Update conf

config.yaml

@@ -365,6 +365,37 @@ boards:
  - CDSC-R
  - SICK-R-PL
  - STS22 (pl)
+ ru:
+ title: Russian
+ language_long: "Russian"
+ has_overall: true
+ acronym: null
+ icon: "🇷🇺"
+ special_icons: null
+ credits: "[Roman Solomatin](https://github.com/Samoed) and SaluteDevices: [Alena Fenogenova](https://github.com/Alenush), [Aleksandr Abramov](https://github.com/Ab1992ao), [Artem Snegirev](https://github.com/artemsnegirev), [Anna Maksimova](https://github.com/anpalmak2003), [Maria Tikhonova](https://github.com/MariyaTikhonova)"
+ tasks:
+ Classification:
+ - GeoreviewClassification
+ - HeadlineClassification
+ - InappropriatenessClassification
+ - KinopoiskClassification
+ - RuReviewsClassification
+ - RuSciBenchGRNTIClassification
+ - RuSciBenchOECDClassification
+ Clustering:
+ - GeoreviewClusteringP2P
+ - RuSciBenchGRNTIClusteringP2P
+ - RuSciBenchOECDClusteringP2P
+ PairClassification:
+ - TERRa
+ Reranking:
+ - RuBQReranking
+ Retrieval:
+ - RiaNewsRetrieval
+ - RuBQRetrieval
+ STS:
+ - RUParaPhraserSTS
+ - RuSTSBenchmarkSTS
  se:
  title: Swedish
  language_long: Swedish
@@ -452,34 +483,3 @@ boards:
  - BrightRetrieval (aops)
  - BrightRetrieval (theoremqa_theorems)
  - BrightRetrieval (theoremqa_questions)
- ru:
- title: Russian
- language_long: "Russian"
- has_overall: true
- acronym: null
- icon: "🇷🇺"
- special_icons: null
- credits: "[Roman Solomatin](https://github.com/Samoed) and SaluteDevices: [Alena Fenogenova](https://github.com/Alenush), [Aleksandr Abramov](https://github.com/Ab1992ao), [Artem Snegirev](https://github.com/artemsnegirev), [Anna Maksimova](https://github.com/anpalmak2003), [Maria Tikhonova](https://github.com/MariyaTikhonova)"
- tasks:
- Classification:
- - GeoreviewClassification
- - HeadlineClassification
- - InappropriatenessClassification
- - KinopoiskClassification
- - RuReviewsClassification
- - RuSciBenchGRNTIClassification
- - RuSciBenchOECDClassification
- Clustering:
- - GeoreviewClusteringP2P
- - RuSciBenchGRNTIClusteringP2P
- - RuSciBenchOECDClusteringP2P
- PairClassification:
- - TERRa
- Reranking:
- - RuBQReranking
- Retrieval:
- - RiaNewsRetrieval
- - RuBQRetrieval
- STS:
- - RUParaPhraserSTS
- - RuSTSBenchmarkSTS

model_meta.yaml

@@ -212,6 +212,15 @@ model_meta:
  is_proprietary: false
  is_sentence_transformers_compatible: true
  uses_instruct: true
+ SFR-Embedding-2_R:
+ link: https://huggingface.co/Salesforce/SFR-Embedding-2_R
+ seq_len: 32768
+ size: 7111
+ dim: 4096
+ is_external: false
+ is_proprietary: false
+ is_sentence_transformers_compatible: true
+ uses_instruct: true
  all-MiniLM-L12-v2:
  link: https://huggingface.co/sentence-transformers/all-MiniLM-L12-v2
  seq_len: 512
@@ -356,6 +365,24 @@ model_meta:
  is_proprietary: false
  is_sentence_transformers_compatible: true
  uses_instruct: false
+ bge-en-icl:
+ link: https://huggingface.co/BAAI/bge-en-icl
+ seq_len: 32768
+ size: 7110
+ dim: 4096
+ is_external: false
+ is_proprietary: false
+ is_sentence_transformers_compatible: true
+ uses_instruct: true
+ bge-multilingual-gemma2:
+ link: https://huggingface.co/BAAI/bge-multilingual-gemma2
+ seq_len: 8192
+ size: 9240
+ dim: 3584
+ is_external: false
+ is_proprietary: false
+ is_sentence_transformers_compatible: false
+ uses_instruct: true
  bge-large-en-v1.5:
  link: https://huggingface.co/BAAI/bge-large-en-v1.5
  seq_len: 512
@@ -514,15 +541,6 @@ model_meta:
  is_proprietary: false
  is_sentence_transformers_compatible: true
  uses_instruct: false
- dfm-sentence-encoder-large-1:
- link: https://huggingface.co/chcaa/dfm-encoder-large-v1
- seq_len: 512
- size: 355
- dim: 1024
- is_external: false # no result in results repo
- is_proprietary: false
- is_sentence_transformers_compatible: true
- uses_instruct: false
  distilbert-base-25lang-cased:
  link: https://huggingface.co/Geotrend/distilbert-base-25lang-cased
  seq_len: 512
@@ -820,6 +838,15 @@ model_meta:
  is_proprietary: false
  is_sentence_transformers_compatible: true
  uses_instruct: true
+ gte-Qwen2-1.5B-instruct:
+ link: https://huggingface.co/Alibaba-NLP/gte-Qwen2-1.5B-instruct
+ seq_len: 32768
+ size: 1780
+ dim: 1536
+ is_external: false
+ is_proprietary: false
+ is_sentence_transformers_compatible: true
+ uses_instruct: true
  gte-Qwen2-7B-instruct:
  link: https://huggingface.co/Alibaba-NLP/gte-Qwen2-7B-instruct
  seq_len: 32768
@@ -1243,6 +1270,24 @@ model_meta:
  is_proprietary: false
  is_sentence_transformers_compatible: true
  uses_instruct: false
+ stella_en_1.5B_v5:
+ link: https://huggingface.co/dunzhang/stella_en_1.5B_v5
+ seq_len: 512
+ size: 1500
+ dim: 8192
+ is_external: false
+ is_proprietary: false
+ is_sentence_transformers_compatible: true
+ uses_instruct: true 
+ stella_en_400M_v5:
+ link: https://huggingface.co/dunzhang/stella_en_400M_v5
+ seq_len: 512
+ size: 400
+ dim: 8192
+ is_external: false
+ is_proprietary: false
+ is_sentence_transformers_compatible: true
+ uses_instruct: true 
  sup-simcse-bert-base-uncased:
  link: https://huggingface.co/princeton-nlp/sup-simcse-bert-base-uncased
  seq_len: 512
@@ -1874,6 +1919,14 @@ models_to_skip:
 - liddlefish/privacy_embedding_bge_small_synthetic
 - mxs980/gte-Qwen2-1.5B-instruct-Q8_0-GGUF
 - leonn71/gte-Qwen2-1.5B-instruct-Q6_K-GGUF
+- niancheng/gte-Qwen2-7B-instruct-Q4_K_M-GGUF
+- cleatherbury/gte-Qwen2-7B-instruct-Q5_K_M-GGUF
+- niancheng/gte-Qwen2-1.5B-instruct-Q4_K_M-GGUF
+- mxs980/b1ade-embed-Q8_0-GGUF
+- chihlunLee/NoInstruct-small-Embedding-v0-Q4_0-GGUF
+- corto-ai/nomic-embed-text-v1
+- bcastle/snowflake-arctic-embed-l-Q8_0-GGUF
+- Intel/neural-embedding-v1
 cross_encoders:
 - FollowIR-7B
 - flan-t5-base