ShushantLLM/distilroberta-ai-job-embeddings

SentenceTransformer based on sentence-transformers/all-distilroberta-v1

This is a sentence-transformers model finetuned from sentence-transformers/all-distilroberta-v1 on the ai-job-embedding-finetuning dataset. 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: sentence-transformers/all-distilroberta-v1
• Maximum Sequence Length: 512 tokens
• Output Dimensionality: 768 dimensions
• Similarity Function: Cosine Similarity
• Training Dataset:
  • ai-job-embedding-finetuning

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, 'architecture': 'RobertaModel'})
  (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("ShushantLLM/distilroberta-ai-job-embeddings")
# Run inference
queries = [
    "Data organization, document analysis, records management",
]
documents = [
    'skills and build your career in a rapidly evolving business climate? Are you looking for a career where professional development is embedded in your employer’s core culture? If so, Chenega Military, Intelligence & Operations Support (MIOS) could be the place for you! Join our team of professionals who support large-scale government operations by leveraging cutting-edge technology and take your career to the next level!\n\nAs one of the newest Chenega companies, Chenega Defense & Aerospace Solutions (CDAS) was developed with the purpose of providing expert Engineering and Technical Support Services to federal customers.\n\nThe Data Analyst will analyze a large variety of documents to ensure proper placement in physical files, perform high-level scanning of master file documents to convert them into an electronic format, and provide meticulous organization and management of case files, including sorting and categorizing documents before scanning.\n\nResponsibilities\n\nWork within the Standard Operating Procedure for the organization of physical files containing documents of various types Establish or maintain physical files, including proper placement of documents as they are createdDisseminate significant amounts of information with attention to detail and accuracyPerform word processing tasksPerform data entry and metadata entry for electronic documentsReconcile inconsistenciesGather information and organize investigative packages, case files, or presentationsObtain additional information from other investigative agencies or databasesVerify information and files against the tracking systemMaintain internal status information on the disposition of designated information and filesDistribute and receive documentsAssist analyst or government official in obtaining or collecting all documents or information to complete case fileProvide administrative information and assistance concerning the case or files to other agencies or organizationsOther duties as assigned\n\n\nQualifications\n\nHigh school diploma or GED equivalent required Must have resided in the United States for at least three out of the last five years or worked for the U.S. in a foreign country as either an employee or contractor in a federal or military capacity for at least three of the last five yearsHaving your own Personally Owned Vehicle (POV) is requiredPossess a demonstrated ability to analyze documents to extract informationGood oral and written communication skillsHave hands-on familiarity with a variety of computer applications,Must have a working knowledge of a variety of computer software applications in word processing, spreadsheets, databases, presentation software (MS Word, Excel, PowerPoint), and OutlookA valid driver’s license is requiredTop Secret clearance required \n\n\nKnowledge, Skills, And Abilities\n\nPossess a demonstrated ability to analyze documents to extract informationGood oral and written communication skillsHave hands-on familiarity with a variety of computer applications, including word processing, database, spreadsheet, and telecommunications softwareMust be a team playerMust be able to work independently and with USMS staff to interpret data rapidly and accurately for proper execution in a records management databaseMust have a working knowledge of a variety of computer software applications in word processing, spreadsheets, databases, presentation software (MS Word, Excel, Access, PowerPoint), and OutlookAbility to work independently on tasks be a self-starter and complete projects with a team as they ariseAttention to detail and the ability to direct the work of others efficiently and effectivelyAbility to consistently deliver high-quality work under extreme pressureAbility to work shiftworkAbility to lift and move boxes up to 25 pounds, including frequently utilizing hands, arms, and legs for file placement and removalExperience with scanning software\n\n\nHow You’ll Grow\n\nAt Chenega MIOS, our professional development plan focuses on helping our team members at every level of their career to identify and use their strengths to do their best work every day. From entry-level employees to senior leaders, we believe there’s always room to learn.\n\nWe offer opportunities to help sharpen skills in addition to hands-on experience in the global, fast-changing business world. From on-the-job learning experiences to formal development programs, our professionals have a variety of opportunities to continue to grow throughout their careers.\n\nBenefits\n\nAt Chenega MIOS, we know that great people make a great organization. We value our team members and offer them a broad range of benefits.\n\nLearn more about what working at Chenega MIOS can mean for you.\n\nChenega MIOS’s culture\n\nOur positive and supportive culture encourages our team members to do their best work every day. We celebrate individuals by recognizing their uniqueness and offering them the flexibility to make daily choices that can help them be healthy, centered, confident, and aware. We offer well-being programs and continuously look for new ways to maintain a culture where we excel and lead healthy, happy lives.\n\nCorporate citizenship\n\nChenega MIOS is led by a purpose to make an impact that matters. This purpose defines who we are and extends to relationships with our clients, our team members, and our communities. We believe that business has the power to inspire and transform. We focus on education, giving, skill-based volunteerism, and leadership to help drive positive social impact in our communities.\n\nLearn more about Chenega’s impact on the world.\n\nChenega MIOS News- https://chenegamios.com/news/\n\nTips from your Talent Acquisition team\n\nWe Want Job Seekers Exploring Opportunities At Chenega MIOS To Feel Prepared And Confident. To Help You With Your Research, We Suggest You Review The Following Links\n\nChenega MIOS web site - www.chenegamios.com\n\nGlassdoor - https://www.glassdoor.com/Overview/Working-at-Chenega-MIOS-EI_IE369514.11,23.htm\n\nLinkedIn - https://www.linkedin.com/company/1472684/\n\nFacebook - https://www.facebook.com/chenegamios/\n\n#DICE\n\n#Chenega Defense & Aerospace Solutions, LLC',
    'skills will be difficult. The more aligned skills they have, the better.Organizational Structure And Impact:Describe the function your group supports from an LOB perspective:Experienced ML engineer to work on universal forecasting models. Focus on ML forecasting, Python and Hadoop. Experience with Python, ARIMA, FB Prophet, Seasonal Naive, Gluon.Data Science Innovation (DSI) is a very unique application. It is truly ML-driven at its heart and our forecasting models originally looked singularly at cash balance forecasting. That has all changed as we have now incorporated approximately 100 additional financial metrics from our new DSI Metrics Farm. This allows future model executions to become a Universal Forecasting Model instead of being limited to just cash forecasting. It’s a very exciting application, especially since the models have been integrated within a Marketplace concept UI that allows Subscriber/Contributor functionality to make information and processing more personal and with greater extensibility across the enterprise. The application architecture is represented by OpenShift, Linux, Oracle, SQL Server, Hadoop, MongoDB, APIs, and a great deal of Python code.Describe the current initiatives that this resource will be impacting:Working toward implementation of Machine Learning Services.Team Background and Preferred Candidate History:Do you only want candidates with a similar background or would you like to see candidates with a diverse industry background?Diverse industry background, finance background preferred. Manager is more focused on the skillset.Describe the dynamic of your team and where this candidate will fit into the overall environment:This person will work with a variety of titles including application architects, web engineers, data engineers, data scientists, application system managers, system integrators, and Quality Engineers.Will work with various teams, but primarily working with one core team - approx 15 - onshore and offshore resources.Candidate Technical and skills profile:Describe the role and the key responsibilities in order of which they will be doing daily:Machine Learning Engineer that work with Data Scientists in a SDLC environment into production.Interviews:Describe interview process (who will be involved, how many interviews, etc.):1 round - 1 hour minimum, panel style',
    "Qualifications\n Data Engineering, Data Modeling, and ETL (Extract Transform Load) skillsData Warehousing and Data Analytics skillsExperience with data-related tools and technologiesStrong problem-solving and analytical skillsExcellent written and verbal communication skillsAbility to work independently and remotelyExperience with cloud platforms (e.g., AWS, Azure) is a plusBachelor's degree in Computer Science, Information Systems, or related field",
]
query_embeddings = model.encode_query(queries)
document_embeddings = model.encode_document(documents)
print(query_embeddings.shape, document_embeddings.shape)
# [1, 768] [3, 768]

# Get the similarity scores for the embeddings
similarities = model.similarity(query_embeddings, document_embeddings)
print(similarities)
# tensor([[ 0.4715, -0.0160,  0.0444]])

Evaluation

Metrics

Triplet

• Datasets: ai-job-validation and ai-job-test
• Evaluated with TripletEvaluator

Metric	ai-job-validation	ai-job-test
cosine_accuracy	1.0	0.9706

Training Details

Training Dataset

ai-job-embedding-finetuning

• Dataset: ai-job-embedding-finetuning at 1de228a
• Size: 810 training samples
• Columns: query, job_description_pos, and job_description_neg
• Approximate statistics based on the first 810 samples:

  	query	job_description_pos	job_description_neg
  type	string	string	string
  details	min: 7 tokens mean: 15.02 tokens max: 38 tokens	min: 10 tokens mean: 350.29 tokens max: 512 tokens	min: 7 tokens mean: 354.92 tokens max: 512 tokens

• Samples:

  query	job_description_pos	job_description_neg
  Senior Data Analyst, monitoring systems, dashboard development, statistical analysis	experience where you can also make an impact on your community. While safety is a serious business, we are a supportive team that is optimizing the remote experience to create strong and fulfilling relationships even when we are physically apart. Our group of hard-working employees thrive in a positive and inclusive environment, where a bias towards action is rewarded. We have raised over $380M in venture capital from investors including Tiger Global, Andreessen Horowitz, Matrix Partners, Meritech Capital Partners, and Initialized Capital. Now surpassing a $3B valuation, Flock is scaling intentionally and seeking the best and brightest to help us meet our goal of reducing crime in the United States by 25% in the next three years. The Opportunity As a Senior Data Analyst on the ML team, you will be responsible for extracting insights aggregated from various data sources, developing dashboards to identify trends and patterns that highlight model performance issues, performing analysis...	SKILLS and EXPERIENCE:3-5+ years of experience domain knowledge with either support of core Banking application experience, Mortgage Servicing or Loan Originations or personal or auto loans within Finance Industry environmentAble to interact with the VP or C-level Business Executives and higher to gather requirements and collaborate with IT; working effectively and independently as well as be collaborative team-oriented team player.Ideally supported Mortgage servicing systems such as Black Knight’s MSP, Sagent, Finastra’s Fusion Servicing Director, Interlinq Loan Servicing (ILS) or other loan servicing platform OR support of other core banking or originations platformSome experience with the following core technologies: T-SQL; SQL Server 2016 or higher; Visual Studio 2017 or higher; SQL Server Data Tools; Team Foundation ServerWorking knowledge of T-SQL programming and scripting, as well as optimization techniques· 3 years of experience with a strong focus on SQL Relational databases, ...
  advanced analytics, financial strategy, data visualization	skills and business acumen to drive impactful results that inform strategic decisions.Commitment to iterative development, with a proven ability to engage and update stakeholders bi-weekly or as necessary, ensuring alignment, feedback incorporation, and transparency throughout the project lifecycle.Project ownership and development from inception to completion, encompassing tasks such as gathering detailed requirements, data preparation, model creation, result generation, and data visualization. Develop insights, methods or tools using various analytic methods such as causal-model approaches, predictive modeling, regressions, machine learning, time series analysis, etc.Handle large amounts of data from multiple and disparate sources, employing advanced Python and SQL techniques to ensure efficiency and accuracyUphold the highest standards of data integrity and security, aligning with both internal and external regulatory requirements and compliance protocols Required Qualifications, C...	experience Life at Visa. Job Description About the Team: VISA is the leader in the payment industry and has been for a long time, but we are also quickly transitioning into a technology company that is fostering an environment for applying the newest technology to solve exciting problems in this area. For a payment system to work well, the risk techniques, performance, and scalability are critical. These techniques and systems benefit from big data, data mining, artificial intelligence, machine learning, cloud computing, & many other advance technologies. At VISA, we have all of these. If you want to be on the cutting edge of the payment space, learn fast, and make a big impact, then the Artificial Intelligence Platform team may be an ideal place for you! Our team needs a Senior Data Engineer with proven knowledge of web application and web service development who will focus on creating new capabilities for the AI Platform while maturing our code base and development processes. You...
  Clinical Operations data analysis, eTMF, EDC implementation, advanced analytics visualization	requirements, and objectives for Clinical initiatives Technical SME for system activities for the clinical system(s), enhancements, and integration projects. Coordinates support activities across vendor(s) Systems include but are not limited to eTMF, EDC, CTMS and Analytics Interfaces with external vendors at all levels to manage the relationship and ensure the proper delivery of services Document Data Transfer Agreements for Data Exchange between BioNTech and Data Providers (CRO, Partner Organizations) Document Data Transformation logic and interact with development team to convert business logic into technical details What you have to offer: Bachelor’s or higher degree in a scientific discipline (e.g., computer science/information systems, engineering, mathematics, natural sciences, medical, or biomedical science) Extensive experience/knowledge of technologies and trends including Visualizations /Advanced Analytics Outstanding analytical skills and result orientation Ab...	Requirements Typically requires 13+ years of professional experience and 6+ years of diversified leadership, planning, communication, organization, and people motivation skills (or equivalent experience). Critical Skills 12+ years of experience in a technology role; proven experience in a leadership role, preferably in a large, complex organization.8+ years Data Engineering, Emerging Technology, and Platform Design experience4+ years Leading large data / technical teams – Data Engineering, Solution Architects, and Business Intelligence Engineers, encouraging a culture of innovation, collaboration, and continuous improvement.Hands-on experience building and delivering Enterprise Data SolutionsExtensive market knowledge and experience with cutting edge Data, Analytics, Data Science, ML and AI technologiesExtensive professional experience with ETL, BI & Data AnalyticsExtensive professional experience with Big Data systems, data pipelines and data processingDeep expertise in Data Archit...

• Loss: MultipleNegativesRankingLoss with these parameters:

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

Evaluation Dataset

ai-job-embedding-finetuning

• Dataset: ai-job-embedding-finetuning at 1de228a
• Size: 101 evaluation samples
• Columns: query, job_description_pos, and job_description_neg
• Approximate statistics based on the first 101 samples:

  	query	job_description_pos	job_description_neg
  type	string	string	string
  details	min: 9 tokens mean: 15.09 tokens max: 28 tokens	min: 15 tokens mean: 346.03 tokens max: 512 tokens	min: 14 tokens mean: 323.36 tokens max: 512 tokens

• Samples:

  query	job_description_pos	job_description_neg
  Azure Data Factory, Databricks, Snowflake architecture	Skills: SQL, PySpark, Databricks, Azure Synapse, Azure Data Factory. Need hands-on coding Requirements:1. Extensive knowledge of any of the big cloud services - Azure, AWS or GCP with practical implementation (like S3, ADLS, Airflow, ADF, Lamda, BigQuery, EC2, Fabric, Databricks or equivalent)2. Strong Hands-on experience in SQL and Python/PySpark programming knowledge. Should be able to write code during an interview with minimal syntax error.3. Strong foundational and architectural knowledge of any of the data warehouses - Snowflake, Redshift. Synapse etc.4. Should be able to drive and deliver projects with little or no guidance. Take ownership, become a self-learner, and have leadership qualities.	experience for yourself, and a better working world for all. Data Analyst, Technology Consulting - Data & Analytics (Data Governance & Controls) - Financial Services Office (Manager) (Multiple Positions), Ernst & Young U.S. LLP, New York, NY. Work with clients to transform the way they use and manage data by architecting data strategies, providing end-to-end solutions that focus on improving their data supply chain, reengineering processes, enhancing risk control, and enabling information intelligence by harnessing latest advanced technologies. Solve complex issues and drive growth across financial services. Define data and analytic strategies by performing assessments, recommending remediation strategies/solutions based on aggregated view of identified gaps, and designing/implementing future state data and analytics solutions. Manage and coach diverse teams of professionals with different backgrounds. Manage cross functional teams, to ensure project task and timeline accountability...
  Big Data Engineer, Spark, Hadoop, AWS GCP	Skills • Expertise and hands-on experience on Spark, and Hadoop echo system components – Must Have • Good and hand-on experience* of any of the Cloud (AWS/GCP) – Must Have • Good knowledge of HiveQL & SparkQL – Must Have Good knowledge of Shell script & Java/Scala/python – Good to Have • Good knowledge of SQL – Good to Have • Good knowledge of migration projects on Hadoop – Good to Have • Good Knowledge of one of the Workflow engines like Oozie, Autosys – Good to Have Good knowledge of Agile Development– Good to Have • Passionate about exploring new technologies – Good to Have • Automation approach – Good to Have Thanks & RegardsShahrukh KhanEmail: [email protected]	Requirements: We're looking for a candidate with exceptional proficiency in Google Sheets. This expertise should include manipulating, analyzing, and managing data within Google Sheets. The candidate should be outstanding at extracting business logic from existing reports and implementing it into new ones. Although a basic understanding of SQL for tasks related to data validation and metrics calculations is beneficial, the primary skill we are seeking is proficiency in Google Sheets. This role will involve working across various cross-functional teams, so strong communication skills are essential. The position requires a meticulous eye for detail, a commitment to delivering high-quality results, and above all, exceptional competency in Google Sheets Google sheet knowledge is preferred.Strong Excel experience without Google will be considered.Data Validation and formulas to extract data are a mustBasic SQL knowledge is required.Strong communications skills are requiredInterview process...
  Energy policy analysis, regulatory impact modeling, distributed energy resource management.	skills, modeling, energy data analysis, and critical thinking are required for a successful candidate. Knowledge of energy systems and distributed solar is required. Reporting to the Senior Manager of Government Affairs, you will work across different teams to model data to inform policy advocacy. The ability to obtain data from multiple sources, including regulatory or legislative hearings, academic articles, and reports, are fundamental to the role. A willingness to perform under deadlines and collaborate within an organization is required. Honesty, accountability, and integrity are a must. Energy Policy & Data Analyst Responsibilities Support Government Affairs team members with energy policy recommendations based on data modelingEvaluate relevant regulatory or legislative filings and model the impacts to Sunnova’s customers and businessAnalyze program proposals (grid services, incentives, net energy metering, fixed charges) and develop recommendations that align with Sunnova’s ...	QualificationsData Engineering, Data Modeling, and ETL (Extract Transform Load) skillsMonitor and support data pipelines and ETL workflowsData Warehousing and Data Analytics skillsExperience with Azure cloud services and toolsStrong problem-solving and analytical skillsProficiency in SQL and other programming languagesExperience with data integration and data migrationExcellent communication and collaboration skillsBachelor's degree in Computer Science, Engineering, or related field Enterprise Required SkillsPython, Big data, Data warehouse, ETL, Development, azure, Azure Data Factory, Azure Databricks, Azure SQL Server, Snowflake, data pipelines Top Skills Details1. 3+ years with ETL Development with Azure stack (Azure Data Factory, Azure Databricks, Azure Blob, Azure SQL). 2. 3+ years with Spark, SQL, and Python. This will show up with working with large sets of data in an enterprise environment. 3. Looking for Proactive individuals who have completed projects from start to complet...

• Loss: MultipleNegativesRankingLoss with these parameters:

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

Training Hyperparameters

Non-Default Hyperparameters

• eval_strategy: steps
• per_device_train_batch_size: 16
• per_device_eval_batch_size: 16
• learning_rate: 2e-05
• num_train_epochs: 1
• warmup_ratio: 0.1
• 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: 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: 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: 42
• data_seed: None
• jit_mode_eval: False
• bf16: False
• 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: False
• ignore_data_skip: False
• fsdp: []
• fsdp_min_num_params: 0
• fsdp_config: {'min_num_params': 0, 'xla': False, 'xla_fsdp_v2': False, 'xla_fsdp_grad_ckpt': False}
• fsdp_transformer_layer_cls_to_wrap: None
• accelerator_config: {'split_batches': False, 'dispatch_batches': None, 'even_batches': True, 'use_seedable_sampler': True, 'non_blocking': False, 'gradient_accumulation_kwargs': None}
• parallelism_config: 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
• project: huggingface
• trackio_space_id: trackio
• 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
• hub_revision: None
• 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
• include_tokens_per_second: False
• include_num_input_tokens_seen: no
• neftune_noise_alpha: None
• optim_target_modules: None
• batch_eval_metrics: False
• eval_on_start: False
• use_liger_kernel: False
• liger_kernel_config: None
• eval_use_gather_object: False
• average_tokens_across_devices: True
• prompts: None
• batch_sampler: no_duplicates
• multi_dataset_batch_sampler: proportional
• router_mapping: {}
• learning_rate_mapping: {}

Training Logs

Epoch	Step	ai-job-validation_cosine_accuracy	ai-job-test_cosine_accuracy
-1	-1	1.0	0.9706

Framework Versions

• Python: 3.12.12
• Sentence Transformers: 5.1.2
• Transformers: 4.57.1
• PyTorch: 2.8.0+cu126
• Accelerate: 1.11.0
• Datasets: 4.0.0
• Tokenizers: 0.22.1

Citation

BibTeX

Sentence Transformers

@inproceedings{reimers-2019-sentence-bert,
    title = "Sentence-BERT: Sentence Embeddings using Siamese BERT-Networks",
    author = "Reimers, Nils and Gurevych, Iryna",
    booktitle = "Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing",
    month = "11",
    year = "2019",
    publisher = "Association for Computational Linguistics",
    url = "https://arxiv.org/abs/1908.10084",
}

MultipleNegativesRankingLoss

@misc{henderson2017efficient,
    title={Efficient Natural Language Response Suggestion for Smart Reply},
    author={Matthew Henderson and Rami Al-Rfou and Brian Strope and Yun-hsuan Sung and Laszlo Lukacs and Ruiqi Guo and Sanjiv Kumar and Balint Miklos and Ray Kurzweil},
    year={2017},
    eprint={1705.00652},
    archivePrefix={arXiv},
    primaryClass={cs.CL}
}