XYZ-embedding-zh / README.md
fangxq's picture
Upload README.md
ae60ac9 verified
|
raw
history blame
16.1 kB
metadata
library_name: sentence-transformers
model-index:
  - name: XYZ-embedding-zh
    results:
      - dataset:
          config: default
          name: MTEB CMedQAv1
          revision: None
          split: test
          type: C-MTEB/CMedQAv1-reranking
        metrics:
          - type: map
            value: 89.61792115239176
          - type: mrr
            value: 91.46722222222222
          - type: main_score
            value: 89.61792115239176
        task:
          type: Reranking
      - dataset:
          config: default
          name: MTEB CMedQAv2
          revision: None
          split: test
          type: C-MTEB/CMedQAv2-reranking
        metrics:
          - type: map
            value: 89.22040591564271
          - type: mrr
            value: 91.2995238095238
          - type: main_score
            value: 89.22040591564271
        task:
          type: Reranking
      - dataset:
          config: default
          name: MTEB CmedqaRetrieval
          revision: None
          split: dev
          type: C-MTEB/CmedqaRetrieval
        metrics:
          - type: map_at_1
            value: 27.939000000000004
          - type: map_at_10
            value: 41.227999999999994
          - type: map_at_100
            value: 43.018
          - type: map_at_1000
            value: 43.120000000000005
          - type: map_at_3
            value: 36.895
          - type: map_at_5
            value: 39.373999999999995
          - type: mrr_at_1
            value: 42.136
          - type: mrr_at_10
            value: 50.394000000000005
          - type: mrr_at_100
            value: 51.288
          - type: mrr_at_1000
            value: 51.324000000000005
          - type: mrr_at_3
            value: 47.887
          - type: mrr_at_5
            value: 49.362
          - type: ndcg_at_1
            value: 42.136
          - type: ndcg_at_10
            value: 47.899
          - type: ndcg_at_100
            value: 54.730999999999995
          - type: ndcg_at_1000
            value: 56.462999999999994
          - type: ndcg_at_3
            value: 42.66
          - type: ndcg_at_5
            value: 44.913
          - type: precision_at_1
            value: 42.136
          - type: precision_at_10
            value: 10.52
          - type: precision_at_100
            value: 1.6070000000000002
          - type: precision_at_1000
            value: 0.183
          - type: precision_at_3
            value: 24.064
          - type: precision_at_5
            value: 17.374000000000002
          - type: recall_at_1
            value: 27.939000000000004
          - type: recall_at_10
            value: 58.29600000000001
          - type: recall_at_100
            value: 86.504
          - type: recall_at_1000
            value: 98.105
          - type: recall_at_3
            value: 42.475
          - type: recall_at_5
            value: 49.454
          - type: main_score
            value: 47.899
        task:
          type: Retrieval
      - dataset:
          config: default
          name: MTEB CovidRetrieval
          revision: None
          split: dev
          type: C-MTEB/CovidRetrieval
        metrics:
          - type: map_at_1
            value: 77.371
          - type: map_at_10
            value: 85.229
          - type: map_at_100
            value: 85.358
          - type: map_at_1000
            value: 85.36
          - type: map_at_3
            value: 84.176
          - type: map_at_5
            value: 84.79299999999999
          - type: mrr_at_1
            value: 77.661
          - type: mrr_at_10
            value: 85.207
          - type: mrr_at_100
            value: 85.33699999999999
          - type: mrr_at_1000
            value: 85.339
          - type: mrr_at_3
            value: 84.229
          - type: mrr_at_5
            value: 84.79299999999999
          - type: ndcg_at_1
            value: 77.766
          - type: ndcg_at_10
            value: 88.237
          - type: ndcg_at_100
            value: 88.777
          - type: ndcg_at_1000
            value: 88.818
          - type: ndcg_at_3
            value: 86.16
          - type: ndcg_at_5
            value: 87.22
          - type: precision_at_1
            value: 77.766
          - type: precision_at_10
            value: 9.841999999999999
          - type: precision_at_100
            value: 1.0070000000000001
          - type: precision_at_1000
            value: 0.101
          - type: precision_at_3
            value: 30.875000000000004
          - type: precision_at_5
            value: 19.073
          - type: recall_at_1
            value: 77.371
          - type: recall_at_10
            value: 97.366
          - type: recall_at_100
            value: 99.684
          - type: recall_at_1000
            value: 100
          - type: recall_at_3
            value: 91.702
          - type: recall_at_5
            value: 94.31
          - type: main_score
            value: 88.237
        task:
          type: Retrieval
      - dataset:
          config: default
          name: MTEB DuRetrieval
          revision: None
          split: dev
          type: C-MTEB/DuRetrieval
        metrics:
          - type: map_at_1
            value: 27.772000000000002
          - type: map_at_10
            value: 84.734
          - type: map_at_100
            value: 87.298
          - type: map_at_1000
            value: 87.32900000000001
          - type: map_at_3
            value: 59.431
          - type: map_at_5
            value: 74.82900000000001
          - type: mrr_at_1
            value: 93.65
          - type: mrr_at_10
            value: 95.568
          - type: mrr_at_100
            value: 95.608
          - type: mrr_at_1000
            value: 95.609
          - type: mrr_at_3
            value: 95.267
          - type: mrr_at_5
            value: 95.494
          - type: ndcg_at_1
            value: 93.65
          - type: ndcg_at_10
            value: 90.794
          - type: ndcg_at_100
            value: 92.88300000000001
          - type: ndcg_at_1000
            value: 93.144
          - type: ndcg_at_3
            value: 90.32
          - type: ndcg_at_5
            value: 89.242
          - type: precision_at_1
            value: 93.65
          - type: precision_at_10
            value: 42.9
          - type: precision_at_100
            value: 4.835
          - type: precision_at_1000
            value: 0.49
          - type: precision_at_3
            value: 80.85
          - type: precision_at_5
            value: 68.14
          - type: recall_at_1
            value: 27.772000000000002
          - type: recall_at_10
            value: 91.183
          - type: recall_at_100
            value: 98.219
          - type: recall_at_1000
            value: 99.55000000000001
          - type: recall_at_3
            value: 60.911
          - type: recall_at_5
            value: 78.31099999999999
          - type: main_score
            value: 90.794
        task:
          type: Retrieval
      - dataset:
          config: default
          name: MTEB EcomRetrieval
          revision: None
          split: dev
          type: C-MTEB/EcomRetrieval
        metrics:
          - type: map_at_1
            value: 54.6
          - type: map_at_10
            value: 64.742
          - type: map_at_100
            value: 65.289
          - type: map_at_1000
            value: 65.29700000000001
          - type: map_at_3
            value: 62.183
          - type: map_at_5
            value: 63.883
          - type: mrr_at_1
            value: 54.6
          - type: mrr_at_10
            value: 64.742
          - type: mrr_at_100
            value: 65.289
          - type: mrr_at_1000
            value: 65.29700000000001
          - type: mrr_at_3
            value: 62.183
          - type: mrr_at_5
            value: 63.883
          - type: ndcg_at_1
            value: 54.6
          - type: ndcg_at_10
            value: 69.719
          - type: ndcg_at_100
            value: 72.148
          - type: ndcg_at_1000
            value: 72.393
          - type: ndcg_at_3
            value: 64.606
          - type: ndcg_at_5
            value: 67.682
          - type: precision_at_1
            value: 54.6
          - type: precision_at_10
            value: 8.53
          - type: precision_at_100
            value: 0.962
          - type: precision_at_1000
            value: 0.098
          - type: precision_at_3
            value: 23.867
          - type: precision_at_5
            value: 15.82
          - type: recall_at_1
            value: 54.6
          - type: recall_at_10
            value: 85.3
          - type: recall_at_100
            value: 96.2
          - type: recall_at_1000
            value: 98.2
          - type: recall_at_3
            value: 71.6
          - type: recall_at_5
            value: 79.10000000000001
          - type: main_score
            value: 69.719
        task:
          type: Retrieval
      - dataset:
          config: default
          name: MTEB MMarcoReranking
          revision: None
          split: dev
          type: C-MTEB/Mmarco-reranking
        metrics:
          - type: map
            value: 35.30260957061897
          - type: mrr
            value: 34.098015873015875
          - type: main_score
            value: 35.30260957061897
        task:
          type: Reranking
      - dataset:
          config: default
          name: MTEB MMarcoRetrieval
          revision: None
          split: dev
          type: C-MTEB/MMarcoRetrieval
        metrics:
          - type: map_at_1
            value: 69.51899999999999
          - type: map_at_10
            value: 78.816
          - type: map_at_100
            value: 79.08500000000001
          - type: map_at_1000
            value: 79.091
          - type: map_at_3
            value: 76.999
          - type: map_at_5
            value: 78.194
          - type: mrr_at_1
            value: 71.80499999999999
          - type: mrr_at_10
            value: 79.29899999999999
          - type: mrr_at_100
            value: 79.532
          - type: mrr_at_1000
            value: 79.537
          - type: mrr_at_3
            value: 77.703
          - type: mrr_at_5
            value: 78.75999999999999
          - type: ndcg_at_1
            value: 71.80499999999999
          - type: ndcg_at_10
            value: 82.479
          - type: ndcg_at_100
            value: 83.611
          - type: ndcg_at_1000
            value: 83.76400000000001
          - type: ndcg_at_3
            value: 79.065
          - type: ndcg_at_5
            value: 81.092
          - type: precision_at_1
            value: 71.80499999999999
          - type: precision_at_10
            value: 9.91
          - type: precision_at_100
            value: 1.046
          - type: precision_at_1000
            value: 0.106
          - type: precision_at_3
            value: 29.727999999999998
          - type: precision_at_5
            value: 18.908
          - type: recall_at_1
            value: 69.51899999999999
          - type: recall_at_10
            value: 93.24
          - type: recall_at_100
            value: 98.19099999999999
          - type: recall_at_1000
            value: 99.36500000000001
          - type: recall_at_3
            value: 84.308
          - type: recall_at_5
            value: 89.119
          - type: main_score
            value: 82.479
        task:
          type: Retrieval
      - dataset:
          config: default
          name: MTEB MedicalRetrieval
          revision: None
          split: dev
          type: C-MTEB/MedicalRetrieval
        metrics:
          - type: map_at_1
            value: 57.8
          - type: map_at_10
            value: 64.215
          - type: map_at_100
            value: 64.78
          - type: map_at_1000
            value: 64.81099999999999
          - type: map_at_3
            value: 62.64999999999999
          - type: map_at_5
            value: 63.57000000000001
          - type: mrr_at_1
            value: 58.099999999999994
          - type: mrr_at_10
            value: 64.371
          - type: mrr_at_100
            value: 64.936
          - type: mrr_at_1000
            value: 64.96600000000001
          - type: mrr_at_3
            value: 62.8
          - type: mrr_at_5
            value: 63.739999999999995
          - type: ndcg_at_1
            value: 57.8
          - type: ndcg_at_10
            value: 67.415
          - type: ndcg_at_100
            value: 70.38799999999999
          - type: ndcg_at_1000
            value: 71.229
          - type: ndcg_at_3
            value: 64.206
          - type: ndcg_at_5
            value: 65.858
          - type: precision_at_1
            value: 57.8
          - type: precision_at_10
            value: 7.75
          - type: precision_at_100
            value: 0.919
          - type: precision_at_1000
            value: 0.099
          - type: precision_at_3
            value: 22.900000000000002
          - type: precision_at_5
            value: 14.540000000000001
          - type: recall_at_1
            value: 57.8
          - type: recall_at_10
            value: 77.5
          - type: recall_at_100
            value: 91.9
          - type: recall_at_1000
            value: 98.6
          - type: recall_at_3
            value: 68.7
          - type: recall_at_5
            value: 72.7
          - type: main_score
            value: 67.415
        task:
          type: Retrieval
      - dataset:
          config: default
          name: MTEB T2Reranking
          revision: None
          split: dev
          type: C-MTEB/T2Reranking
        metrics:
          - type: map
            value: 69.06615146698508
          - type: mrr
            value: 79.7588755091294
          - type: main_score
            value: 69.06615146698508
        task:
          type: Reranking
      - dataset:
          config: default
          name: MTEB T2Retrieval
          revision: None
          split: dev
          type: C-MTEB/T2Retrieval
        metrics:
          - type: map_at_1
            value: 28.084999999999997
          - type: map_at_10
            value: 78.583
          - type: map_at_100
            value: 82.14399999999999
          - type: map_at_1000
            value: 82.204
          - type: map_at_3
            value: 55.422000000000004
          - type: map_at_5
            value: 67.973
          - type: mrr_at_1
            value: 91.014
          - type: mrr_at_10
            value: 93.381
          - type: mrr_at_100
            value: 93.45400000000001
          - type: mrr_at_1000
            value: 93.45599999999999
          - type: mrr_at_3
            value: 92.99300000000001
          - type: mrr_at_5
            value: 93.234
          - type: ndcg_at_1
            value: 91.014
          - type: ndcg_at_10
            value: 85.931
          - type: ndcg_at_100
            value: 89.31
          - type: ndcg_at_1000
            value: 89.869
          - type: ndcg_at_3
            value: 87.348
          - type: ndcg_at_5
            value: 85.929
          - type: precision_at_1
            value: 91.014
          - type: precision_at_10
            value: 42.495
          - type: precision_at_100
            value: 5.029999999999999
          - type: precision_at_1000
            value: 0.516
          - type: precision_at_3
            value: 76.248
          - type: precision_at_5
            value: 63.817
          - type: recall_at_1
            value: 28.084999999999997
          - type: recall_at_10
            value: 84.88
          - type: recall_at_100
            value: 95.902
          - type: recall_at_1000
            value: 98.699
          - type: recall_at_3
            value: 57.113
          - type: recall_at_5
            value: 71.251
          - type: main_score
            value: 85.931
        task:
          type: Retrieval
      - dataset:
          config: default
          name: MTEB VideoRetrieval
          revision: None
          split: dev
          type: C-MTEB/VideoRetrieval
        metrics:
          - type: map_at_1
            value: 66.4
          - type: map_at_10
            value: 75.86
          - type: map_at_100
            value: 76.185
          - type: map_at_1000
            value: 76.188
          - type: map_at_3
            value: 74.167
          - type: map_at_5
            value: 75.187
          - type: mrr_at_1
            value: 66.4
          - type: mrr_at_10
            value: 75.86
          - type: mrr_at_100
            value: 76.185
          - type: mrr_at_1000
            value: 76.188
          - type: mrr_at_3
            value: 74.167
          - type: mrr_at_5
            value: 75.187
          - type: ndcg_at_1
            value: 66.4
          - type: ndcg_at_10
            value: 80.03099999999999
          - type: ndcg_at_100
            value: 81.459
          - type: ndcg_at_1000
            value: 81.527
          - type: ndcg_at_3
            value: 76.621
          - type: ndcg_at_5
            value: 78.446
          - type: precision_at_1
            value: 66.4
          - type: precision_at_10
            value: 9.29
          - type: precision_at_100
            value: 0.992
          - type: precision_at_1000
            value: 0.1
          - type: precision_at_3
            value: 27.900000000000002
          - type: precision_at_5
            value: 17.62
          - type: recall_at_1
            value: 66.4
          - type: recall_at_10
            value: 92.9
          - type: recall_at_100
            value: 99.2
          - type: recall_at_1000
            value: 99.7
          - type: recall_at_3
            value: 83.7
          - type: recall_at_5
            value: 88.1
          - type: main_score
            value: 80.03099999999999
        task:
          type: Retrieval
pipeline_tag: sentence-similarity
tags:
  - sentence-transformers
  - feature-extraction
  - sentence-similarity
  - mteb

XYZ-embedding-zh

This is a sentence-transformers model: It maps sentences & paragraphs to a 1792 dimensional dense vector space and can be used for tasks like clustering or semantic search.

Usage (Sentence-Transformers)

Using this model becomes easy when you have sentence-transformers installed:

pip install -U sentence-transformers

Then you can use the model like this:

from sentence_transformers import SentenceTransformer
sentences = ["This is an example sentence", "Each sentence is converted"]

model = SentenceTransformer('fangxq/XYZ-embedding-zh')
embeddings = model.encode(sentences)
print(embeddings)

Evaluation Results

For an automated evaluation of this model, see the Sentence Embeddings Benchmark: https://seb.sbert.net

Full Model Architecture

SentenceTransformer(
  (0): Transformer({'max_seq_length': 512, 'do_lower_case': False}) with Transformer model: BertModel 
  (1): Pooling({'word_embedding_dimension': 1024, 'pooling_mode_cls_token': False, 'pooling_mode_mean_tokens': True, 'pooling_mode_max_tokens': False, 'pooling_mode_mean_sqrt_len_tokens': False, 'pooling_mode_weightedmean_tokens': False, 'pooling_mode_lasttoken': False, 'include_prompt': True})
  (2): Dense({'in_features': 1024, 'out_features': 1792, 'bias': True, 'activation_function': 'torch.nn.modules.linear.Identity'})
)

Citing & Authors