zeroshot's picture
Update README.md
1429ba1
metadata
license: mit
language:
  - en
tags:
  - sparse sparsity quantized onnx embeddings int8
  - mteb
model-index:
  - name: bge-large-en-v1.5-sparse
    results:
      - task:
          type: STS
        dataset:
          type: mteb/biosses-sts
          name: MTEB BIOSSES
          config: default
          split: test
          revision: d3fb88f8f02e40887cd149695127462bbcf29b4a
        metrics:
          - type: cos_sim_pearson
            value: 87.73305831153709
          - type: cos_sim_spearman
            value: 85.64351771070989
          - type: euclidean_pearson
            value: 86.06880877736519
          - type: euclidean_spearman
            value: 85.60676988543395
          - type: manhattan_pearson
            value: 85.69108036145253
          - type: manhattan_spearman
            value: 85.05314281283421
      - task:
          type: STS
        dataset:
          type: mteb/sickr-sts
          name: MTEB SICK-R
          config: default
          split: test
          revision: a6ea5a8cab320b040a23452cc28066d9beae2cee
        metrics:
          - type: cos_sim_pearson
            value: 85.61833776000717
          - type: cos_sim_spearman
            value: 80.73718686921521
          - type: euclidean_pearson
            value: 83.9368704709159
          - type: euclidean_spearman
            value: 80.64477415487963
          - type: manhattan_pearson
            value: 83.92383757341743
          - type: manhattan_spearman
            value: 80.59625506933862
      - task:
          type: STS
        dataset:
          type: mteb/sts12-sts
          name: MTEB STS12
          config: default
          split: test
          revision: a0d554a64d88156834ff5ae9920b964011b16384
        metrics:
          - type: cos_sim_pearson
            value: 83.81272888013494
          - type: cos_sim_spearman
            value: 76.07038564455931
          - type: euclidean_pearson
            value: 80.33676600912023
          - type: euclidean_spearman
            value: 75.86575335744111
          - type: manhattan_pearson
            value: 80.36973770593211
          - type: manhattan_spearman
            value: 75.88787860200954
      - task:
          type: STS
        dataset:
          type: mteb/sts13-sts
          name: MTEB STS13
          config: default
          split: test
          revision: 7e90230a92c190f1bf69ae9002b8cea547a64cca
        metrics:
          - type: cos_sim_pearson
            value: 85.58781524090651
          - type: cos_sim_spearman
            value: 86.80508359626748
          - type: euclidean_pearson
            value: 85.22891409219575
          - type: euclidean_spearman
            value: 85.78295876926319
          - type: manhattan_pearson
            value: 85.2193177032458
          - type: manhattan_spearman
            value: 85.74049940198427
      - task:
          type: STS
        dataset:
          type: mteb/sts14-sts
          name: MTEB STS14
          config: default
          split: test
          revision: 6031580fec1f6af667f0bd2da0a551cf4f0b2375
        metrics:
          - type: cos_sim_pearson
            value: 84.0862821699066
          - type: cos_sim_spearman
            value: 81.67856196476185
          - type: euclidean_pearson
            value: 83.38475353138897
          - type: euclidean_spearman
            value: 81.45279784228292
          - type: manhattan_pearson
            value: 83.29235221714131
          - type: manhattan_spearman
            value: 81.3971683104493
      - task:
          type: STS
        dataset:
          type: mteb/sts15-sts
          name: MTEB STS15
          config: default
          split: test
          revision: ae752c7c21bf194d8b67fd573edf7ae58183cbe3
        metrics:
          - type: cos_sim_pearson
            value: 87.44459051393112
          - type: cos_sim_spearman
            value: 88.74673154561383
          - type: euclidean_pearson
            value: 88.13112382236628
          - type: euclidean_spearman
            value: 88.56241954487271
          - type: manhattan_pearson
            value: 88.11098632041256
          - type: manhattan_spearman
            value: 88.55607051247829
      - task:
          type: STS
        dataset:
          type: mteb/sts16-sts
          name: MTEB STS16
          config: default
          split: test
          revision: 4d8694f8f0e0100860b497b999b3dbed754a0513
        metrics:
          - type: cos_sim_pearson
            value: 82.8825746257794
          - type: cos_sim_spearman
            value: 84.6066555379785
          - type: euclidean_pearson
            value: 84.12438131112606
          - type: euclidean_spearman
            value: 84.75862802179907
          - type: manhattan_pearson
            value: 84.12791217960807
          - type: manhattan_spearman
            value: 84.7739597139034
      - task:
          type: STS
        dataset:
          type: mteb/sts17-crosslingual-sts
          name: MTEB STS17 (en-en)
          config: en-en
          split: test
          revision: af5e6fb845001ecf41f4c1e033ce921939a2a68d
        metrics:
          - type: cos_sim_pearson
            value: 89.19971502207773
          - type: cos_sim_spearman
            value: 89.75109780507901
          - type: euclidean_pearson
            value: 89.5913898113725
          - type: euclidean_spearman
            value: 89.20244860773123
          - type: manhattan_pearson
            value: 89.68755363801112
          - type: manhattan_spearman
            value: 89.3105024782381
      - task:
          type: STS
        dataset:
          type: mteb/sts22-crosslingual-sts
          name: MTEB STS22 (en)
          config: en
          split: test
          revision: 6d1ba47164174a496b7fa5d3569dae26a6813b80
        metrics:
          - type: cos_sim_pearson
            value: 61.73885819503523
          - type: cos_sim_spearman
            value: 64.09521607825829
          - type: euclidean_pearson
            value: 64.22116001518724
          - type: euclidean_spearman
            value: 63.84189650719827
          - type: manhattan_pearson
            value: 64.23930191730729
          - type: manhattan_spearman
            value: 63.7536172795383
      - task:
          type: STS
        dataset:
          type: mteb/stsbenchmark-sts
          name: MTEB STSBenchmark
          config: default
          split: test
          revision: b0fddb56ed78048fa8b90373c8a3cfc37b684831
        metrics:
          - type: cos_sim_pearson
            value: 85.68505574064375
          - type: cos_sim_spearman
            value: 86.87614324154406
          - type: euclidean_pearson
            value: 86.96751967489614
          - type: euclidean_spearman
            value: 86.78979082790067
          - type: manhattan_pearson
            value: 86.92578795715433
          - type: manhattan_spearman
            value: 86.74076104131726
      - task:
          type: PairClassification
        dataset:
          type: mteb/sprintduplicatequestions-pairclassification
          name: MTEB SprintDuplicateQuestions
          config: default
          split: test
          revision: d66bd1f72af766a5cc4b0ca5e00c162f89e8cc46
        metrics:
          - type: cos_sim_accuracy
            value: 99.80990099009901
          - type: cos_sim_ap
            value: 95.00187845875503
          - type: cos_sim_f1
            value: 90.37698412698413
          - type: cos_sim_precision
            value: 89.66535433070865
          - type: cos_sim_recall
            value: 91.10000000000001
          - type: dot_accuracy
            value: 99.63366336633663
          - type: dot_ap
            value: 87.6642728041652
          - type: dot_f1
            value: 81.40803173029252
          - type: dot_precision
            value: 80.7276302851524
          - type: dot_recall
            value: 82.1
          - type: euclidean_accuracy
            value: 99.8079207920792
          - type: euclidean_ap
            value: 94.88531851782375
          - type: euclidean_f1
            value: 90.49019607843137
          - type: euclidean_precision
            value: 88.75
          - type: euclidean_recall
            value: 92.30000000000001
          - type: manhattan_accuracy
            value: 99.81188118811882
          - type: manhattan_ap
            value: 94.87944331919043
          - type: manhattan_f1
            value: 90.5
          - type: manhattan_precision
            value: 90.5
          - type: manhattan_recall
            value: 90.5
          - type: max_accuracy
            value: 99.81188118811882
          - type: max_ap
            value: 95.00187845875503
          - type: max_f1
            value: 90.5
      - task:
          type: PairClassification
        dataset:
          type: mteb/twittersemeval2015-pairclassification
          name: MTEB TwitterSemEval2015
          config: default
          split: test
          revision: 70970daeab8776df92f5ea462b6173c0b46fd2d1
        metrics:
          - type: cos_sim_accuracy
            value: 86.3861238600465
          - type: cos_sim_ap
            value: 74.50058066578084
          - type: cos_sim_f1
            value: 69.25949774629748
          - type: cos_sim_precision
            value: 67.64779874213836
          - type: cos_sim_recall
            value: 70.94986807387863
          - type: dot_accuracy
            value: 81.57000655659535
          - type: dot_ap
            value: 59.10193583653485
          - type: dot_f1
            value: 58.39352155832786
          - type: dot_precision
            value: 49.88780852655198
          - type: dot_recall
            value: 70.3957783641161
          - type: euclidean_accuracy
            value: 86.37420277761221
          - type: euclidean_ap
            value: 74.41671247141966
          - type: euclidean_f1
            value: 69.43907156673114
          - type: euclidean_precision
            value: 64.07853636769299
          - type: euclidean_recall
            value: 75.77836411609499
          - type: manhattan_accuracy
            value: 86.30267628300649
          - type: manhattan_ap
            value: 74.34438603336339
          - type: manhattan_f1
            value: 69.41888619854721
          - type: manhattan_precision
            value: 64.13870246085011
          - type: manhattan_recall
            value: 75.64643799472296
          - type: max_accuracy
            value: 86.3861238600465
          - type: max_ap
            value: 74.50058066578084
          - type: max_f1
            value: 69.43907156673114
      - task:
          type: PairClassification
        dataset:
          type: mteb/twitterurlcorpus-pairclassification
          name: MTEB TwitterURLCorpus
          config: default
          split: test
          revision: 8b6510b0b1fa4e4c4f879467980e9be563ec1cdf
        metrics:
          - type: cos_sim_accuracy
            value: 88.87530562347187
          - type: cos_sim_ap
            value: 85.69496469410068
          - type: cos_sim_f1
            value: 77.96973052787007
          - type: cos_sim_precision
            value: 74.8900865125514
          - type: cos_sim_recall
            value: 81.3135201724669
          - type: dot_accuracy
            value: 86.70780455621532
          - type: dot_ap
            value: 80.03489678512908
          - type: dot_f1
            value: 73.26376129933124
          - type: dot_precision
            value: 70.07591733445804
          - type: dot_recall
            value: 76.75546658453958
          - type: euclidean_accuracy
            value: 88.85978189156674
          - type: euclidean_ap
            value: 85.67894953317325
          - type: euclidean_f1
            value: 78.04295942720763
          - type: euclidean_precision
            value: 75.67254845241538
          - type: euclidean_recall
            value: 80.56667693255312
          - type: manhattan_accuracy
            value: 88.88306748942446
          - type: manhattan_ap
            value: 85.66556510677526
          - type: manhattan_f1
            value: 78.06278290950576
          - type: manhattan_precision
            value: 74.76912231230173
          - type: manhattan_recall
            value: 81.65999384046813
          - type: max_accuracy
            value: 88.88306748942446
          - type: max_ap
            value: 85.69496469410068
          - type: max_f1
            value: 78.06278290950576

bge-large-en-v1.5-sparse

Usage

This is the sparse ONNX variant of the bge-small-en-v1.5 embeddings model accelerated with Sparsify for quantization/pruning and DeepSparseSentenceTransformers for inference.

pip install -U deepsparse-nightly[sentence_transformers]
from deepsparse.sentence_transformers import DeepSparseSentenceTransformer
model = DeepSparseSentenceTransformer('neuralmagic/bge-large-en-v1.5-sparse', export=False)

# Our sentences we like to encode
sentences = ['This framework generates embeddings for each input sentence',
    'Sentences are passed as a list of string.',
    'The quick brown fox jumps over the lazy dog.']

# Sentences are encoded by calling model.encode()
embeddings = model.encode(sentences)

# Print the embeddings
for sentence, embedding in zip(sentences, embeddings):
    print("Sentence:", sentence)
    print("Embedding:", embedding.shape)
    print("")

For general questions on these models and sparsification methods, reach out to the engineering team on our community Slack.