Add SetFit model

1_Pooling/config.json

@@ -1,7 +1,7 @@
 {
-  "word_embedding_dimension": 384,
-  "pooling_mode_cls_token": true,
-  "pooling_mode_mean_tokens": false,
+  "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,

README.md

@@ -5,7 +5,7 @@ tags:
 - sentence-transformers
 - text-classification
 - generated_from_setfit_trainer
-base_model: BAAI/bge-small-en-v1.5
+base_model: sentence-transformers/paraphrase-mpnet-base-v2
 metrics:
 - accuracy
 widget:
@@ -390,25 +390,11 @@ widget:
     \ employer.,"
 pipeline_tag: text-classification
 inference: true
-model-index:
-- name: SetFit with BAAI/bge-small-en-v1.5
-  results:
-  - task:
-      type: text-classification
-      name: Text Classification
-    dataset:
-      name: Unknown
-      type: unknown
-      split: test
-    metrics:
-    - type: accuracy
-      value: 0.875
-      name: Accuracy
 ---
 
-# SetFit with BAAI/bge-small-en-v1.5
+# SetFit with sentence-transformers/paraphrase-mpnet-base-v2
 
-This is a [SetFit](https://github.com/huggingface/setfit) model that can be used for Text Classification. This SetFit model uses [BAAI/bge-small-en-v1.5](https://huggingface.co/BAAI/bge-small-en-v1.5) as the Sentence Transformer embedding model. A [LogisticRegression](https://scikit-learn.org/stable/modules/generated/sklearn.linear_model.LogisticRegression.html) instance is used for classification.
+This is a [SetFit](https://github.com/huggingface/setfit) model that can be used for Text Classification. This SetFit model uses [sentence-transformers/paraphrase-mpnet-base-v2](https://huggingface.co/sentence-transformers/paraphrase-mpnet-base-v2) as the Sentence Transformer embedding model. A [LogisticRegression](https://scikit-learn.org/stable/modules/generated/sklearn.linear_model.LogisticRegression.html) instance is used for classification.
 
 The model has been trained using an efficient few-shot learning technique that involves:
 
@@ -419,7 +405,7 @@ The model has been trained using an efficient few-shot learning technique that i
 
 ### Model Description
 - **Model Type:** SetFit
-- **Sentence Transformer body:** [BAAI/bge-small-en-v1.5](https://huggingface.co/BAAI/bge-small-en-v1.5)
+- **Sentence Transformer body:** [sentence-transformers/paraphrase-mpnet-base-v2](https://huggingface.co/sentence-transformers/paraphrase-mpnet-base-v2)
 - **Classification head:** a [LogisticRegression](https://scikit-learn.org/stable/modules/generated/sklearn.linear_model.LogisticRegression.html) instance
 - **Maximum Sequence Length:** 512 tokens
 - **Number of Classes:** 2 classes
@@ -439,13 +425,6 @@ The model has been trained using an efficient few-shot learning technique that i
 | 0     | <ul><li>"Driver Manager - Days - Driver Manager - Olathe, KS\n\nThe role | An Driver Manager is responsible for\n\nLearning the business from the ground up through our hands-on training program that includes exposure to driver and equipment management, customer service, and transportation logistics.Maintaining a high level of engagement and cultivating positive working relationships with your fleet of assigned drivers.Reviewing scheduled pick-up and delivery appointments daily.Confirming that each driver is fully informed of all customer and company expectations on every load at the point of dispatch.Coordinating drivers’ scheduled home time, downtime, current status, and predicted time available.Addressing safety non-conformance and violations/incidents.Escalate matters impacting on-time delivery to appropriate departments as they occur.\n\nCandidates should expect to spend the majority of their day making/receiving calls.\n\nThe requirements | This will be a perfect fit for you if...\n\nSelf-motivated with a desire to learn about the growing transportation industry.Strong multi-tasking ability.You can type at least 40-45 wpm (preferred).Computer proficient and able to navigate between multiple programs.Excellent written and oral communication skills.\n\nThe details | What are the hours, pay, and location? \n\nThis is a full-time, in-office position located in south Olathe, KS.Starting salary of $50,000 + Incentives Schedule: Must be Flexible 5:00a-5:00p.\nThe perks | What's in it for you?\n\nA casual-dress, smoke-free work environmentHealth, dental, life, and disability insurance coverage401(k) plan with company matchPaid time off and additional time each anniversaryDiscounted gym memberships, mobile phone services, tires\n\nJob Type: Full-time\n\nTransAm is committed to the principles of equal employment opportunity and nondiscrimination.,"</li><li>'Senior Scientist, In Vivo Ocular Pharmacology - Description:\n\nJohnson & Johnson is recruiting for a Senior Scientist, Specialty Ophthalmology Discovery located in Spring House PA to support drug discovery of novel therapeutics for retinal disease.\n\nAt Johnson & Johnson,\u202fwe believe health is everything. Our strength in healthcare innovation empowers us to build a\u202fworld where complex diseases are prevented, treated, and cured,\u202fwhere treatments are smarter and less invasive, and\u202fsolutions are personal. Through our expertise in Innovative Medicine and MedTech, we are uniquely positioned to innovate across the full spectrum of healthcare solutions today to deliver the breakthroughs of tomorrow, and profoundly impact health for humanity. Learn more at https://www.jnj.com/.\u202f\n\nWe are seeking a highly motivated and talented lab-based Senior Scientist to join our Retinal Diseases Discovery team located in Spring House, PA (USA). The successful candidate will join a dynamic, multi-disciplinary team of exploratory scientists and play a key role in the evaluation of new drug concepts using relevant model systems. He/she will use emerging developments in these fields to generate novel target ideas and provide technical and strategic input with the goal of progressing novel drug candidates into translational development. The qualified candidate will work in cross-functional teams, including translational medicine, discovery, biomarker teams, program project teams and disease-specific working groups, to shape a discovery and development strategy for novel drug candidates and will bring forward novel translational animal models and innovative technologies to support the discovery pipeline.\n\nThis role is laboratory-based and requires an established background in retinal cell biology, neuroscience, or metabolism. Responsibilities include the execution of discovery projects to develop new retinal disease therapies and expand the retina portfolio. The selected candidate will contribute to research programs through validation of new targets and design and execution of team-based research to help drive project advancements.\n\nCore Responsibilities\n\nDesign and conduct experiments to support new target validation, analysis, and interpretation of results, with focus on in vivo assays/models to support discovery and translational drug development programs.Establish robust discovery and/or pharmacology data packages aimed at the progression of our differentiated assets into the clinic.Evolve project strategy in collaboration with the core team and functional partners and present scientific progress to discovery translational science, governance and leadership teams.Establish and execute against timelines to enable project progression while working in a fast-paced and highly matrixed environment.Support collaborations with academic investigators, including advising on experimentation and analysis of results, and conducting complementary experiments.Contribute to the preparation and submission of technical reports, patent applications, and manuscripts as appropriate.Ensure compliance with all company training, documentation, and ensure safe laboratory working practices.\n\nQualifications:\n\n A Ph.D. degree in the biological sciences (or equivalent) with 1 year of post-doctoral experience in the pharmaceutical industry or academic environment, or B.S. or M.S. degree in the biological sciences with a minimum of 8 years of experience in relevant pharmaceutical industry setting.Demonstrated experience in vivo studies, especially creation and characterization of animal models of retinal disease. Expertise in performing pharmacological and mechanistic studies in models of eye/ocular disease is highly desired.Established background in cellular and molecular biology and competency with in vitro techniques is preferred.A background in ophthalmic drug discovery and translational models of ocular disease is preferred.Experience with validation of new target concepts is required, with keen knowledge of experimental design, underlying scientific and biological principles, and data analysis and interpretation (i.e., from hypothesis through planning and execution of experiments in support of retinal disease-related programs).Established record of scientific accomplishments directed toward the discovery and development of therapeutic agents, including strong publication record in journals, oral presentations within and outside industry, and participation in professional societies is required.Must be highly motivated, with excellent organizational skills, and capable of working collaboratively in a fast-paced and highly matrixed environment.Ability to forge and foster collaborations (internal and external), deliver scientific content to diverse audiences, and agility and adaptability working across multiple projects is highly desirable.\n\nThe base pay range for this position is $104,000 to $166,750.\n\nThe Company maintains highly competitive, performance-based compensation programs. Under current guidelines, this position is eligible for an annual performance bonus in accordance with the terms of the applicable plan. The annual performance bonus is a cash bonus intended to provide an incentive to achieve annual targeted results by rewarding for individual and the corporation’s performance over a calendar/ performance year. Bonuses are awarded at the Company’s discretion on an individual basis.\n\nEmployees may be eligible to participate in Company employee benefit programs such as health insurance, savings plan, pension plan, disability plan, vacation pay, sick time, holiday pay, and work, personal and family time off in accordance with the terms of the applicable plans. Additional information can be found through the link below.\n\n\u202f\n\nEligible for benefits to include medical, dental, vision and time off as well as any others as provided for in the applicable Collective Bargaining Agreement.\n\nFor additional general information on company benefits, please go to: - https://www.careers.jnj.com/employee-benefits\n\nJohnson & Johnson is an Affirmative Action and Equal Opportunity Employer. All qualified applicants will receive consideration for employment without regard to race, color, religion, sex, sexual orientation, gender identity, age, national origin, or protected veteran status and will not be discriminated against on the basis of disability.\n\n,'</li><li>"Public Relations Manager - Bullhorn is the global leader in software for the staffing industry. After more than 20 years, more than 10,000 companies rely on Bullhorn’s cloud-based platform to power their staffing processes from start to finish. Led by the original co-founder, partnered with venture capital, and powered by seasoned leaders across a global workforce with an eye toward innovation, Bullhorn has had year over year growth, making it the market leader in the recruitment software space while allowing for new opportunities for over 29% of our employees to advance their careers in the past 12 months.\n\nWe are a remote-first organization and over 38% of our employees reside outside the United States. Headquartered in Boston, we also have offices in London, Brighton, Rotterdam, Frankfurt and Sydney (just in case you’re in the area to stop by). Whether you’re local or remote, our vision is to ensure every employee has a sense of belonging, a voice that is heard, and a clear path for success. Your incredible experience as an employee will consist of flexible work hours to ensure a positive work-life balance and use Zoom, Slack, and other tools to stay connected.\n\nReporting into the Director, Global Content and Communications, the PR / Communications Manager will be responsible for strengthening Bullhorn’s position as a leader in the market, demonstrating our expertise in staffing as well as our growth as a technology business. This role will craft Bullhorn’s point of view on key topics and then work independently and with teammates to get those messages into market through a combination of earned and owned media – all to raise awareness of the business and advance our thought leadership efforts.\n\nA key part of this role will be partnering with our PR agency to identify story angles that highlight Bullhorn’s unique strengths and point of view. Are you obsessed with keeping up with the latest trends in AI? Do you comb the business section of your favourite news outlet trying to parse what broader trends mean to your customers? The Communications Manager needs to have a finger on the pulse of the staffing industry and an interest in tracking the economic and technology trends that will impact it.\n\nAs the lead for Bullhorn’s public relations efforts, this person will also consult with international marketing colleagues on PR efforts in markets outside the U.S.\n\nOccasional (<10%) travel may be required. Ability to work on East Coast hours preferred. Please submit a portfolio or writing samples with your resume.\n\nResponsibilities Include\n\nDeveloping, overseeing, and measuring the performance of Bullhorn’s earned media strategyHandling incoming media inquiries and occasional proactive pitchingManaging PR agency relationshipsManaging executive thought leadership opportunities, including owned contentCultivating a bench of internal experts and customer spokespeopleWriting and editing press releases, bylines, and media comments, when not supported by agencySupporting Bullhorn’s thought leadership strategy by identifying opportunities to publicly promote our GRID research program and other research deliverables\n\nThis Job Might Be a Fit For You If\n\nYou have proven experience in communications or public relations. Agency experience preferred; in-house experience a plus.You’re a strong writer who understands what makes a story compelling and relatable.You have experience working with executives and are comfortable adapting to different communication styles and voices.You have strong stakeholder management skills, can headline challenges and establish priorities, and unlock the power of the group.You thrive in fast-paced settings, with the ability to build relationships remotely.\n\nYou Might Be a Fit For Bullhorn If\n\nYou love working in an agile environment and can roll with the punchesYou take ownership of your work and continuously strive for improvement\n\nWhat We Offer...\n\nBenefits eligibility effective DAY ONE including Medical, Dental, Vision, 401(k), 401(k) Match, and moreUnlimited VacationMental health benefits (EAP & 98point6)Full Access to LinkedIn LearningQuarterly paid volunteer daysLucrative Employee Referral Program (eligible for prior to your first day)Career development opportunities up/across Bullhorn\n\nBullhorn's core purpose is to create an incredible customer experience, which starts with first creating an incredible employee experience. Our vision is for every employee to have a sense of belonging, a voice that is heard, and a clear path for success. We are committed to building diverse and inclusive teams, and our culture is shaped by our five core values: Ownership, Energy, Speed & Agility, Service, and Being Human.\n\nWe’re looking for real-life humans, each with their own unique set of thoughts, beliefs, cultures, identities, and a background and body that is completely individual. We also love humans who have taken less traditional paths of education and believe that experience and learning come in many forms. Together, all these unique individuals make Bullhorn stronger. If you’re reading this, you’re probably applying for/considering applying for a job with us, and we want you to know that Bullhorn is an equal opportunity employer. For us, that means we always have, and will always, strive to be as inclusive as possible in all aspects of employment and that we do not and will not tolerate discrimination of any kind.\n\n,"</li></ul> |
 | 1     | <ul><li>'Intermediate Data Scientist - The School of Data Science (SDS) at the University of Virginia (UVA) seeks an Intermediate Data Scientist to work in collaboration with Don Brown, PhD and Sana Syed, MD, MS, focusing on understanding gut structure and function in common gastrointestinal (GI) diseases using cutting-edge machine learning and AI methods. The overarching goal of this work is to personalize care for pediatric patients suffering from chronic GI disease by improving diagnostics, predicting future disease complications, and identifying better disease biomarkers and novel drug targets. Details about the Gastro Science Lab and the Syed lab can be found at https://gastrodatasciencelab.org/ and https://med.virginia.edu/sana-syed-lab/.\n\nThis is a one year restricted position continuation is based on the availability of funding and satisfactory performance.\n\nData Scientists provide sophisticated data management and analysis to support University projects or programs. They focus primarily on high-level data projections and statistical analysis. They manage the design and programming of all data entry forms and the training and supervision of project research coders, student workers, and volunteers. They oversee regular assessments of reliability, submit data on a monthly basis, and assist with literature searches pertinent to various research project topics.\n\nThe Successful Candidate Will\n\nWork in a professional manner and have a strong willingness to learn and improve.Promote a culture of excellence by supporting others and generating new ideas to drive the lab forward.Act as a champion for the lab’s research at local, regional, and national conferences.Drive the collection of new data and the refinement of existing data for new purposes.Independently and creatively analyze data to test or refine hypotheses.Explore and examine data from multiple disparate sources in order to identify, analyze, and report trends in the data.Develop and execute of statistical mathematical and predictive models.Visualize and report data findings creatively in a variety of visual formats to support research presentations, manuscripts, and media write-ups.Establish links across existing data sources and find new interesting data correlations.Lead projects in concept formulation, determination of appropriate statistical methodology, data analysis, research evaluation, and final research reporting.Collaborate across faculty and staff to provide actionable data-driven insights.Formulate and define analytic scope and objectives through research and fact-finding as a self-starter.Be a leader of a lab data science team and provide guidance to less experienced data analysts/scientists.\n\nQualifications\n\nMaster\'s Degree and at least 3 years of relevant experience.Strong Organization and time line management skills .Experience in AI/ML modeling approaches such as: metabolic modeling, convolutional neural networks, and Gradient-weighted Class Activation Mapping.Understand all phases of the analytic process including data collection, preparation, modeling, evaluation, and deployment.\n\nAnticipated hiring range: $100,000 - $120,000 / annual\n\nTo Apply\n\nPlease visit UVA job board: https://jobs.virginia.edu and search for “R0056431”\n\nComplete An Application And Attach\n\nCover LetterCurriculum Vitae \n\nPlease note that multiple documents can be uploaded in the box.\n\nINTERNAL APPLICANTS: Please search for "find jobs" on your workday home page and apply using the internal job board.\n\nReview of applications will begin January 22, 2024 and continue until the position is filled.\n\nFor questions about the position, please contact: Adam Greene, Research Program Officer ([email protected]) For questions about the application process, please contact: Rhiannon O\'Coin ([email protected])\n\nFor more information about the School of Data Science, please see www.datascience.virginia.edu\n\nFor more information about the University of Virginia and the Charlottesville community, please see www.virginia.edu/life/charlottesville and www.embarkuva.com\n\nThe selected candidate will be required to complete a background check at the time of the offer per University policy.\n\nPHYSICAL DEMANDS This is primarily a sedentary job involving extensive use of desktop computers. The job does occasionally require traveling some distance to attend meetings, and programs.\n\nThe University of Virginia, including the UVA Health System which represents the UVA Medical Center, Schools of Medicine and Nursing, UVA Physician’s Group and the Claude Moore Health Sciences Library, are fundamentally committed to the diversity of our faculty and staff. We believe diversity is excellence expressing itself through every person\'s perspectives and lived experiences. We are equal opportunity and affirmative action employers. All qualified applicants will receive consideration for employment without regard to age, color, disability, gender identity or expression, marital status, national or ethnic origin, political affiliation, race, religion, sex (including pregnancy), sexual orientation, veteran status, and family medical or genetic information.,'</li><li>"Artificial Intelligence Engineer - Company Description Shake - social networking \n Role Description This is a part-time hybrid role for an AI Software Engineer at SHAKE. As an AI Software Engineer, you will be responsible for the day-to-day tasks associated with pattern recognition, computer science, neural networks, software development, and natural language processing (NLP). This role is remote work.\n Qualifications Strong knowledge and experience in pattern recognition, computer science, and neural networksProficiency in software development, with a focus on AI technologiesExperience in natural language processing (NLP)Ability to work independently and remotelyExcellent problem-solving and analytical skillsStrong communication and collaboration skillsMaster's or Ph.D. in Computer Science, AI, or related fieldsRelevant industry certifications (e.g., TensorFlow, PyTorch) are a plus,"</li><li>'Senior Staff Data Scientist (Remote) - Company Description\n\nVericast is a big data company. We receive on average over 100 billion intent signals daily, which assist in generating a deep understanding of a person’s interest and in-market signals across 1,300 interest topics. This is coupled with strong geographic targeting, as over 30 billion location signals are collected daily from over one million retail stores and over 120 million households.\n\nData Science plays a crucial role in delivering our solutions today and will play a more prominent role in our future. A typical data science project has a solid mathematical foundation, an exploratory dimension, and a data-driven workflow. This is also true at Vericast. Our data science projects have strong foundations on machine learning, data engineering, and modeling. We are building a privacy-centric future of digital advertising by focusing on web content. We are connecting web content to consumer interest and action, ultimately driving which ads are shown on a webpage.\n\nTo continue our journey, we are seeking data science experts who are passionate about using cutting edge technology and conceiving innovative methods to solve unique and complex problems. As a Senior Staff Data Scientist at Vericast, your contributions will help us stay at the forefront of the AdTech industry.\n\nJob Description\n\nA Senior Staff Data Scientist is a hands-on expert who is passionate about all aspects of data science and can contribute by designing, conducting, and incorporating analyses of large-scale data from a wide variety of sources. This involves converting ambiguous requirements to concrete solutions for exploring data, designing and/or applying appropriate algorithms, documenting the findings, and incorporating the analysis into end-to-end solutions, systems, and platforms. Effective communication with other job disciplines is required. Contributions are expected at a level of results above and beyond entry-level and mid-level Data Scientists.\n\nKey Duties & Responsibilities\n\nHave a wider impact by providing insights and effective leadership into data science, digital media, and data engineering. This individual will have the hands-on skills to be an individual contributor and the experience for mentoring and leading other data scientists (25%)Act often as a technical lead, determining approach, objectives, requirements, features, milestones, implementation tasks, and tradeoffs of end-to-end large scale data science projects, platforms, and systems (25%)Act as a subject matter expert in data science (ML/AI) algorithms and underlying technologies (programming languages and systems) (15%)Design, conduct, and incorporate analyses of large-scale data from a wide variety of sources (15%)Work within the scrum practices in team projects (10%)Contribute to hiring process by screening higher level candidates, team interviews, manager candidates, i.e., act as a "Bar Raiser" (10%)\n\nQualifications\n\nEducation\n\nBachelor\'s Degree in a quantitative discipline (Computer Science, Mathematics, Engineering, Statistics) (Required)Master\'s Degree in a quantitative discipline (Computer Science, Mathematics, Engineering, Statistics) (Desired)Doctorate Degree (Preferred)In lieu of the above education requirements, a combination of experience and education will be considered.\n\nExperience\n\n8 - 10 years Relevant Experience (Required)\n\nKnowledge/Skills/Abilities\n\nStrong analytical skills, with expertise and solid understanding of multiple statistical/analytical machine learning techniques applied at large scale.Technical proficiency in ML algorithms, scalable ML platforms, languages, and tools (Python, Spark, ML/Ops) in a corporate setting is highly desirable.Ability to communicate effectively across multi-disciplinary teams (e.g., data science, engineering and product management, org leadership).Prior experience in applying Data Science in Digital Marketing Technology, Graph Theory, Privacy and Geolocation Data is a plus.\n\nAdditional Information\n\nSalary:$160,000-175,000\n\nThe ultimate compensation offered for the position will depend upon several factors such as skill level, cost of living, experience, and responsibilities.\n\nVericast offers a generous total rewards benefits package that includes medical, dental and vision coverage, 401K and flexible PTO. A wide variety of additional benefits like life insurance, employee assistance and pet insurance are also available, not to mention smart and friendly coworkers!\n\nAt Vericast, we don’t just accept differences - we celebrate them, we support them, and we thrive on them for the benefit of our employees, our clients, and our community.\u202fAs an Equal Opportunity employer, Vericast considers applicants for all positions without regard to race, color, creed, religion, national origin or ancestry, sex, sexual orientation, gender identity, age, disability, genetic information, veteran status, or any other classifications protected by law. Applicants who have disabilities may request that accommodations be made in order to complete the selection process by contacting our Talent Acquisition team at [email protected]. EEO is the law. To review your rights under Equal Employment Opportunity please visit: www.dol.gov/ofccp/regs/compliance/posters/pdf/eeopost.pdf.\n\n,'</li></ul>                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                           |
 
-## Evaluation
-
-### Metrics
-| Label   | Accuracy |
-|:--------|:---------|
-| **all** | 0.875    |
-
 ## Uses
 
 ### Direct Use for Inference
@@ -511,7 +490,7 @@ Join my client on this thrilling journey and contribute to shaping the future of
 
 ### Training Hyperparameters
 - batch_size: (8, 8)
-- num_epochs: (1, 1)
+- num_epochs: (4, 4)
 - max_steps: -1
 - sampling_strategy: oversampling
 - body_learning_rate: (2e-05, 1e-05)
@@ -529,7 +508,10 @@ Join my client on this thrilling journey and contribute to shaping the future of
 ### Training Results
 | Epoch | Step | Training Loss | Validation Loss |
 |:-----:|:----:|:-------------:|:---------------:|
-| 0.025 | 1    | 0.2118        | -               |
+| 0.025 | 1    | 0.1975        | -               |
+| 1.25  | 50   | 0.0018        | -               |
+| 2.5   | 100  | 0.0002        | -               |
+| 3.75  | 150  | 0.0002        | -               |
 
 ### Framework Versions
 - Python: 3.10.12

config.json

@@ -1,31 +1,24 @@
 {
-  "_name_or_path": "BAAI/bge-small-en-v1.5",
+  "_name_or_path": "sentence-transformers/paraphrase-mpnet-base-v2",
   "architectures": [
-    "BertModel"
+    "MPNetModel"
   ],
   "attention_probs_dropout_prob": 0.1,
-  "classifier_dropout": null,
+  "bos_token_id": 0,
+  "eos_token_id": 2,
   "hidden_act": "gelu",
   "hidden_dropout_prob": 0.1,
-  "hidden_size": 384,
-  "id2label": {
-    "0": "LABEL_0"
-  },
+  "hidden_size": 768,
   "initializer_range": 0.02,
-  "intermediate_size": 1536,
-  "label2id": {
-    "LABEL_0": 0
-  },
-  "layer_norm_eps": 1e-12,
-  "max_position_embeddings": 512,
-  "model_type": "bert",
+  "intermediate_size": 3072,
+  "layer_norm_eps": 1e-05,
+  "max_position_embeddings": 514,
+  "model_type": "mpnet",
   "num_attention_heads": 12,
   "num_hidden_layers": 12,
-  "pad_token_id": 0,
-  "position_embedding_type": "absolute",
+  "pad_token_id": 1,
+  "relative_attention_num_buckets": 32,
   "torch_dtype": "float32",
   "transformers_version": "4.39.0",
-  "type_vocab_size": 2,
-  "use_cache": true,
-  "vocab_size": 30522
+  "vocab_size": 30527
 }

config_sentence_transformers.json

@@ -1,8 +1,8 @@
 {
   "__version__": {
-    "sentence_transformers": "2.2.2",
-    "transformers": "4.28.1",
-    "pytorch": "1.13.0+cu117"
+    "sentence_transformers": "2.0.0",
+    "transformers": "4.7.0",
+    "pytorch": "1.9.0+cu102"
   },
   "prompts": {},
   "default_prompt_name": null,

config_setfit.json

@@ -1,4 +1,4 @@
 {
-  "normalize_embeddings": false,
-  "labels": null
+  "labels": null,
+  "normalize_embeddings": false
 }

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:4b3c62677b15ea2dfb7bcc6ba3d9f3145f6833b0ed021a26f879a3ce35117bc4
-size 133462128
+oid sha256:bbaf4cdb573975e4a85ee5366c3b07c83c659b7794827a392045682320c62afe
+size 437967672

model_head.pkl

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:20496e5fac4a8233d533c838fa76b63d6cc9eb2e0ca37f87082fe716a024419b
-size 3935
+oid sha256:d22834ee6ea68a0802e80a78d597f52c814e84471f9cfc2073e4e998a159b371
+size 7007

modules.json

@@ -10,11 +10,5 @@
     "name": "1",
     "path": "1_Pooling",
     "type": "sentence_transformers.models.Pooling"
-  },
-  {
-    "idx": 2,
-    "name": "2",
-    "path": "2_Normalize",
-    "type": "sentence_transformers.models.Normalize"
   }
 ]

sentence_bert_config.json

@@ -1,4 +1,4 @@
 {
   "max_seq_length": 512,
-  "do_lower_case": true
+  "do_lower_case": false
 }

special_tokens_map.json

@@ -1,27 +1,41 @@
 {
+  "bos_token": {
+    "content": "<s>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
   "cls_token": {
-    "content": "[CLS]",
+    "content": "<s>",
     "lstrip": false,
     "normalized": false,
     "rstrip": false,
     "single_word": false
   },
-  "mask_token": {
-    "content": "[MASK]",
+  "eos_token": {
+    "content": "</s>",
     "lstrip": false,
     "normalized": false,
     "rstrip": false,
     "single_word": false
   },
+  "mask_token": {
+    "content": "<mask>",
+    "lstrip": true,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
   "pad_token": {
-    "content": "[PAD]",
+    "content": "<pad>",
     "lstrip": false,
     "normalized": false,
     "rstrip": false,
     "single_word": false
   },
   "sep_token": {
-    "content": "[SEP]",
+    "content": "</s>",
     "lstrip": false,
     "normalized": false,
     "rstrip": false,

tokenizer_config.json

@@ -1,57 +1,59 @@
 {
   "added_tokens_decoder": {
     "0": {
-      "content": "[PAD]",
+      "content": "<s>",
       "lstrip": false,
       "normalized": false,
       "rstrip": false,
       "single_word": false,
       "special": true
     },
-    "100": {
-      "content": "[UNK]",
+    "1": {
+      "content": "<pad>",
       "lstrip": false,
       "normalized": false,
       "rstrip": false,
       "single_word": false,
       "special": true
     },
-    "101": {
-      "content": "[CLS]",
+    "2": {
+      "content": "</s>",
       "lstrip": false,
       "normalized": false,
       "rstrip": false,
       "single_word": false,
       "special": true
     },
-    "102": {
-      "content": "[SEP]",
+    "104": {
+      "content": "[UNK]",
       "lstrip": false,
       "normalized": false,
       "rstrip": false,
       "single_word": false,
       "special": true
     },
-    "103": {
-      "content": "[MASK]",
-      "lstrip": false,
+    "30526": {
+      "content": "<mask>",
+      "lstrip": true,
       "normalized": false,
       "rstrip": false,
       "single_word": false,
       "special": true
     }
   },
+  "bos_token": "<s>",
   "clean_up_tokenization_spaces": true,
-  "cls_token": "[CLS]",
+  "cls_token": "<s>",
   "do_basic_tokenize": true,
   "do_lower_case": true,
-  "mask_token": "[MASK]",
+  "eos_token": "</s>",
+  "mask_token": "<mask>",
   "model_max_length": 512,
   "never_split": null,
-  "pad_token": "[PAD]",
-  "sep_token": "[SEP]",
+  "pad_token": "<pad>",
+  "sep_token": "</s>",
   "strip_accents": null,
   "tokenize_chinese_chars": true,
-  "tokenizer_class": "BertTokenizer",
+  "tokenizer_class": "MPNetTokenizer",
   "unk_token": "[UNK]"
 }

vocab.txt

@@ -1,3 +1,7 @@
+<s>
+<pad>
+</s>
+<unk>
 [PAD]
 [unused0]
 [unused1]
@@ -30520,3 +30524,4 @@ necessitated
 ##：
 ##？
 ##～
+<mask>