metadata

base_model: BAAI/bge-base-en-v1.5
datasets: []
language: []
library_name: sentence-transformers
metrics:
  - cosine_accuracy@1
  - cosine_accuracy@3
  - cosine_accuracy@5
  - cosine_accuracy@10
  - cosine_precision@1
  - cosine_precision@3
  - cosine_precision@5
  - cosine_precision@10
  - cosine_recall@1
  - cosine_recall@3
  - cosine_recall@5
  - cosine_recall@10
  - cosine_ndcg@10
  - cosine_mrr@10
  - cosine_map@100
pipeline_tag: sentence-similarity
tags:
  - sentence-transformers
  - sentence-similarity
  - feature-extraction
  - generated_from_trainer
  - dataset_size:160
  - loss:MatryoshkaLoss
  - loss:MultipleNegativesRankingLoss
widget:
  - source_sentence: >-
      Priya Softweb has specific guidelines for managing the arrival of
      international shipments. To ensure smooth customs clearance, the company
      requires an authorization letter from the client, written on their company
      letterhead. This letter must clearly state that the shipment is "Not for
      commercial purposes" to prevent the application of duty charges by the
      customs office. All international shipments should be addressed to Keyur
      Patel at Priya Softweb Solutions Pvt. Ltd., with the company's full
      address and contact information clearly indicated. Employees are advised
      to contact the HR department for the correct format of the authorization
      letter and to inform Keyur Patel about the expected arrival of such
      shipments. These procedures streamline the handling of international
      shipments and help avoid potential customs-related delays or
      complications.
    sentences:
      - >-
        Female employees at Priya Softweb are allowed to wear:- Formal
        trousers/jeans and shirts- Sarees- Formal skirts- T-shirts with collars-
        Chudidars & Kurtis- Salwar SuitsHowever, they are not allowed to wear:-
        Round neck, deep neck, cold shoulder, and fancy T-shirts- Low waist
        jeans, short T-shirts, and short shirts- Transparent wear- Wear with
        deep-cut sleeves- Capris- Slippers- Visible tattoos & piercingsPriya
        Softweb emphasizes a professional appearance for its employees while
        providing flexibility in choosing appropriate attire within the defined
        guidelines.
      - >-
        Priya Softweb has specific guidelines for managing the arrival of
        international shipments. To ensure smooth customs clearance, the company
        requires an authorization letter from the client, written on their
        company letterhead. This letter must clearly state that the shipment is
        "Not for commercial purposes" to prevent the application of duty charges
        by the customs office. All international shipments should be addressed
        to Keyur Patel at Priya Softweb Solutions Pvt. Ltd., with the company's
        full address and contact information clearly indicated. Employees are
        advised to contact the HR department for the correct format of the
        authorization letter and to inform Keyur Patel about the expected
        arrival of such shipments. These procedures streamline the handling of
        international shipments and help avoid potential customs-related delays
        or complications.
      - >-
        Priya Softweb has a structured onboarding process for new employees.
        Upon joining, new hires undergo an induction program conducted by the HR
        department. This program introduces them to the company's culture,
        values, processes, and policies, ensuring they are well-acquainted with
        the work environment and expectations. HR also facilitates introductions
        to the relevant department and sends out a company-wide email announcing
        the new employee's arrival. Additionally, new employees are required to
        complete quarterly Ethics & Compliance training to familiarize
        themselves with the company's ethical standards and compliance
        requirements. This comprehensive onboarding approach helps new employees
        integrate seamlessly into the company and quickly become productive
        members of the team.
  - source_sentence: >-
      The sanctioning and approving authority for Casual Leave, Sick Leave, and
      Privilege Leave at Priya Softweb is the Leader/Manager.
    sentences:
      - >-
        Even if an employee utilizes the 'Hybrid' Work From Home model for only
        half a day, a full count is deducted from their monthly allowance of 4
        WFH days. This clarifies that any utilization of the 'Hybrid' model,
        regardless of the duration, is considered a full WFH day and counts
        towards the monthly limit.
      - >-
        The sanctioning and approving authority for Casual Leave, Sick Leave,
        and Privilege Leave at Priya Softweb is the Leader/Manager.
      - >-
        To be eligible for gratuity at Priya Softweb, an employee must have
        completed a minimum of 5 continuous years of service. This ensures that
        only long-term employees are entitled to this benefit.
  - source_sentence: >-
      Priya Softweb utilizes Employee Agreements/Bonds as a mechanism to retain
      talent within the company. These agreements are implemented in various
      situations, including: * **Retention:** When the company seeks to retain
      valuable employees who have resigned, a 15-month bond may be applied based
      on the company's requirements. * **Freshers:** New employees with 0 to 1
      year of experience are generally subject to an 18-month bond. *
      **Rejoining:** When former employees are rehired, a 15-month bond is
      typically implemented. These bond periods vary based on the specific
      circumstances and aim to ensure a certain level of commitment from
      employees, especially in roles that require significant investment in
      training and development.
    sentences:
      - >-
        To claim gratuity, employees must submit an application form to the
        Accounts department. This formal process ensures proper documentation
        and timely processing of the gratuity payment.
      - >-
        Priya Softweb acknowledges the efforts of employees who work late hours.
        Employees working more than 11 hours on weekdays are eligible for
        reimbursement of up to Rs. 250/- for their dinner expenses. However,
        this reimbursement is subject to approval from their Department Head.
        This policy recognizes the extra effort put in by employees working
        extended hours and provides some financial compensation for their meals.
      - >-
        Priya Softweb utilizes Employee Agreements/Bonds as a mechanism to
        retain talent within the company. These agreements are implemented in
        various situations, including: * **Retention:** When the company seeks
        to retain valuable employees who have resigned, a 15-month bond may be
        applied based on the company's requirements. * **Freshers:** New
        employees with 0 to 1 year of experience are generally subject to an
        18-month bond. * **Rejoining:** When former employees are rehired, a
        15-month bond is typically implemented. These bond periods vary based on
        the specific circumstances and aim to ensure a certain level of
        commitment from employees, especially in roles that require significant
        investment in training and development.
  - source_sentence: >-
      Chewing tobacco, gutka, gum, or smoking within the office premises is
      strictly prohibited at Priya Softweb. Bringing such substances inside the
      office will lead to penalties and potentially harsh decisions from
      management. This strict policy reflects Priya Softweb's commitment to a
      healthy and clean work environment.
    sentences:
      - >-
        Chewing tobacco, gutka, gum, or smoking within the office premises is
        strictly prohibited at Priya Softweb. Bringing such substances inside
        the office will lead to penalties and potentially harsh decisions from
        management. This strict policy reflects Priya Softweb's commitment to a
        healthy and clean work environment.
      - >-
        In situations of 'Bad Weather', the HR department at Priya Softweb will
        enable the 'Work From Home' option within the OMS system based on the
        severity of the weather and potential safety risks for employees
        commuting to the office. This proactive approach prioritizes employee
        safety and allows for flexible work arrangements during adverse weather
        events.
      - Priya Softweb employees are entitled to 5 Casual Leaves (CL) per year.
  - source_sentence: >-
      Priya Softweb prioritizes the health and wellness of its employees. The
      company strongly prohibits chewing tobacco, gutka, gum, or smoking within
      the office premises. Penalties and harsh decisions from management await
      anyone found bringing such substances into the office. Furthermore,
      carrying food to the desk is not permitted. Employees are encouraged to
      use the terrace dining facility for lunch, snacks, and dinner. Priya
      Softweb also emphasizes cleanliness and orderliness in the workspace.
      Employees are responsible for maintaining their designated work areas,
      keeping them clean, organized, and free from unnecessary items. Spitting
      gutka, gum, or tobacco in the washrooms is strictly prohibited. These
      policies contribute to a healthier and more pleasant work environment for
      everyone.
    sentences:
      - >-
        Priya Softweb prioritizes the health and wellness of its employees. The
        company strongly prohibits chewing tobacco, gutka, gum, or smoking
        within the office premises. Penalties and harsh decisions from
        management await anyone found bringing such substances into the office.
        Furthermore, carrying food to the desk is not permitted. Employees are
        encouraged to use the terrace dining facility for lunch, snacks, and
        dinner. Priya Softweb also emphasizes cleanliness and orderliness in the
        workspace. Employees are responsible for maintaining their designated
        work areas, keeping them clean, organized, and free from unnecessary
        items. Spitting gutka, gum, or tobacco in the washrooms is strictly
        prohibited. These policies contribute to a healthier and more pleasant
        work environment for everyone.
      - >-
        The Performance Appraisal at Priya Softweb is solely based on the
        employee's performance evaluation. The evaluation score is compiled by
        the Team Leader/Project Manager, who also gives the final rating to the
        team member. Detailed recommendations are provided by the TL/PM, and
        increment or promotion is granted accordingly. This process ensures that
        performance is the primary factor driving salary revisions and
        promotions.
      - >-
        Priya Softweb actively promotes diversity in its hiring practices. The
        company focuses on recruiting individuals from a wide range of
        backgrounds, including different races, ethnicities, religions,
        political beliefs, education levels, socio-economic backgrounds,
        geographical locations, languages, and cultures. This commitment to
        diversity enriches the company culture and brings in a variety of
        perspectives and experiences.
model-index:
  - name: SentenceTransformer based on BAAI/bge-base-en-v1.5
    results:
      - task:
          type: information-retrieval
          name: Information Retrieval
        dataset:
          name: dim 768
          type: dim_768
        metrics:
          - type: cosine_accuracy@1
            value: 1
            name: Cosine Accuracy@1
          - type: cosine_accuracy@3
            value: 1
            name: Cosine Accuracy@3
          - type: cosine_accuracy@5
            value: 1
            name: Cosine Accuracy@5
          - type: cosine_accuracy@10
            value: 1
            name: Cosine Accuracy@10
          - type: cosine_precision@1
            value: 1
            name: Cosine Precision@1
          - type: cosine_precision@3
            value: 0.33333333333333326
            name: Cosine Precision@3
          - type: cosine_precision@5
            value: 0.20000000000000004
            name: Cosine Precision@5
          - type: cosine_precision@10
            value: 0.10000000000000002
            name: Cosine Precision@10
          - type: cosine_recall@1
            value: 1
            name: Cosine Recall@1
          - type: cosine_recall@3
            value: 1
            name: Cosine Recall@3
          - type: cosine_recall@5
            value: 1
            name: Cosine Recall@5
          - type: cosine_recall@10
            value: 1
            name: Cosine Recall@10
          - type: cosine_ndcg@10
            value: 1
            name: Cosine Ndcg@10
          - type: cosine_mrr@10
            value: 1
            name: Cosine Mrr@10
          - type: cosine_map@100
            value: 1
            name: Cosine Map@100
      - task:
          type: information-retrieval
          name: Information Retrieval
        dataset:
          name: dim 512
          type: dim_512
        metrics:
          - type: cosine_accuracy@1
            value: 1
            name: Cosine Accuracy@1
          - type: cosine_accuracy@3
            value: 1
            name: Cosine Accuracy@3
          - type: cosine_accuracy@5
            value: 1
            name: Cosine Accuracy@5
          - type: cosine_accuracy@10
            value: 1
            name: Cosine Accuracy@10
          - type: cosine_precision@1
            value: 1
            name: Cosine Precision@1
          - type: cosine_precision@3
            value: 0.33333333333333326
            name: Cosine Precision@3
          - type: cosine_precision@5
            value: 0.20000000000000004
            name: Cosine Precision@5
          - type: cosine_precision@10
            value: 0.10000000000000002
            name: Cosine Precision@10
          - type: cosine_recall@1
            value: 1
            name: Cosine Recall@1
          - type: cosine_recall@3
            value: 1
            name: Cosine Recall@3
          - type: cosine_recall@5
            value: 1
            name: Cosine Recall@5
          - type: cosine_recall@10
            value: 1
            name: Cosine Recall@10
          - type: cosine_ndcg@10
            value: 1
            name: Cosine Ndcg@10
          - type: cosine_mrr@10
            value: 1
            name: Cosine Mrr@10
          - type: cosine_map@100
            value: 1
            name: Cosine Map@100
      - task:
          type: information-retrieval
          name: Information Retrieval
        dataset:
          name: dim 256
          type: dim_256
        metrics:
          - type: cosine_accuracy@1
            value: 1
            name: Cosine Accuracy@1
          - type: cosine_accuracy@3
            value: 1
            name: Cosine Accuracy@3
          - type: cosine_accuracy@5
            value: 1
            name: Cosine Accuracy@5
          - type: cosine_accuracy@10
            value: 1
            name: Cosine Accuracy@10
          - type: cosine_precision@1
            value: 1
            name: Cosine Precision@1
          - type: cosine_precision@3
            value: 0.33333333333333326
            name: Cosine Precision@3
          - type: cosine_precision@5
            value: 0.20000000000000004
            name: Cosine Precision@5
          - type: cosine_precision@10
            value: 0.10000000000000002
            name: Cosine Precision@10
          - type: cosine_recall@1
            value: 1
            name: Cosine Recall@1
          - type: cosine_recall@3
            value: 1
            name: Cosine Recall@3
          - type: cosine_recall@5
            value: 1
            name: Cosine Recall@5
          - type: cosine_recall@10
            value: 1
            name: Cosine Recall@10
          - type: cosine_ndcg@10
            value: 1
            name: Cosine Ndcg@10
          - type: cosine_mrr@10
            value: 1
            name: Cosine Mrr@10
          - type: cosine_map@100
            value: 1
            name: Cosine Map@100
      - task:
          type: information-retrieval
          name: Information Retrieval
        dataset:
          name: dim 128
          type: dim_128
        metrics:
          - type: cosine_accuracy@1
            value: 1
            name: Cosine Accuracy@1
          - type: cosine_accuracy@3
            value: 1
            name: Cosine Accuracy@3
          - type: cosine_accuracy@5
            value: 1
            name: Cosine Accuracy@5
          - type: cosine_accuracy@10
            value: 1
            name: Cosine Accuracy@10
          - type: cosine_precision@1
            value: 1
            name: Cosine Precision@1
          - type: cosine_precision@3
            value: 0.33333333333333326
            name: Cosine Precision@3
          - type: cosine_precision@5
            value: 0.20000000000000004
            name: Cosine Precision@5
          - type: cosine_precision@10
            value: 0.10000000000000002
            name: Cosine Precision@10
          - type: cosine_recall@1
            value: 1
            name: Cosine Recall@1
          - type: cosine_recall@3
            value: 1
            name: Cosine Recall@3
          - type: cosine_recall@5
            value: 1
            name: Cosine Recall@5
          - type: cosine_recall@10
            value: 1
            name: Cosine Recall@10
          - type: cosine_ndcg@10
            value: 1
            name: Cosine Ndcg@10
          - type: cosine_mrr@10
            value: 1
            name: Cosine Mrr@10
          - type: cosine_map@100
            value: 1
            name: Cosine Map@100
      - task:
          type: information-retrieval
          name: Information Retrieval
        dataset:
          name: dim 64
          type: dim_64
        metrics:
          - type: cosine_accuracy@1
            value: 1
            name: Cosine Accuracy@1
          - type: cosine_accuracy@3
            value: 1
            name: Cosine Accuracy@3
          - type: cosine_accuracy@5
            value: 1
            name: Cosine Accuracy@5
          - type: cosine_accuracy@10
            value: 1
            name: Cosine Accuracy@10
          - type: cosine_precision@1
            value: 1
            name: Cosine Precision@1
          - type: cosine_precision@3
            value: 0.33333333333333326
            name: Cosine Precision@3
          - type: cosine_precision@5
            value: 0.20000000000000004
            name: Cosine Precision@5
          - type: cosine_precision@10
            value: 0.10000000000000002
            name: Cosine Precision@10
          - type: cosine_recall@1
            value: 1
            name: Cosine Recall@1
          - type: cosine_recall@3
            value: 1
            name: Cosine Recall@3
          - type: cosine_recall@5
            value: 1
            name: Cosine Recall@5
          - type: cosine_recall@10
            value: 1
            name: Cosine Recall@10
          - type: cosine_ndcg@10
            value: 1
            name: Cosine Ndcg@10
          - type: cosine_mrr@10
            value: 1
            name: Cosine Mrr@10
          - type: cosine_map@100
            value: 1
            name: Cosine Map@100

SentenceTransformer based on BAAI/bge-base-en-v1.5

This is a sentence-transformers model finetuned from BAAI/bge-base-en-v1.5. 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: BAAI/bge-base-en-v1.5
Maximum Sequence Length: 512 tokens
Output Dimensionality: 768 tokens
Similarity Function: Cosine Similarity

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': True}) with Transformer model: BertModel 
  (1): Pooling({'word_embedding_dimension': 768, 'pooling_mode_cls_token': True, 'pooling_mode_mean_tokens': False, 'pooling_mode_max_tokens': False, 'pooling_mode_mean_sqrt_len_tokens': False, 'pooling_mode_weightedmean_tokens': False, 'pooling_mode_lasttoken': False, 'include_prompt': True})
  (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("kr-manish/fine-tune-embedding-bge-base-HrPolicy_vfinal")
# Run inference
sentences = [
    'Priya Softweb prioritizes the health and wellness of its employees. The company strongly prohibits chewing tobacco, gutka, gum, or smoking within the office premises. Penalties and harsh decisions from management await anyone found bringing such substances into the office. Furthermore, carrying food to the desk is not permitted. Employees are encouraged to use the terrace dining facility for lunch, snacks, and dinner. Priya Softweb also emphasizes cleanliness and orderliness in the workspace. Employees are responsible for maintaining their designated work areas, keeping them clean, organized, and free from unnecessary items. Spitting gutka, gum, or tobacco in the washrooms is strictly prohibited. These policies contribute to a healthier and more pleasant work environment for everyone.',
    'Priya Softweb prioritizes the health and wellness of its employees. The company strongly prohibits chewing tobacco, gutka, gum, or smoking within the office premises. Penalties and harsh decisions from management await anyone found bringing such substances into the office. Furthermore, carrying food to the desk is not permitted. Employees are encouraged to use the terrace dining facility for lunch, snacks, and dinner. Priya Softweb also emphasizes cleanliness and orderliness in the workspace. Employees are responsible for maintaining their designated work areas, keeping them clean, organized, and free from unnecessary items. Spitting gutka, gum, or tobacco in the washrooms is strictly prohibited. These policies contribute to a healthier and more pleasant work environment for everyone.',
    "The Performance Appraisal at Priya Softweb is solely based on the employee's performance evaluation. The evaluation score is compiled by the Team Leader/Project Manager, who also gives the final rating to the team member. Detailed recommendations are provided by the TL/PM, and increment or promotion is granted accordingly. This process ensures that performance is the primary factor driving salary revisions and promotions.",
]
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

Information Retrieval

Dataset: dim_768
Evaluated with InformationRetrievalEvaluator

Metric	Value
cosine_accuracy@1	1.0
cosine_accuracy@3	1.0
cosine_accuracy@5	1.0
cosine_accuracy@10	1.0
cosine_precision@1	1.0
cosine_precision@3	0.3333
cosine_precision@5	0.2
cosine_precision@10	0.1
cosine_recall@1	1.0
cosine_recall@3	1.0
cosine_recall@5	1.0
cosine_recall@10	1.0
cosine_ndcg@10	1.0
cosine_mrr@10	1.0
cosine_map@100	1.0

Information Retrieval

Dataset: dim_512
Evaluated with InformationRetrievalEvaluator

Metric	Value
cosine_accuracy@1	1.0
cosine_accuracy@3	1.0
cosine_accuracy@5	1.0
cosine_accuracy@10	1.0
cosine_precision@1	1.0
cosine_precision@3	0.3333
cosine_precision@5	0.2
cosine_precision@10	0.1
cosine_recall@1	1.0
cosine_recall@3	1.0
cosine_recall@5	1.0
cosine_recall@10	1.0
cosine_ndcg@10	1.0
cosine_mrr@10	1.0
cosine_map@100	1.0

Information Retrieval

Dataset: dim_256
Evaluated with InformationRetrievalEvaluator

Metric	Value
cosine_accuracy@1	1.0
cosine_accuracy@3	1.0
cosine_accuracy@5	1.0
cosine_accuracy@10	1.0
cosine_precision@1	1.0
cosine_precision@3	0.3333
cosine_precision@5	0.2
cosine_precision@10	0.1
cosine_recall@1	1.0
cosine_recall@3	1.0
cosine_recall@5	1.0
cosine_recall@10	1.0
cosine_ndcg@10	1.0
cosine_mrr@10	1.0
cosine_map@100	1.0

Information Retrieval

Dataset: dim_128
Evaluated with InformationRetrievalEvaluator

Metric	Value
cosine_accuracy@1	1.0
cosine_accuracy@3	1.0
cosine_accuracy@5	1.0
cosine_accuracy@10	1.0
cosine_precision@1	1.0
cosine_precision@3	0.3333
cosine_precision@5	0.2
cosine_precision@10	0.1
cosine_recall@1	1.0
cosine_recall@3	1.0
cosine_recall@5	1.0
cosine_recall@10	1.0
cosine_ndcg@10	1.0
cosine_mrr@10	1.0
cosine_map@100	1.0

Information Retrieval

Dataset: dim_64
Evaluated with InformationRetrievalEvaluator

Metric	Value
cosine_accuracy@1	1.0
cosine_accuracy@3	1.0
cosine_accuracy@5	1.0
cosine_accuracy@10	1.0
cosine_precision@1	1.0
cosine_precision@3	0.3333
cosine_precision@5	0.2
cosine_precision@10	0.1
cosine_recall@1	1.0
cosine_recall@3	1.0
cosine_recall@5	1.0
cosine_recall@10	1.0
cosine_ndcg@10	1.0
cosine_mrr@10	1.0
cosine_map@100	1.0

Training Details

Training Dataset

Unnamed Dataset

Size: 160 training samples
Columns: positive and anchor
Approximate statistics based on the first 1000 samples:
positive anchor
type string string
details
min: 16 tokens
mean: 90.76 tokens
max: 380 tokens

min: 16 tokens
mean: 90.76 tokens
max: 380 tokens

	positive	anchor
type	string	string
details	min: 16 tokens mean: 90.76 tokens max: 380 tokens	min: 16 tokens mean: 90.76 tokens max: 380 tokens

Samples:

positive	anchor
`The general timings for the Marketing team vary: BD works from 1:00 PM to 10:00 PM or 3:00 PM to 12:00 AM, while BA/SEO works from 11:00 AM to 8:00 PM.`	`The general timings for the Marketing team vary: BD works from 1:00 PM to 10:00 PM or 3:00 PM to 12:00 AM, while BA/SEO works from 11:00 AM to 8:00 PM.`
`Priya Softweb acknowledges the efforts of employees who work late hours. Employees working more than 11 hours on weekdays are eligible for reimbursement of up to Rs. 250/- for their dinner expenses. However, this reimbursement is subject to approval from their Department Head. This policy recognizes the extra effort put in by employees working extended hours and provides some financial compensation for their meals.`	`Priya Softweb acknowledges the efforts of employees who work late hours. Employees working more than 11 hours on weekdays are eligible for reimbursement of up to Rs. 250/- for their dinner expenses. However, this reimbursement is subject to approval from their Department Head. This policy recognizes the extra effort put in by employees working extended hours and provides some financial compensation for their meals.`
`While Priya Softweb allows employees to keep their cell phones during work hours for emergency purposes, excessive personal mobile phone usage and lengthy calls within the office premises are strictly prohibited. Excessive use may result in disciplinary actions. This policy aims to strike a balance between allowing accessibility for emergencies and maintaining a productive work environment free from distractions.`	`While Priya Softweb allows employees to keep their cell phones during work hours for emergency purposes, excessive personal mobile phone usage and lengthy calls within the office premises are strictly prohibited. Excessive use may result in disciplinary actions. This policy aims to strike a balance between allowing accessibility for emergencies and maintaining a productive work environment free from distractions.`

Loss: MatryoshkaLoss with these parameters:

{
    "loss": "MultipleNegativesRankingLoss",
    "matryoshka_dims": [
        768,
        512,
        256,
        128,
        64
    ],
    "matryoshka_weights": [
        1,
        1,
        1,
        1,
        1
    ],
    "n_dims_per_step": -1
}

Training Hyperparameters

Non-Default Hyperparameters

eval_strategy: epoch
per_device_train_batch_size: 16
per_device_eval_batch_size: 16
gradient_accumulation_steps: 16
learning_rate: 3e-05
num_train_epochs: 15
lr_scheduler_type: cosine
warmup_ratio: 0.1
fp16: True
load_best_model_at_end: True
optim: adamw_torch_fused

All Hyperparameters

Click to expand

overwrite_output_dir: False
do_predict: False
eval_strategy: epoch
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: 16
eval_accumulation_steps: None
learning_rate: 3e-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: 15
max_steps: -1
lr_scheduler_type: cosine
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_fused
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
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: batch_sampler
multi_dataset_batch_sampler: proportional

Training Logs

Epoch	Step	Training Loss	dim_128_cosine_map@100	dim_256_cosine_map@100	dim_512_cosine_map@100	dim_64_cosine_map@100	dim_768_cosine_map@100
0	0	-	1.0	1.0	1.0	1.0	1.0
1.0	1	-	1.0	1.0	1.0	1.0	1.0
2.0	3	-	1.0	1.0	1.0	1.0	1.0
3.0	4	-	1.0	1.0	1.0	1.0	1.0
4.0	6	-	1.0	1.0	1.0	1.0	1.0
5.0	8	-	1.0	1.0	1.0	1.0	1.0
6.0	9	-	1.0	1.0	1.0	1.0	1.0
6.4	10	0.0767	-	-	-	-	-
7.0	11	-	1.0	1.0	1.0	1.0	1.0
8.0	12	-	1.0	1.0	1.0	1.0	1.0
9.0	13	-	1.0	1.0	1.0	1.0	1.0
10.0	15	-	1.0	1.0	1.0	1.0	1.0

The bold row denotes the saved checkpoint.

Framework Versions

Python: 3.10.12
Sentence Transformers: 3.0.1
Transformers: 4.41.2
PyTorch: 2.1.2+cu121
Accelerate: 0.32.1
Datasets: 2.19.1
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",
}

MatryoshkaLoss

@misc{kusupati2024matryoshka,
    title={Matryoshka Representation Learning}, 
    author={Aditya Kusupati and Gantavya Bhatt and Aniket Rege and Matthew Wallingford and Aditya Sinha and Vivek Ramanujan and William Howard-Snyder and Kaifeng Chen and Sham Kakade and Prateek Jain and Ali Farhadi},
    year={2024},
    eprint={2205.13147},
    archivePrefix={arXiv},
    primaryClass={cs.LG}
}

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}
}