shaurya23102/insurance-embedding-model

SentenceTransformer based on sentence-transformers/all-mpnet-base-v2

This is a sentence-transformers model finetuned from sentence-transformers/all-mpnet-base-v2 on the json dataset. It maps sentences & paragraphs to a 768-dimensional dense vector space and can be used for retrieval.

Model Details

Model Description

• Model Type: Sentence Transformer
• Base model: sentence-transformers/all-mpnet-base-v2
• Maximum Sequence Length: 384 tokens
• Output Dimensionality: 768 dimensions
• Similarity Function: Cosine Similarity
• Supported Modality: Text
• Training Dataset:
  • json

Model Sources

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

Full Model Architecture

SentenceTransformer(
  (0): Transformer({'transformer_task': 'feature-extraction', 'modality_config': {'text': {'method': 'forward', 'method_output_name': 'last_hidden_state'}}, 'module_output_name': 'token_embeddings', 'architecture': 'MPNetModel'})
  (1): Pooling({'embedding_dimension': 768, 'pooling_mode': 'mean', '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 = [
    'What does the insurer consider as fraud for the purpose of this policy?',
    'in accordance with the terms and conditions of the chosen policy 6 Fraud If any claim made by the insured person is in any respect fraudulent or if any false statement or declaration is made or used in support thereof or if any fraudulent means or devices are used by the insured person or anyone acting on his her behalf to obtain any benefit under this policy all benefits under this policy and the premium paid shall be forfeited Any amount already paid against claims made under this policy but which are found fraudulent later shall be repaid by all recipient s policyholder s who has made that particular claim who shall be jointly and severally liable for such repayment to the insurer For the purpose of this clause the expression fraud means any of the following acts committed by the Insured Person or by his agent or the hospital doctor any other party acting on behalf of the insured person with intent to deceive the insurer or to induce the insurer to issue an insurance Policy a the suggestion as a fact of that which is not true and which the Insured Person does not believe to be true',
    'activity The Insured Person agrees that choosing to utilize any of the wellness services or any information or advise rendered by Our Health Service Providers or Network Providers or the Company will be solely at the Insured Person s discretion and own risk and should not be used to diagnose or identify treatment for a medical or mental health condition The Wellness Points earned by the Insured person through the Wellness Program can be carried forward for a maximum of 3 years and shall have to be redeemed at the end of the 3rd Policy Year In case the Insured Person does not wish to redeem the wellness points earned the same will be forfeited In case of expiry of Policy the accrued wellness points may be carried forward for a period not exceeding three months There shall not be any cash reimbursement or redemption available against the wellness points accumulated by an Insured Person We or Our Health Service Providers or Our Network Providers do not warrant the validity accuracy completeness safety quality or applicability of the content or anything said or written or any suggestions provided in the course of providing the wellness services We or our affiliates',
]
embeddings = model.encode(sentences)
print(embeddings.shape)
# [3, 768]

# Get the similarity scores for the embeddings
similarities = model.similarity(embeddings, embeddings)
print(similarities)
# tensor([[1.0000, 0.6955, 0.1400],
#         [0.6955, 1.0000, 0.2254],
#         [0.1400, 0.2254, 1.0000]])

Training Details

Training Dataset

json

• Dataset: json
• Size: 91 training samples
• Columns: anchor, positive, negative_1, and negative_2
• Approximate statistics based on the first 91 samples:

  	anchor	positive	negative_1	negative_2
  type	string	string	string	string
  modality	text	text	text	text
  details	min: 10 tokens mean: 20.18 tokens max: 38 tokens	min: 62 tokens mean: 233.59 tokens max: 384 tokens	min: 206 tokens mean: 228.08 tokens max: 384 tokens	min: 204 tokens mean: 223.13 tokens max: 272 tokens

• Samples:

  anchor	positive	negative_1	negative_2
  What is the maximum number of wellness points an Insured Person can earn under the 'Tele consultations' activity?	Policy Start Date the Insured Person can earn a maximum of 400 wellness points under this activity 2 Advisory on Preventive Health Check up The reports of the Preventive Health Check up of the Insured Person if referred to our tele consultation platform example IL Hello Doctor for medical advisory and opinion shall reward the Insured Person with a maximum of 300 wellness points 3 Medical Vault The Insured Person has to save relevant medical records diagnostic reports prescriptions preventive health check up reports in the medical vault on the mobile application This activity shall reward the Insured Person a maximum of 300 wellness points 4 First usage of Chat with Health Expert Health Coach Service The Insured Person shall be rewarded 100 wellness points on the first time usage of the chat functionality on our mobile application The Insured Person can virtually chat with health experts like physiotherapists counsellors dieticians etc under this service 5 Tele consultations The Insured...	conditions This Cover can be availed only on Cashless basis and is limited to once per Policy Year per Insured Person Maximum of 2 Preventive Health Check up coupons shall be provided per Policy Year for a floater Policy This Cover can be availed through our mobile application or via utilization of Health Check up coupons provided with the Policy kit by calling at our Toll free number 1800 2666 The Network Provider Health Service Provider shall be assigned by Us post receiving Insured Person s request to avail a Health Check up under this Cover Utilization of this Preventive Health Check up will not impact the Annual Sum Insured or eligibility for Guaranteed Cumulative Bonus and or Super No Claim Bonus if opted and available Un utilised Health Check up package will not be carried forward to the next Policy Year and it will be the Insured Person s choice and responsibility to utilise the same with in the designated Policy Period We shall not be liable to provide any reminders or notific...	services shall be provided through our Empaneled Health Service Provider subject to availability at the time of appointment This Optional Cover shall also include e consultation given by a General Medical Practitioner or Specialist or Super Specialist Medical Practitioner or AYUSH Medical Practitioner through a virtual mode of communication such as but not limited to chat email video online portal or mobile application Physiotherapy sessions and counselling availed for psychiatric ailments or mental health issues shall be excluded from the scope of this Optional Cover as the same are covered under optional cover 7 iv Physiotherapy sessions and optional cover 7 v e counselling respectively ii Routine Diagnostic and Minor Procedure Cover We shall cover medical expenses incurred for outpatient diagnostic tests recommended by Medical Practitioner under our cashless network available in the mobile application in relation to any Illness contracted or Injury suffered by the Insured Person dur...
  What wellness points will I be awarded for viewing my E card and verifying the details on the same?	above the wellness points to be awarded shall be doubled provided that both the Insured Persons complete their respective wellness activities Detailed explanation of Table A has been mentioned below A Onboarding 1 Addition of Policy Details The Insured Person shall be awarded 500 welcome wellness points on downloading our mobile application and registering the policy details 2 E card Verification The insured person shall be awarded 300 wellness points to view the E card verify the details mentioned on the same and confirm to the Company about the same The wellness points awarded for onboarding i e for addition of Policy details and E card verification shall only be onetime for the first year of the Policy and not for any subsequent renewals thereof B Health Assessment 1 Health Risk Assessment HRA The Health Risk Assessment HRA questionnaire is a tool for evaluation of the Insured Person s health and quality of life by reviewing the personal lifestyle practices affecting the Insured Per...	required by Us or Our In house claim processing team to investigate the Claim or Our obligation to make payment for it The relevant documents can be sent to ICICI Lombard Health Care ICICI Bank Tower Plot no 12 Financial district Nanakramguda Gachibowli Hyderabad 5000032 3 CLAIM SERVICE GUARANTEE We provide You Claim Service Guarantee as follows a For Reimbursement Claims We shall make the payment of admissible claim as per terms amp conditions of Policy OR communicate non admissibility of claim within 14 days after You submit complete set of documents amp information in respect of the claim In case We fail to make the payment of admissible claim or to communicate non admissibility of claim within this time period We shall pay 2 interest over and above the rate defined as per IRDA Protection of Policyholder s Interest Regulations 2017 b For Cashless Claims If You notify pre authorization request for cashless facility through any of Our empanelled network hospitals along with complete s...	the disability j Any other document as may be required by the Us If You are covered under any health and accident insurance policy of other insurance company and become entitled to Claim under such policy then You can submit to Us the copies of the above listed documents medical records provided they are duly certified by such insurance company or any hospital where You are getting treated as applicable Note The cover under this extension shall terminate in the event of Your Claim becoming admissible hereunder In consequence thereof no benefit shall be payable under this extension of the policy thereafter 7 BeFit All benefits under the BeFit cover can be availed only on cashless basis via our mobile application and are subject to the terms conditions and exclusions and the availability of Sum Insured under the Cover BeFit cover can only be opted by Insured Person s up to the age of 65 years All services shall be provided through our Empaneled Health Service Provider subject to availabi...
  What is the time frame within which I need to provide information to the claim processing team after hospitalisation for reimbursement settlement?	ordinate with Our claim team to provide cashless facility We will consider Your request after having obtained accurate and complete information for the Illness or Injury for which cashless Hospitalisation facility is sought by You and We will confirm Your request in writing B For Reimbursement Settlement i You shall give notice to Us or Our In house claim processing team by calling the toll free number 1800 2666 or emailing us at customersupport icicilombard com as specified in the Policy provided to You and also in writing at Our address with particulars as below Policy number Your Name Your relationship with the Proposer Nature of Illness or Injury Name and address of the attending Medical Practitioner and the Hospital Any other information that may be relevant to the Illness Injury Hospitalisation The above information needs to be provided to Us or Our In house claim processing team immediately and in any event within 10 days of Hospitalisation failing which We will have the right t...	the disability j Any other document as may be required by the Us If You are covered under any health and accident insurance policy of other insurance company and become entitled to Claim under such policy then You can submit to Us the copies of the above listed documents medical records provided they are duly certified by such insurance company or any hospital where You are getting treated as applicable Note The cover under this extension shall terminate in the event of Your Claim becoming admissible hereunder In consequence thereof no benefit shall be payable under this extension of the policy thereafter 7 BeFit All benefits under the BeFit cover can be availed only on cashless basis via our mobile application and are subject to the terms conditions and exclusions and the availability of Sum Insured under the Cover BeFit cover can only be opted by Insured Person s up to the age of 65 years All services shall be provided through our Empaneled Health Service Provider subject to availabi...	under this extension shall terminate in the event of Your Claim becoming admissible hereunder In consequence thereof no benefit shall be payable to You under this Optional Cover of the Policy thereafter 6 Personal Accident We will pay You or Your Nominee legal heir as the case may be the Sum Insured as specified against this Optional Cover in the Policy Schedule on occurrence of any Insured Event as specifically described hereunder arising due to an Injury sustained by You during the Policy Year Insured Event Accidental Death We will pay Your Nominee legal heir as the case may be the Sum Insured as specified against this Optional Cover in the Policy Schedule on the unfortunate event of Your death provided such death results solely and directly from an Injury sustained within a period of twelve months from the date of Accident resulting in such Injury Provided that the date of occurrence of the Accident falls within the Policy Year Insured Event Permanent Total Disablement PTD resulting...

• Loss: MultipleNegativesRankingLoss with these parameters:

  {
      "scale": 20.0,
      "similarity_fct": "cos_sim",
      "gather_across_devices": false,
      "directions": [
          "query_to_doc"
      ],
      "partition_mode": "joint",
      "hardness_mode": null,
      "hardness_strength": 0.0
  }
  

Evaluation Dataset

json

• Dataset: json
• Size: 11 evaluation samples
• Columns: anchor, positive, negative_1, and negative_2
• Approximate statistics based on the first 11 samples:

  	anchor	positive	negative_1	negative_2
  type	string	string	string	string
  modality	text	text	text	text
  details	min: 13 tokens mean: 15.55 tokens max: 19 tokens	min: 204 tokens mean: 233.45 tokens max: 384 tokens	min: 206 tokens mean: 234.73 tokens max: 384 tokens	min: 209 tokens mean: 224.55 tokens max: 249 tokens

• Samples:

  anchor	positive	negative_1	negative_2
  What is the maximum period for New Born Baby Cover?	will not be covered under this Cover c This Cover is available only under a family floater Policy d This Cover is available for You or Your spouse provided You and Your spouse both are covered under the same family floater Policy e We will not cover ectopic pregnancy under this Cover the same shall be covered under In patient Treatment 10 New Born Baby Cover We will cover the Medical Expenses incurred by You on Hospitalization of a New born Baby during each Policy Year of Policy Period subject to the limits as specified in the Policy Schedule This Optional Cover will be provided only if Maternity Cover is applicable to You This Optional Cover will cover Medical Expenses incurred on the New born Baby during Hospitalization for a minimum period of consecutive 24 hours for a maximum period up to 90 days from the date of birth of the baby 11 Voluntary Co Payment The Insured Person has the choice to opt for Voluntary Co payment and avail subsequent discount on premium In case Voluntary Co p...	forming part thereof The Policy contains details of the extent of cover available to You what is excluded from the cover and the terms amp conditions on which the Policy is issued to You Proposer Policyholder means the person s or the entity named in the Policy Schedule who executed the Policy Schedule and is are responsible for payment of premium s Policy Period means the period commencing from the Policy Period Start Date Time and ending at the Policy Period End Date Time of the Policy and as specifically appearing in the Policy Schedule Policy Year means a period of twelve months beginning from the Policy Period Start Date and ending on the last day of such twelve month period For the purpose of subsequent years Policy Year shall mean a period of twelve months beginning from the end of the previous Policy Year and lapsing on the last day of such twelve month period till the Policy Period End Date as specified in the Policy Schedule Single Private Room means an air conditioned room i...	BENEFITS COVERED UNDER THE POLICY The coverage mentioned below differs between the various plan offerings and the wordings of only the relevant covers opted by the Insured Person and as mentioned in the Policy schedule will be applicable The Company hereby agrees subject to the terms conditions and exclusions herein contained or otherwise expressed for the period and to the extent of the Sum Insured as specified in the Schedule to this Policy The Policy covers Reasonable and Customary Charges incurred towards medical treatment taken during the Policy Period for an Illness Accident or condition described below if this is contracted or sustained by an Insured Person during the Policy Period and subject always to the Sum Insured any subsidiary limit specified in the schedule of Benefits the terms conditions limitations and exclusions mentioned in the Policy and eligibility as per the insurance plan opted by Insured and stated in the Schedule A Basic Cover The payment under this Basic Cove...
  What services cannot be provided on an outpatient basis?	basis subject to availability of our empaneled Service Provider s Kindly visit our website for cities locations where such services are available f Treatment availed is not categorized under AYUSH or any form of non allopathic treatment g Such treatment cannot be provided on outpatient basis However in case of unavailability of our empaneled Service Provider in the Insured Person s location in case the Insured Person intends to avail the services of Non network Provider and claims for reimbursement a prior approval from Company needs to be taken before availing such services In case the Insured Person breaches the conditions of approval or fails to take the prior written approval from Company we are not liable to settle any claim under this section For the purpose of this Cover Home Care Treatment shall include a Diagnostic tests underwent at home as advised by Medical Practitioner b Medicines prescribed in writing by a Medical Practitioner c Consultation charges of the Medical Practit...	the disability j Any other document as may be required by the Us If You are covered under any health and accident insurance policy of other insurance company and become entitled to Claim under such policy then You can submit to Us the copies of the above listed documents medical records provided they are duly certified by such insurance company or any hospital where You are getting treated as applicable Note The cover under this extension shall terminate in the event of Your Claim becoming admissible hereunder In consequence thereof no benefit shall be payable under this extension of the policy thereafter 7 BeFit All benefits under the BeFit cover can be availed only on cashless basis via our mobile application and are subject to the terms conditions and exclusions and the availability of Sum Insured under the Cover BeFit cover can only be opted by Insured Person s up to the age of 65 years All services shall be provided through our Empaneled Health Service Provider subject to availabi...	by each of You shall be conditions precedent to admission of Our liability You are requested to go through our list of de listed excluded providers which is available on our website As the list is dynamic please refer to the latest list The claim pay out would be adjudicated in following sequence i If a room accommodation has been opted for where the room rent or category is higher than the eligible limit as applicable for the Insured Person then the associated medical expenses payable shall be pro rated as per applicable limits a Associated medical expenses means those expenses as listed below which vary in accordance with the room rent or room category or ICU Charges in a Hospital i Room boarding nursing and operation theatre expenses as charged by the Hospital where the Insured Person availed treatment ii Fees charged by surgeon anesthetist Medical Practitioner iii Investigation expenses ii Zone based Co payment shall be applicable in all cases where treatment is taken in a zone hig...
  What does the insurer consider as fraud for the purpose of this policy?	in accordance with the terms and conditions of the chosen policy 6 Fraud If any claim made by the insured person is in any respect fraudulent or if any false statement or declaration is made or used in support thereof or if any fraudulent means or devices are used by the insured person or anyone acting on his her behalf to obtain any benefit under this policy all benefits under this policy and the premium paid shall be forfeited Any amount already paid against claims made under this policy but which are found fraudulent later shall be repaid by all recipient s policyholder s who has made that particular claim who shall be jointly and severally liable for such repayment to the insurer For the purpose of this clause the expression fraud means any of the following acts committed by the Insured Person or by his agent or the hospital doctor any other party acting on behalf of the insured person with intent to deceive the insurer or to induce the insurer to issue an insurance Policy a the su...	activity The Insured Person agrees that choosing to utilize any of the wellness services or any information or advise rendered by Our Health Service Providers or Network Providers or the Company will be solely at the Insured Person s discretion and own risk and should not be used to diagnose or identify treatment for a medical or mental health condition The Wellness Points earned by the Insured person through the Wellness Program can be carried forward for a maximum of 3 years and shall have to be redeemed at the end of the 3rd Policy Year In case the Insured Person does not wish to redeem the wellness points earned the same will be forfeited In case of expiry of Policy the accrued wellness points may be carried forward for a period not exceeding three months There shall not be any cash reimbursement or redemption available against the wellness points accumulated by an Insured Person We or Our Health Service Providers or Our Network Providers do not warrant the validity accuracy comple...	relevant information sought by the company in the proposal form and other connected documents to enable it to take informed decision in the context of underwriting the risk 2 Condition Precedent to Admission of Liability The terms and conditions of the policy must be fulfilled by the insured person for the Company to make any payment for claim s arising under the policy 3 Claim Settlement provision for Penal lnterest I The Company shall settle or reject a claim as the case may be within 30 days from the date of receipt of last necessary document II ln the case of delay in the payment of a claim the Company shall be liable to pay interest to the policyholder from the date of receipt of last necessary document to the date of payment of claim at a rate 2 above the bank rate III However where the circumstances of a claim warrant an investigation in the opinion of the Company it shall initiate and complete such investigation at the earliest in any case not later than 30 days from the date o...

• Loss: MultipleNegativesRankingLoss with these parameters:

  {
      "scale": 20.0,
      "similarity_fct": "cos_sim",
      "gather_across_devices": false,
      "directions": [
          "query_to_doc"
      ],
      "partition_mode": "joint",
      "hardness_mode": null,
      "hardness_strength": 0.0
  }
  

Training Hyperparameters

Non-Default Hyperparameters

• learning_rate: 2e-05
• warmup_steps: 0.1
• weight_decay: 0.01
• load_best_model_at_end: True

All Hyperparameters

Click to expand

• per_device_train_batch_size: 8
• num_train_epochs: 3
• max_steps: -1
• learning_rate: 2e-05
• lr_scheduler_type: linear
• lr_scheduler_kwargs: None
• warmup_steps: 0.1
• optim: adamw_torch_fused
• optim_args: None
• weight_decay: 0.01
• adam_beta1: 0.9
• adam_beta2: 0.999
• adam_epsilon: 1e-08
• optim_target_modules: None
• gradient_accumulation_steps: 1
• average_tokens_across_devices: True
• max_grad_norm: 1.0
• label_smoothing_factor: 0.0
• bf16: False
• fp16: False
• bf16_full_eval: False
• fp16_full_eval: False
• tf32: None
• gradient_checkpointing: False
• gradient_checkpointing_kwargs: None
• torch_compile: False
• torch_compile_backend: None
• torch_compile_mode: None
• use_liger_kernel: False
• liger_kernel_config: None
• use_cache: False
• neftune_noise_alpha: None
• torch_empty_cache_steps: None
• auto_find_batch_size: False
• log_on_each_node: True
• logging_nan_inf_filter: True
• include_num_input_tokens_seen: no
• log_level: passive
• log_level_replica: warning
• disable_tqdm: False
• project: huggingface
• trackio_space_id: None
• trackio_bucket_id: None
• trackio_static_space_id: None
• per_device_eval_batch_size: 8
• prediction_loss_only: True
• eval_on_start: False
• eval_do_concat_batches: True
• eval_use_gather_object: False
• eval_accumulation_steps: None
• include_for_metrics: []
• batch_eval_metrics: False
• save_only_model: False
• save_on_each_node: False
• enable_jit_checkpoint: False
• push_to_hub: False
• hub_private_repo: None
• hub_model_id: None
• hub_strategy: every_save
• hub_always_push: False
• hub_revision: None
• load_best_model_at_end: True
• ignore_data_skip: False
• restore_callback_states_from_checkpoint: False
• full_determinism: False
• seed: 42
• data_seed: None
• use_cpu: False
• accelerator_config: {'split_batches': False, 'dispatch_batches': None, 'even_batches': True, 'use_seedable_sampler': True, 'non_blocking': False, 'gradient_accumulation_kwargs': None}
• parallelism_config: None
• dataloader_drop_last: False
• dataloader_num_workers: 0
• dataloader_pin_memory: True
• dataloader_persistent_workers: False
• dataloader_prefetch_factor: None
• remove_unused_columns: True
• label_names: None
• train_sampling_strategy: random
• length_column_name: length
• ddp_find_unused_parameters: None
• ddp_bucket_cap_mb: None
• ddp_broadcast_buffers: False
• ddp_static_graph: None
• ddp_backend: None
• ddp_timeout: 1800
• fsdp: None
• fsdp_config: None
• deepspeed: None
• debug: []
• skip_memory_metrics: True
• do_predict: False
• resume_from_checkpoint: None
• warmup_ratio: None
• local_rank: -1
• prompts: None
• batch_sampler: batch_sampler
• multi_dataset_batch_sampler: proportional
• router_mapping: {}
• learning_rate_mapping: {}

Training Logs

Epoch	Step	Training Loss	Validation Loss
0.8333	10	0.6838	-
1.0	12	-	0.4639
1.6667	20	0.1799	-
2.0	24	-	0.3426
2.5	30	0.2323	-
3.0	36	-	0.3644

• The bold row denotes the saved checkpoint.

Training Time

• Training: 1.7 minutes

Framework Versions

• Python: 3.12.13
• Sentence Transformers: 5.5.1
• Transformers: 5.10.2
• PyTorch: 2.11.0+cu128
• Accelerate: 1.13.0
• Datasets: 4.0.0
• Tokenizers: 0.22.2

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{oord2019representationlearningcontrastivepredictive,
      title={Representation Learning with Contrastive Predictive Coding},
      author={Aaron van den Oord and Yazhe Li and Oriol Vinyals},
      year={2019},
      eprint={1807.03748},
      archivePrefix={arXiv},
      primaryClass={cs.LG},
      url={https://arxiv.org/abs/1807.03748},
}