ArianeS21/reranker-msmarco-v1.1-ms-marco-TinyBERT-L-6-cmnrl

CrossEncoder based on cross-encoder/ms-marco-TinyBERT-L6

This is a Cross Encoder model finetuned from cross-encoder/ms-marco-TinyBERT-L6 on the ms_marco dataset using the sentence-transformers library. It computes scores for pairs of texts, which can be used for text reranking and semantic search.

Model Details

Model Description

• Model Type: Cross Encoder
• Base model: cross-encoder/ms-marco-TinyBERT-L6
• Maximum Sequence Length: 512 tokens
• Number of Output Labels: 1 label
• Training Dataset:
  • ms_marco
• Language: en

Model Sources

• Documentation: Sentence Transformers Documentation
• Documentation: Cross Encoder Documentation
• Repository: Sentence Transformers on GitHub
• Hugging Face: Cross Encoders on Hugging Face

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 CrossEncoder

# Download from the 🤗 Hub
model = CrossEncoder("ArianeS21/reranker-msmarco-v1.1-ms-marco-TinyBERT-L-6-cmnrl")
# Get scores for pairs of texts
pairs = [
    ['What types of cells are polyps', 'A colon polyp is a small clump of cells that forms on the lining of the colon. Most colon polyps are harmless. But over time, some colon polyps can develop into colon cancer, which is often fatal when found in its later stages. Anyone can develop colon polyps. '],
    ['course required to become a scientist', "Learn how to become a biological scientist. Research the education, training information and experience required for starting a career in biological science. The first educational step to becoming a biological scientist is to acquire a 4-year bachelor's in biology. An aspiring biological scientist takes classes in biology, chemistry, physics, mathematics and computer science. General education requirements might include English, a foreign language, communications, social sciences and arts. Classes that require laboratory work are important due to the skills acquired in them"],
    ['province definition canada', 'province. Use province in a sentence. LINK / CITE ADD TO WORD LIST. noun. The definition of a province is a specific area or location within a country or state that is often a subset of a larger union. Ontario is an example of a province of Canada'],
    ['after miscarriage how long to get period', 'If you experience any bleeding period after this 20 days, it will probably be the real period. This normally comes at about 4 to 6 weeks after the miscarriage. This period will be heavy as compared to the normal light periods that you usually get, and may be accompanied by severe cramping.'],
    ['what is cardiac disease', 'The term heart disease is often used interchangeably with the term cardiovascular disease.. Cardiovascular disease generally refers to conditions that involve narrowed or blocked blood vessels that can lead to a heart attack, chest pain (angina) or stroke.'],
]
scores = model.predict(pairs)
print(scores.shape)
# (5,)

# Or rank different texts based on similarity to a single text
ranks = model.rank(
    'What types of cells are polyps',
    [
        'A colon polyp is a small clump of cells that forms on the lining of the colon. Most colon polyps are harmless. But over time, some colon polyps can develop into colon cancer, which is often fatal when found in its later stages. Anyone can develop colon polyps. ',
        "Learn how to become a biological scientist. Research the education, training information and experience required for starting a career in biological science. The first educational step to becoming a biological scientist is to acquire a 4-year bachelor's in biology. An aspiring biological scientist takes classes in biology, chemistry, physics, mathematics and computer science. General education requirements might include English, a foreign language, communications, social sciences and arts. Classes that require laboratory work are important due to the skills acquired in them",
        'province. Use province in a sentence. LINK / CITE ADD TO WORD LIST. noun. The definition of a province is a specific area or location within a country or state that is often a subset of a larger union. Ontario is an example of a province of Canada',
        'If you experience any bleeding period after this 20 days, it will probably be the real period. This normally comes at about 4 to 6 weeks after the miscarriage. This period will be heavy as compared to the normal light periods that you usually get, and may be accompanied by severe cramping.',
        'The term heart disease is often used interchangeably with the term cardiovascular disease.. Cardiovascular disease generally refers to conditions that involve narrowed or blocked blood vessels that can lead to a heart attack, chest pain (angina) or stroke.',
    ]
)
# [{'corpus_id': ..., 'score': ...}, {'corpus_id': ..., 'score': ...}, ...]

Evaluation

Metrics

Cross Encoder Reranking

• Datasets: NanoMSMARCO_R100, NanoNFCorpus_R100 and NanoNQ_R100
• Evaluated with CrossEncoderRerankingEvaluator with these parameters:

  {
      "at_k": 10,
      "always_rerank_positives": true
  }
  

Metric	NanoMSMARCO_R100	NanoNFCorpus_R100	NanoNQ_R100
map	0.5740 (+0.0844)	0.3119 (+0.0509)	0.5853 (+0.1657)
mrr@10	0.5685 (+0.0910)	0.5150 (+0.0152)	0.5939 (+0.1672)
ndcg@10	0.6375 (+0.0971)	0.3247 (-0.0003)	0.6343 (+0.1336)

Cross Encoder Nano BEIR

• Dataset: NanoBEIR_R100_mean
• Evaluated with CrossEncoderNanoBEIREvaluator with these parameters:

  {
      "dataset_names": [
          "msmarco",
          "nfcorpus",
          "nq"
      ],
      "rerank_k": 100,
      "at_k": 10,
      "always_rerank_positives": true
  }
  

Metric	Value
map	0.4904 (+0.1003)
mrr@10	0.5591 (+0.0911)
ndcg@10	0.5322 (+0.0768)

Training Details

Training Dataset

ms_marco

• Dataset: ms_marco at a47ee7a
• Size: 64,538 training samples
• Columns: query, positive, negative_1, negative_2, negative_3, negative_4, and negative_5
• Approximate statistics based on the first 1000 samples:

  	query	positive	negative_1	negative_2	negative_3	negative_4	negative_5
  type	string	string	string	string	string	string	string
  details	min: 11 characters mean: 33.45 characters max: 105 characters	min: 70 characters mean: 430.82 characters max: 1001 characters	min: 116 characters mean: 425.24 characters max: 1004 characters	min: 117 characters mean: 416.19 characters max: 916 characters	min: 55 characters mean: 410.99 characters max: 873 characters	min: 75 characters mean: 411.65 characters max: 921 characters	min: 121 characters mean: 416.45 characters max: 906 characters

• Samples:

  query	positive	negative_1	negative_2	negative_3	negative_4	negative_5
  age requirement for relisting military	17 years of age for voluntary military service; soldiers cannot be deployed until the age of 18; no conscription; 3 years of secondary education required; must be a citizen of NZ, the UK, Australia, Canada, or the US, and resident of NZ for the previous 5 years (2013). 18 years of age for voluntary military service; the armed forces law of 31 December 2004 did not specify a minimum age for enlistment, but the government claimed that no one younger than 18 was being recruited; mandatory retirement age 45 (enlisted), 50 (NCOs), and 55 (officers) (2012).	18 years of age for compulsory military service; 16 years of age for volunteers; 17 years of age for Law Enforcement Forces; 15 years of age for Basij Forces (Popular Mobilization Army); conscript military service obligation is 18 months; women exempt from military service (2012). 18 years of age for voluntary military service; the armed forces law of 31 December 2004 did not specify a minimum age for enlistment, but the government claimed that no one younger than 18 was being recruited; mandatory retirement age 45 (enlisted), 50 (NCOs), and 55 (officers) (2012).	You must meet all the GMC (General Military Course) membership requirements and: 1 Be a United States citizen. 2 Be of legal age as required by the state in which you will be attending ROTC or 17 years old with parent or guardian consent. 3 Be in good academic standing.	It truly depends on whether or not you are a non-prior or prior service member, looking at enlisting into the active duty or reserve components, etc. In the United States military system, you may enlist into any branch at the age of 17 years old with parental consent, or 18 years old without parental consent. Age waivers, although rare, are possible. In the United States Air Force, the maximum age allowed to enlist and ship to basic training is 27 years old. In the Air Force Reserves, the maximum age is 34 years old. In the Air National Guard, the maximum age is 39	1 You must be a U.S. citizen or resident alien. 2 You must be at least 17 years old (17-year old applicants require parental consent). 3 You must (with very few exceptions) have a high school diploma. 4 You must pass a physical medical exam. Before you visit your local recruiter, be sure you meet the minimum qualifications for serving in the U.S. Armed Forces. Some qualifications are required by all five services: 1	To enlist in the Army Reserve you must be: 1 Between the ages of 17–35 (must ship to basic training prior to 35th birthday). 2 A U.S. citizen or registered alien – U.S. Citizens or Permanent Resident Aliens (people who have an INS I-151/I-551 “Green Card”) may join the U.S. Military. 3 Healthy and in good physical condition. 1 Between the ages of 17–35 (must ship to basic training prior to 35th birthday). 2 A U.S. citizen or registered alien – U.S. Citizens or Permanent Resident Aliens (people who have an INS I-151/I-551 “Green Card”) may join the U.S. Military.
  pro football hall of fame nominees	Email. Brett Favre, Terrell Owens and Alan Faneca highlight the list of first-year Pro Football Hall of Fame nominees for the class of 2016 announced Wednesday night. Other first-year nominees include Lawyer Milloy, Brian Westbrook, Clinton Portis, John Carney, Keith Bulluck and Mike Vrabel.	Favre was among the 108 modern-day nominees for the Pro Football Hall of Fame class of 2016 announced Wednesday night. The NFL's career leader in nearly every passing category when he retired after the 2010 season and winner of the 1997 Super Bowl with Green Bay, Favre leads a strong list of first-time nominees.	Who's headed to Canton? The Pro Football Hall of Fame unveiled 103 modern-era nominees for the Class of 2016, including Brett Favre in his first year of eligibility.	The Pro Football Hall of Fame officially unveiled its nominees for the Class of 2016, and there's at least one lock among the newest additions to the 108-man list — a graybeard by the name of Brett Favre.	The complete list of nominees, which can be viewed on the Hall of Fame's official website, includes 11 first-timers, including running backs Brian Westbrook and Clinton Portis, along with linebackers Keith Bulluck and Mike Vrabel.	Other pro football players to make the Hall of Fame 2016 nominees list include wide receiver Terrell Owens, who played 15 seasons in the NFL. As pointed out by USA Today columnist Jarrett Bell, Owens could run into some trouble getting in to the Hall of Fame because of his troubles with the law.
  what is quartz made of	1 The mineral of which quartz is made is called silica, with the above composition SiO2. 2 I hope I was able to give some hints as to what quartz is and how it can be further divided into two subgroups based on its visible or invisible crystal structure (macro ... and 3 ... micro... ). So if you have looked for a simple straightforward answer, here it is: Quartz is made of Silicon Si and oxygen O. But not everything is so simple... Follow on... I would like to make a digression here, but at the end we will have, I hope, better knowledge about what quartz is.	Silestone is a compound made of natural quartz, which makes it extraordinarily hard and resilient. This is the one and only countertop made of quartz with bacteriostatic protection, giving it hygienic properties that are not to be found anywhere else on the market. It is an excellent surface for kitchen countertops, bathrooms, floors and wall cladding using the minimum number of joints. Man-Made Quartz. Zodiaq is 93% pure crystal quartz, which gives Zodiaq a radiance without the imperfections common in natural stone. It is incredibly durable and resists both heat and scratches.	A quartz clock is a clock that uses an electronic oscillator that is regulated by a quartz crystal to keep time. This crystal oscillator creates a signal with very precise frequency, so that quartz clocks are at least an order of magnitude more accurate than mechanical clocks.	1 The mineral of which quartz is made is called silica, with the above composition SiO2. 2 I hope I was able to give some hints as to what quartz is and how it can be further divided into two subgroups based on its visible or invisible crystal structure (macro ... and ... So if you have looked for a simple straightforward answer, here it is: Quartz is made of Silicon Si and oxygen O. But not everything is so simple... Follow on... I would like to make a digression here, but at the end we will have, I hope, better knowledge about what quartz is.	Today we look at quartz countertops, otherwise known as engineered stone countertops. Material: Quartz. How It's Made: Quartz countertops are man-made engineered stone countertops formed by combining 90% ground quartz (a natural hard mineral) with 8-10% resins, polymers, and pigments. This forms a very hard granite-like surface. Quartz is the second most abundant material in the earth's crust (which is good), but the acrylic resins used in quartz countertops are petroleum by-products, and they often contain alumina trihydrate fillers made from bauxite ore, which is mined primarily under toxic conditions in developing countries.	In terms of marketing, quartz counter companies are smart to capitalize on this word because of those implications. And it is true. But not completely. In simple terms, quartz counters are 90% made up of stone-like materials and 10% of binders that are either plastic-like (polymeric) or cement-based. Of those stone-like materials, quartz is only one of many substances. According to Breton (more on them later) materials, other materials include:

• Loss: CachedMultipleNegativesRankingLoss with these parameters:

  {
      "scale": 10.0,
      "num_negatives": 5,
      "activation_fn": "torch.nn.modules.activation.Sigmoid",
      "mini_batch_size": 32
  }
  

Evaluation Dataset

ms_marco

• Dataset: ms_marco at a47ee7a
• Size: 10,000 evaluation samples
• Columns: query, positive, negative_1, negative_2, negative_3, negative_4, and negative_5
• Approximate statistics based on the first 1000 samples:

  	query	positive	negative_1	negative_2	negative_3	negative_4	negative_5
  type	string	string	string	string	string	string	string
  details	min: 11 characters mean: 34.79 characters max: 100 characters	min: 105 characters mean: 434.18 characters max: 889 characters	min: 91 characters mean: 419.47 characters max: 888 characters	min: 66 characters mean: 416.66 characters max: 861 characters	min: 95 characters mean: 418.5 characters max: 992 characters	min: 97 characters mean: 412.56 characters max: 954 characters	min: 74 characters mean: 416.09 characters max: 927 characters

• Samples:

  query	positive	negative_1	negative_2	negative_3	negative_4	negative_5
  What types of cells are polyps	A colon polyp is a small clump of cells that forms on the lining of the colon. Most colon polyps are harmless. But over time, some colon polyps can develop into colon cancer, which is often fatal when found in its later stages. Anyone can develop colon polyps.	Types of cancer in the colon and rectum. Several types of cancer can start in the colon or rectum. Adenocarcinomas: More than 95% of colorectal cancers are a type of cancer known as adenocarcinomas. These cancers start in cells that form glands that make mucus to lubricate the inside of the colon and rectum. A polyp is a benign, non-cancerous tumor. Some polyps can change into cancer but not all do. The chance of changing into a cancer depends on the kind of polyp. The 2 main types of polyps are: 1 Adenomatous polyps (adenomas) are polyps that can change into cancer.	Adenomatous polyps, or adenomas, are polyps that grow on the lining of the colon and which carry a high risk of cancer. The adenomatous polyp is considered pre-malignant, i.e., likely to develop into colon cancer. The other types of polyps that can occur in the colon are the hyperplastic and inflammatory polyps. They are unlikely to develop into colorectal cancer.	Polyps are often non-cancerous growths but some can develop into cancer. The two most common types of polyps found in the colon and rectum include: 1 Hyperplastic and inflammatory polyps. 2 Usually these polyps do not carry a risk of developing into cancer. 3 However, large hyperplastic polyps, especially on the right side of the colon, are of concern and should be completely removed	A polyp is a projection (growth) of tissue from the inner lining of the colon into the lumen (hollow center) of the colon. Different types of polyps look different under the microscope. Polyps are benign (non-cancerous) growths, but cancer can start in some types of polyps.	Not all colon polyps are the same. There are different histologic types, that is, the cells that make up the polyp have different characteristics when viewed under the microscope. They also vary in size, number, and location. Most importantly, they vary in their tendency to become cancerous (malignant).
  course required to become a scientist	Learn how to become a biological scientist. Research the education, training information and experience required for starting a career in biological science. The first educational step to becoming a biological scientist is to acquire a 4-year bachelor's in biology. An aspiring biological scientist takes classes in biology, chemistry, physics, mathematics and computer science. General education requirements might include English, a foreign language, communications, social sciences and arts. Classes that require laboratory work are important due to the skills acquired in them	Most entry-level forensic positions require a bachelor’s degree in forensic or natural science. Depending on which type of forensic scientist you’d like to become, courses that should be taken during these initial four years may include: 1 Biochemistry. 2 Toxicology. 3 Criminal Justice. 4 Physical Science. 5 Chemistry.	A: In order to become a forensic scientist, you will need an education in forensic sciences. If you are just looking for an entry level in the criminal justice sector, you can go for a 6 month certificate or diploma. A 4 year bachelor's degree can also get you entry-level positions with forensic departments. A: A forensic scientist needs to have a proper education in their respective field. You first need a bachelor's degree in forensic sciences or any other related science topic like psychology, medicine etc. after the completion of the degree it is recommended by most employers that you have a graduate degree as well.	A: If you want to become a forensic scientist for the police you should have a four year undergraduate degree in a field like chemistry, biology or related natural science like pre-med, biochemistry or microbiology. A: A forensic scientist needs to have a proper education in their respective field. You first need a bachelor's degree in forensic sciences or any other related science topic like psychology, medicine etc. after the completion of the degree it is recommended by most employers that you have a graduate degree as well.	Scientists generally work for government agencies, universities, and private laboratories. Some may work normal business hours, while others who are conducting research or have deadlines may work unusual hours. Requirements to become a scientist depend on the type of science you want to study.	Entry-level forensic scientists usually have a bachelor's degree in forensic science or a related field, such as biology, chemistry or physics. A forensic science major typically includes those basic sciences, plus classes in pharmacology, statistics, computer modeling, biochemistry and criminal justice.
  province definition canada	province. Use province in a sentence. LINK / CITE ADD TO WORD LIST. noun. The definition of a province is a specific area or location within a country or state that is often a subset of a larger union. Ontario is an example of a province of Canada	By Susan Munroe. Definition: A territory in Canada is one of three political units which, with ten provinces, make up Canada. The main differences between the powers of provinces and those of territories are the power of the provinces to amend their constitutions and the power to control the management and sale of public lands	Ontario, Quebec, New Brunswick, and Nova Scotia are the original provinces, formed when British North American colonies federated on July 1, 1867, into the Dominion of Canada and by stages began accruing the indicia of sovereignty from the United Kingdom. The provinces and territories of Canada combine to make up the world's second-largest country by area. In 1867, three provinces of British North America — New Brunswick, Nova Scotia and the Province of Canada (which, on the formation of Canada, was divided into Ontario and Quebec)—were united to form the new nation	Definition of PROVINCE for Kids. 1. : a part of a country having a government of its own (as one of the divisions of Canada). 2. provinces plural: the part or parts of a country far from the capital or chief city. function, office, duty, province mean the acts or operations expected of a person or thing. function implies a definite end or purpose or a particular kind of work. office is typically applied to the function or service associated with a trade or profession or a special relationship to others.	The major difference between a Canadian province and a territory is that provinces receive their power and authority from the Constitution Act, 1867 (formerly called the British North America Act, 1867), whereas territorial governments have powers delegated to them by the federal government. The provinces and territories of Canada combine to make up the world's second-largest country by area. In 1867, three provinces of British North America — New Brunswick, Nova Scotia and the Province of Canada (which, on the formation of Canada, was divided into Ontario and Quebec)—were united to form the new nation	Full Definition of PROVINCE. 1. a: a country or region brought under the control of the ancient Roman government b: an administrative district or division of a country c plural: all of a country except the metropolises. 2. function, office, duty, province mean the acts or operations expected of a person or thing. function implies a definite end or purpose or a particular kind of work. office is typically applied to the function or service associated with a trade or profession or a special relationship to others.

• Loss: CachedMultipleNegativesRankingLoss with these parameters:

  {
      "scale": 10.0,
      "num_negatives": 5,
      "activation_fn": "torch.nn.modules.activation.Sigmoid",
      "mini_batch_size": 32
  }
  

Training Hyperparameters

Non-Default Hyperparameters

• eval_strategy: steps
• per_device_train_batch_size: 32
• per_device_eval_batch_size: 32
• learning_rate: 2e-05
• num_train_epochs: 1
• warmup_ratio: 0.1
• seed: 12
• bf16: True
• load_best_model_at_end: True
• 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: 32
• per_device_eval_batch_size: 32
• 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: 0.0
• 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: 12
• data_seed: None
• jit_mode_eval: False
• use_ipex: False
• bf16: True
• fp16: False
• 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: True
• 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: no_duplicates
• multi_dataset_batch_sampler: proportional

Training Logs

Epoch	Step	Training Loss	Validation Loss	NanoMSMARCO_R100_ndcg@10	NanoNFCorpus_R100_ndcg@10	NanoNQ_R100_ndcg@10	NanoBEIR_R100_mean_ndcg@10
-1	-1	-	-	0.6636 (+0.1231)	0.3523 (+0.0273)	0.6943 (+0.1937)	0.5701 (+0.1147)
0.0005	1	1.2039	-	-	-	-	-
0.0496	100	1.6169	-	-	-	-	-
0.0992	200	1.3581	-	-	-	-	-
0.1487	300	1.3087	-	-	-	-	-
0.1983	400	1.3009	1.2809	0.6375 (+0.0971)	0.3247 (-0.0003)	0.6343 (+0.1336)	0.5322 (+0.0768)
0.2479	500	1.2715	-	-	-	-	-
0.2975	600	1.3229	-	-	-	-	-
0.3471	700	1.324	-	-	-	-	-
0.3966	800	1.3319	1.3450	0.6148 (+0.0744)	0.3242 (-0.0009)	0.6529 (+0.1522)	0.5306 (+0.0752)
0.4462	900	1.4677	-	-	-	-	-
0.4958	1000	1.7681	-	-	-	-	-
0.5454	1100	1.8034	-	-	-	-	-
0.5949	1200	1.5998	1.4563	0.5059 (-0.0345)	0.3180 (-0.0071)	0.5745 (+0.0739)	0.4661 (+0.0108)
0.6445	1300	1.4844	-	-	-	-	-
0.6941	1400	1.4353	-	-	-	-	-
0.7437	1500	1.4714	-	-	-	-	-
0.7933	1600	1.4229	1.4313	0.5913 (+0.0509)	0.3180 (-0.0071)	0.6100 (+0.1093)	0.5064 (+0.0510)
0.8428	1700	1.4229	-	-	-	-	-
0.8924	1800	1.4379	-	-	-	-	-
0.9420	1900	1.441	-	-	-	-	-
0.9916	2000	1.4296	1.4060	0.5829 (+0.0425)	0.3120 (-0.0131)	0.6404 (+0.1398)	0.5118 (+0.0564)
-1	-1	-	-	0.6375 (+0.0971)	0.3247 (-0.0003)	0.6343 (+0.1336)	0.5322 (+0.0768)

• The bold row denotes the saved checkpoint.

Framework Versions

• Python: 3.10.15
• Sentence Transformers: 4.1.0
• Transformers: 4.48.3
• PyTorch: 2.6.0+cu118
• Accelerate: 1.6.0
• Datasets: 3.3.2
• Tokenizers: 0.21.0

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",
}