
     size: null
-  udever-bloom-1b1:
     dim: 1536
     is_external: true
-    is_proprietary: false
-    is_sentence_transformers_compatible: true
+    is_proprietary: true
+    is_sentence_transformers_compatible: false
+  udever-bloom-1b1:
     link: https://huggingface.co/izhx/udever-bloom-1b1
     seq_len: 2048
     size: null
-  udever-bloom-560m:
-    dim: 1024
+    dim: 1536
     is_external: true
     is_proprietary: false
     is_sentence_transformers_compatible: true
+  udever-bloom-560m:
     link: https://huggingface.co/izhx/udever-bloom-560m
     seq_len: 2048
     size: null
-  universal-sentence-encoder-multilingual-3:
-    dim: 512
+    dim: 1024
     is_external: true
     is_proprietary: false
     is_sentence_transformers_compatible: true
+  universal-sentence-encoder-multilingual-3:
     link: https://huggingface.co/vprelovac/universal-sentence-encoder-multilingual-3
     seq_len: 512
     size: null
-  universal-sentence-encoder-multilingual-large-3:
     dim: 512
     is_external: true
     is_proprietary: false
     is_sentence_transformers_compatible: true
+  universal-sentence-encoder-multilingual-large-3:
     link: https://huggingface.co/vprelovac/universal-sentence-encoder-multilingual-large-3
     seq_len: 512
     size: null
-  unsup-simcse-bert-base-uncased:
-    dim: 768
+    dim: 512
     is_external: true
     is_proprietary: false
     is_sentence_transformers_compatible: true
+  unsup-simcse-bert-base-uncased:
     link: https://huggingface.co/princeton-nlp/unsup-simcse-bert-base-uncased
     seq_len: 512
     size: 110
-  use-cmlm-multilingual:
     dim: 768
     is_external: true
     is_proprietary: false
     is_sentence_transformers_compatible: true
+  use-cmlm-multilingual:
     link: https://huggingface.co/sentence-transformers/use-cmlm-multilingual
     seq_len: 512
     size: 472
+    dim: 768
+    is_external: true
+    is_proprietary: false
+    is_sentence_transformers_compatible: true
   voyage-2:
+    link: https://docs.voyageai.com/embeddings/
+    seq_len: 1024
+    size: null
     dim: 1024
     is_external: true
     is_proprietary: true
     is_sentence_transformers_compatible: false
+  voyage-code-2:
     link: https://docs.voyageai.com/embeddings/
-    seq_len: 1024
+    seq_len: 16000
     size: null
-  voyage-code-2:
     dim: 1536
     is_external: true
     is_proprietary: true
     is_sentence_transformers_compatible: false
+  voyage-large-2-instruct:
     link: https://docs.voyageai.com/embeddings/
     seq_len: 16000
     size: null
-  voyage-large-2-instruct:
     dim: 1024
     is_external: true
     is_proprietary: false
     is_sentence_transformers_compatible: false
+  voyage-law-2:
     link: https://docs.voyageai.com/embeddings/
-    seq_len: 16000
+    seq_len: 4000
     size: null
-  voyage-law-2:
     dim: 1024
     is_external: true
     is_proprietary: true
     is_sentence_transformers_compatible: false
+  voyage-lite-01-instruct:
     link: https://docs.voyageai.com/embeddings/
     seq_len: 4000
     size: null
-  voyage-lite-01-instruct:
     dim: 1024
     is_external: true
     is_proprietary: true
     is_sentence_transformers_compatible: false
+  voyage-lite-02-instruct:
     link: https://docs.voyageai.com/embeddings/
     seq_len: 4000
-    size: null
-  voyage-lite-02-instruct:
+    size: 1220
     dim: 1024
     is_external: true
     is_proprietary: true
     is_sentence_transformers_compatible: false
-    link: https://docs.voyageai.com/embeddings/
-    seq_len: 4000
-    size: 1220
   xlm-roberta-base:
-    dim: 768
-    is_external: true
-    is_proprietary: false
-    is_sentence_transformers_compatible: true
     link: https://huggingface.co/xlm-roberta-base
     seq_len: 514
     size: 279
-  xlm-roberta-large:
-    dim: 1024
+    dim: 768
     is_external: true
     is_proprietary: false
     is_sentence_transformers_compatible: true
+  xlm-roberta-large:
     link: https://huggingface.co/xlm-roberta-large
     seq_len: 514
     size: 560
+    dim: 1024
+    is_external: true
+    is_proprietary: false
+    is_sentence_transformers_compatible: true
 models_to_skip:
 - michaelfeil/ct2fast-e5-large-v2
 - McGill-NLP/LLM2Vec-Sheared-LLaMA-mntp-unsup-simcse