last-checkpoint/README.md

metadata

language:
  - en
library_name: sentence-transformers
tags:
  - sentence-transformers
  - sentence-similarity
  - feature-extraction
  - generated_from_trainer
  - dataset_size:689221
  - loss:MultipleNegativesRankingLoss
  - loss:CoSENTLoss
  - loss:GISTEmbedLoss
  - loss:OnlineContrastiveLoss
  - loss:MultipleNegativesSymmetricRankingLoss
base_model: bobox/DeBERTaV3-small-GeneralSentenceTransformer
datasets:
  - sentence-transformers/all-nli
  - sentence-transformers/stsb
  - tals/vitaminc
  - nyu-mll/glue
  - allenai/scitail
  - sentence-transformers/xsum
  - sentence-transformers/sentence-compression
widget:
  - source_sentence: >-
      A man in a Santa Claus costume is sitting on a wooden chair holding a
      microphone and a stringed instrument.
    sentences:
      - The man is is near the ball.
      - The man is wearing a costume.
      - People are having a picnic.
  - source_sentence: A street vendor selling his art.
    sentences:
      - A man is selling things on the street.
      - A woman is walking outside.
      - A clown is talking into a microphone.
  - source_sentence: A boy looks surly as his father looks at the camera.
    sentences:
      - a boy looks at his farther
      - >-
        A dark-haired girl in a spotted shirt is pointing at the picture while
        sitting next to a boy wearing a purple shirt and jeans.
      - Man and woman stop and chat with each other.
  - source_sentence: >-
      Which company provided streetcar connections between downtown and the
      hospital?
    sentences:
      - >-
        In 1914 developers Billings & Meyering acquired the tract, completed
        street development, provided the last of the necessary municipal
        improvements including water service, and began marketing the property
        with fervor.
      - >-
        The war was fought primarily along the frontiers between New France and
        the British colonies, from Virginia in the South to Nova Scotia in the
        North.
      - >-
        On the basis of CST, Burnet developed a theory of how an immune response
        is triggered according to the self/nonself distinction: "self"
        constituents (constituents of the body) do not trigger destructive
        immune responses, while "nonself" entities (pathogens, an allograft)
        trigger a destructive immune response.
  - source_sentence: What language did Tesla study while in school?
    sentences:
      - >-
        Because of the complexity of medications including specific indications,
        effectiveness of treatment regimens, safety of medications (i.e., drug
        interactions) and patient compliance issues (in the hospital and at
        home) many pharmacists practicing in hospitals gain more education and
        training after pharmacy school through a pharmacy practice residency and
        sometimes followed by another residency in a specific area.
      - >-
        Rev. Jimmy Creech was defrocked after a highly publicized church trial
        in 1999 on account of his participation in same-sex union ceremonies.
      - Tesla was the fourth of five children.
pipeline_tag: sentence-similarity

SentenceTransformer based on bobox/DeBERTaV3-small-GeneralSentenceTransformer

This is a sentence-transformers model finetuned from bobox/DeBERTaV3-small-GeneralSentenceTransformer on the nli-pairs, sts-label, vitaminc-pairs, qnli-contrastive, scitail-pairs-qa, scitail-pairs-pos, xsum-pairs and compression-pairs 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/DeBERTaV3-small-GeneralSentenceTransformer
• Maximum Sequence Length: 512 tokens
• Output Dimensionality: 768 tokens
• Similarity Function: Cosine Similarity
• Training Datasets:
  • nli-pairs
  • sts-label
  • vitaminc-pairs
  • qnli-contrastive
  • scitail-pairs-qa
  • scitail-pairs-pos
  • xsum-pairs
  • compression-pairs
• 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/DeBERTaV3-small-GeneralSentenceTransformer-keepTraining-checkpoints-tmp")
# Run inference
sentences = [
    'What language did Tesla study while in school?',
    'Tesla was the fourth of five children.',
    'Rev. Jimmy Creech was defrocked after a highly publicized church trial in 1999 on account of his participation in same-sex union ceremonies.',
]
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

nli-pairs

• Dataset: nli-pairs at d482672
• Size: 150,000 training samples
• Columns: sentence1 and sentence2
• Approximate statistics based on the first 1000 samples:

  	sentence1	sentence2
  type	string	string
  details	min: 5 tokens mean: 16.62 tokens max: 62 tokens	min: 4 tokens mean: 9.46 tokens max: 29 tokens

• Samples:

  sentence1	sentence2
  A person on a horse jumps over a broken down airplane.	A person is outdoors, on a horse.
  Children smiling and waving at camera	There are children present
  A boy is jumping on skateboard in the middle of a red bridge.	The boy does a skateboarding trick.

• Loss: MultipleNegativesRankingLoss with these parameters:

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

sts-label

• Dataset: sts-label at ab7a5ac
• Size: 5,749 training samples
• Columns: sentence1, sentence2, and score
• Approximate statistics based on the first 1000 samples:

  	sentence1	sentence2	score
  type	string	string	float
  details	min: 6 tokens mean: 9.81 tokens max: 27 tokens	min: 5 tokens mean: 9.74 tokens max: 25 tokens	min: 0.0 mean: 0.54 max: 1.0

• Samples:

  sentence1	sentence2	score
  A plane is taking off.	An air plane is taking off.	1.0
  A man is playing a large flute.	A man is playing a flute.	0.76
  A man is spreading shreded cheese on a pizza.	A man is spreading shredded cheese on an uncooked pizza.	0.76

• Loss: CoSENTLoss with these parameters:

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

vitaminc-pairs

• Dataset: vitaminc-pairs at be6febb
• Size: 75,142 training samples
• Columns: label, sentence1, and sentence2
• Approximate statistics based on the first 1000 samples:

  	label	sentence1	sentence2
  type	int	string	string
  details	1: 100.00%	min: 7 tokens mean: 17.36 tokens max: 57 tokens	min: 7 tokens mean: 37.74 tokens max: 224 tokens

• Samples:

  label	sentence1	sentence2
  1	Baron Waddington was the Home Secretary during the Poll Tax Riots .	Baron Waddington of Read ( Home Secretary during the Poll Tax Riots and former Lord Privy Seal and Leader of the House of Lords ) , Phil Willis , Liberal Democrat MP for Harrogate & Knaresborough , Liberal Democrats official site .
  1	Captopril inhibits the catabolism of endogenous opioids and this contributes to its hypotensive action , while naloxone , a mu -opiate antagonist , opposes this .	Captopril inhibits the catabolism of endogenous opioids and this contributes to its hypotensive action , and naloxone a mu -opiate antagonist , opposes this ( 7 ) Goldfrank 's toxicologic emergencies , Lewis R. Goldfrank , Neal Flomenbaum , page 953.Meyler 's Side Effects of Analgesics and Anti-inflammatory Drugs , Jeffrey K. Aronson , page 120.
  1	In under 60 seconds to the end of the first quarter , The Patriots made a more than 79-yard offensive into their opponent 's side .	The Patriots countered on their next drive marching 80 yards culminating with a 4-yard touchdown run by Brady ( in the process Brady tied Curtis Martin 's club record for rushing touchdowns in the playoffs ) with less than a minute remaining in the first quarter .

• Loss: GISTEmbedLoss with these parameters:

  {'guide': SentenceTransformer(
    (0): Transformer({'max_seq_length': 512, 'do_lower_case': True}) with Transformer model: BertModel 
    (1): Pooling({'word_embedding_dimension': 384, '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': 0.05}
  

qnli-contrastive

• Dataset: qnli-contrastive at bcdcba7
• Size: 104,743 training 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: 13.68 tokens max: 36 tokens	min: 7 tokens mean: 34.78 tokens max: 178 tokens	0: 100.00%

• Samples:

  sentence1	sentence2	label
  How many Protestants live in Greece?	Greek citizens who are Roman Catholic are estimated to be at around 50,000 with the Roman Catholic immigrant community in the country approximately 200,000.	0
  What restricted the 1870s Child labour in Australia?	Child labour was restricted by compulsorry schooling.	0
  Whose assumption of power ended the era of stagnation?	Under his rule, the Russian SFSR and the rest of the Soviet Union went through an era of stagnation.	0

• Loss: OnlineContrastiveLoss

scitail-pairs-qa

• Dataset: scitail-pairs-qa at 0cc4353
• Size: 14,987 training samples
• Columns: sentence2 and sentence1
• Approximate statistics based on the first 1000 samples:

  	sentence2	sentence1
  type	string	string
  details	min: 7 tokens mean: 15.72 tokens max: 40 tokens	min: 6 tokens mean: 14.86 tokens max: 41 tokens

• Samples:

  sentence2	sentence1
  Concave lenses can correct myopia.	What type of lenses can correct myopia?
  Bases normally have a bitter taste.	What type of taste do bases normally have?
  An acorn growing into an adult tree will take the longest time to observe.	Which of these will take the LONGEST time to observe?

• Loss: MultipleNegativesRankingLoss with these parameters:

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

scitail-pairs-pos

• Dataset: scitail-pairs-pos at 0cc4353
• Size: 8,600 training samples
• Columns: sentence1 and sentence2
• Approximate statistics based on the first 1000 samples:

  	sentence1	sentence2
  type	string	string
  details	min: 7 tokens mean: 23.51 tokens max: 67 tokens	min: 7 tokens mean: 15.66 tokens max: 39 tokens

• Samples:

  sentence1	sentence2
  One gram of carbohydrate provides four calories.	One gram of carbohydrates provides four calories of energy.
  Our eyes can see a certain spectrum of light, called visible light, starting with red and ending with violet.	Violet and red are two types of visible light.
  Adult-onset Diabetes Former term for noninsulin-dependent or type II diabetes.	Type 2 diabetes is also known as noninsulin-dependent or adult-onset diabetes.

• Loss: MultipleNegativesRankingLoss with these parameters:

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

xsum-pairs

• Dataset: xsum-pairs at 788ddaf
• Size: 150,000 training samples
• Columns: sentence1 and sentence2
• Approximate statistics based on the first 1000 samples:

  	sentence1	sentence2
  type	string	string
  details	min: 38 tokens mean: 350.51 tokens max: 512 tokens	min: 11 tokens mean: 26.76 tokens max: 60 tokens

• Samples:

  sentence1	sentence2
  President Barack Obama has worked to improve relations with the communist government in Havana, culminating in his historic visit in March 2016. The president-elect threatened in a tweet to put an end to the detente following the death of Fidel Castro. But the White House bristled at Mr Trump's warning, saying the president was not concerned about the threat. White House Press Secretary Josh Earnest told reporters that reversing the policy would be "a significant economic blow" to Cubans and was "not as easy as a stroke of a pen". President-elect Trump tweeted he would "terminate" Mr Obama's policy on normalising relations with Cuba as thousands of Cubans queued to pay their respects to , who died on Friday. They gathered in Havana's Revolution Square as part of farewell commemorations which will last until Tuesday night, when foreign leaders are due to arrive in Cuba to pay their respects. A cortege will then transport his ashes east across the island to Santiago de Cuba, reversing the route Castro took during the Cuban revolution. They will be laid to rest on Sunday in the city's Santa Ifigenia cemetery. In his tweet, Mr Trump said that if "Cuba is unwilling to make a better deal for the Cuban people, the Cuban/American people and the US as a whole, I will terminate deal". Mr Trump, who takes office in January, said during the election campaign that he would reverse President Barack Obama's rapprochement with Cuba. Under Mr Obama, diplomatic ties were restored in 2015 after being severed in 1961. Some trade restrictions have been eased and the White House has been lobbying the US Congress to terminate an economic embargo that has been in place for decades. What does a Trump presidency mean for US-Cuba relations? Mr Trump's team has accused the Obama administration of giving too much away to Cuba without receiving enough in return. His communications director, Jason Miller, said Mr Trump was seeking "freedom in Cuba for the Cubans and a good deal for Americans where we aren't played for fools". But the White House said that better ties with Cuba served US interests and that reversing the changes would deal "a significant economic blow" to the people of Cuba. "After five decades of not seeing results, the president believed it was time to try something different," said White House spokesman Josh Earnest. In a separate development the first scheduled commercial flight from the US to Havana in more than 50 years has departed from Miami. The American Airlines (AA) flight on Monday morning was the first of a new service to the Cuban capital which will fly from Florida four times a day. "It's a monumental day of great historic relevance with Miami being the epicentre of the Cuban-American community and American's hub for the region," AA Vice President Ralph Lopez was quoted by The Miami Herald as saying. Several airlines began routes to other parts of Cuba earlier this year, with many more flights and destinations in the offing.	Donald Trump says he will end the thaw between the US and Cuba if the country does not offer a "better deal".
  It will allow members of the public to make complaints about the conduct of Scotland's charities The new measure was recommended by the Fundraising Working Group in Scotland. The Scottish government said it was important confidence was maintained in the country's charities, but stressed that the majority operated to high standards. Communities Secretary Angela Constance said: "Scotland's charities benefit from a great deal of public trust and it's important that that confidence is maintained. "This new phone number and website will be run by Scottish Fundraising Complaints - set up by the Scottish Council for Voluntary Organisations and the Scottish Charity Regulator - and will give people information about how to raise any concerns they may have." Members of the public who are concerned about the fundraising tactics of a charity in Scotland can call 0808 164 2520 or visit the website. Cross-border charities, where charities operate in Scotland but are registered in England and Wales, will continue to be regulated by the Fundraising Regulator.	A new phone number and website set up to protect the public against aggressive fundraisers has gone live.
  Carla Whitlock was sprayed with drain cleaner containing sulphuric acid in Southampton on 18 September. Billy Midmore, 23, who was found guilty of causing grievous bodily harm with intent, was jailed for 15 years with a further five years on licence. Geoffrey Midmore, 27, had previously pleaded guilty to the same charge and was jailed for nine years. Southampton Crown Court heard the acid was thrown in the 37-year-old's face after a drug deal went wrong. Sentencing the pair, Judge Peter Ralls QC said: "Your behaviour displays a level of medieval barbarism that is appalling. "You used a weapon that was pernicious and evil. You planned for this, which adds to the culpability." Billy Midmore, who denied any involvement in the attack, admitted sending Ms Whitlock threatening texts after his brother was robbed of drugs and cash worth £2,000. Brothers jailed for Southampton acid attack - as it happened Prosecutors said he held the mother-of-six responsible after she helped broker the deal between the brothers and a Southampton drug dealer. Train CCTV showed him giving a high-five and fist-bumping a friend hours after the attack. The brothers, from London, were arrested in Gillingham, Kent, on 29 September, following a two-week nationwide manhunt. During his trial, Billy Midmore admitted coming to Southampton from London to sell crack cocaine and heroin. The court also heard Geoffrey Midmore had sent a photograph of the drain cleaner on WhatsApp to an acquaintance, with the words: "This is one face melter." The court heard Ms Whitlock needed surgery after the attack and remained in "significant pain". A consultant said her right eye was prone to "breakdown and infection" and her eyesight was unlikely to return. The judge told the brothers: "You bought the drain cleaner because you intended to pour it in her face. "To describe it as the 'face melter' could only have one interpretation." After sentencing, Ch Insp Debra Masson, of Hampshire Constabulary, described the brothers as "men of violence" who "made people's lives a misery". "The evidence shown in court of their behaviour directly after the attack, coupled with their concerted efforts to evade capture, served to expose them as the dangerous criminals they are," she said. "Although things will never be the same for Carla, we hope that the sentence handed to Geoffrey and Billy Midmore today goes some way to giving her closure and allows her to feel that justice has been delivered."	Two brothers have been jailed for an acid attack which left a woman scarred and blind in one eye.

• Loss: MultipleNegativesSymmetricRankingLoss with these parameters:

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

compression-pairs

• Dataset: compression-pairs at 605bc91
• Size: 180,000 training samples
• Columns: sentence1 and sentence2
• Approximate statistics based on the first 1000 samples:

  	sentence1	sentence2
  type	string	string
  details	min: 10 tokens mean: 31.89 tokens max: 125 tokens	min: 5 tokens mean: 10.21 tokens max: 28 tokens

• Samples:

  sentence1	sentence2
  The USHL completed an expansion draft on Monday as 10 players who were on the rosters of USHL teams during the 2009-10 season were selected by the League's two newest entries, the Muskegon Lumberjacks and Dubuque Fighting Saints.	USHL completes expansion draft
  Major League Baseball Commissioner Bud Selig will be speaking at St. Norbert College next month.	Bud Selig to speak at St. Norbert College
  It's fresh cherry time in Michigan and the best time to enjoy this delicious and nutritious fruit.	It's cherry time

• Loss: MultipleNegativesSymmetricRankingLoss with these parameters:

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

Evaluation Datasets

nli-pairs

• Dataset: nli-pairs at d482672
• Size: 6,808 evaluation samples
• Columns: sentence1 and sentence2
• Approximate statistics based on the first 1000 samples:

  	sentence1	sentence2
  type	string	string
  details	min: 5 tokens mean: 17.64 tokens max: 63 tokens	min: 4 tokens mean: 9.67 tokens max: 29 tokens

• Samples:

  sentence1	sentence2
  Two women are embracing while holding to go packages.	Two woman are holding packages.
  Two young children in blue jerseys, one with the number 9 and one with the number 2 are standing on wooden steps in a bathroom and washing their hands in a sink.	Two kids in numbered jerseys wash their hands.
  A man selling donuts to a customer during a world exhibition event held in the city of Angeles	A man selling donuts to a customer.

• Loss: MultipleNegativesRankingLoss with these parameters:

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

qnli-contrastive

• Dataset: qnli-contrastive at bcdcba7
• Size: 5,463 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: 14.13 tokens max: 36 tokens	min: 4 tokens mean: 36.58 tokens max: 225 tokens	0: 100.00%

• Samples:

  sentence1	sentence2	label
  What came into force after the new constitution was herald?	As of that day, the new constitution heralding the Second Republic came into force.	0
  What is the first major city in the stream of the Rhine?	The most important tributaries in this area are the Ill below of Strasbourg, the Neckar in Mannheim and the Main across from Mainz.	0
  What is the minimum required if you want to teach in Canada?	In most provinces a second Bachelor's Degree such as a Bachelor of Education is required to become a qualified teacher.	0

• Loss: OnlineContrastiveLoss

Training Hyperparameters

Non-Default Hyperparameters

• eval_strategy: steps
• per_device_train_batch_size: 20
• per_device_eval_batch_size: 16
• learning_rate: 5e-06
• weight_decay: 1e-10
• lr_scheduler_type: cosine
• warmup_ratio: 0.33
• save_safetensors: False
• fp16: True
• push_to_hub: True
• hub_model_id: bobox/DeBERTaV3-small-GeneralSentenceTransformer-keepTraining-checkpoints-tmp
• hub_strategy: checkpoint
• 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: 20
• per_device_eval_batch_size: 16
• per_gpu_train_batch_size: None
• per_gpu_eval_batch_size: None
• gradient_accumulation_steps: 1
• eval_accumulation_steps: None
• learning_rate: 5e-06
• weight_decay: 1e-10
• adam_beta1: 0.9
• adam_beta2: 0.999
• adam_epsilon: 1e-08
• max_grad_norm: 1.0
• num_train_epochs: 3
• max_steps: -1
• lr_scheduler_type: cosine
• lr_scheduler_kwargs: {}
• 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/DeBERTaV3-small-GeneralSentenceTransformer-keepTraining-checkpoints-tmp
• hub_strategy: checkpoint
• 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
• batch_sampler: no_duplicates
• multi_dataset_batch_sampler: proportional

Training Logs

Epoch	Step	Training Loss	nli-pairs loss	qnli-contrastive loss
None	0	-	0.1391	0.0068
0.1500	5170	0.2058	0.1433	0.0066
0.3000	10340	0.1978	0.1448	0.0053
0.4500	15510	0.2122	0.1443	0.0063
0.6000	20680	0.1918	0.1494	0.0053
0.7501	25850	0.2103	0.1488	0.0082
0.9001	31020	0.2056	0.1513	0.0039
1.0501	36190	0.2067	0.1501	0.0079
1.2001	41360	0.1987	0.1485	0.0084
1.3501	46530	0.1987	0.1517	0.0056
1.5001	51700	0.205	0.1490	0.0062
1.6501	56870	0.183	0.1458	0.0061
1.8001	62040	0.1763	0.1418	0.0080
1.9502	67210	0.1982	0.1401	0.0066

Framework Versions

• Python: 3.10.13
• Sentence Transformers: 3.0.1
• Transformers: 4.41.2
• PyTorch: 2.1.2
• Accelerate: 0.30.1
• Datasets: 2.19.2
• 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",
}

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}
}

CoSENTLoss

@online{kexuefm-8847,
    title={CoSENT: A more efficient sentence vector scheme than Sentence-BERT},
    author={Su Jianlin},
    year={2022},
    month={Jan},
    url={https://kexue.fm/archives/8847},
}

GISTEmbedLoss

@misc{solatorio2024gistembed,
    title={GISTEmbed: Guided In-sample Selection of Training Negatives for Text Embedding Fine-tuning}, 
    author={Aivin V. Solatorio},
    year={2024},
    eprint={2402.16829},
    archivePrefix={arXiv},
    primaryClass={cs.LG}
}