checkpoint-1498/README.md

metadata

base_model: bobox/DeBERTa-small-ST-v1-test-step3
datasets:
  - tals/vitaminc
  - allenai/scitail
  - allenai/sciq
  - allenai/qasc
  - sentence-transformers/msmarco-msmarco-distilbert-base-v3
  - sentence-transformers/natural-questions
  - sentence-transformers/trivia-qa
  - sentence-transformers/gooaq
  - google-research-datasets/paws
language:
  - en
library_name: sentence-transformers
metrics:
  - pearson_cosine
  - spearman_cosine
  - pearson_manhattan
  - spearman_manhattan
  - pearson_euclidean
  - spearman_euclidean
  - pearson_dot
  - spearman_dot
  - pearson_max
  - spearman_max
  - cosine_accuracy
  - cosine_accuracy_threshold
  - cosine_f1
  - cosine_f1_threshold
  - cosine_precision
  - cosine_recall
  - cosine_ap
  - dot_accuracy
  - dot_accuracy_threshold
  - dot_f1
  - dot_f1_threshold
  - dot_precision
  - dot_recall
  - dot_ap
  - manhattan_accuracy
  - manhattan_accuracy_threshold
  - manhattan_f1
  - manhattan_f1_threshold
  - manhattan_precision
  - manhattan_recall
  - manhattan_ap
  - euclidean_accuracy
  - euclidean_accuracy_threshold
  - euclidean_f1
  - euclidean_f1_threshold
  - euclidean_precision
  - euclidean_recall
  - euclidean_ap
  - max_accuracy
  - max_accuracy_threshold
  - max_f1
  - max_f1_threshold
  - max_precision
  - max_recall
  - max_ap
pipeline_tag: sentence-similarity
tags:
  - sentence-transformers
  - sentence-similarity
  - feature-extraction
  - generated_from_trainer
  - dataset_size:163205
  - loss:CachedGISTEmbedLoss
widget:
  - source_sentence: who is the president of chil
    sentences:
      - >-
        An endoscopic retrograde cholangiopancreatogram (ERCP) is usually done
        by a gastroenterologist. This is a doctor who has special training in
        diseases of the digestive system. The doctor must be trained in
        endoscopy. A thin, flexible fiber-optic endoscope (scope) is used.
      - "No. Cleveland Cavaliers guard, Kyrie Irving, is not the son of former NBA great Julius Erving. Kyrie Irving's father is former Australian professional basketball player Drederick Irving. Kyrie's mother, Elizabeth (who was a standout athlete herself), died when Kyrie was just four. Since then, dad Drederick raised him as a single parent and served as Kyrieâ\x80\x99s first basketball coach."
      - >-
        Veronica Verónica Michelle Bachelet jeria is The president Of. Chile
        she was the first Female president Of chile from 2006 march - 11 2010
        march, 11 and was-re elected for the-2014 2018. Term'chile s presidents
        are not-re electable. immediatelyhe was the first female President of
        Chile from 2006 March 11 - 2010 March 11, and was re-elected for the
        2014-2018 term.
  - source_sentence: >-
      More than 273 people have died from the 2019-20 coronavirus outside
      mainland China .
    sentences:
      - >-
        More than 3,700 people have died : around 3,100 in mainland China and
        around 550 in all other countries combined .
      - >-
        More than 3,200 people have died : almost 3,000 in mainland China and
        around 275 in other countries .
      - more than 4,900 deaths have been attributed to COVID-19 .
  - source_sentence: >-
      The movement of an air mass over earth's surface causes local weather
      changes.
    sentences:
      - The movement of an air mass over Earth's surface causes
      - Which best explains why children resemble their parents?
      - >-
        What ensures that seeds germinate only when conditions for seedling
        survival are optimal?
  - source_sentence: >-
      Heirloom seeds come from plants that were traditionally grown in human
      populations, as opposed to the seeds used for large-scale agricultural
      production.
    sentences:
      - What do you call a substance that is not an acid or a base?
      - >-
        What type of seeds come from plants that were traditionally grown in
        human populations, as opposed to the seeds used for large-scale
        agricultural production?
      - >-
        High consumption of saturated fats is linked to an increased risk of
        what disease?
  - source_sentence: >-
      2012 was the 300th anniversary of the world's first industrial
      steam-powered machine, which was a?
    sentences:
      - >-
        Armed Forces Radio Vietnam | Video | C-SPAN.org Supreme Court November
        11, 2006 Armed Forces Radio in Vietnam Adrian Cronauer talked about
        being a disc jockey in Vietnam and the movie based on his experiences,
        “Good Morning, Vietnam!” The ninth… read more Armed Forces Radio in
        Vietnam Adrian Cronauer talked about being a disc jockey in Vietnam and
        the movie based on his experiences, “Good Morning, Vietnam!” The ninth
        annual conference of the World War II Veterans Committee for the first
        time expanded to include the Vietnam War under the new umbrella
        organization, the American Veterans Center. close Transcript type
      - >-
        The Worlds First Steam Engine 300th Anniversary - YouTube The Worlds
        First Steam Engine 300th Anniversary Want to watch this again later?
        Sign in to add this video to a playlist. Need to report the video? Sign
        in to report inappropriate content. Rating is available when the video
        has been rented. This feature is not available right now. Please try
        again later. Published on Nov 14, 2012 The replica Newcomen Pumping
        Engine at the Black Country Living Museum in Dudley has been brought
        back to life for the 300th anniversary of the first recorded practical
        application of steam power. The self-acting valve gear is a clanking
        cacophony of joy! The restoration team were helped by Guy Martin and the
        engine was featured in the Channel 4 documentary series 'How Britain
        Worked' (as was my own miniature Newcomen engine model, used by Guy to
        demonstrate how the engine works!) See my other vids for some footage of
        my little 'un in action! Category
      - >-
        YouTube | Logopedia | Fandom powered by Wikia 2015–present 2005–2011 The
        logo consists of the black word "You" and a red rounded rectangle with
        the word "Tube" in it next to it. This logo is still being used on some
        other pages. Logo with the slogan "Broadcast Yourself". Notice that the
        red square looks different in this variation. Add a photo to this
        gallery 2011–2013 This modification of the YouTube logo was introduced
        in July 2011 as a part of the Cosmic Panda experiment. It officially
        became the new logo a few months later. It has the red square in a
        darker color this time. Also, starting in 2012, the slogan "Broadcast
        Yourself" was retired. 2013–2015 On December 19, 2013, the red rectangle
        was made lighter in color. Also, the word "You" was made more black and
        the shadow behind the word "Tube" was removed. This is still used as a
        secondary logo. Alternate Version, only for social media.
model-index:
  - name: SentenceTransformer based on bobox/DeBERTa-small-ST-v1-test-step3
    results:
      - task:
          type: semantic-similarity
          name: Semantic Similarity
        dataset:
          name: sts test
          type: sts-test
        metrics:
          - type: pearson_cosine
            value: 0.884780047363058
            name: Pearson Cosine
          - type: spearman_cosine
            value: 0.9069954294787459
            name: Spearman Cosine
          - type: pearson_manhattan
            value: 0.9057450598837923
            name: Pearson Manhattan
          - type: spearman_manhattan
            value: 0.9018185140584661
            name: Spearman Manhattan
          - type: pearson_euclidean
            value: 0.904848497388399
            name: Pearson Euclidean
          - type: spearman_euclidean
            value: 0.9006308603323997
            name: Spearman Euclidean
          - type: pearson_dot
            value: 0.8686782237189175
            name: Pearson Dot
          - type: spearman_dot
            value: 0.8651062350096334
            name: Spearman Dot
          - type: pearson_max
            value: 0.9057450598837923
            name: Pearson Max
          - type: spearman_max
            value: 0.9069954294787459
            name: Spearman Max
      - task:
          type: binary-classification
          name: Binary Classification
        dataset:
          name: allNLI dev
          type: allNLI-dev
        metrics:
          - type: cosine_accuracy
            value: 0.736328125
            name: Cosine Accuracy
          - type: cosine_accuracy_threshold
            value: 0.7933480739593506
            name: Cosine Accuracy Threshold
          - type: cosine_f1
            value: 0.6428571428571429
            name: Cosine F1
          - type: cosine_f1_threshold
            value: 0.6018309593200684
            name: Cosine F1 Threshold
          - type: cosine_precision
            value: 0.5236363636363637
            name: Cosine Precision
          - type: cosine_recall
            value: 0.8323699421965318
            name: Cosine Recall
          - type: cosine_ap
            value: 0.6338667537371122
            name: Cosine Ap
          - type: dot_accuracy
            value: 0.73046875
            name: Dot Accuracy
          - type: dot_accuracy_threshold
            value: 290.1109313964844
            name: Dot Accuracy Threshold
          - type: dot_f1
            value: 0.6099009900990099
            name: Dot F1
          - type: dot_f1_threshold
            value: 200.0629425048828
            name: Dot F1 Threshold
          - type: dot_precision
            value: 0.463855421686747
            name: Dot Precision
          - type: dot_recall
            value: 0.8901734104046243
            name: Dot Recall
          - type: dot_ap
            value: 0.5977502254208901
            name: Dot Ap
          - type: manhattan_accuracy
            value: 0.7421875
            name: Manhattan Accuracy
          - type: manhattan_accuracy_threshold
            value: 273.52783203125
            name: Manhattan Accuracy Threshold
          - type: manhattan_f1
            value: 0.6501128668171559
            name: Manhattan F1
          - type: manhattan_f1_threshold
            value: 362.5467529296875
            name: Manhattan F1 Threshold
          - type: manhattan_precision
            value: 0.5333333333333333
            name: Manhattan Precision
          - type: manhattan_recall
            value: 0.8323699421965318
            name: Manhattan Recall
          - type: manhattan_ap
            value: 0.6336272156038607
            name: Manhattan Ap
          - type: euclidean_accuracy
            value: 0.744140625
            name: Euclidean Accuracy
          - type: euclidean_accuracy_threshold
            value: 12.976269721984863
            name: Euclidean Accuracy Threshold
          - type: euclidean_f1
            value: 0.6419213973799127
            name: Euclidean F1
          - type: euclidean_f1_threshold
            value: 17.44091033935547
            name: Euclidean F1 Threshold
          - type: euclidean_precision
            value: 0.5157894736842106
            name: Euclidean Precision
          - type: euclidean_recall
            value: 0.8497109826589595
            name: Euclidean Recall
          - type: euclidean_ap
            value: 0.6359674505468691
            name: Euclidean Ap
          - type: max_accuracy
            value: 0.744140625
            name: Max Accuracy
          - type: max_accuracy_threshold
            value: 290.1109313964844
            name: Max Accuracy Threshold
          - type: max_f1
            value: 0.6501128668171559
            name: Max F1
          - type: max_f1_threshold
            value: 362.5467529296875
            name: Max F1 Threshold
          - type: max_precision
            value: 0.5333333333333333
            name: Max Precision
          - type: max_recall
            value: 0.8901734104046243
            name: Max Recall
          - type: max_ap
            value: 0.6359674505468691
            name: Max Ap
      - task:
          type: binary-classification
          name: Binary Classification
        dataset:
          name: Qnli dev
          type: Qnli-dev
        metrics:
          - type: cosine_accuracy
            value: 0.693359375
            name: Cosine Accuracy
          - type: cosine_accuracy_threshold
            value: 0.6543508768081665
            name: Cosine Accuracy Threshold
          - type: cosine_f1
            value: 0.6733668341708542
            name: Cosine F1
          - type: cosine_f1_threshold
            value: 0.5232125520706177
            name: Cosine F1 Threshold
          - type: cosine_precision
            value: 0.556786703601108
            name: Cosine Precision
          - type: cosine_recall
            value: 0.8516949152542372
            name: Cosine Recall
          - type: cosine_ap
            value: 0.7124557023921706
            name: Cosine Ap
          - type: dot_accuracy
            value: 0.666015625
            name: Dot Accuracy
          - type: dot_accuracy_threshold
            value: 256.9267272949219
            name: Dot Accuracy Threshold
          - type: dot_f1
            value: 0.6719242902208202
            name: Dot F1
          - type: dot_f1_threshold
            value: 186.3544921875
            name: Dot F1 Threshold
          - type: dot_precision
            value: 0.535175879396985
            name: Dot Precision
          - type: dot_recall
            value: 0.902542372881356
            name: Dot Recall
          - type: dot_ap
            value: 0.6584233742794683
            name: Dot Ap
          - type: manhattan_accuracy
            value: 0.6953125
            name: Manhattan Accuracy
          - type: manhattan_accuracy_threshold
            value: 357.7731018066406
            name: Manhattan Accuracy Threshold
          - type: manhattan_f1
            value: 0.6757281553398058
            name: Manhattan F1
          - type: manhattan_f1_threshold
            value: 370.83966064453125
            name: Manhattan F1 Threshold
          - type: manhattan_precision
            value: 0.6236559139784946
            name: Manhattan Precision
          - type: manhattan_recall
            value: 0.7372881355932204
            name: Manhattan Recall
          - type: manhattan_ap
            value: 0.7226095638737486
            name: Manhattan Ap
          - type: euclidean_accuracy
            value: 0.6953125
            name: Euclidean Accuracy
          - type: euclidean_accuracy_threshold
            value: 16.447265625
            name: Euclidean Accuracy Threshold
          - type: euclidean_f1
            value: 0.6785714285714286
            name: Euclidean F1
          - type: euclidean_f1_threshold
            value: 17.252986907958984
            name: Euclidean F1 Threshold
          - type: euclidean_precision
            value: 0.6380597014925373
            name: Euclidean Precision
          - type: euclidean_recall
            value: 0.7245762711864406
            name: Euclidean Recall
          - type: euclidean_ap
            value: 0.7183967614546444
            name: Euclidean Ap
          - type: max_accuracy
            value: 0.6953125
            name: Max Accuracy
          - type: max_accuracy_threshold
            value: 357.7731018066406
            name: Max Accuracy Threshold
          - type: max_f1
            value: 0.6785714285714286
            name: Max F1
          - type: max_f1_threshold
            value: 370.83966064453125
            name: Max F1 Threshold
          - type: max_precision
            value: 0.6380597014925373
            name: Max Precision
          - type: max_recall
            value: 0.902542372881356
            name: Max Recall
          - type: max_ap
            value: 0.7226095638737486
            name: Max Ap

SentenceTransformer based on bobox/DeBERTa-small-ST-v1-test-step3

This is a sentence-transformers model finetuned from bobox/DeBERTa-small-ST-v1-test-step3 on the negation-triplets, vitaminc-pairs, scitail-pairs-qa, scitail-pairs-pos, xsum-pairs, sciq_pairs, qasc_pairs, openbookqa_pairs, msmarco_pairs, nq_pairs, trivia_pairs, gooaq_pairs, paws-pos and global_dataset datasets. It maps sentences & paragraphs to a 768-dimensional dense vector space and can be used for semantic textual similarity, semantic search, paraphrase mining, text classification, clustering, and more.

Model Details

Model Description

• Model Type: Sentence Transformer
• Base model: bobox/DeBERTa-small-ST-v1-test-step3
• Maximum Sequence Length: 512 tokens
• Output Dimensionality: 768 tokens
• Similarity Function: Cosine Similarity
• Training Datasets:
  • negation-triplets
  • vitaminc-pairs
  • scitail-pairs-qa
  • scitail-pairs-pos
  • xsum-pairs
  • sciq_pairs
  • qasc_pairs
  • openbookqa_pairs
  • msmarco_pairs
  • nq_pairs
  • trivia_pairs
  • gooaq_pairs
  • paws-pos
  • global_dataset
• Language: en

Model Sources

• Documentation: Sentence Transformers Documentation
• Repository: Sentence Transformers on GitHub
• Hugging Face: Sentence Transformers on Hugging Face

Full Model Architecture

SentenceTransformer(
  (0): Transformer({'max_seq_length': 512, 'do_lower_case': False}) with Transformer model: DebertaV2Model 
  (1): Pooling({'word_embedding_dimension': 768, '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})
)

Usage

Direct Usage (Sentence Transformers)

First install the Sentence Transformers library:

pip install -U sentence-transformers

Then you can load this model and run inference.

from sentence_transformers import SentenceTransformer

# Download from the 🤗 Hub
model = SentenceTransformer("bobox/DeBERTa2-0.9B-ST-v2-checkpoints-tmp")
# Run inference
sentences = [
    "2012 was the 300th anniversary of the world's first industrial steam-powered machine, which was a?",
    "The Worlds First Steam Engine 300th Anniversary - YouTube The Worlds First Steam Engine 300th Anniversary Want to watch this again later? Sign in to add this video to a playlist. Need to report the video? Sign in to report inappropriate content. Rating is available when the video has been rented. This feature is not available right now. Please try again later. Published on Nov 14, 2012 The replica Newcomen Pumping Engine at the Black Country Living Museum in Dudley has been brought back to life for the 300th anniversary of the first recorded practical application of steam power. The self-acting valve gear is a clanking cacophony of joy! The restoration team were helped by Guy Martin and the engine was featured in the Channel 4 documentary series 'How Britain Worked' (as was my own miniature Newcomen engine model, used by Guy to demonstrate how the engine works!) See my other vids for some footage of my little 'un in action! Category",
    'YouTube | Logopedia | Fandom powered by Wikia 2015–present 2005–2011 The logo consists of the black word "You" and a red rounded rectangle with the word "Tube" in it next to it. This logo is still being used on some other pages. Logo with the slogan "Broadcast Yourself". Notice that the red square looks different in this variation. Add a photo to this gallery 2011–2013 This modification of the YouTube logo was introduced in July 2011 as a part of the Cosmic Panda experiment. It officially became the new logo a few months later. It has the red square in a darker color this time. Also, starting in 2012, the slogan "Broadcast Yourself" was retired. 2013–2015 On December 19, 2013, the red rectangle was made lighter in color. Also, the word "You" was made more black and the shadow behind the word "Tube" was removed. This is still used as a secondary logo. Alternate Version, only for social media.',
]
embeddings = model.encode(sentences)
print(embeddings.shape)
# [3, 768]

# Get the similarity scores for the embeddings
similarities = model.similarity(embeddings, embeddings)
print(similarities.shape)
# [3, 3]

Evaluation

Metrics

Semantic Similarity

• Dataset: sts-test
• Evaluated with EmbeddingSimilarityEvaluator

Metric	Value
pearson_cosine	0.8848
spearman_cosine	0.907
pearson_manhattan	0.9057
spearman_manhattan	0.9018
pearson_euclidean	0.9048
spearman_euclidean	0.9006
pearson_dot	0.8687
spearman_dot	0.8651
pearson_max	0.9057
spearman_max	0.907

Binary Classification

• Dataset: allNLI-dev
• Evaluated with BinaryClassificationEvaluator

Metric	Value
cosine_accuracy	0.7363
cosine_accuracy_threshold	0.7933
cosine_f1	0.6429
cosine_f1_threshold	0.6018
cosine_precision	0.5236
cosine_recall	0.8324
cosine_ap	0.6339
dot_accuracy	0.7305
dot_accuracy_threshold	290.1109
dot_f1	0.6099
dot_f1_threshold	200.0629
dot_precision	0.4639
dot_recall	0.8902
dot_ap	0.5978
manhattan_accuracy	0.7422
manhattan_accuracy_threshold	273.5278
manhattan_f1	0.6501
manhattan_f1_threshold	362.5468
manhattan_precision	0.5333
manhattan_recall	0.8324
manhattan_ap	0.6336
euclidean_accuracy	0.7441
euclidean_accuracy_threshold	12.9763
euclidean_f1	0.6419
euclidean_f1_threshold	17.4409
euclidean_precision	0.5158
euclidean_recall	0.8497
euclidean_ap	0.636
max_accuracy	0.7441
max_accuracy_threshold	290.1109
max_f1	0.6501
max_f1_threshold	362.5468
max_precision	0.5333
max_recall	0.8902
max_ap	0.636

Binary Classification

• Dataset: Qnli-dev
• Evaluated with BinaryClassificationEvaluator

Metric	Value
cosine_accuracy	0.6934
cosine_accuracy_threshold	0.6544
cosine_f1	0.6734
cosine_f1_threshold	0.5232
cosine_precision	0.5568
cosine_recall	0.8517
cosine_ap	0.7125
dot_accuracy	0.666
dot_accuracy_threshold	256.9267
dot_f1	0.6719
dot_f1_threshold	186.3545
dot_precision	0.5352
dot_recall	0.9025
dot_ap	0.6584
manhattan_accuracy	0.6953
manhattan_accuracy_threshold	357.7731
manhattan_f1	0.6757
manhattan_f1_threshold	370.8397
manhattan_precision	0.6237
manhattan_recall	0.7373
manhattan_ap	0.7226
euclidean_accuracy	0.6953
euclidean_accuracy_threshold	16.4473
euclidean_f1	0.6786
euclidean_f1_threshold	17.253
euclidean_precision	0.6381
euclidean_recall	0.7246
euclidean_ap	0.7184
max_accuracy	0.6953
max_accuracy_threshold	357.7731
max_f1	0.6786
max_f1_threshold	370.8397
max_precision	0.6381
max_recall	0.9025
max_ap	0.7226

Training Details

Training Datasets

negation-triplets

• Dataset: negation-triplets
• Size: 7,500 training samples
• Columns: anchor, entailment, and negative
• Approximate statistics based on the first 1000 samples:

  	anchor	entailment	negative
  type	string	string	string
  details	min: 4 tokens mean: 21.77 tokens max: 89 tokens	min: 4 tokens mean: 13.89 tokens max: 43 tokens	min: 4 tokens mean: 14.24 tokens max: 43 tokens

• Samples:

  anchor	entailment	negative
  Exploring inland to the north is difficult on foot, The streets become very steep.	It is difficult to explore inland on foot.	It is easy to explore inland by foot.
  Michele is the lead singer in 14 of the top 20 selling Glee songs as of 2010 .	Michele is featured lead singer in 14 of the top 20 selling Glee songs as of 2010 .	Michele is featured lead singer in none of the top 20 selling Glee songs as of 2010.
  Its theme is show business , based on Hollywood in the 1930s and 1940s .	Spanning 135 acres ( 546,000 m ) in size , its theme is show business , drawing inspiration from the heyday of Hollywood in the 1930s and 1940s .	Covering 25 acres ( 101,000 m ) in size , its theme is agriculture , drawing inspiration from the modern era of Silicon Valley in the 2000s and 2010s .

• Loss: CachedGISTEmbedLoss with these parameters:

  {'guide': SentenceTransformer(
    (0): Transformer({'max_seq_length': 512, 'do_lower_case': False}) with Transformer model: BertModel 
    (1): Pooling({'word_embedding_dimension': 768, 'pooling_mode_cls_token': True, 'pooling_mode_mean_tokens': False, '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): Normalize()
  ), 'temperature': 0.025}
  

vitaminc-pairs

• Dataset: vitaminc-pairs at be6febb
• Size: 7,500 training samples
• Columns: claim and evidence
• Approximate statistics based on the first 1000 samples:

  	claim	evidence
  type	string	string
  details	min: 7 tokens mean: 17.12 tokens max: 56 tokens	min: 8 tokens mean: 38.83 tokens max: 187 tokens

• Samples:

  claim	evidence
  Prokhorov offered Joseph Tsai , the executive vice-chairman of the Alibaba Group , a stake in Brooklyn Nets .	In late 2017 , there were multiple reports of an agreement for Prokhorov to sell a 49 % stake in the team to Joseph Tsai , the executive vice chairman of the Alibaba Group , with an option for Tsai to become the majority owner .
  Vartan 's father was of Armenian descent .	Vartan 's father was Bulgarian-born and of part Armenian and Hungarian descent and Vartan 's mother is an American Jew originally from Poland ; Vartan has said about his French background that “ The funny thing is I ’ m actually a Polish Jew who happens to be born in France .
  Big Sean was born on 25 March 1987 .	Sean Michael Leonard Anderson ( born March 25 , 1987 ) , known professionally as Big Sean , is an American hip hop recording artist , comedian , singer-songwriter .

• Loss: CachedGISTEmbedLoss with these parameters:

  {'guide': SentenceTransformer(
    (0): Transformer({'max_seq_length': 512, 'do_lower_case': False}) with Transformer model: BertModel 
    (1): Pooling({'word_embedding_dimension': 768, 'pooling_mode_cls_token': True, 'pooling_mode_mean_tokens': False, '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): Normalize()
  ), 'temperature': 0.025}
  

scitail-pairs-qa

• Dataset: scitail-pairs-qa at 0cc4353
• Size: 7,118 training samples
• Columns: sentence1 and sentence2
• Approximate statistics based on the first 1000 samples:

  	sentence1	sentence2
  type	string	string
  details	min: 7 tokens mean: 15.71 tokens max: 41 tokens	min: 6 tokens mean: 14.76 tokens max: 41 tokens

• Samples:

  sentence1	sentence2
  The hair cells sense(s) the movement of liquid in ear canals.	What senses the movement of liquid in ear canals?
  Evidence allows theories to be widely accepted.	What allows theories to be widely accepted?
  The field of study known as mathematics is called the language of science.	What field of study is called the language of science?

• Loss: CachedGISTEmbedLoss with these parameters:

  {'guide': SentenceTransformer(
    (0): Transformer({'max_seq_length': 512, 'do_lower_case': False}) with Transformer model: BertModel 
    (1): Pooling({'word_embedding_dimension': 768, 'pooling_mode_cls_token': True, 'pooling_mode_mean_tokens': False, '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): Normalize()
  ), 'temperature': 0.025}
  

scitail-pairs-pos

• Dataset: scitail-pairs-pos at 0cc4353
• Size: 4,300 training samples
• Columns: sentence1 and sentence2
• Approximate statistics based on the first 1000 samples:

  	sentence1	sentence2
  type	string	string
  details	min: 8 tokens mean: 23.62 tokens max: 67 tokens	min: 7 tokens mean: 15.55 tokens max: 38 tokens

• Samples:

  sentence1	sentence2
  Plants convert carbon dioxide into oxygen as a byproduct of photosynthesis.	Oxygen is made by trees and other plants during photosynthesis.
  Kinetic energy is the energy of bodies in motion.	Kinetic engergy is the energy of anything in motion.
  Digestion begins in the mouth where salivary amylase starts the breakdown of carbohydrates.	Carbohydrate digestion begins in the mouth.

• Loss: CachedGISTEmbedLoss with these parameters:

  {'guide': SentenceTransformer(
    (0): Transformer({'max_seq_length': 512, 'do_lower_case': False}) with Transformer model: BertModel 
    (1): Pooling({'word_embedding_dimension': 768, 'pooling_mode_cls_token': True, 'pooling_mode_mean_tokens': False, '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): Normalize()
  ), 'temperature': 0.025}
  

xsum-pairs

• Dataset: xsum-pairs
• Size: 7,500 training samples
• Columns: summary and document
• Approximate statistics based on the first 1000 samples:

  	summary	document
  type	string	string
  details	min: 8 tokens mean: 25.65 tokens max: 44 tokens	min: 49 tokens mean: 221.16 tokens max: 421 tokens

• Samples:

  summary	document
  A council has received complaints about people urinating and defecating outdoors after a village's public toilets were closed down.	The toilets at Staffin on Skye were among a number of Highland Council-run sites shut this year to save money. The council has been in talks with Staffin Community Trust about alternative facilities. In the meantime, two complaints have been received public urination and defecation in and around the village. The nearest public toilet is in Kilmuir about 11 miles (17km) from Staffin. Residents have told BBC Alba of people using ditches, a nearby quarry and the rear of the closed toilet block. Highland Council it had cost £6,700 a year to run Staffin's loos. A spokeswoman said: "We cannot confirm that people have been urinating and defecating in the area - we have found no evidence to support these claims. "There have been two complaints made, one to environmental health and one to roads and community works. "In addition to this we have had three enquiries about Staffin toilets." She added: "The council has been in discussions with Staffin Community Trust and have agreed in principle to operate a seasonal - April to October - Highland Comfort Scheme within the community hall."
  More than 70 workers have been taken off a North Sea oil platform after it suffered a loss of power.	The coastguard was first alerted to the issue on the Bruce installation, east of Shetland, just after 20:00 on Thursday. A total of 76 people from the 121 on board were taken off the platform and flown to neighbouring installations in coastguard helicopters. Forty-five workers stayed on the platform in a bid to restore power. BP said work to restore power was ongoing. A Coastguard spokesman said: "Following a loss of power on the Bruce platform off Aberdeen late on Thursday 22nd June, HM Coastguard rescue helicopters from Shetland and Inverness were tasked to partially down-man the platform. "Seventy six non-essential personnel were transferred overnight by helicopter to other platforms in the area."
  Shaun Hutchinson has agreed a two-year deal with League One club Millwall after his recent release by Fulham.	The defender, 25, who spent two years at Craven Cottage, will officially join the Lions when his Fulham contract expires at the end of the month. "Coming to work at a place where you are so wanted is a great feeling," he told the Lions' website. "I can't wait to get going and am looking forward to putting on the Millwall shirt." Newcastle-born Hutchinson made 121 league appearances for Motherwell before Fulham signed him from the Scottish side after a number of clubs - including Millwall - had shown an interest in him. However, he struggled to establish himself as a first-team regular, making 34 league appearances during his time in west London. "Shaun is a player that as a club we have been aware of for quite a period of time," said Millwall manager Neil Harris. "We tracked him diligently when he was at Motherwell and when he was moving down south we bid for him. "We monitored him last season knowing that he was coming out of contract. "He is a no-nonsense defender who likes to head it in both boxes and is very aggressive in his approach." Find all the latest football transfers on our dedicated page.

• Loss: CachedGISTEmbedLoss with these parameters:

  {'guide': SentenceTransformer(
    (0): Transformer({'max_seq_length': 512, 'do_lower_case': False}) with Transformer model: BertModel 
    (1): Pooling({'word_embedding_dimension': 768, 'pooling_mode_cls_token': True, 'pooling_mode_mean_tokens': False, '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): Normalize()
  ), 'temperature': 0.025}
  

sciq_pairs

• Dataset: sciq_pairs at 2c94ad3
• Size: 5,547 training samples
• Columns: sentence1 and sentence2
• Approximate statistics based on the first 1000 samples:

  	sentence1	sentence2
  type	string	string
  details	min: 7 tokens mean: 17.08 tokens max: 54 tokens	min: 2 tokens mean: 90.28 tokens max: 512 tokens

• Samples:

  sentence1	sentence2
  A rounded hollow carved in the side of a mountain by a glacier is known as?	A cirque is a rounded hollow carved in the side of a mountain by a glacier. The highest cliff of a cirque is called the headwall.
  What is a group of lions called?	Lions live in social groups called prides . Adult females in the pride hunt cooperatively, which is more efficient than hunting alone. Then they share the food with the rest of the pride. For their part, adult males defend the pride’s territory from other predators.
  What is a measure of the average amount of energy of motion, or kinetic energy, a system contains called?	There are other units in chemistry that are important, and we will cover others in the course of the entire book. One of the fundamental quantities in science is temperature. Temperature is a measure of the average amount of energy of motion, or kinetic energy, a system contains. Temperatures are expressed using scales that use units called degrees, and there are several temperature scales in use. In the United States, the commonly used temperature scale is the Fahrenheit scale (symbolized by °F and spoken as “degrees Fahrenheit”). On this scale, the freezing point of liquid water (the temperature at which liquid water turns to solid ice) is 32°F, and the boiling point of water (the temperature at which liquid water turns to steam) is 212°F. Science also uses other scales to express temperature. The Celsius scale (symbolized by °C and spoken as “degrees Celsius”) is a temperature scale where 0°C is the freezing point of water and 100°C is the boiling point of water; the scale is divided into 100 divisions between these two landmarks and extended higher and lower. By comparing the Fahrenheit and Celsius scales, a conversion between the two scales can be determined: °C=(°F–32) × 59° Saylor URL: http://www. saylor. org/books.

• Loss: CachedGISTEmbedLoss with these parameters:

  {'guide': SentenceTransformer(
    (0): Transformer({'max_seq_length': 512, 'do_lower_case': False}) with Transformer model: BertModel 
    (1): Pooling({'word_embedding_dimension': 768, 'pooling_mode_cls_token': True, 'pooling_mode_mean_tokens': False, '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): Normalize()
  ), 'temperature': 0.025}
  

qasc_pairs

• Dataset: qasc_pairs at a34ba20
• Size: 3,863 training samples
• Columns: sentence1 and sentence2
• Approximate statistics based on the first 1000 samples:

  	sentence1	sentence2
  type	string	string
  details	min: 4 tokens mean: 11.39 tokens max: 25 tokens	min: 15 tokens mean: 33.65 tokens max: 74 tokens

• Samples:

  sentence1	sentence2
  Meat, fish and chicken are used to do what by the human body?	protein is used to repair cells by the human body. Meat, fish and chicken are rich sources of protein. Meat, fish and chicken are used to repair cells by the human body.
  A mutation in the sex cells of a parent can cause a new trait to appear in the parent 's what?	a mutation in the sex cells of a parent can cause a new trait to appear in the parent 's offspring. Child' is ambiguous between 'offspring' and 'immature offspring'. A mutation in the sex cells of a parent can cause a new trait to appear in the parent's child.
  Weathering means breaking down what from larger whole into smaller pieces by weather?	weathering means breaking down rocks from larger whole into smaller pieces by weather. Igneous rocks are made from lava that hardens. weathering means breaking down something made from lava from larger whole into smaller pieces by weather

• Loss: CachedGISTEmbedLoss with these parameters:

  {'guide': SentenceTransformer(
    (0): Transformer({'max_seq_length': 512, 'do_lower_case': False}) with Transformer model: BertModel 
    (1): Pooling({'word_embedding_dimension': 768, 'pooling_mode_cls_token': True, 'pooling_mode_mean_tokens': False, '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): Normalize()
  ), 'temperature': 0.025}
  

openbookqa_pairs

• Dataset: openbookqa_pairs
• Size: 2,261 training samples
• Columns: question and fact
• Approximate statistics based on the first 1000 samples:

  	question	fact
  type	string	string
  details	min: 3 tokens mean: 13.94 tokens max: 78 tokens	min: 5 tokens mean: 11.67 tokens max: 31 tokens

• Samples:

  question	fact
  Skills are learned characteristics. To get better at doing something, you must stretch yourself in ways that	skills are learned characteristics
  the lunar surface contains	the moon 's surface contains flat areas
  natural disasters can cause animals to	natural disasters can cause animals to leave an environment

• Loss: CachedGISTEmbedLoss with these parameters:

  {'guide': SentenceTransformer(
    (0): Transformer({'max_seq_length': 512, 'do_lower_case': False}) with Transformer model: BertModel 
    (1): Pooling({'word_embedding_dimension': 768, 'pooling_mode_cls_token': True, 'pooling_mode_mean_tokens': False, '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): Normalize()
  ), 'temperature': 0.025}
  

msmarco_pairs

• Dataset: msmarco_pairs at 28ff31e
• Size: 7,500 training samples
• Columns: sentence1 and sentence2
• Approximate statistics based on the first 1000 samples:

  	sentence1	sentence2
  type	string	string
  details	min: 4 tokens mean: 8.62 tokens max: 31 tokens	min: 16 tokens mean: 76.31 tokens max: 195 tokens

• Samples:

  sentence1	sentence2
  what county is youngsville nc in	My Home Town. Youngsville is a growing town in southwestern Franklin County, rich in history and full of promise. Its location in the prosperous Triangle region of North Carolina has meant that Youngsville has seen its share of growth, but without losing the rural charm that has made it a desired destination for families and businesses alike.
  what college did allan houston play for	Wade Houston. Wade Houston (born October 9, 1944) is an American former college basketball player and coach. He was an assistant coach under Denny Crum at the University of Louisville for 13 years until 1989 when he was named the head coach of the University of Tennessee.
  what does the rock	Webster Dictionary(5.00 / 1 vote)Rate this definition: 1 Rock(noun) see Roc. 2 Rock(noun) a distaff used in spinning; the staff or frame about which flax is arranged, and from which the thread is drawn in spinning. 3 Rock(noun) a large concreted mass of stony material; a large fixed stone or crag.

• Loss: CachedGISTEmbedLoss with these parameters:

  {'guide': SentenceTransformer(
    (0): Transformer({'max_seq_length': 512, 'do_lower_case': False}) with Transformer model: BertModel 
    (1): Pooling({'word_embedding_dimension': 768, 'pooling_mode_cls_token': True, 'pooling_mode_mean_tokens': False, '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): Normalize()
  ), 'temperature': 0.025}
  

nq_pairs

• Dataset: nq_pairs at f9e894e
• Size: 7,500 training samples
• Columns: sentence1 and sentence2
• Approximate statistics based on the first 1000 samples:

  	sentence1	sentence2
  type	string	string
  details	min: 9 tokens mean: 11.66 tokens max: 25 tokens	min: 17 tokens mean: 131.66 tokens max: 512 tokens

• Samples:

  sentence1	sentence2
  where did they film last man on earth	The Last Man on Earth (TV series) The main recording location for the series is a 20th Century Fox studio in Chatsworth, California.[35][36][37]
  when was the first soviet atomic bomb tested	Soviet atomic bomb project On 29 August 1949, the Soviet Union secretly conducted its first successful weapon test (First Lightning), based on the U.S. design at the Semipalatinsk in Kazakhstan.[2]
  which two chambers make the united kingdom a bicameral type of government	Parliament of the United Kingdom The parliament is bicameral, consisting of an upper house (the House of Lords) and a lower house (the House of Commons).[4] The Sovereign forms the third component of the legislature (the Queen-in-Parliament).[5][6] The House of Lords includes two different types of members: the Lords Spiritual, consisting of the most senior bishops of the Church of England, and the Lords Temporal, consisting mainly of life peers, appointed by the Sovereign on the advice of the Prime Minister,[7] and of 92 hereditary peers, sitting either by virtue of holding a royal office, or by being elected by their fellow hereditary peers. Prior to the opening of the Supreme Court in October 2009, the House of Lords also performed a judicial role through the Law Lords.

• Loss: CachedGISTEmbedLoss with these parameters:

  {'guide': SentenceTransformer(
    (0): Transformer({'max_seq_length': 512, 'do_lower_case': False}) with Transformer model: BertModel 
    (1): Pooling({'word_embedding_dimension': 768, 'pooling_mode_cls_token': True, 'pooling_mode_mean_tokens': False, '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): Normalize()
  ), 'temperature': 0.025}
  

trivia_pairs

• Dataset: trivia_pairs at a7c36e3
• Size: 6,012 training samples
• Columns: query and answer
• Approximate statistics based on the first 1000 samples:

  	query	answer
  type	string	string
  details	min: 7 tokens mean: 17.19 tokens max: 84 tokens	min: 17 tokens mean: 203.25 tokens max: 499 tokens

• Samples:

  query	answer
  Which resort was advertised in a travel poster as '............ is so bracing'?	Kath - Postcards From the Past - Other Seaside Places                                                                          Kath - Postcards from the Past -                                        Mablethorpe, Skegness, Sutton-on-Sea and Ingoldmells              Here are my postcards of Mablethorpe, Skegness,Sutton-on-Sea and Ingoldmells which together with Cleethorpes (see Cleethorpes separate pages) make up the major seaside resorts of Lincolnshire. These are typical British seaside resorts with all the usual expected amenities. Ingoldmells is a small coastal village and holiday resort three miles to the north of Skegness and is the home to the original Butlin's Holiday Camp built in 1936 which s still very popular today. Mablethorpe is located between Cleethorpes and Skegness and boasts all the usual attractions. It is home to the popular Golden Sands Holiday Park and there is a seal sanctuary situated at North End which has rescued hundreds of seals over the years. Skegness is popular with holiday makers and day-trippers from the Midlands, it was once a small port but from 1877 it was developed into a holiday resort. In 1908 it acquired it's famous "Jolly Fisherman" mascot taken from a Railway travel poster entitled "Skegness is so Bracing". Sutton-on-Sea, a few miles south of Mablethorpe, is one of the smaller seaside resorts being described as a "charming and tranquil village", it has golden sands, the usual attractions and golf can be played at nearby Sandilands. Click on the link for a larger image.
  Amman is the capital city of which country?	Jordan Facts on Largest Cities, Populations, Symbols - Worldatlas.com Ethnicity: Arab 98%, Circassian 1%, Armenian 1% GDP total: $38.67 billion (2012 est.) GDP per capita: $6,000 (2012 est.) Language: Arabic (official), English widely understood among upper and middle classes Largest Cities: (by population) Amman, Irbid Name: Aramaic Yarden in origin, meaning "down-flowing," or "one who descends," and is named after the River Jordan National Day: May 25 Religion: Sunni Muslim 92%, Christian 6% (majority Greek Orthodox, but some Greek and Roman Catholics, Syrian Orthodox, Coptic Orthodox, Armenian Orthodox, and Protestant denominations), other 2% (several small Shia Muslim and Druze populations)
  A famous sports car named after Argentina's hot dusty North wind, is the?	Tamerlane's Thoughts: Cars named after winds Cars named after winds Bora (ancient Greek name for north wind) Ghibli (Libyan wind) Khamsin (hot dusty wind in North Africa and Arabian Peninsula) Mistral (wind from the north that blows over the northwest coast of the Mediterranean) Shamal (summer wind over Iraq and Persian Gulf) Pagani: Zonda (Andean wind in Argentina) VW Passat ("trade wind" in German) Santana (Santa Ana wind (subject to debate)) Scirocco (warm wind in Mediterranean) Yugo (it is not derived from the word Yugoslavia; rather, yugo is a southeasterly wind on the Adriatic) 10 comments: Anonymous said... What about the Ford Zephyr and the Zonda and the Austin Maestro or if you want a helicopter try the Chinook for size, wind would seem popular for names kashgar216 said... Thanks for playing! I'll add the Zephyr. The Diablo is a stretch. There is a wind called that but were the Lambo people thinking-- let's name our next supercar after an obscure Northern California wind? I'm also debating the Maestro thing. Check back soon for an update!

• Loss: CachedGISTEmbedLoss with these parameters:

  {'guide': SentenceTransformer(
    (0): Transformer({'max_seq_length': 512, 'do_lower_case': False}) with Transformer model: BertModel 
    (1): Pooling({'word_embedding_dimension': 768, 'pooling_mode_cls_token': True, 'pooling_mode_mean_tokens': False, '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): Normalize()
  ), 'temperature': 0.025}
  

gooaq_pairs

• Dataset: gooaq_pairs at b089f72
• Size: 7,500 training samples
• Columns: sentence1 and sentence2
• Approximate statistics based on the first 1000 samples:

  	sentence1	sentence2
  type	string	string
  details	min: 8 tokens mean: 11.58 tokens max: 20 tokens	min: 12 tokens mean: 57.14 tokens max: 132 tokens

• Samples:

  sentence1	sentence2
  are starkist tuna cans bpa free?	The StarKist Chunk Light canned tuna we tested averaged 3 ppb of BPA, but BPA levels in the same brand in a plastic pouch weren't measurable. ... We tested two products that their manufacturers claimed were packaged in BPA-free cans and found the chemical in both of the foods.
  is a 2gb graphics card enough?	Generally speaking, for 1080p gaming, 2GB of video memory is an adequate minimum, but 4GB is much better. In cards under $300 nowadays, you'll see graphics memory ranging from 1GB up to 8GB.
  how much does it cost to jump your phone?	To get Jump, sign up to pay the full price of your new phone in 24 equal monthly installments. Then sign up for the Jump program, which costs an additional $9 to $12 per month, depending on your phone.

• Loss: CachedGISTEmbedLoss with these parameters:

  {'guide': SentenceTransformer(
    (0): Transformer({'max_seq_length': 512, 'do_lower_case': False}) with Transformer model: BertModel 
    (1): Pooling({'word_embedding_dimension': 768, 'pooling_mode_cls_token': True, 'pooling_mode_mean_tokens': False, '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): Normalize()
  ), 'temperature': 0.025}
  

paws-pos

• Dataset: paws-pos at 161ece9
• Size: 7,500 training samples
• Columns: sentence1 and sentence2
• Approximate statistics based on the first 1000 samples:

  	sentence1	sentence2
  type	string	string
  details	min: 9 tokens mean: 25.78 tokens max: 56 tokens	min: 9 tokens mean: 25.7 tokens max: 57 tokens

• Samples:

  sentence1	sentence2
  Quintus Caecilius Metellus Macedonicus was the second son of Roman politician and general Lucius Caecilius Metellus Diadematus .	Quintus Caecilius Metellus Macedonicus was the second son of the Roman politician , General Lucius Caecilius Metellus Diadematus .
  She moved to Switzerland when she was a few months old , then to France , but mostly grew up in Paris .	She moved to Switzerland when she was a few months old , then grew up to France , but largely in Paris .
  The NBA season from 1979 to 80 was the 34th season of the National Basketball Association .	The 1979 -- 80 National Basketball Association season was the 34th season of the NBA .

• Loss: CachedGISTEmbedLoss with these parameters:

  {'guide': SentenceTransformer(
    (0): Transformer({'max_seq_length': 512, 'do_lower_case': False}) with Transformer model: BertModel 
    (1): Pooling({'word_embedding_dimension': 768, 'pooling_mode_cls_token': True, 'pooling_mode_mean_tokens': False, '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): Normalize()
  ), 'temperature': 0.025}
  

global_dataset

• Dataset: global_dataset
• Size: 81,604 training samples
• Columns: sentence1 and sentence2
• Approximate statistics based on the first 1000 samples:

  	sentence1	sentence2
  type	string	string
  details	min: 4 tokens mean: 35.5 tokens max: 335 tokens	min: 2 tokens mean: 59.34 tokens max: 415 tokens

• Samples:

  sentence1	sentence2
  Great egrets get food by doing what?	Heterotrophs get food by eating other living things.. Great egrets are heterotrophs . Great egrets get food by eating other livings things.
  is oil price going to increase	The current downward swing in oil prices has raised a similar specter of low oil prices for a prolonged period. So far, this year oil prices have dropped by more than 57% from last year's prices. However, the decline may be temporary. According to analysts, oil prices will rise back up again in 2017.
  Vascular plants have a dominant sporophyte generation.	Do nonvascular or vascular plants have a dominant sporophyte generation?

• Loss: CachedGISTEmbedLoss with these parameters:

  {'guide': SentenceTransformer(
    (0): Transformer({'max_seq_length': 512, 'do_lower_case': False}) with Transformer model: BertModel 
    (1): Pooling({'word_embedding_dimension': 768, 'pooling_mode_cls_token': True, 'pooling_mode_mean_tokens': False, '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): Normalize()
  ), 'temperature': 0.025}
  

Evaluation Datasets

vitaminc-pairs

• Dataset: vitaminc-pairs at be6febb
• Size: 128 evaluation samples
• Columns: claim and evidence
• Approximate statistics based on the first 1000 samples:

  	claim	evidence
  type	string	string
  details	min: 9 tokens mean: 21.42 tokens max: 41 tokens	min: 11 tokens mean: 35.55 tokens max: 79 tokens

• Samples:

  claim	evidence
  Dragon Con had over 5000 guests .	Among the more than 6000 guests and musical performers at the 2009 convention were such notables as Patrick Stewart , William Shatner , Leonard Nimoy , Terry Gilliam , Bruce Boxleitner , James Marsters , and Mary McDonnell .
  COVID-19 has reached more than 185 countries .	As of , more than cases of COVID-19 have been reported in more than 190 countries and 200 territories , resulting in more than deaths .
  In March , Italy had 3.6x times more cases of coronavirus than China .	As of 12 March , among nations with at least one million citizens , Italy has the world 's highest per capita rate of positive coronavirus cases at 206.1 cases per million people ( 3.6x times the rate of China ) and is the country with the second-highest number of positive cases as well as of deaths in the world , after China .

• Loss: CachedGISTEmbedLoss with these parameters:

  {'guide': SentenceTransformer(
    (0): Transformer({'max_seq_length': 512, 'do_lower_case': False}) with Transformer model: BertModel 
    (1): Pooling({'word_embedding_dimension': 768, 'pooling_mode_cls_token': True, 'pooling_mode_mean_tokens': False, '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): Normalize()
  ), 'temperature': 0.025}
  

negation-triplets

• Dataset: negation-triplets
• Size: 128 evaluation samples
• Columns: anchor, entailment, and negative
• Approximate statistics based on the first 1000 samples:

  	anchor	entailment	negative
  type	string	string	string
  details	min: 8 tokens mean: 13.98 tokens max: 31 tokens	min: 6 tokens mean: 12.47 tokens max: 21 tokens	min: 7 tokens mean: 12.7 tokens max: 22 tokens

• Samples:

  anchor	entailment	negative
  A bathroom with a toilet, sink, and shower.	A full bathroom with a wicker laundry basket.	An empty bathroom with no laundry basket.
  A person in a helmet standing by their motorcycle.	A motorcyclist stands next to a motorcycle at a lookout over a beach.	A motorcyclist stands nowhere near a motorcycle at a lookout over a beach.
  A bathroom with a poster of an ugly face above the toilette.	A bathroom with a white toilet and sink.	A bathroom with a black toilet and sink.

• Loss: CachedGISTEmbedLoss with these parameters:

  {'guide': SentenceTransformer(
    (0): Transformer({'max_seq_length': 512, 'do_lower_case': False}) with Transformer model: BertModel 
    (1): Pooling({'word_embedding_dimension': 768, 'pooling_mode_cls_token': True, 'pooling_mode_mean_tokens': False, '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): Normalize()
  ), 'temperature': 0.025}
  

scitail-pairs-pos

• Dataset: scitail-pairs-pos at 0cc4353
• Size: 128 evaluation samples
• Columns: sentence1 and sentence2
• Approximate statistics based on the first 1000 samples:

  	sentence1	sentence2
  type	string	string
  details	min: 9 tokens mean: 20.28 tokens max: 56 tokens	min: 8 tokens mean: 15.48 tokens max: 23 tokens

• Samples:

  sentence1	sentence2
  humans normally have 23 pairs of chromosomes.	Humans typically have 23 pairs pairs of chromosomes.
  A solution is a homogenous mixture of two or more substances that exist in a single phase.	Solution is the term for a homogeneous mixture of two or more substances.
  Upwelling The physical process in near-shore ocean systems of rising of nutrients and colder bottom waters to the surface because of constant wind patterns along the shoreline.	Upwelling is the term for when deep ocean water rises to the surface.

• Loss: CachedGISTEmbedLoss with these parameters:

  {'guide': SentenceTransformer(
    (0): Transformer({'max_seq_length': 512, 'do_lower_case': False}) with Transformer model: BertModel 
    (1): Pooling({'word_embedding_dimension': 768, 'pooling_mode_cls_token': True, 'pooling_mode_mean_tokens': False, '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): Normalize()
  ), 'temperature': 0.025}
  

scitail-pairs-qa

• Dataset: scitail-pairs-qa at 0cc4353
• Size: 128 evaluation samples
• Columns: sentence1 and sentence2
• Approximate statistics based on the first 1000 samples:

  	sentence1	sentence2
  type	string	string
  details	min: 7 tokens mean: 16.12 tokens max: 36 tokens	min: 8 tokens mean: 15.41 tokens max: 31 tokens

• Samples:

  sentence1	sentence2
  The hair cells sense(s) the movement of liquid in ear canals.	What senses the movement of liquid in ear canals?
  A metallic bond is the force of attraction between a positive metal ion and valence electrons.	What is the force of attraction between a positive metal ion and valence electrons?
  High consumption of saturated fats is linked to an increased risk of cardiovascular disease.	High consumption of saturated fats is linked to an increased risk of what disease?

• Loss: CachedGISTEmbedLoss with these parameters:

  {'guide': SentenceTransformer(
    (0): Transformer({'max_seq_length': 512, 'do_lower_case': False}) with Transformer model: BertModel 
    (1): Pooling({'word_embedding_dimension': 768, 'pooling_mode_cls_token': True, 'pooling_mode_mean_tokens': False, '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): Normalize()
  ), 'temperature': 0.025}
  

xsum-pairs

• Dataset: xsum-pairs
• Size: 128 evaluation samples
• Columns: summary and document
• Approximate statistics based on the first 1000 samples:

  	summary	document
  type	string	string
  details	min: 13 tokens mean: 25.59 tokens max: 41 tokens	min: 70 tokens mean: 217.06 tokens max: 347 tokens

• Samples:

  summary	document
  A charity which has supported fishermen for more than 100 years has sold its base in Cornwall.	The Fishermen's Mission in Newlyn opened in 1903, but the charity said the building was no longer cost-effective to run. It sold the centre to a local businessman, and said it hoped to use the proceeds to benefit the area. The mission provides practical and spiritual support for fishermen and their families. Julian Waring, from the charity, said that in the 1960s and 1970s there was a demand to house fishermen, feed them fresh meals and provide clothing in emergencies. He said: "Those demands have changed. We're still here in Newlyn and a memorial room will remain, but there's no need for accommodation or a canteen. If accommodation is needed we'll house them in a B&B. "To manage a building comes at a huge cost and if it's not used to its full potential it needs to be reassessed." Mr Waring said 88p of every £1 donated to the charity was spent helping fishermen and the move would allow the charity to "better serve the county as a whole" rather than just Newlyn. In 2014 the mission provided emergency grants of £500 to fishermen in Cornwall who were unable to work due to prolonged winter storms. The mission has had a presence in Cornwall since 1896. It was initially based in Penzance and moved to Newlyn in 1903.
  The recovery in Scotland's labour market has continued into the second half of 2015, according to the latest Bank of Scotland report on jobs.	But, it said there were signs of the upturn slowing. July saw a rise in demand for staff and an increase in average starting salaries, the report said. However, in each case the rates of improvement were well below the highs reached one year ago. The Bank of Scotland Labour Market Barometer for July was measured at 58.2. That is well above the 50 "no-change" level, pointing to a further improvement in overall labour market conditions north of the border. But, the latest reading was the lowest since May 2013, and well below last July's survey-record high of 67.3. The equivalent UK index recorded a six-month low of 61.1 at the start of the third quarter. Donald MacRae, chief Economist at Bank of Scotland, said: "Scotland's labour market continued to improve in July. "The number of people appointed to both permanent and temporary jobs rose in the month but the number of vacancies for permanent jobs increased at the slowest pace in just over two years. "Salary inflation remained solid although slowing to a five-month low. "These results show an economy demonstrating both confidence and growth in the second half of 2015." The report also said:
  A man arrested after a man died in a shooting at a pool party in Surrey has been released on bail.	Ricardo Hunter from Coulsdon in south London, died from a single gunshot wound at the private event in Headley, near Epsom. Two others were injured in the shooting just after 02:30 BST on Monday. The 38-year-old man from London, arrested on suspicion of murder and attempted murder, was bailed until 8 September, Surrey Police said. A 36-year-old woman was shot in the leg and taken to hospital while another man was treated for minor shoulder wounds. On Wednesday a woman, 30, arrested on suspicion of assisting an offender was released on conditional bail until September.

• Loss: CachedGISTEmbedLoss with these parameters:

  {'guide': SentenceTransformer(
    (0): Transformer({'max_seq_length': 512, 'do_lower_case': False}) with Transformer model: BertModel 
    (1): Pooling({'word_embedding_dimension': 768, 'pooling_mode_cls_token': True, 'pooling_mode_mean_tokens': False, '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): Normalize()
  ), 'temperature': 0.025}
  

sciq_pairs

• Dataset: sciq_pairs at 2c94ad3
• Size: 128 evaluation samples
• Columns: sentence1 and sentence2
• Approximate statistics based on the first 1000 samples:

  	sentence1	sentence2
  type	string	string
  details	min: 8 tokens mean: 16.67 tokens max: 32 tokens	min: 2 tokens mean: 84.85 tokens max: 512 tokens

• Samples:

  sentence1	sentence2
  Psip can be positive or negative relative to what kind of pressure?
  How many pathways do plants have for carbon fixation?	Plants have evolved three pathways for carbon fixation. The most common pathway combines one molecule of CO 2 with a 5-carbon sugar called ribulose biphosphate (RuBP). The enzyme which catalyzes this reaction, ribulose-1,5-bisphosphate carboxylase oxygenase (nicknamed RuBisCo ), is the most abundant enzyme on earth! The resulting 6-carbon molecule is unstable, so it immediately splits into two much more stable 3-carbon phosphoglycerate molecules. The 3 carbons in the first stable molecule of this pathway give this largest group of plants the name “C-3. ”.
  What is the amount of force pushing against a given area?	Pressure is defined as the amount of force pushing against a given area. How much pressure a gas exerts depends on the amount of gas. The more gas particles there are, the greater the pressure.

• Loss: CachedGISTEmbedLoss with these parameters:

  {'guide': SentenceTransformer(
    (0): Transformer({'max_seq_length': 512, 'do_lower_case': False}) with Transformer model: BertModel 
    (1): Pooling({'word_embedding_dimension': 768, 'pooling_mode_cls_token': True, 'pooling_mode_mean_tokens': False, '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): Normalize()
  ), 'temperature': 0.025}
  

qasc_pairs

• Dataset: qasc_pairs at a34ba20
• Size: 128 evaluation samples
• Columns: sentence1 and sentence2
• Approximate statistics based on the first 1000 samples:

  	sentence1	sentence2
  type	string	string
  details	min: 6 tokens mean: 11.69 tokens max: 21 tokens	min: 20 tokens mean: 34.49 tokens max: 59 tokens

• Samples:

  sentence1	sentence2
  Where are the structures where proteins are made located on?	Ribosomes are structures in the cytoplasm where proteins are made.. Rough endoplasmic reticulum has ribosomes attached d. structures where proteins are made are located on rough endoplasmic reticulum
  What is something that can be used to impede electrical transference?	an electrical insulator slows the transfer of electricity. Sulfur is a good electrical insulator. Sulfur slows the transfer of electricity.
  What are lakes formed by?	lakes are formed by precipitation and runoff. Clouds form and precipitation occurs. lakes are formed by clouds

• Loss: CachedGISTEmbedLoss with these parameters:

  {'guide': SentenceTransformer(
    (0): Transformer({'max_seq_length': 512, 'do_lower_case': False}) with Transformer model: BertModel 
    (1): Pooling({'word_embedding_dimension': 768, 'pooling_mode_cls_token': True, 'pooling_mode_mean_tokens': False, '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): Normalize()
  ), 'temperature': 0.025}
  

openbookqa_pairs

• Dataset: openbookqa_pairs
• Size: 128 evaluation samples
• Columns: question and fact
• Approximate statistics based on the first 1000 samples:

  	question	fact
  type	string	string
  details	min: 3 tokens mean: 13.98 tokens max: 47 tokens	min: 4 tokens mean: 11.78 tokens max: 28 tokens

• Samples:

  question	fact
  The thermal production of a stove is generically used for	a stove generates heat for cooking usually
  What creates a valley?	a valley is formed by a river flowing
  when it turns day and night on a planet, what cause this?	a planet rotating causes cycles of day and night on that planet

• Loss: CachedGISTEmbedLoss with these parameters:

  {'guide': SentenceTransformer(
    (0): Transformer({'max_seq_length': 512, 'do_lower_case': False}) with Transformer model: BertModel 
    (1): Pooling({'word_embedding_dimension': 768, 'pooling_mode_cls_token': True, 'pooling_mode_mean_tokens': False, '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): Normalize()
  ), 'temperature': 0.025}
  

msmarco_pairs

• Dataset: msmarco_pairs at 28ff31e
• Size: 128 evaluation samples
• Columns: sentence1 and sentence2
• Approximate statistics based on the first 1000 samples:

  	sentence1	sentence2
  type	string	string
  details	min: 4 tokens mean: 8.79 tokens max: 20 tokens	min: 18 tokens mean: 78.0 tokens max: 179 tokens

• Samples:

  sentence1	sentence2
  how much american dollars does it cost to join weight watchers?	Beyond that, many cable services offer free, on-demand workout videos, as do various websites, like YouTube. If you join a weight-loss program: If you opt for Weight Watchers, what you spend will depend on whether you're attending in-person meetings ($42.95 a month) or joining the organization online ($18.95 a month).
  what causes the leaves of an oak tree to curl	A. Leaves of oak trees can be attacked by erophytid (pronounced Arrow-Fie-Tid) mites which cause a curling reaction. You can spray with diazinon insecticide or just let nature take its course. The curled leaves are still functional and will support tree growth. All leaves will not be affected. 2. Q. We have a spectacular oak tree in our yard that had an infestation of oak galls last year.
  what is tramadol hcl	Tramadol hydrochloride is a centrally acting synthetic analgesic in an extended-release formulation. The chemical name is (±) cis-2-[(dimethylamino)methyl]­-1-(3-methoxyphenyl) cyclohexanol hydrochloride. Its structural formula is: The molecular weight of tramadol hydrochloride is 299.8. It is a white, bitter, crystalline and odorless powder that is readily soluble in water and ethanol and has a pKa of 9.41. The n-octanol/water log partition coefficient (logP) is 1.35 at pH 7.

• Loss: CachedGISTEmbedLoss with these parameters:

  {'guide': SentenceTransformer(
    (0): Transformer({'max_seq_length': 512, 'do_lower_case': False}) with Transformer model: BertModel 
    (1): Pooling({'word_embedding_dimension': 768, 'pooling_mode_cls_token': True, 'pooling_mode_mean_tokens': False, '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): Normalize()
  ), 'temperature': 0.025}
  

nq_pairs

• Dataset: nq_pairs at f9e894e
• Size: 128 evaluation samples
• Columns: sentence1 and sentence2
• Approximate statistics based on the first 1000 samples:

  	sentence1	sentence2
  type	string	string
  details	min: 10 tokens mean: 11.39 tokens max: 18 tokens	min: 41 tokens mean: 146.72 tokens max: 392 tokens

• Samples:

  sentence1	sentence2
  where does the queen mary 2 sail to	RMS Queen Mary 2 Like her predecessor Queen Elizabeth 2 she is built for crossing the Atlantic Ocean, though she is regularly used for cruising; in the winter season she cruises from New York to the Caribbean on twelve- or thirteen-day tours. Queen Mary 2's 30-knot (56 km/h; 35 mph) open ocean speed sets the ship apart from cruise ships, such as MS Oasis of the Seas, which has a service speed of 22.6 knots (41.9 km/h; 26.0 mph); QM2's normal service speed is 26 knots (48 km/h; 30 mph).[14] While the hull of a cruise ship will typically have a block coefficient of 0.73 (1.0 would represent a rectangular block) Queen Mary 2 is more fine-lined, with a block coefficient of 0.61.[15]
  who sings lean with it rock with it	Lean wit It, Rock wit It "Lean wit It, Rock wit It" is a song by Atlanta rap group Dem Franchize Boyz from their album On Top of Our Game. The recording features Peanut and Charlay and was produced by Maurice "Parlae" Gleaton.
  which division of the nervous system consists of the brain spinal cord and optic nerves	Central nervous system The central nervous system (CNS) is the part of the nervous system consisting of the brain and spinal cord. The central nervous system is so named because it integrates information it receives from, and coordinates and influences the activity of, all parts of the bodies of bilaterally symmetric animals—that is, all multicellular animals except sponges and radially symmetric animals such as jellyfish—and it contains the majority of the nervous system. Many consider the retina[2] and the optic nerve (cranial nerve II),[3][4] as well as the olfactory nerves (cranial nerve I) and olfactory epithelium[5] as parts of the CNS, synapsing directly on brain tissue without intermediate ganglia. As such, the olfactory epithelium is the only central nervous tissue in direct contact with the environment, which opens up for therapeutic treatments. [5] The CNS is contained within the dorsal body cavity, with the brain housed in the cranial cavity and the spinal cord in the spinal canal. In vertebrates, the brain is protected by the skull, while the spinal cord is protected by the vertebrae.[6] The brain and spinal cord are both enclosed in the meninges.[6] In central nervous system, the interneuronal space is filled with large amount of supporting non nervous cells called neuroglial cells.

• Loss: CachedGISTEmbedLoss with these parameters:

  {'guide': SentenceTransformer(
    (0): Transformer({'max_seq_length': 512, 'do_lower_case': False}) with Transformer model: BertModel 
    (1): Pooling({'word_embedding_dimension': 768, 'pooling_mode_cls_token': True, 'pooling_mode_mean_tokens': False, '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): Normalize()
  ), 'temperature': 0.025}
  

trivia_pairs

• Dataset: trivia_pairs at a7c36e3
• Size: 128 evaluation samples
• Columns: query and answer
• Approximate statistics based on the first 1000 samples:

  	query	answer
  type	string	string
  details	min: 9 tokens mean: 16.9 tokens max: 51 tokens	min: 30 tokens mean: 200.84 tokens max: 379 tokens

• Samples:

  query	answer
  "Which actor spoke the line ""I love the smell of napalm in the morning* in the 1979 film Apocalypse Now?"	Smell of Napalm Scene - Apocalypse Now - YouTube Smell of Napalm Scene - Apocalypse Now Want to watch this again later? Sign in to add this video to a playlist. Need to report the video? Sign in to report inappropriate content. Rating is available when the video has been rented. This feature is not available right now. Please try again later. Uploaded on Nov 8, 2008 This is my favorite scene from one of my favorite movies, "Apocalypse Now" in which Kilgore talks about the smell of napalm. This scene gives birth to one of the most famous movie quotes of all time, "I love the smell of napalm in the morning." Lt. Col. Bill Kilgore is played by the actor Robert Duvall. Category
  After winning £152,000 in 1961, whose autobiography was entitled 'Spend, Spend, Spend'?	Viv Nicholson goes from 'spend, spend, spend' to 'sell, sell, sell': Sorry tale of pools winner who's auctioning off her memorabilia now she's in a care home
  In which 'New England' state is Harvard University situated?	What state are the New England Patriots from?

• Loss: CachedGISTEmbedLoss with these parameters:

  {'guide': SentenceTransformer(
    (0): Transformer({'max_seq_length': 512, 'do_lower_case': False}) with Transformer model: BertModel 
    (1): Pooling({'word_embedding_dimension': 768, 'pooling_mode_cls_token': True, 'pooling_mode_mean_tokens': False, '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): Normalize()
  ), 'temperature': 0.025}
  

gooaq_pairs

• Dataset: gooaq_pairs at b089f72
• Size: 128 evaluation samples
• Columns: sentence1 and sentence2
• Approximate statistics based on the first 1000 samples:

  	sentence1	sentence2
  type	string	string
  details	min: 8 tokens mean: 11.62 tokens max: 24 tokens	min: 13 tokens mean: 55.83 tokens max: 107 tokens

• Samples:

  sentence1	sentence2
  we are never ever getting back together country?	"We Are Never Ever Getting Back Together" is a song recorded by American singer-songwriter Taylor Swift for her fourth studio album, Red (2012). Swift co-wrote the song with its producers, Max Martin and Shellback. ... It also topped the US Hot Country Songs for ten weeks, Swift's longest reign to date.
  can an optometrist treat lazy eye?	Your eye doctor may recommend eye patch therapy in addition to corrective lenses. Strabismus surgery is usually required if the amblyopia is due to a large eye turn. This type of surgery aligns the eyes and corrects the problem within the eye muscles. After the surgery the eyes will able to focus better.
  ok google are the everly brothers still alive?	Don Everly is now 80, while Phil Everly died in January 2014 at age 74.

• Loss: CachedGISTEmbedLoss with these parameters:

  {'guide': SentenceTransformer(
    (0): Transformer({'max_seq_length': 512, 'do_lower_case': False}) with Transformer model: BertModel 
    (1): Pooling({'word_embedding_dimension': 768, 'pooling_mode_cls_token': True, 'pooling_mode_mean_tokens': False, '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): Normalize()
  ), 'temperature': 0.025}
  

paws-pos

• Dataset: paws-pos at 161ece9
• Size: 128 evaluation samples
• Columns: sentence1 and sentence2
• Approximate statistics based on the first 1000 samples:

  	sentence1	sentence2
  type	string	string
  details	min: 10 tokens mean: 25.72 tokens max: 42 tokens	min: 10 tokens mean: 25.55 tokens max: 41 tokens

• Samples:

  sentence1	sentence2
  They were there to enjoy us and they were there to pray for us .	They were there for us to enjoy and they were there for us to pray .
  After the end of the war in June 1902 , Higgins left Southampton in the `` SSBavarian '' in August , returning to Cape Town the following month .	In August , after the end of the war in June 1902 , Higgins Southampton left the `` SSBavarian '' and returned to Cape Town the following month .
  From the merger of the Four Rivers Council and the Audubon Council , the Shawnee Trails Council was born .	Shawnee Trails Council was formed from the merger of the Four Rivers Council and the Audubon Council .

• Loss: CachedGISTEmbedLoss with these parameters:

  {'guide': SentenceTransformer(
    (0): Transformer({'max_seq_length': 512, 'do_lower_case': False}) with Transformer model: BertModel 
    (1): Pooling({'word_embedding_dimension': 768, 'pooling_mode_cls_token': True, 'pooling_mode_mean_tokens': False, '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): Normalize()
  ), 'temperature': 0.025}
  

global_dataset

• Dataset: global_dataset
• Size: 663 evaluation samples
• Columns: sentence1 and sentence2
• Approximate statistics based on the first 1000 samples:

  	sentence1	sentence2
  type	string	string
  details	min: 4 tokens mean: 31.01 tokens max: 347 tokens	min: 2 tokens mean: 58.72 tokens max: 392 tokens

• Samples:

  sentence1	sentence2
  Seeds can be dispersed through all sorts of creative ways, such as	seed dispersal is when the seeds of a plant are moved from the plant to a new environment
  It was part of the Hanover Township , then Chatham Township , before being recorded in 1899 as Florham Park .	It was part of Hanover Township , then Chatham Township before being incorporated as Florham Park in 1899 .
  Pivac joined the Scarlets as assistant coach in July before Easterby announced his decision to move. He takes over the head coach role with immediate effect, though former Ireland international Easterby will remain at the Scarlets until October. "Joining the Scarlets was an exciting challenge for me," said Pivac. "I am honoured that the Scarlets have the faith and belief in me to take the squad forward and build on the good work and solid foundations that Simon and his team have put in place." Former Auckland coach Pivac said Easterby played a key role in his recruitment and he was also influenced by Llanelli's famous 9-3 win over the All Blacks in the 1970s. "There were two factors why I chose Scarlets, the fact that Simon Easterby jumped on a plane and came to New Zealand rather than a lot of talks going on for a long period of time. "[And] As a young boy growing up listening to the All Blacks play Llanelli in 1972 on the radio back home, I've never forgotten that moment. "Knowing the Scarlets has a proud history, like the union I've come from, was important to me and coming to an area where they live and breathe rugby like home." In addition to his role at Auckland, New Zealander Pivac coached Fiji to the Pacific Tri-Nations and was also coach of the side which won the 2005 Rugby World Cup Sevens.	Scarlets have confirmed Wayne Pivac will take over as head coach from Simon Easterby who has been appointed Ireland forwards coach.

• Loss: CachedGISTEmbedLoss with these parameters:

  {'guide': SentenceTransformer(
    (0): Transformer({'max_seq_length': 512, 'do_lower_case': False}) with Transformer model: BertModel 
    (1): Pooling({'word_embedding_dimension': 768, 'pooling_mode_cls_token': True, 'pooling_mode_mean_tokens': False, '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): Normalize()
  ), 'temperature': 0.025}
  

Training Hyperparameters

Non-Default Hyperparameters

• eval_strategy: steps
• per_device_train_batch_size: 96
• per_device_eval_batch_size: 128
• gradient_accumulation_steps: 2
• learning_rate: 4.5e-05
• weight_decay: 0.001
• num_train_epochs: 5
• lr_scheduler_type: cosine_with_min_lr
• lr_scheduler_kwargs: {'num_cycles': 0.5, 'min_lr': 9e-06}
• warmup_ratio: 0.33
• save_safetensors: False
• fp16: True
• push_to_hub: True
• hub_model_id: bobox/DeBERTa2-0.9B-ST-v2-checkpoints-tmp
• hub_strategy: all_checkpoints
• batch_sampler: no_duplicates

All Hyperparameters

Click to expand

• overwrite_output_dir: False
• do_predict: False
• eval_strategy: steps
• prediction_loss_only: True
• per_device_train_batch_size: 96
• per_device_eval_batch_size: 128
• per_gpu_train_batch_size: None
• per_gpu_eval_batch_size: None
• gradient_accumulation_steps: 2
• eval_accumulation_steps: None
• torch_empty_cache_steps: None
• learning_rate: 4.5e-05
• weight_decay: 0.001
• adam_beta1: 0.9
• adam_beta2: 0.999
• adam_epsilon: 1e-08
• max_grad_norm: 1.0
• num_train_epochs: 5
• max_steps: -1
• lr_scheduler_type: cosine_with_min_lr
• lr_scheduler_kwargs: {'num_cycles': 0.5, 'min_lr': 9e-06}
• warmup_ratio: 0.33
• warmup_steps: 0
• log_level: passive
• log_level_replica: warning
• log_on_each_node: True
• logging_nan_inf_filter: True
• save_safetensors: False
• save_on_each_node: False
• save_only_model: False
• restore_callback_states_from_checkpoint: False
• no_cuda: False
• use_cpu: False
• use_mps_device: False
• seed: 42
• data_seed: None
• jit_mode_eval: False
• use_ipex: False
• bf16: False
• fp16: True
• fp16_opt_level: O1
• half_precision_backend: auto
• bf16_full_eval: False
• fp16_full_eval: False
• tf32: None
• local_rank: 0
• ddp_backend: None
• tpu_num_cores: None
• tpu_metrics_debug: False
• debug: []
• dataloader_drop_last: False
• dataloader_num_workers: 0
• dataloader_prefetch_factor: None
• past_index: -1
• disable_tqdm: False
• remove_unused_columns: True
• label_names: None
• load_best_model_at_end: False
• ignore_data_skip: False
• fsdp: []
• fsdp_min_num_params: 0
• fsdp_config: {'min_num_params': 0, 'xla': False, 'xla_fsdp_v2': False, 'xla_fsdp_grad_ckpt': False}
• fsdp_transformer_layer_cls_to_wrap: None
• accelerator_config: {'split_batches': False, 'dispatch_batches': None, 'even_batches': True, 'use_seedable_sampler': True, 'non_blocking': False, 'gradient_accumulation_kwargs': None}
• deepspeed: None
• label_smoothing_factor: 0.0
• optim: adamw_torch
• optim_args: None
• adafactor: False
• group_by_length: False
• length_column_name: length
• ddp_find_unused_parameters: None
• ddp_bucket_cap_mb: None
• ddp_broadcast_buffers: False
• dataloader_pin_memory: True
• dataloader_persistent_workers: False
• skip_memory_metrics: True
• use_legacy_prediction_loop: False
• push_to_hub: True
• resume_from_checkpoint: None
• hub_model_id: bobox/DeBERTa2-0.9B-ST-v2-checkpoints-tmp
• hub_strategy: all_checkpoints
• hub_private_repo: False
• hub_always_push: False
• gradient_checkpointing: False
• gradient_checkpointing_kwargs: None
• include_inputs_for_metrics: False
• eval_do_concat_batches: True
• fp16_backend: auto
• push_to_hub_model_id: None
• push_to_hub_organization: None
• mp_parameters:
• auto_find_batch_size: False
• full_determinism: False
• torchdynamo: None
• ray_scope: last
• ddp_timeout: 1800
• torch_compile: False
• torch_compile_backend: None
• torch_compile_mode: None
• dispatch_batches: None
• split_batches: None
• include_tokens_per_second: False
• include_num_input_tokens_seen: False
• neftune_noise_alpha: None
• optim_target_modules: None
• batch_eval_metrics: False
• eval_on_start: False
• eval_use_gather_object: False
• batch_sampler: no_duplicates
• multi_dataset_batch_sampler: proportional

Training Logs

Click to expand

Epoch	Step	Training Loss	negation-triplets loss	vitaminc-pairs loss	scitail-pairs-pos loss	sciq pairs loss	trivia pairs loss	xsum-pairs loss	openbookqa pairs loss	msmarco pairs loss	nq pairs loss	global dataset loss	paws-pos loss	scitail-pairs-qa loss	qasc pairs loss	gooaq pairs loss	Qnli-dev_max_ap	allNLI-dev_max_ap	sts-test_spearman_cosine
0.0129	11	0.1802	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
0.0257	22	0.1573	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
0.0386	33	0.1184	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
0.0515	44	0.1456	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
0.0643	55	0.2102	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
0.0772	66	0.151	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
0.0901	77	0.1356	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
0.1029	88	0.2292	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
0.1158	99	0.126	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
0.1287	110	0.1942	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
0.1415	121	0.2089	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
0.1544	132	0.1225	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
0.1673	143	0.2504	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
0.1801	154	0.1454	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
0.1930	165	0.2052	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
0.2058	176	0.1321	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
0.2187	187	0.1975	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
0.2316	198	0.1615	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
0.2444	209	0.2174	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
0.2503	214	-	1.0606	2.7611	0.0596	0.0132	0.2116	0.0044	1.3197	0.1359	0.0413	0.2187	0.0459	0.0000	0.0490	0.1187	0.7328	0.6432	0.9079
0.2573	220	0.1545	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
0.2702	231	0.1201	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
0.2830	242	0.1389	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
0.2959	253	0.1657	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
0.3088	264	0.237	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
0.3216	275	0.1094	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
0.3345	286	0.196	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
0.3474	297	0.2164	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
0.3602	308	0.1793	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
0.3731	319	0.2878	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
0.3860	330	0.1189	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
0.3988	341	0.1475	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
0.4117	352	0.1019	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
0.4246	363	0.1587	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
0.4374	374	0.2483	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
0.4503	385	0.1427	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
0.4632	396	0.1199	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
0.4760	407	0.2037	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
0.4889	418	0.1317	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
0.5006	428	-	1.0632	2.7318	0.0605	0.0029	0.2094	0.0045	1.2716	0.1622	0.0400	0.2122	0.0461	0.0000	0.0601	0.1271	0.7294	0.6397	0.9063
0.5018	429	0.1293	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
0.5146	440	0.1902	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
0.5275	451	0.1429	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
0.5404	462	0.2446	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
0.5532	473	0.1623	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
0.5661	484	0.0707	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
0.5789	495	0.1557	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
0.5918	506	0.2016	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
0.6047	517	0.1018	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
0.6175	528	0.1821	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
0.6304	539	0.1437	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
0.6433	550	0.1112	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
0.6561	561	0.12	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
0.6690	572	0.0933	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
0.6819	583	0.0939	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
0.6947	594	0.2064	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
0.7076	605	0.131	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
0.7205	616	0.161	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
0.7333	627	0.213	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
0.7462	638	0.1853	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
0.7509	642	-	1.0684	2.7074	0.0593	0.0030	0.2113	0.0052	1.2897	0.1479	0.0423	0.2030	0.0460	0.0000	0.0563	0.1147	0.7257	0.6432	0.9073
0.7591	649	0.1919	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
0.7719	660	0.1395	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
0.7848	671	0.2047	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
0.7977	682	0.1421	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
0.8105	693	0.1227	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
0.8234	704	0.1235	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
0.8363	715	0.2004	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
0.8491	726	0.1568	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
0.8620	737	0.1598	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
0.8749	748	0.1328	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
0.8877	759	0.0999	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
0.9006	770	0.1058	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
0.9135	781	0.1673	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
0.9263	792	0.1905	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
0.9392	803	0.1463	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
0.9520	814	0.1294	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
0.9649	825	0.1312	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
0.9778	836	0.1308	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
0.9906	847	0.1076	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
1.0012	856	-	1.0481	2.6587	0.0596	0.0031	0.2101	0.0056	1.2992	0.1593	0.0405	0.2027	0.0462	0.0000	0.0542	0.1140	0.7226	0.6377	0.9070
1.0035	858	0.1085	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
1.0164	869	0.2214	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
1.0292	880	0.1214	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
1.0421	891	0.1049	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
1.0550	902	0.1897	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
1.0678	913	0.1273	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
1.0807	924	0.1474	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
1.0936	935	0.1313	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
1.1064	946	0.1769	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
1.1193	957	0.143	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
1.1322	968	0.1968	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
1.1450	979	0.1771	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
1.1579	990	0.1822	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
1.1708	1001	0.2467	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
1.1836	1012	0.1419	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
1.1965	1023	0.1782	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
1.2094	1034	0.1297	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
1.2222	1045	0.1972	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
1.2351	1056	0.1491	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
1.2480	1067	0.1721	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
1.2515	1070	-	1.0599	2.6560	0.0610	0.0029	0.2088	0.0053	1.2817	0.1446	0.0420	0.2135	0.0460	0.0000	0.0509	0.1112	0.7247	0.6369	0.9072
1.2608	1078	0.1279	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
1.2737	1089	0.106	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
1.2865	1100	0.1597	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
1.2994	1111	0.192	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
1.3123	1122	0.165	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
1.3251	1133	0.1472	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
1.3380	1144	0.1528	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
1.3509	1155	0.202	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
1.3637	1166	0.1974	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
1.3766	1177	0.2229	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
1.3895	1188	0.1104	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
1.4023	1199	0.1544	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
1.4152	1210	0.0875	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
1.4281	1221	0.1607	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
1.4409	1232	0.2026	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
1.4538	1243	0.185	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
1.4667	1254	0.1114	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
1.4795	1265	0.2033	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
1.4924	1276	0.1216	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
1.5018	1284	-	1.0499	2.6942	0.0598	0.0033	0.2039	0.0053	1.3073	0.1548	0.0390	0.2039	0.0455	0.0000	0.0437	0.1066	0.7248	0.6358	0.9076
1.5053	1287	0.1108	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
1.5181	1298	0.188	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
1.5310	1309	0.1731	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
1.5439	1320	0.2191	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
1.5567	1331	0.146	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
1.5696	1342	0.1045	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
1.5825	1353	0.1901	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
1.5953	1364	0.1898	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
1.6082	1375	0.0942	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
1.6211	1386	0.1809	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
1.6339	1397	0.1083	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
1.6468	1408	0.1277	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
1.6596	1419	0.1039	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
1.6725	1430	0.0933	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
1.6854	1441	0.11	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
1.6982	1452	0.2423	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
1.7111	1463	0.1085	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
1.7240	1474	0.1678	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
1.7368	1485	0.1799	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
1.7497	1496	0.1811	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
1.7520	1498	-	1.0360	2.6751	0.0565	0.0044	0.1985	0.0060	1.2448	0.1428	0.0401	0.2027	0.0459	0.0000	0.0569	0.1081	0.7226	0.6360	0.9070

Framework Versions

• Python: 3.10.14
• Sentence Transformers: 3.0.1
• Transformers: 4.44.0
• PyTorch: 2.4.0
• Accelerate: 0.33.0
• Datasets: 2.21.0
• Tokenizers: 0.19.1

Citation

BibTeX

Sentence Transformers

@inproceedings{reimers-2019-sentence-bert,
    title = "Sentence-BERT: Sentence Embeddings using Siamese BERT-Networks",
    author = "Reimers, Nils and Gurevych, Iryna",
    booktitle = "Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing",
    month = "11",
    year = "2019",
    publisher = "Association for Computational Linguistics",
    url = "https://arxiv.org/abs/1908.10084",
}