README.md

metadata

language:
  - en
license: apache-2.0
tags:
  - sentence-transformers
  - sentence-similarity
  - feature-extraction
  - generated_from_trainer
  - dataset_size:800
  - loss:MultipleNegativesRankingLoss
base_model: intfloat/e5-base-v2
widget:
  - source_sentence: >-
      For the following multiple choice question, select one correct answer. Let
      s think step by step. Question In a postoperative patient with a urinary
      diversion, the nurse should monitor the urine volume every hour. Below how
      many ml h of urine may indicate that the patient is dehydrated or has some
      type of internal obstruction or loss ? Options A. 200 ml h. B. 100 ml h.
      C. 80 ml h. D. 50 ml h. E. 30 ml h.
    sentences:
      - >-
        Our approach shows that gene expression can be explained by a modest
        number of co localized transcription factors, however, information on
        cell type specific binding is crucial for understanding combinatorial
        gene regulation.
      - >-
        We have developed a rapid, simple, sensitive and specific method to
        quantify β antithrombin activity using 1μL of plasma. β antithrombin
        significantly increases in patients with ischemic cerebrovascular
        disease during the acute event, probably by its release from the
        vasculature.
      - >-
        A postoperative patient with a urinary diversion requires close
        monitoring of urine output to ensure that the diversion is functioning
        properly and that the patient is not experiencing any complications.
        Monitoring urine volume every hour is a crucial aspect of postoperative
        care in this scenario. To determine the correct answer, let s analyze
        each option A. 200 ml h This is a relatively high urine output, and it
        would not typically indicate dehydration or internal obstruction. In
        fact, a urine output of 200 ml h is generally considered adequate and
        may even be higher than the average urine output for a healthy adult. B.
        100 ml h This is also a relatively high urine output and would not
        typically indicate dehydration or internal obstruction. A urine output
        of 100 ml h is still within the normal range and would not raise
        concerns about dehydration or obstruction. C. 80 ml h While this is a
        slightly lower urine output, it is still within the normal range and
        would not necessarily indicate dehydration or internal obstruction. D.
        50 ml h This is a lower urine output, and it may start to raise concerns
        about dehydration or internal obstruction. However, it is still not the
        lowest option, and the nurse may need to consider other factors before
        determining the cause of the low urine output. E. 30 ml h This is the
        lowest urine output option, and it would likely indicate that the
        patient is dehydrated or has some type of internal obstruction or loss.
        A urine output of 30 ml h is generally considered low and would require
        immediate attention from the nurse to determine the cause and take
        corrective action. Considering the options, the correct answer is E. 30
        ml h. A urine output of 30 ml h is a critical threshold that may
        indicate dehydration or internal obstruction, and the nurse should take
        immediate action to assess the patient s fluid status and the
        functioning of the urinary diversion. Answer E.
  - source_sentence: In tumor lysis syndrome all of the following are seen except
    sentences:
      - >-
        The results indicated that some polymorphic variations of drug metabolic
        and transporter genes may be potential biomarkers for clinical outcome
        of gemcitabine based therapy in patients with locally advanced
        pancreatic cancer.
      - >-
        Variations in the prevalence of depressive symptoms occurred between
        centres, not always related to levels of illness. There was no
        consistent relationship between proportions of symptoms in well persons
        and cases for all centres. Few symptoms were present in 60 of the older
        population stereotypes of old age were not upheld.
      - >-
        Tumor lysis syndrome Caused by destruction of large number of rapidly
        proliferating neoplastic cells. It frequently leads to ARF It is
        characterized by Hypocalcemia Hyperkalemia Lactic acidosis Hyperuricemia
        Hyperphosphatemia Most frequently associated with treatment of Burkitt
        lymphoma ALL CLL Solid tumors
  - source_sentence: >-
      Does prevalence of central venous occlusion in patients with chronic
      defibrillator lead?
    sentences:
      - >-
        Intraoperative small dose IV haloperidol is effective against post
        operative nausea and vomiting with no significant effect on overall QoR.
        It may also attenuate the analgesic effects of morphine PCA.
      - >-
        Intubation is generally done with the help of endotracheal tube ETT .
        The internal diameter of ETT used ranges between 3 and 8 mm depending on
        the age, sex, and size of nares of the patient. Potex north and south
        polar performed Rae tubes RAE right angled ETT and flexo metallic tubes
        are commonly used. Out of them, North Pole Rae tube is preferred in case
        of ankylosis patient due to the direction of the curve of ETT which
        favors its placement in restricted mouth opening as in case of
        ankylosis.
      - >-
        The low prevalence of subclavian vein occlusion or severe stenosis among
        defibrillator recipients found in this study suggests that the placement
        of additional transvenous leads in a patient who already has a
        ventricular defibrillator is feasible in a high percentage of patients
        93 .
  - source_sentence: >-
      Is mode of presentation of B3 breast core biopsies screen detected or
      symptomatic a distinguishing factor in the final histopathologic result or
      risk of diagnosis of malignancy?
    sentences:
      - >-
        This observation may indicate a considerable difference in
        cardiovascular risk between genotype groups as a result of an increase
        in FVIIa after a fat rich diet.
      - >-
        Mode of patient presentation with a screen detected or symptomatic
        lesion was not a distinguishing factor for breast histopathologic
        subclassification or for the final cancer diagnosis in patients whose
        breast core biopsy was classified as B3.
      - >-
        Ans. is a i.e., Apaf 1o One of these proteins is cytochrome c, well
        known for its role in mitochondrial respiration. In the cytosol,
        cytochrome C binds to a protein called Apaf 1 apoptosis activating
        factor 1 , and the complex activates caspase 9. Bc1 2 and Bcl x may also
        directly inhibit Apaf 1 activation, and their loss from cells may permit
        activation of Apaf 1 .
  - source_sentence: >-
      Is the Danish National Hospital Register a valuable study base for
      epidemiologic research in febrile seizures?
    sentences:
      - >-
        Interstitial cystitis IC is a condition that causes discomfort or pain
        in the bladder and a need to urinate frequently and urgently. It is far
        more common in women than in men. The symptoms vary from person to
        person. Some people may have pain without urgency or frequency. Others
        have urgency and frequency without pain. Women s symptoms often get
        worse during their periods. They may also have pain with sexual
        intercourse. The cause of IC isn t known. There is no one test to tell
        if you have it. Doctors often run tests to rule out other possible
        causes of symptoms. There is no cure for IC, but treatments can help
        most people feel better. They include Distending, or inflating, the
        bladder Bathing the inside of the bladder with a drug solution Oral
        medicines Electrical nerve stimulation Physical therapy Lifestyle
        changes Bladder training In rare cases, surgery NIH National Institute
        of Diabetes and Digestive and Kidney Diseases
      - >-
        Ans. is c i.e., Presence of depression Good prognostic factors Acute
        onset late onset onset after 35 years of age Presence of precipitating
        stressor Good premorbid adjustment catatonic best prognosis Paranoid 2nd
        best sho duration 6 months Married Positive symptoms Presence of
        depression family history of mood disorder first episode pyknic fat
        physique female sex good treatment compliance good response to treatment
        good social suppo presence of confusion or perplexity normal brain CT
        Scan outpatient treatment.
      - >-
        The Danish National Hospital Register is a valuable tool for
        epidemiologic research in febrile seizures.
pipeline_tag: sentence-similarity
library_name: sentence-transformers
metrics:
  - cosine_accuracy
model-index:
  - name: MPNet base trained on AllNLI triplets
    results:
      - task:
          type: triplet
          name: Triplet
        dataset:
          name: eval dataset
          type: eval-dataset
        metrics:
          - type: cosine_accuracy
            value: 1
            name: Cosine Accuracy
      - task:
          type: triplet
          name: Triplet
        dataset:
          name: test dataset
          type: test-dataset
        metrics:
          - type: cosine_accuracy
            value: 0.97
            name: Cosine Accuracy

MPNet base trained on AllNLI triplets

This is a sentence-transformers model finetuned from intfloat/e5-base-v2. 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: intfloat/e5-base-v2
• Maximum Sequence Length: 512 tokens
• Output Dimensionality: 768 dimensions
• Similarity Function: Cosine Similarity
• Language: en
• License: apache-2.0

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: BertModel 
  (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})
  (2): Normalize()
)

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("sentence_transformers_model_id")
# Run inference
sentences = [
    'Is the Danish National Hospital Register a valuable study base for epidemiologic research in febrile seizures?',
    'The Danish National Hospital Register is a valuable tool for epidemiologic research in febrile seizures.',
    'Ans. is c i.e., Presence of depression Good prognostic factors Acute onset late onset onset after 35 years of age Presence of precipitating stressor Good premorbid adjustment catatonic best prognosis Paranoid 2nd best sho duration 6 months Married Positive symptoms Presence of depression family history of mood disorder first episode pyknic fat physique female sex good treatment compliance good response to treatment good social suppo presence of confusion or perplexity normal brain CT Scan outpatient treatment.',
]
embeddings = model.encode(sentences)
print(embeddings.shape)
# [3, 768]

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

Evaluation

Metrics

Triplet

• Datasets: eval-dataset and test-dataset
• Evaluated with TripletEvaluator

Metric	eval-dataset	test-dataset
cosine_accuracy	1.0	0.97

Training Details

Training Dataset

Unnamed Dataset

• Size: 800 training samples
• Columns: sentence1, sentence2, and label
• Approximate statistics based on the first 800 samples:

  	sentence1	sentence2	label
  type	string	string	float
  details	min: 5 tokens mean: 22.88 tokens max: 205 tokens	min: 4 tokens mean: 81.77 tokens max: 512 tokens	min: 1.0 mean: 1.0 max: 1.0

• Samples:

  sentence1	sentence2	label
  Triad of biotin deficiency is	Dermatitis, glossitis, Alopecia 407 H 314 Basic pathology 8th Biotin deficiency clinical features Adult Mental changes depression, hallucination , paresthesia, anorexia, nausea, A scaling, seborrheic and erythematous rash may occur around the eye, nose, mouth, as well as extremities 407 H Infant hypotonia, lethargy, apathy, alopecia and a characteristic rash that includes the ears.Symptoms of biotin deficiency includes Anaemia, loss of apepite dermatitis, glossitis 150 U. Satyanarayan Symptoms of biotin deficiency Dermatitis spectacle eyed appearance due to circumocular alopecia, pallor of skin membrane, depression, Lassitude, somnolence, anemia and hypercholesterolaemia 173 Rana Shinde 6th	1.0
  Drug responsible for the below condition	Thalidomide given to pregnant lady can lead to hypoplasia of limbs called as Phocomelia .	1.0
  Is benefit from procarbazine , lomustine , and vincristine in oligodendroglial tumors associated with mutation of IDH?	IDH mutational status identified patients with oligodendroglial tumors who did and did not benefit from alkylating agent chemotherapy with RT. Although patients with codeleted tumors lived longest, patients with noncodeleted IDH mutated tumors also lived longer after CRT.	1.0

• Loss: MultipleNegativesRankingLoss with these parameters:

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

Evaluation Dataset

Unnamed Dataset

• Size: 100 evaluation samples
• Columns: question, answer, and hard_negative
• Approximate statistics based on the first 100 samples:

  	question	answer	hard_negative
  type	string	string	NoneType
  details	min: 5 tokens mean: 22.52 tokens max: 103 tokens	min: 10 tokens mean: 83.51 tokens max: 403 tokens

• Samples:

  question	answer	hard_negative
  Hutchinsons secondaries In skull are due to tumors in	Adrenal neuroblastomas are malig8nant neoplasms arising from sympathetic neuroblsts in Medulla of adrenal gland Neuroblastoma is a cancer that develops from immature nerve cells found in several areas of the body.Neuroblastoma most commonly arises in and around the adrenalglands, which have similar origins to nerve cells and sit atop the kidneys.	None
  Proliferative glomerular deposits in the kidney are found in	IgA nephropathy or Berger s disease immune complex mediated glomerulonephritis defined by the presence of diffuse mesangial IgA deposits often associated with mesangial hypercellularity. Male preponderance, peak incidence in the second and third decades of life.Clinical and laboratory findings Two most common presentations recurrent episodes of macroscopic hematuria during or immediately following an upper respiratory infection often accompanied by proteinuria or persistent asymptomatic microscopic hematuriaIgA deposited in the mesangium is typically polymeric and of the IgA1 subclass. IgM, IgG, C3, or immunoglobulin light chains may be codistributed with IgAPresence of elevated serum IgA levels in 20 50 of patients, IgA deposition in skin biopsies in 15 55 of patients, elevated levels of secretory IgA and IgA fibronectin complexesIgA nephropathy is a benign disease mostly, 5 30 of patients go into a complete remission, with others having hematuria but well preserved renal functionAbou...	None
  Does meconium aspiration induce oxidative injury in the hippocampus of newborn piglets?	Our data thus suggest that oxidative injury associated with pulmonary, but not systemic, hemodynamic disturbances may contribute to hippocampal damage after meconium aspiration in newborns.	None

• Loss: MultipleNegativesRankingLoss with these parameters:

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

Training Hyperparameters

Non-Default Hyperparameters

• do_predict: True
• eval_strategy: steps
• per_device_train_batch_size: 16
• per_device_eval_batch_size: 16
• num_train_epochs: 1
• warmup_ratio: 0.1
• fp16: True
• load_best_model_at_end: True
• batch_sampler: no_duplicates

All Hyperparameters

Click to expand

• overwrite_output_dir: False
• do_predict: True
• eval_strategy: steps
• prediction_loss_only: True
• per_device_train_batch_size: 16
• 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
• torch_empty_cache_steps: None
• learning_rate: 5e-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: 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: 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: False
• 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	eval-dataset_cosine_accuracy	test-dataset_cosine_accuracy
0	0	1.0	-
1.0	25	-	0.97

Framework Versions

• Python: 3.11.10
• Sentence Transformers: 3.3.0
• Transformers: 4.46.2
• PyTorch: 2.5.1+cu124
• Accelerate: 1.1.1
• Datasets: 3.1.0
• Tokenizers: 0.20.3

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}
}