metadata

language:
  - en
tags:
  - sentence-transformers
  - sentence-similarity
  - feature-extraction
  - generated_from_trainer
  - dataset_size:4731012
  - loss:MultipleNegativesRankingLoss
  - loss:CachedMultipleNegativesRankingLoss
  - loss:SoftmaxLoss
  - loss:CosineSimilarityLoss
base_model: tasksource/ModernBERT-base-nli
widget:
  - source_sentence: Christa McAuliffe taught social studies at Concord High School.
    sentences:
      - >-
        The Football League play-offs for the 1994 -- 95 season were held in May
        1995 , with the finals taking place at Wembley Stadium in London ..
        Football League play-offs. Football League play-offs. 1994 Football
        League play-offs. Wembley Stadium. Wembley Stadium ( 1923 ). London.
        London. The play-off semi-finals were played over two legs and were
        contested by the teams who finished in 2nd , 3rd , 4th and 5th place in
        the Football League First Division and Football League Second Division
        and the 3rd , 4th , 5th , and 6th placed teams in the Football League
        Third Division table .. Football League First Division. 1994–95 Football
        League First Division. Football League Second Division. 1994–95 Football
        League Second Division. Football League Third Division. 1994–95 Football
        League Third Division. The winners of the semi-finals progressed through
        to the finals , with the winner of these matches gaining promotion for
        the following season .. following season. 1995-96 in English football
      - >-
        Sir Alexander Mackenzie Elementary is a public elementary school in
        Vancouver , British Columbia part of School District 39 Vancouver ..
        Vancouver. Vancouver, British Columbia. British Columbia. British
        Columbia. School District 39 Vancouver. School District 39 Vancouver.
        elementary school. elementary school
      - >-
        Help Wanted -LRB- Hataraku Hito  : Hard Working People in Japan , Job
        Island  : Hard Working People in Europe -RRB- is a game that features a
        collection of various , Wii Remote-based minigames .. Wii. Wii. Wii
        Remote. Wii Remote. The game is developed and published by Hudson Soft
        and was released in Japan for Nintendo 's Wii on November 27 , 2008 , in
        Europe on March 13 , 2009 , in Australia on March 27 , 2009 , and in
        North America on May 12 , 2009 .. Hudson Soft. Hudson Soft. Wii. Wii.
        Nintendo. Nintendo
  - source_sentence: >-
      The researchers asked children of different ages to use words to form
      semantic correspondence. For example, when children see the words eagle,
      bear and robin, they combine them best according to their meaning. The
      results showed that older participants were more likely to develop
      different types of false memory than younger participants. Because there
      are many forms of classification in their minds. For example, young
      children classify eagles and robins as birds, while older children
      classify eagles and bears as predators. Compared with children, they have
      a concept of predators in their minds.
    sentences:
      - Extractive Industries Transparency Initiative is an organization
      - Mason heard a pun
      - >-
        Older children are more likely to have false memories than younger ones
        conforms to the context.
  - source_sentence: >-
      Version 0.5 is released today. The biggest change is that this version
      finally has upload progress.

      Download it here:

      Or go to for more information about this project.

      Changelog:

      * Refactored the authentication_controller

      * Put before_filter :authorize in ApplicationController (and using
      skip_before_filter in other controllers if necessary)

      * Using 'unless' instead of 'if not'

      * Using find_by() instead of find(:first)

      * Upload progress (yay!)

      Forums |

      Admin
    sentences:
      - This example wikipedia comment contains an insult.
      - 'This text is about: hardware update'
      - The example summary is factually consistent with the full article.
  - source_sentence: >-
      Make sure to make it to the Brew House in Pella, IA tomorrow @ 3 to meet
      with @user supporters! #SemST
    sentences:
      - This example is ANT.
      - This example is valid question.
      - This example is favor.
  - source_sentence: >-
      Also at increased risk are those whose immune systems suppressed by
      medications or by diseases such as cancer, diabetes and AIDS.
    sentences:
      - In 1995, the last survey, those numbers were equal.
      - >-
        Also at increased risk are those with suppressed immune systems due to
        illness or medicines.
      - Singapore stocks close 0.54 pct higher
datasets:
  - tomaarsen/natural-questions-hard-negatives
  - tomaarsen/gooaq-hard-negatives
  - bclavie/msmarco-500k-triplets
  - sentence-transformers/gooaq
  - sentence-transformers/natural-questions
  - tasksource/merged-2l-nli
  - tasksource/merged-3l-nli
  - tasksource/zero-shot-label-nli
  - MoritzLaurer/dataset_train_nli
  - google-research-datasets/paws
  - nyu-mll/glue
  - mwong/fever-evidence-related
  - tasksource/sts-companion
pipeline_tag: sentence-similarity
library_name: sentence-transformers

SentenceTransformer based on tasksource/ModernBERT-base-nli

This is a sentence-transformers model finetuned from tasksource/ModernBERT-base-nli on the tomaarsen/natural-questions-hard-negatives, tomaarsen/gooaq-hard-negatives, bclavie/msmarco-500k-triplets, sentence-transformers/gooaq, sentence-transformers/natural-questions, merged-2l-nli, merged-3l-nli, zero-shot-label-nli, dataset_train_nli, paws/labeled_final, glue/mrpc, glue/qqp, fever-evidence-related, glue/stsb, sick/relatedness and sts-companion 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: tasksource/ModernBERT-base-nli
Maximum Sequence Length: 2048 tokens
Output Dimensionality: 768 dimensions
Similarity Function: Cosine Similarity
Training Datasets:
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': 2048, 'do_lower_case': False}) with Transformer model: ModernBertModel 
  (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})
)

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("tasksource/ModernBERT-base-embed")
# Run inference
sentences = [
    'Also at increased risk are those whose immune systems suppressed by medications or by diseases such as cancer, diabetes and AIDS.',
    'Also at increased risk are those with suppressed immune systems due to illness or medicines.',
    'In 1995, the last survey, those numbers were equal.',
]
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]

Training Details

Training Datasets

tomaarsen/natural-questions-hard-negatives

Dataset: tomaarsen/natural-questions-hard-negatives at 52dfa09
Size: 96,658 training samples
Columns: query, answer, negative_1, negative_2, negative_3, negative_4, and negative_5

Approximate statistics based on the first 1000 samples:

	query	answer	negative_1	negative_2	negative_3	negative_4	negative_5
type	string	string	string	string	string	string	string
details	min: 10 tokens mean: 12.52 tokens max: 26 tokens	min: 17 tokens mean: 137.85 tokens max: 556 tokens	min: 23 tokens mean: 144.11 tokens max: 1035 tokens	min: 13 tokens mean: 142.73 tokens max: 832 tokens	min: 15 tokens mean: 146.37 tokens max: 649 tokens	min: 19 tokens mean: 145.79 tokens max: 549 tokens	min: 19 tokens mean: 142.01 tokens max: 574 tokens

Samples:

query	answer	negative_1	negative_2	negative_3	negative_4	negative_5
`when did richmond last play in a preliminary final`	Richmond Football Club Richmond began 2017 with 5 straight wins, a feat it had not achieved since 1995. A series of close losses hampered the Tigers throughout the middle of the season, including a 5-point loss to the Western Bulldogs, 2-point loss to Fremantle, and a 3-point loss to the Giants. Richmond ended the season strongly with convincing victories over Fremantle and St Kilda in the final two rounds, elevating the club to 3rd on the ladder. Richmond's first final of the season against the Cats at the MCG attracted a record qualifying final crowd of 95,028; the Tigers won by 51 points. Having advanced to the first preliminary finals for the first time since 2001, Richmond defeated Greater Western Sydney by 36 points in front of a crowd of 94,258 to progress to the Grand Final against Adelaide, their first Grand Final appearance since 1982. The attendance was 100,021, the largest crowd to a grand final since 1986. The Crows led at quarter time and led by as many as 13, but the Tig...	Brisbane Bears However, the club was still struggling off-field. One of the Bears' biggest problems was its lack of support (both on and off the field) in Melbourne, the location of most of its away matches. In mid-1996, the struggling Fitzroy Football Club collapsed due to financial pressures and was seeking to merge its assets with another club. When a merger with North Melbourne in forming the North Fitzroy Kangaroos failed to win the support of the other AFL clubs, a deal for a merger was done between Fitzroy and the Bears. The new team was known as the Brisbane Lions, based at the Gabba, with Northey as the coach of the merged club. As such, the history of the Brisbane Bears as an individual entity ended after the 1996 season, with ten seasons of competition and the third-place finish in 1996 as its best performance. The Bears last match as a separate entity was a preliminary final on Saturday 21 September 1996 at the Melbourne Cricket Ground (where the Bears played their first VF...	`Virginia Tech–West Virginia football rivalry Virginia Tech held the trophy in six of the nine years in which it was contested, but West Virginia leads the all-time series 28–23–1. The last game was played on September 3, 2017 at FedEx Field in Landover, MD; Virginia Tech won 31–24.`	`Martin Truex Jr. To start off the Round of 12, Truex scored his 6th win of the season at Charlotte after leading 91 out of 334 laps to secure a spot for the Round of 8. Just two weeks later, he scored another win at Kansas despite having a restart violation early in the race.`	Adelaide Football Club Star midfielder for many years Patrick Dangerfield left the club at the end of the 2015 season (a season in which he won the club's best and fairest) and Don Pyke, a former premiership player and assistant coach with West Coast who had also been an assistant coach at Adelaide from 2005 to 2006, was appointed Adelaide's senior coach for at least three years.[9] Adelaide was widely tipped to slide out of the finals in 2016[27][28][29] but the Crows proved to be one of the successes of the season, comfortably qualifying for a home elimination final and defeating North Melbourne by 62 points, before being eliminated the next week by eventual beaten grand finalists, Sydney in the semi-finals. The club had a dominant 2017 season, winning their opening six games and never falling below second place for the entire season. Adelaide claimed their second McClelland Trophy as minor premiers.[30] The Adelaide Crows entered the 2017 finals series as favourites for the premiers...	Battle of Appomattox Court House The Battle of Appomattox Court House (Virginia, U.S.), fought on the morning of April 9, 1865, was one of the last battles of the American Civil War (1861â€“1865). It was the final engagement of Confederate States Army General-in-Chief, Robert E. Lee, and his Army of Northern Virginia before it surrendered to the Union Army of the Potomac under the Commanding General of the United States, Ulysses S. Grant. Lee, having abandoned the Confederate capital of Richmond, Virginia, after the nine and one-half month Siege of Petersburg and Richmond, retreated west, hoping to join his army with the remaining Confederate forces in North Carolina, the Army of Tennessee under Gen. Joseph E. Johnston. Union infantry and cavalry forces under Gen. Philip Sheridan pursued and cut off the Confederates' retreat at the central Virginia village of Appomattox Court House. Lee launched a last-ditch attack to break through the Union forces to his front, assuming the Union forc...
`who sang what in the world's come over you`	Jack Scott (singer) At the beginning of 1960, Scott again changed record labels, this time to Top Rank Records.[1] He then recorded four Billboard Hot 100 hits – "What in the World's Come Over You" (#5), "Burning Bridges" (#3) b/w "Oh Little One" (#34), and "It Only Happened Yesterday" (#38).[1] "What in the World's Come Over You" was Scott's second gold disc winner.[6] Scott continued to record and perform during the 1960s and 1970s.[1] His song "You're Just Gettin' Better" reached the country charts in 1974.[1] In May 1977, Scott recorded a Peel session for BBC Radio 1 disc jockey, John Peel.	`Lover, You Should've Come Over "Lover, You Should've Come Over" is the seventh track on Jeff Buckley's album Grace. Inspired by the ending of the relationship between Buckley and Rebecca Moore,[1] it concerns the despondency of a young man growing older, finding that his actions represent a perspective he feels that he should have outgrown. Biographer and critic David Browne describes the lyrics as "confused and confusing" and the music as "a languid beauty."[1]`	`It's Christmas (All Over The World) "It's Christmas (All Over The World)" is a song recorded by Scottish singer Sheena Easton. It was released in November 1985 as the theme song from the soundtrack of Santa Claus: The Movie. The song was written by Bill House and John Hobbs.`	`The End of the World (Skeeter Davis song) "The End of the World" is a country pop song written by Arthur Kent and Sylvia Dee, for American singer Skeeter Davis. It had success in the 1960s and spawned many covers.`	`Israel Kamakawiwoʻole His voice became famous outside Hawaii when his album Facing Future was released in 1993. His medley of "Somewhere Over the Rainbow/What a Wonderful World" was released on his albums Ka ʻAnoʻi and Facing Future. It was subsequently featured in several films, television programs, and television commercials.`	`Make the World Go Away "Make the World Go Away'" is a country-popular music song composed by Hank Cochran. It has become a Top 40 popular success three times: for Timi Yuro (during 1963), for Eddy Arnold (1965), and for the brother-sister duo Donny and Marie Osmond (1975). The original version of the song was recorded by Ray Price during 1963. It has remained a country crooner standard ever since.`
`who produces the most wool in the world`	Wool Global wool production is about 2 million tonnes per year, of which 60% goes into apparel. Wool comprises ca 3% of the global textile market, but its value is higher owing to dying and other modifications of the material.[1] Australia is a leading producer of wool which is mostly from Merino sheep but has been eclipsed by China in terms of total weight.[30] New Zealand (2016) is the third-largest producer of wool, and the largest producer of crossbred wool. Breeds such as Lincoln, Romney, Drysdale, and Elliotdale produce coarser fibers, and wool from these sheep is usually used for making carpets.	Baa, Baa, Black Sheep As with many nursery rhymes, attempts have been made to find origins and meanings for the rhyme, most which have no corroborating evidence.[1] Katherine Elwes Thomas in The Real Personages of Mother Goose (1930) suggested that the rhyme referred to resentment at the heavy taxation on wool.[5] This has particularly been taken to refer to the medieval English "Great" or "Old Custom" wool tax of 1275, which survived until the fifteenth century.[1] More recently the rhyme has been connected to the slave trade, particularly in the southern United States.[6] This explanation was advanced during debates over political correctness and the use and reform of nursery rhymes in the 1980s, but has no supporting historical evidence.[7] Rather than being negative, the wool of black sheep may have been prized as it could be made into dark cloth without dyeing.[6]	`Raymond Group Raymond Group is an Indian branded fabric and fashion retailer, incorporated in 1925. It produces suiting fabric, with a capacity of producing 31 million meters of wool and wool-blended fabrics. Gautam Singhania is the chairman and managing director of the Raymond group.[3]`	Silk in the Indian subcontinent Silk in the Indian subcontinent is a luxury good. In India, about 97% of the raw mulberry silk is produced in the five Indian states of Karnataka, Andhra Pradesh, Tamil Nadu, West Bengal and Jammu and Kashmir.[1] Mysore and North Bangalore, the upcoming site of a US$20 million "Silk City", contribute to a majority of silk production.[2] Another emerging silk producer is Tamil Nadu where mulberry cultivation is concentrated in Salem, Erode and Dharmapuri districts. Hyderabad, Andhra Pradesh and Gobichettipalayam, Tamil Nadu were the first locations to have automated silk reeling units.[3] yoyo quantity:::	F. W. Woolworth Company The two Woolworth brothers pioneered and developed merchandising, direct purchasing, sales, and customer service practices commonly used today. Despite its growing to be one of the largest retail chains in the world through most of the 20th century, increased competition led to its decline beginning in the 1980s, while its sporting goods division grew. The chain went out of business in July 1997, when the company decided to focus primarily on sporting goods and renamed itself Venator Group. By 2001, the company focused exclusively on the sporting goods market, changing its name to the present Foot Locker, Inc., changing its ticker symbol from its familiar Z in 2003 to its present ticker (NYSE:Â FL).	Silk Silk's absorbency makes it comfortable to wear in warm weather and while active. Its low conductivity keeps warm air close to the skin during cold weather. It is often used for clothing such as shirts, ties, blouses, formal dresses, high fashion clothes, lining, lingerie, pajamas, robes, dress suits, sun dresses and Eastern folk costumes. For practical use, silk is excellent as clothing that protects from many biting insects that would ordinarily pierce clothing, such as mosquitoes and horseflies.

Loss: MultipleNegativesRankingLoss with these parameters:

{
    "scale": 20.0,
    "similarity_fct": "cos_sim"
}

tomaarsen/gooaq-hard-negatives

Dataset: tomaarsen/gooaq-hard-negatives at 87594a1
Size: 200,000 training samples
Columns: question, answer, negative_1, negative_2, negative_3, negative_4, and negative_5

Approximate statistics based on the first 1000 samples:

	question	answer	negative_1	negative_2	negative_3	negative_4	negative_5
type	string	string	string	string	string	string	string
details	min: 8 tokens mean: 11.99 tokens max: 22 tokens	min: 15 tokens mean: 57.82 tokens max: 138 tokens	min: 14 tokens mean: 57.42 tokens max: 125 tokens	min: 14 tokens mean: 56.84 tokens max: 120 tokens	min: 14 tokens mean: 57.08 tokens max: 155 tokens	min: 15 tokens mean: 57.54 tokens max: 129 tokens	min: 14 tokens mean: 58.23 tokens max: 195 tokens

Samples:

question	answer	negative_1	negative_2	negative_3	negative_4	negative_5
`is toprol xl the same as metoprolol?`	`Metoprolol succinate is also known by the brand name Toprol XL. It is the extended-release form of metoprolol. Metoprolol succinate is approved to treat high blood pressure, chronic chest pain, and congestive heart failure.`	`Secondly, metoprolol and metoprolol ER have different brand-name equivalents: Brand version of metoprolol: Lopressor. Brand version of metoprolol ER: Toprol XL.`	`Pill with imprint 1 is White, Round and has been identified as Metoprolol Tartrate 25 mg.`	`Interactions between your drugs No interactions were found between Allergy Relief and metoprolol. This does not necessarily mean no interactions exist. Always consult your healthcare provider.`	`Metoprolol is a type of medication called a beta blocker. It works by relaxing blood vessels and slowing heart rate, which improves blood flow and lowers blood pressure. Metoprolol can also improve the likelihood of survival after a heart attack.`	`Metoprolol starts to work after about 2 hours, but it can take up to 1 week to fully take effect. You may not feel any different when you take metoprolol, but this doesn't mean it's not working. It's important to keep taking your medicine.`
`are you experienced cd steve hoffman?`	`The Are You Experienced album was apparently mastered from the original stereo UK master tapes (according to Steve Hoffman - one of the very few who has heard both the master tapes and the CDs produced over the years). ... The CD booklets were a little sparse, but at least they stayed true to the album's original design.`	`I Saw the Light. Showcasing the unique talent and musical influence of country-western artist Hank Williams, this candid biography also sheds light on the legacy of drug abuse and tormented relationships that contributes to the singer's legend.`	`(Read our ranking of his top 10.) And while Howard dresses the part of director, any notion of him as a tortured auteur or dictatorial taskmasker — the clichés of the Hollywood director — are tossed aside. He's very nice.`	`He was a music star too. Where're you people born and brought up? We 're born and brought up here in Anambra State at Nkpor town, near Onitsha.`	`At the age of 87 he has now retired from his live shows and all the traveling involved. And although he still picks up his Martin Guitar and does a show now and then, his life is now devoted to writing his memoirs.`	`The owner of the mysterious voice behind all these videos is a man who's seen a lot, visiting a total of 56 intimate celebrity spaces over the course of five years. His name is Joe Sabia — that's him in the photo — and he's currently the VP of creative development at Condé Nast Entertainment.`
`how are babushka dolls made?`	`Matryoshka dolls are made of wood from lime, balsa, alder, aspen, and birch trees; lime is probably the most common wood type. ... After cutting, the trees are stripped of most of their bark, although a few inner rings of bark are left to bind the wood and keep it from splitting.`	`A quick scan of the auction and buy-it-now listings on eBay finds porcelain doll values ranging from around $5 and $10 to several thousand dollars or more but no dolls listed above $10,000.`	`Japanese dolls are called as ningyō in Japanese and literally translates to 'human form'.`	`Matyoo: All Fresno Girl dolls come just as real children are born.`	`As of September 2016, there are over 100 characters. The main toy line includes 13-inch Dolls, the mini-series, and a variety of mini play-sets and plush dolls as well as Lalaloopsy Littles, smaller siblings of the 13-inch dolls. A spin-off known as "Lala-Oopsies" came out in late 2012.`	`LOL dolls are little baby dolls that come wrapped inside a surprise toy ball. Each ball has layers that contain stickers, secret messages, mix and match accessories–and finally–a doll. ... The doll on the ball is almost never the doll inside. Dolls are released in series, so not every doll is available all the time.`

Loss: MultipleNegativesRankingLoss with these parameters:

{
    "scale": 20.0,
    "similarity_fct": "cos_sim"
}

bclavie/msmarco-500k-triplets

Dataset: bclavie/msmarco-500k-triplets at cb1a85c
Size: 200,000 training samples
Columns: query, positive, and negative

Approximate statistics based on the first 1000 samples:

	query	positive	negative
type	string	string	string
details	min: 4 tokens mean: 9.31 tokens max: 31 tokens	min: 20 tokens mean: 82.19 tokens max: 216 tokens	min: 18 tokens mean: 78.99 tokens max: 209 tokens

Samples:

query	positive	negative
`the most important factor that influences k+ secretion is __________.`	`The regulation of K+ distribution between the intracellular and extracellular space is referred to as internal K+ balance. The most important factors regulating this movement under normal conditions are insulin and catecholamines (1).`	`They are both also important for secretion and flow of bile: 1 Cholecystokinin: The name of this hormone describes its effect on the biliary system-cholecysto = gallbladder and kinin = movement. 2 Secretin: This hormone is secreted in response to acid in the duodenum.`
`how much did the mackinac bridge cost to build`	`The cost to design the project was $3,500,000 (Steinman Company). The cost to construct the bridge was $70, 268,500. Two primary contractors were hired to build the bridge: American Bridge for superstructure - $44,532,900; and Merritt-Chapman and Scott of New York for the foundations - $25,735,600.`	`When your child needs a dental tooth bridge, you need to know the average cost so you can factor the price into your budget. Several factors affect the price of a bridge, which can run between $700 to $1,500 per tooth. If you have insurance or your child is covered by Medicaid, part of the cost may be covered.`
`when do concussion symptoms appear`	`Then you can get advice on what to do next. For milder symptoms, the doctor may recommend rest and ask you to watch your child closely for changes, such as a headache that gets worse. Symptoms of a concussion don't always show up right away, and can develop within 24 to 72 hours after an injury.`	`Concussion: A traumatic injury to soft tissue, usually the brain, as a result of a violent blow, shaking, or spinning. A brain concussion can cause immediate but temporary impairment of brain functions, such as thinking, vision, equilibrium, and consciousness. After a person has had a concussion, he or she is at increased risk for recurrence. Moreover, after a person has several concussions, less of a blow can cause injury, and the person can require more time to recover.`

Loss: MultipleNegativesRankingLoss with these parameters:

{
    "scale": 20.0,
    "similarity_fct": "cos_sim"
}

sentence-transformers/gooaq

Dataset: sentence-transformers/gooaq at b089f72
Size: 200,000 training samples
Columns: question and answer
Approximate statistics based on the first 1000 samples:
question answer
type string string
details
min: 8 tokens
mean: 12.19 tokens
max: 22 tokens

min: 13 tokens
mean: 58.34 tokens
max: 124 tokens

	question	answer
type	string	string
details	min: 8 tokens mean: 12.19 tokens max: 22 tokens	min: 13 tokens mean: 58.34 tokens max: 124 tokens

Samples:

question	answer
`is toprol xl the same as metoprolol?`	`Metoprolol succinate is also known by the brand name Toprol XL. It is the extended-release form of metoprolol. Metoprolol succinate is approved to treat high blood pressure, chronic chest pain, and congestive heart failure.`
`are you experienced cd steve hoffman?`	`The Are You Experienced album was apparently mastered from the original stereo UK master tapes (according to Steve Hoffman - one of the very few who has heard both the master tapes and the CDs produced over the years). ... The CD booklets were a little sparse, but at least they stayed true to the album's original design.`
`how are babushka dolls made?`	`Matryoshka dolls are made of wood from lime, balsa, alder, aspen, and birch trees; lime is probably the most common wood type. ... After cutting, the trees are stripped of most of their bark, although a few inner rings of bark are left to bind the wood and keep it from splitting.`

Loss: CachedMultipleNegativesRankingLoss with these parameters:

{
    "scale": 20.0,
    "similarity_fct": "cos_sim"
}

sentence-transformers/natural-questions

Dataset: sentence-transformers/natural-questions at f9e894e
Size: 100,231 training samples
Columns: query and answer
Approximate statistics based on the first 1000 samples:
query answer
type string string
details
min: 10 tokens
mean: 12.47 tokens
max: 23 tokens

min: 17 tokens
mean: 138.32 tokens
max: 556 tokens

	query	answer
type	string	string
details	min: 10 tokens mean: 12.47 tokens max: 23 tokens	min: 17 tokens mean: 138.32 tokens max: 556 tokens

Samples:

query	answer
`when did richmond last play in a preliminary final`	Richmond Football Club Richmond began 2017 with 5 straight wins, a feat it had not achieved since 1995. A series of close losses hampered the Tigers throughout the middle of the season, including a 5-point loss to the Western Bulldogs, 2-point loss to Fremantle, and a 3-point loss to the Giants. Richmond ended the season strongly with convincing victories over Fremantle and St Kilda in the final two rounds, elevating the club to 3rd on the ladder. Richmond's first final of the season against the Cats at the MCG attracted a record qualifying final crowd of 95,028; the Tigers won by 51 points. Having advanced to the first preliminary finals for the first time since 2001, Richmond defeated Greater Western Sydney by 36 points in front of a crowd of 94,258 to progress to the Grand Final against Adelaide, their first Grand Final appearance since 1982. The attendance was 100,021, the largest crowd to a grand final since 1986. The Crows led at quarter time and led by as many as 13, but the Tig...
`who sang what in the world's come over you`	Jack Scott (singer) At the beginning of 1960, Scott again changed record labels, this time to Top Rank Records.[1] He then recorded four Billboard Hot 100 hits – "What in the World's Come Over You" (#5), "Burning Bridges" (#3) b/w "Oh Little One" (#34), and "It Only Happened Yesterday" (#38).[1] "What in the World's Come Over You" was Scott's second gold disc winner.[6] Scott continued to record and perform during the 1960s and 1970s.[1] His song "You're Just Gettin' Better" reached the country charts in 1974.[1] In May 1977, Scott recorded a Peel session for BBC Radio 1 disc jockey, John Peel.
`who produces the most wool in the world`	Wool Global wool production is about 2 million tonnes per year, of which 60% goes into apparel. Wool comprises ca 3% of the global textile market, but its value is higher owing to dying and other modifications of the material.[1] Australia is a leading producer of wool which is mostly from Merino sheep but has been eclipsed by China in terms of total weight.[30] New Zealand (2016) is the third-largest producer of wool, and the largest producer of crossbred wool. Breeds such as Lincoln, Romney, Drysdale, and Elliotdale produce coarser fibers, and wool from these sheep is usually used for making carpets.

Loss: CachedMultipleNegativesRankingLoss with these parameters:

{
    "scale": 20.0,
    "similarity_fct": "cos_sim"
}

merged-2l-nli

Dataset: merged-2l-nli at af845c6
Size: 425,243 training samples
Columns: sentence1, sentence2, and label
Approximate statistics based on the first 1000 samples:
sentence1 sentence2 label
type string string int
details
min: 4 tokens
mean: 83.27 tokens
max: 1202 tokens

min: 5 tokens
mean: 16.7 tokens
max: 126 tokens

0: ~52.90%
1: ~47.10%

	sentence1	sentence2	label
type	string	string	int
details	min: 4 tokens mean: 83.27 tokens max: 1202 tokens	min: 5 tokens mean: 16.7 tokens max: 126 tokens	0: ~52.90% 1: ~47.10%

Samples:

sentence1	sentence2	label
`In 1783 , the Sunni Al-Khalifa family captured Bahrain from the Persians .`	`The is a geographical/political entity`	`0`
::stage Egg:: Newt eggs are encased in a gel-like substance rather than a hard shell. Adult females release eggs one at a time and store them in clusters ranging from a handful to several dozen in size. Adults often take an active role in defending their eggs after depositing them. Mothers may curl their body around the eggs to provide protection. Some newt species even wrap leaves around each egg individually to camouflage them, according to San Diego Zoo. Newt eggs are small: some measure only a millimeter or two in diameter. Mom usually anchors her eggs to underwater plants and other structures to keep them safe. ::stage Tadpole:: Newts that hatch from submerged eggs usually emerge as aquatic larvae with fishlike tails and gills that allow them to breathe beneath the water's surface. Not all newt species have an aquatic or 'tadpole' phase. This tadpole stage tends to be short, except in fully aquatic species. Eastern newt (Notophthalmus viridescens) larvae spend only a few months as...	`Tadpole thing is a newt's terrestrial larval phase known as.`	`0`
`Target You are now a valid target, you nasty little shit! 86.176.169.49`	`This example wikipedia comment contains an insult.`	`1`

Loss: SoftmaxLoss

merged-3l-nli

Dataset: merged-3l-nli at e311b1f
Size: 564,204 training samples
Columns: sentence1, sentence2, and label
Approximate statistics based on the first 1000 samples:
sentence1 sentence2 label
type string string int
details
min: 5 tokens
mean: 110.76 tokens
max: 2048 tokens

min: 5 tokens
mean: 28.37 tokens
max: 485 tokens

0: ~36.00%
1: ~32.70%
2: ~31.30%

	sentence1	sentence2	label
type	string	string	int
details	min: 5 tokens mean: 110.76 tokens max: 2048 tokens	min: 5 tokens mean: 28.37 tokens max: 485 tokens	0: ~36.00% 1: ~32.70% 2: ~31.30%

Samples:

sentence1	sentence2	label
`Iceland does not have a high latitude.`	`Iceland . Iceland is warmed by the Gulf Stream and has a temperate climate , despite a high latitude just outside the Arctic Circle . Its high latitude and marine influence still keeps summers chilly , with most of the archipelago having a tundra climate .`	`2`
`The populist, by contrast, panders to his audience, figuring out what it likes and then delivering it in heaps.`	`Populists hate the audience and antagonizes them; so their support, as a tyrant's, is similarly lacking.`	`2`
`The prison sentence of that convict will end after 2 months.`	`Before 212 days, the prison sentence of that convict will end.`	`1`

Loss: SoftmaxLoss

zero-shot-label-nli

Dataset: zero-shot-label-nli at b363c89
Size: 1,090,333 training samples
Columns: label, sentence1, and sentence2
Approximate statistics based on the first 1000 samples:
label sentence1 sentence2
type int string string
details
0: ~50.20%
2: ~49.80%

min: 3 tokens
mean: 66.27 tokens
max: 2048 tokens

min: 7 tokens
mean: 8.07 tokens
max: 17 tokens

	label	sentence1	sentence2
type	int	string	string
details	0: ~50.20% 2: ~49.80%	min: 3 tokens mean: 66.27 tokens max: 2048 tokens	min: 7 tokens mean: 8.07 tokens max: 17 tokens

Samples:

label	sentence1	sentence2
`0`	`The LAY MAN! Just to let you know you are missed and thought off. Do have a great day. And if you can send me bimbo and ugo's numbers, ill appreciate. Safe`	`This example is ham.`
`2`	`Crisp: oh really!!!!`	`This example is Automotive.`
`2`	`Insurance policies should be simple .`	`This example is negative.`

Loss: SoftmaxLoss

dataset_train_nli

Dataset: dataset_train_nli at 1e00964
Size: 1,018,733 training samples
Columns: sentence1, sentence2, and label
Approximate statistics based on the first 1000 samples:
sentence1 sentence2 label
type string string int
details
min: 4 tokens
mean: 95.56 tokens
max: 1152 tokens

min: 8 tokens
mean: 14.05 tokens
max: 38 tokens

0: ~50.60%
1: ~49.40%

	sentence1	sentence2	label
type	string	string	int
details	min: 4 tokens mean: 95.56 tokens max: 1152 tokens	min: 8 tokens mean: 14.05 tokens max: 38 tokens	0: ~50.60% 1: ~49.40%

Samples:

sentence1	sentence2	label
`where is nayagara falls located`	`The example utterance is a query about music.`	`1`
`Druyun gets nine-month prison sentence A former top Air Force acquisition executive today was sentenced to nine months in prison for conspiring to help Boeing Co. win a multibillion-dollar Pentagon contract.`	`This example news text is about world news`	`1`
`Writing on the #39;wall #39; n Last edition of the Far Eastern Economic Review is shown on the streets of Hong Kong. The weekly news magazine is to fold in its current form with the loss of 80 jobs, the magazine #39;s publisher Dow Jones said yesterday.`	`This example news text is about science and technology`	`1`

Loss: SoftmaxLoss

paws/labeled_final

Dataset: paws/labeled_final at 161ece9
Size: 49,401 training samples
Columns: sentence1, sentence2, and label
Approximate statistics based on the first 1000 samples:
sentence1 sentence2 label
type string string int
details
min: 10 tokens
mean: 27.44 tokens
max: 51 tokens

min: 8 tokens
mean: 27.44 tokens
max: 51 tokens

0: ~55.60%
1: ~44.40%

	sentence1	sentence2	label
type	string	string	int
details	min: 10 tokens mean: 27.44 tokens max: 51 tokens	min: 8 tokens mean: 27.44 tokens max: 51 tokens	0: ~55.60% 1: ~44.40%

Samples:

sentence1	sentence2	label
`In Paris , in October 1560 , he secretly met the English ambassador , Nicolas Throckmorton , asking him for a passport to return to England through Scotland .`	`In October 1560 , he secretly met with the English ambassador , Nicolas Throckmorton , in Paris , and asked him for a passport to return to Scotland through England .`	`0`
`The NBA season of 1975 -- 76 was the 30th season of the National Basketball Association .`	`The 1975 -- 76 season of the National Basketball Association was the 30th season of the NBA .`	`1`
`There are also specific discussions , public profile debates and project discussions .`	`There are also public discussions , profile specific discussions , and project discussions .`	`0`

Loss: SoftmaxLoss

glue/mrpc

Dataset: glue/mrpc at bcdcba7
Size: 3,668 training samples
Columns: sentence1, sentence2, and label
Approximate statistics based on the first 1000 samples:
sentence1 sentence2 label
type string string int
details
min: 10 tokens
mean: 27.55 tokens
max: 49 tokens

min: 12 tokens
mean: 27.25 tokens
max: 48 tokens

0: ~33.70%
1: ~66.30%

	sentence1	sentence2	label
type	string	string	int
details	min: 10 tokens mean: 27.55 tokens max: 49 tokens	min: 12 tokens mean: 27.25 tokens max: 48 tokens	0: ~33.70% 1: ~66.30%

Samples:

sentence1	sentence2	label
`Amrozi accused his brother , whom he called " the witness " , of deliberately distorting his evidence .`	`Referring to him as only " the witness " , Amrozi accused his brother of deliberately distorting his evidence .`	`1`
`Yucaipa owned Dominick 's before selling the chain to Safeway in 1998 for $ 2.5 billion .`	`Yucaipa bought Dominick 's in 1995 for $ 693 million and sold it to Safeway for $ 1.8 billion in 1998 .`	`0`
`They had published an advertisement on the Internet on June 10 , offering the cargo for sale , he added .`	`On June 10 , the ship 's owners had published an advertisement on the Internet , offering the explosives for sale .`	`1`

Loss: SoftmaxLoss

glue/qqp

Dataset: glue/qqp at bcdcba7
Size: 363,846 training samples
Columns: sentence1, sentence2, and label
Approximate statistics based on the first 1000 samples:
sentence1 sentence2 label
type string string int
details
min: 4 tokens
mean: 15.63 tokens
max: 50 tokens

min: 6 tokens
mean: 15.55 tokens
max: 76 tokens

0: ~62.70%
1: ~37.30%

	sentence1	sentence2	label
type	string	string	int
details	min: 4 tokens mean: 15.63 tokens max: 50 tokens	min: 6 tokens mean: 15.55 tokens max: 76 tokens	0: ~62.70% 1: ~37.30%

Samples:

sentence1	sentence2	label
`How can I stop my laptop from hibernating in windows 10?`	`How do I shutdown windows 10 instead of hibernating it?`	`0`
`Is it worth the cost if ever I fix my gap teeth?`	`Is it worth it to fix teeth gap?`	`1`
`Why is USA the biggest threat to the global economy and Germany is not?`	`What is the biggest threat to the global economy over the next year (in 2011)?`	`0`

Loss: SoftmaxLoss

fever-evidence-related

Dataset: fever-evidence-related at 14aba00
Size: 403,218 training samples
Columns: sentence1, sentence2, and label
Approximate statistics based on the first 1000 samples:
sentence1 sentence2 label
type string string int
details
min: 7 tokens
mean: 13.58 tokens
max: 59 tokens

min: 33 tokens
mean: 344.03 tokens
max: 2048 tokens

0: ~32.00%
1: ~68.00%

	sentence1	sentence2	label
type	string	string	int
details	min: 7 tokens mean: 13.58 tokens max: 59 tokens	min: 33 tokens mean: 344.03 tokens max: 2048 tokens	0: ~32.00% 1: ~68.00%

Samples:

sentence1	sentence2	label
`The Last of Us Part II had the developer Naughty Dog.`	Bishop Asbury Cottage is a 17th-century cottage on Newton Road , Great Barr , England , known for being the boyhood home of Francis Asbury -LRB- 1745 -- 1816 -RRB- , one of the first two bishops of the Methodist Episcopal Church -LRB- now The United Methodist Church -RRB- in the United States .. Cottage. Cottage. Great Barr. Great Barr. England. England. Francis Asbury. Francis Asbury. bishops. Bishop ( Methodism ). Methodist Episcopal Church. Methodist Episcopal Church. The United Methodist Church. The United Methodist Church. It is now a museum in his memory .	`1`
`Boomerang (1992 film) was released on July.`	`Petr Alekseyevich Bezobrazov -LRB- 29 January 1845 -- 17 July 1906 -RRB- was an admiral in the Imperial Russian Navy .. Imperial Russian Navy. Imperial Russian Navy`	`1`
`G-Dragon was the first Korean solo artist to a type of tour.`	`The Scott Viking 2 was the first British high performance two seat sailplane , flying a few days before the outbreak of World War II .. World War II. World War II. Only one was built ; it was used in radar station trials in the Summer of 1940 .`	`1`

Loss: SoftmaxLoss

glue/stsb

Dataset: glue/stsb at bcdcba7
Size: 5,749 training samples
Columns: sentence1, sentence2, and label
Approximate statistics based on the first 1000 samples:
sentence1 sentence2 label
type string string float
details
min: 6 tokens
mean: 15.0 tokens
max: 48 tokens

min: 6 tokens
mean: 15.02 tokens
max: 51 tokens

min: 0.0
mean: 2.73
max: 5.0

	sentence1	sentence2	label
type	string	string	float
details	min: 6 tokens mean: 15.0 tokens max: 48 tokens	min: 6 tokens mean: 15.02 tokens max: 51 tokens	min: 0.0 mean: 2.73 max: 5.0

Samples:

sentence1	sentence2	label
`Syria peace plan conditions “unacceptable,” opposition says`	`Syria peace dashed as deadline passes`	`2.0`
`Romney picks Ryan as vice presidential running mate: source`	`Romney to tap Ryan as vice presidential running mate`	`5.0`
`Death toll rises to 6 as Storm Xaver batters northern Europe`	`Storm death toll rises as wind, rain batters north. Europe`	`3.200000047683716`

Loss: CosineSimilarityLoss with these parameters:

{
    "loss_fct": "torch.nn.modules.loss.MSELoss"
}

sick/relatedness

Dataset: sick/relatedness
Size: 4,439 training samples
Columns: sentence1, sentence2, and label
Approximate statistics based on the first 1000 samples:
sentence1 sentence2 label
type string string float
details
min: 6 tokens
mean: 12.08 tokens
max: 30 tokens

min: 6 tokens
mean: 11.86 tokens
max: 23 tokens

min: 1.0
mean: 3.5
max: 5.0

	sentence1	sentence2	label
type	string	string	float
details	min: 6 tokens mean: 12.08 tokens max: 30 tokens	min: 6 tokens mean: 11.86 tokens max: 23 tokens	min: 1.0 mean: 3.5 max: 5.0

Samples:

sentence1	sentence2	label
`The man is standing on a rocky mountain and gray clouds are in the background`	`A black topless person is packing a pile of rocks and a front of clouds are in the background`	`2.9000000953674316`
`A man is standing on a dirt hill next to a black jeep`	`A man in a hat is standing outside of a green vehicle`	`2.5999999046325684`
`A man is talking on a cell phone`	`A man is making a phone call`	`4.300000190734863`

Loss: CosineSimilarityLoss with these parameters:

{
    "loss_fct": "torch.nn.modules.loss.MSELoss"
}

sts-companion

Dataset: sts-companion at fd8beff
Size: 5,289 training samples
Columns: label, sentence1, and sentence2
Approximate statistics based on the first 1000 samples:
label sentence1 sentence2
type float string string
details
min: 0.0
mean: 3.02
max: 5.0

min: 5 tokens
mean: 17.69 tokens
max: 60 tokens

min: 6 tokens
mean: 16.26 tokens
max: 51 tokens

	label	sentence1	sentence2
type	float	string	string
details	min: 0.0 mean: 3.02 max: 5.0	min: 5 tokens mean: 17.69 tokens max: 60 tokens	min: 6 tokens mean: 16.26 tokens max: 51 tokens

Samples:

label	sentence1	sentence2
`4.25`	`It then appointed a task force to formulate the necessary changes in tax and spending policies.`	`He has appointed a working party to make the necessary changes to the policies of public spending and fiscal policies.`
`4.25`	`festive social event, celebration`	`an occasion on which people can assemble for social interaction and entertainment.`
`3.6`	`Who'd have thought an American hero could be a Canadian? NYT: Man Who Sheltered Americans in Tehran, Dies at 88`	`John Sheardown, Canadian Who Sheltered Americans in Tehran, Dies at 88`

Loss: CosineSimilarityLoss with these parameters:

{
    "loss_fct": "torch.nn.modules.loss.MSELoss"
}

Evaluation Datasets

merged-2l-nli

Dataset: merged-2l-nli at af845c6
Size: 4,053 evaluation samples
Columns: sentence1, sentence2, and label
Approximate statistics based on the first 1000 samples:
sentence1 sentence2 label
type string string int
details
min: 6 tokens
mean: 75.82 tokens
max: 1219 tokens

min: 5 tokens
mean: 13.5 tokens
max: 158 tokens

0: ~51.00%
1: ~49.00%

	sentence1	sentence2	label
type	string	string	int
details	min: 6 tokens mean: 75.82 tokens max: 1219 tokens	min: 5 tokens mean: 13.5 tokens max: 158 tokens	0: ~51.00% 1: ~49.00%

Samples:

sentence1	sentence2	label
`What happens to the norm when a number is multiplied by p?`	`While completing Q (roughly, filling the gaps) with respect to the absolute value yields the field of real numbers, completing with respect to the p-adic norm`	−
`The abode of the Greek gods was on the summit of Mount Olympus, in Thessaly.`	`Mount Olympus is in Thessaly.`	`1`
`The drain is clogged with hair. It has to be cleaned.`	`The hair has to be cleaned.`	`0`

Loss: SoftmaxLoss

merged-3l-nli

Dataset: merged-3l-nli at e311b1f
Size: 2,872 evaluation samples
Columns: sentence1, sentence2, and label
Approximate statistics based on the first 1000 samples:
sentence1 sentence2 label
type string string int
details
min: 6 tokens
mean: 167.67 tokens
max: 2048 tokens

min: 2 tokens
mean: 23.6 tokens
max: 430 tokens

0: ~38.20%
1: ~31.30%
2: ~30.50%

	sentence1	sentence2	label
type	string	string	int
details	min: 6 tokens mean: 167.67 tokens max: 2048 tokens	min: 2 tokens mean: 23.6 tokens max: 430 tokens	0: ~38.20% 1: ~31.30% 2: ~30.50%

Samples:

sentence1	sentence2	label
`But if Congress opts for debt over taxation, you can count on thoughtless commentators to denounce the interest payments on that debt as a second, and separate, outrage.`	`Everybody considers the interest on the national debt an outrage.`	`1`
`The 1997 KNVB Cup Final was a football match between Roda JC and Heerenveen on 8 May 1997 at De Kuip, Rotterdam. It was the final match of the 1996–97 KNVB Cup competition and the 79th KNVB Cup final. Roda won 4–2 after goals from Gerald Sibon, Ger Senden, Eric van der Luer and Maarten Schops. It was the side's first KNVB Cup trophy.`	`Roda JC kept the Cup trophy at their headquarters.`	`1`
`Discover Financial Services, Inc. is an American financial services company, which issues the Discover Card and operates the Discover and Pulse networks, and owns Diners Club International. Discover Card is the third largest credit card brand in the United States, when measured by cards in force, with nearly 50 million cardholders.`	`Discover Card is a way to build credit for less than 50 million cardholders`	`0`

Loss: SoftmaxLoss

zero-shot-label-nli

Dataset: zero-shot-label-nli at b363c89
Size: 14,419 evaluation samples
Columns: label, sentence1, and sentence2
Approximate statistics based on the first 1000 samples:
label sentence1 sentence2
type int string string
details
0: ~51.40%
2: ~48.60%

min: 3 tokens
mean: 72.92 tokens
max: 2048 tokens

min: 7 tokens
mean: 8.01 tokens
max: 23 tokens

	label	sentence1	sentence2
type	int	string	string
details	0: ~51.40% 2: ~48.60%	min: 3 tokens mean: 72.92 tokens max: 2048 tokens	min: 7 tokens mean: 8.01 tokens max: 23 tokens

Samples:

label	sentence1	sentence2
`2`	`Police suspected that Shaichat , 20 , had been abducted either by Palestinians or by Israeli Arabs . Nobody claimed responsibility for Schaichat 's death , but police suspect that the 20-year-old soldier was abducted either by Palestinians or Israeli Arabs .`	`This example is equivalent.`
`2`	`Can immorality be achieved by blocking death genes? Can immortality be achieved by blocking death genes?`	`This example is not_duplicate.`
`2`	`can a minor sit at a bar in nj`	`This example is False.`

Loss: SoftmaxLoss

dataset_train_nli

Dataset: dataset_train_nli at 1e00964
Size: 1,018,733 evaluation samples
Columns: sentence1, sentence2, and label
Approximate statistics based on the first 1000 samples:
sentence1 sentence2 label
type string string int
details
min: 4 tokens
mean: 96.94 tokens
max: 1020 tokens

min: 8 tokens
mean: 13.86 tokens
max: 38 tokens

0: ~52.50%
1: ~47.50%

	sentence1	sentence2	label
type	string	string	int
details	min: 4 tokens mean: 96.94 tokens max: 1020 tokens	min: 8 tokens mean: 13.86 tokens max: 38 tokens	0: ~52.50% 1: ~47.50%

Samples:

sentence1	sentence2	label
Ecoplug MAX® ECOPLUG MAX® is an efficient method to prevent regroth from leaf trees. - Provides 100 percent effective on all brushwood - Can be used all year round - Kills all unwanted leaf tree - Minimizes chemical diffusion - Kills the entire root system of the treated tree/stump - Fully selective method reduce chemical use up to 90% compared to previously used methods. - Can be used all year around. - Will exterminate: Alder, elm, aspen, birch, beech, lime, maple, mountain ash,sallow, poplar, ash, cherry, bird cherry, oak and more broad leafed trees - Minimize the use of chemicals during treatment of trees and stumps. - The product will kill off the entire root system, but only the root system. Neither people, animals or the enviromnent will be exposed to our product..	`This text is about: root extermination`	`0`
`can you start f. m. eight hundred and ninety radio channel`	`The intent of this example utterance is a datetime query.`	`1`
`never again swings between false sentiment and unfunny madcap comedy and , along the way , expects the audience to invest in the central relationship as some kind of marriage of true minds .`	`The sentiment in this example rotten tomatoes movie review is negative`	`0`

Loss: SoftmaxLoss

paws/labeled_final

Dataset: paws/labeled_final at 161ece9
Size: 8,000 evaluation samples
Columns: sentence1, sentence2, and label
Approximate statistics based on the first 1000 samples:
sentence1 sentence2 label
type string string int
details
min: 9 tokens
mean: 27.86 tokens
max: 51 tokens

min: 9 tokens
mean: 27.83 tokens
max: 52 tokens

0: ~54.90%
1: ~45.10%

	sentence1	sentence2	label
type	string	string	int
details	min: 9 tokens mean: 27.86 tokens max: 51 tokens	min: 9 tokens mean: 27.83 tokens max: 52 tokens	0: ~54.90% 1: ~45.10%

Samples:

sentence1	sentence2	label
`Bradd Crellin represented BARLA Cumbria on a tour of Australia with 6 other players representing Britain , also on a tour of Australia .`	`Bradd Crellin also represented BARLA Great Britain on a tour through Australia on a tour through Australia with 6 other players representing Cumbria .`	`0`
`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 .`	`1`
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 .	`1`

Loss: SoftmaxLoss

glue/mrpc

Dataset: glue/mrpc at bcdcba7
Size: 408 evaluation samples
Columns: sentence1, sentence2, and label
Approximate statistics based on the first 408 samples:
sentence1 sentence2 label
type string string int
details
min: 14 tokens
mean: 27.92 tokens
max: 46 tokens

min: 11 tokens
mean: 27.24 tokens
max: 43 tokens

0: ~31.62%
1: ~68.38%

	sentence1	sentence2	label
type	string	string	int
details	min: 14 tokens mean: 27.92 tokens max: 46 tokens	min: 11 tokens mean: 27.24 tokens max: 43 tokens	0: ~31.62% 1: ~68.38%

Samples:

sentence1	sentence2	label
`He said the foodservice pie business doesn 't fit the company 's long-term growth strategy .`	`" The foodservice pie business does not fit our long-term growth strategy .`	`1`
`Magnarelli said Racicot hated the Iraqi regime and looked forward to using his long years of training in the war .`	`His wife said he was " 100 percent behind George Bush " and looked forward to using his years of training in the war .`	`0`
`The dollar was at 116.92 yen against the yen , flat on the session , and at 1.2891 against the Swiss franc , also flat .`	`The dollar was at 116.78 yen JPY = , virtually flat on the session , and at 1.2871 against the Swiss franc CHF = , down 0.1 percent .`	`0`

Loss: SoftmaxLoss

glue/qqp

Dataset: glue/qqp at bcdcba7
Size: 40,430 evaluation samples
Columns: sentence1, sentence2, and label
Approximate statistics based on the first 1000 samples:
sentence1 sentence2 label
type string string int
details
min: 3 tokens
mean: 15.77 tokens
max: 48 tokens

min: 6 tokens
mean: 16.05 tokens
max: 69 tokens

0: ~62.00%
1: ~38.00%

	sentence1	sentence2	label
type	string	string	int
details	min: 3 tokens mean: 15.77 tokens max: 48 tokens	min: 6 tokens mean: 16.05 tokens max: 69 tokens	0: ~62.00% 1: ~38.00%

Samples:

sentence1	sentence2	label
`What happens to a question on Quora if it is marked as needing further improvement?`	`If Quora doesn't understand my question and marks it as needing improvement, can others still see it?`	`1`
`What does the open blue circle in Facebook Messenger mean?`	`"what does ""delivered"" mean on Facebook messenger?"`	`0`
`How do I cool my mind?`	`What is the best way to be cool?`	`0`

Loss: SoftmaxLoss

fever-evidence-related

Dataset: fever-evidence-related at 14aba00
Size: 54,578 evaluation samples
Columns: sentence1, sentence2, and label
Approximate statistics based on the first 1000 samples:
sentence1 sentence2 label
type string string int
details
min: 7 tokens
mean: 13.66 tokens
max: 26 tokens

min: 33 tokens
mean: 282.39 tokens
max: 1713 tokens

0: ~28.10%
1: ~71.90%

	sentence1	sentence2	label
type	string	string	int
details	min: 7 tokens mean: 13.66 tokens max: 26 tokens	min: 33 tokens mean: 282.39 tokens max: 1713 tokens	0: ~28.10% 1: ~71.90%

Samples:

sentence1	sentence2	label
`Colin Kaepernick became a starting quarterback during the 49ers 63rd season in the National Football League.`	RapidAdvance is a technology-powered financial services company that provides working capital to small and mid-sized businesses in the United States .. United States. United States. financial services. financial services. working capital. working capital. small and mid-sized businesses. Small and medium-sized enterprises. It offers small business loan programs for business owners in a variety of industries , including traditional retail establishments , brand name chain restaurants , automotive repair , manufacturing , trucking , and professional service providers .. Founded in 2005 and headquartered in Bethesda , Maryland , the company was acquired by Dan Gilbert 's Rockbridge Growth Equity , LLC in 2013 .. It is part of Rock Ventures `` family '' of companies that include the Cleveland Cavaliers , Fathead , Quicken Loans and Genius .. Rock Ventures. Rock Ventures. Cleveland Cavaliers. Cleveland Cavaliers. Fathead. Fathead ( brand ). Quicken Loans. Quicken Loans. Genius. Genius	`1`
`Colin Kaepernick became a starting quarterback during the 49ers 63rd season in the National Football League.`	Arthur Herbert Copeland -LRB- June 22 , 1898 Rochester , New York -- July 6 , 1970 -RRB- was an American mathematician .. American. United States. He graduated from Harvard University in 1926 and taught at Rice University and the University of Michigan .. Rice University. Rice University. University of Michigan. University of Michigan. Harvard University. Harvard University. His main interest was in the foundations of probability .. probability. probability theory. He worked with Paul Erdos on the Copeland-Erdos constant .. Copeland-Erdos constant. Copeland-Erdos constant. Paul Erdos. Paul Erdos. His son , Arthur Herbert Copeland , Jr. , is also a mathematician .	`1`
`Tilda Swinton is a vegan.`	Michael Ronald Taylor -LRB- 1 June 1938 , Ealing , West London - 19 January 1969 -RRB- was a British jazz composer , pianist and co-songwriter for the band Cream .. Ealing. Ealing. London. London. British. United Kingdom. Cream. Cream ( band ). Mike Taylor was brought up by his grandparents in London and Kent , and joined the RAF for his national service .. London. London. Having rehearsed and written extensively throughout the early 1960s , he recorded two albums for the Lansdowne series produced by Denis Preston : Pendulum -LRB- 1966 -RRB- with drummer Jon Hiseman , bassist Tony Reeves and saxophonist Dave Tomlin -RRB- and Trio -LRB- 1967 -RRB- with Hiseman and bassists Jack Bruce and Ron Rubin .. Denis Preston. Denis Preston. Jon Hiseman. Jon Hiseman. Dave Tomlin. Dave Tomlin ( musician ). Jack Bruce. Jack Bruce. They were issued on UK Columbia .. Columbia. Columbia Graphophone Company. During his brief recording career , several of Taylor 's pieces were played and recorded by his ...	`1`

Loss: SoftmaxLoss

glue/stsb

Dataset: glue/stsb at bcdcba7
Size: 1,500 evaluation samples
Columns: sentence1, sentence2, and label
Approximate statistics based on the first 1000 samples:
sentence1 sentence2 label
type string string float
details
min: 5 tokens
mean: 16.46 tokens
max: 44 tokens

min: 6 tokens
mean: 16.47 tokens
max: 44 tokens

min: 0.0
mean: 2.35
max: 5.0

	sentence1	sentence2	label
type	string	string	float
details	min: 5 tokens mean: 16.46 tokens max: 44 tokens	min: 6 tokens mean: 16.47 tokens max: 44 tokens	min: 0.0 mean: 2.35 max: 5.0

Samples:

sentence1	sentence2	label
`The room used for defecation is almost always referred to by euphemism.`	`I'm English, and would probably use 'toilet' most of the time, and always in the context of a private home.`	`1.600000023841858`
`The two-year note US2YT=RR fell 5/32 in price, taking its yield to 1.23 percent from 1.16 percent late on Monday.`	`The benchmark 10-year note US10YT=RR lost 11/32 in price, taking its yield to 3.21 percent from 3.17 percent late on Monday.`	`2.0`
`I use Elinchrom Skyports, but if money is not an issue then go for PocketWizards.`	`Or just go with the ultra-cheap YongNuo RF-602, which give you a lot of bang for the buck.`	`1.2000000476837158`

Loss: CosineSimilarityLoss with these parameters:

{
    "loss_fct": "torch.nn.modules.loss.MSELoss"
}

sick/relatedness

Dataset: sick/relatedness
Size: 495 evaluation samples
Columns: sentence1, sentence2, and label
Approximate statistics based on the first 495 samples:
sentence1 sentence2 label
type string string float
details
min: 6 tokens
mean: 12.69 tokens
max: 32 tokens

min: 6 tokens
mean: 12.15 tokens
max: 30 tokens

min: 1.0
mean: 3.59
max: 5.0

Samples:

sentence1	sentence2	label
`The young boys are playing outdoors and the man is smiling nearby`	`There is no boy playing outdoors and there is no man smiling`	`3.5999999046325684`
`A person in a black jacket is doing tricks on a motorbike`	`A skilled person is riding a bicycle on one wheel`	`3.4000000953674316`
`Four children are doing backbends in the gym`	`Four girls are doing backbends and playing outdoors`	`3.799999952316284`

Loss: CosineSimilarityLoss with these parameters:

{
    "loss_fct": "torch.nn.modules.loss.MSELoss"
}

sts-companion

Dataset: sts-companion at fd8beff
Size: 5,289 evaluation samples
Columns: label, sentence1, and sentence2
Approximate statistics based on the first 1000 samples:
label sentence1 sentence2
type float string string
details
min: 0.0
mean: 3.24
max: 5.0

min: 5 tokens
mean: 19.56 tokens
max: 81 tokens

min: 5 tokens
mean: 17.21 tokens
max: 72 tokens

Samples:

label	sentence1	sentence2
`3.8`	`After all, it is by no means certain that the proposed definition of equitable price is better than any other, because the various definitions that are currently in use in the Member States are all perfectly satisfactory.`	`In fact, it is not absolutely certain that the definition of price that is proposed is better than another, because the different currently in the Member States all fully.`
`2.0`	`rslw: no, why would i hate them?`	`why do you hate america so much?`
`3.0`	`Families of #Newtown Victims Look for Answers on #Gun Violence #NRA`	`Families of Newtown School Massacre Victims Organize Against Gun Violence`

Loss: CosineSimilarityLoss with these parameters:

{
    "loss_fct": "torch.nn.modules.loss.MSELoss"
}

Training Hyperparameters

Non-Default Hyperparameters

per_device_train_batch_size: 24
learning_rate: 2e-05
weight_decay: 1e-06
num_train_epochs: 1
warmup_ratio: 0.1
fp16: True

All Hyperparameters

Click to expand

overwrite_output_dir: False
do_predict: False
eval_strategy: no
prediction_loss_only: True
per_device_train_batch_size: 24
per_device_eval_batch_size: 8
per_gpu_train_batch_size: None
per_gpu_eval_batch_size: None
gradient_accumulation_steps: 1
eval_accumulation_steps: None
torch_empty_cache_steps: None
learning_rate: 2e-05
weight_decay: 1e-06
adam_beta1: 0.9
adam_beta2: 0.999
adam_epsilon: 1e-08
max_grad_norm: 1.0
num_train_epochs: 1
max_steps: -1
lr_scheduler_type: linear
lr_scheduler_kwargs: {}
warmup_ratio: 0.1
warmup_steps: 0
log_level: passive
log_level_replica: warning
log_on_each_node: True
logging_nan_inf_filter: True
save_safetensors: True
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: False
resume_from_checkpoint: None
hub_model_id: None
hub_strategy: every_save
hub_private_repo: None
hub_always_push: False
gradient_checkpointing: False
gradient_checkpointing_kwargs: None
include_inputs_for_metrics: False
include_for_metrics: []
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
use_liger_kernel: False
eval_use_gather_object: False
average_tokens_across_devices: False
prompts: None
batch_sampler: batch_sampler
multi_dataset_batch_sampler: proportional

Training Logs

Epoch	Step	Training Loss
0.0067	500	10.6192
0.0134	1000	1.9196
0.0202	1500	1.0304
0.0269	2000	0.9269
0.0336	2500	0.7738
0.0403	3000	0.7092
0.0471	3500	0.6571
0.0538	4000	0.6408
0.0605	4500	0.6348
0.0672	5000	0.5927
0.0739	5500	0.5848
0.0807	6000	0.5542
0.0874	6500	0.558
0.0941	7000	0.5394
0.1008	7500	0.5632
0.1076	8000	0.5037
0.1143	8500	0.5278

Framework Versions

Python: 3.11.4
Sentence Transformers: 3.3.1
Transformers: 4.48.0.dev0
PyTorch: 2.4.0+cu121
Accelerate: 1.0.1
Datasets: 2.20.0
Tokenizers: 0.21.0

Citation

BibTeX

Sentence Transformers and SoftmaxLoss

@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",
}

MultipleNegativesRankingLoss

@misc{henderson2017efficient,
    title={Efficient Natural Language Response Suggestion for Smart Reply},
    author={Matthew Henderson and Rami Al-Rfou and Brian Strope and Yun-hsuan Sung and Laszlo Lukacs and Ruiqi Guo and Sanjiv Kumar and Balint Miklos and Ray Kurzweil},
    year={2017},
    eprint={1705.00652},
    archivePrefix={arXiv},
    primaryClass={cs.CL}
}

CachedMultipleNegativesRankingLoss

@misc{gao2021scaling,
    title={Scaling Deep Contrastive Learning Batch Size under Memory Limited Setup},
    author={Luyu Gao and Yunyi Zhang and Jiawei Han and Jamie Callan},
    year={2021},
    eprint={2101.06983},
    archivePrefix={arXiv},
    primaryClass={cs.LG}
}