Recipes¶

You have the choice of four different recipe options. Upon selecting each recipe, the corresponding configuration user interface (UI) will be presented.
- Classification - Classify the documents in dataset.
- Summarization - Summarize the documents in dataset.
- PII Redaction - Hide the PII/replace with characters ** in documents in dataset.
- Q&A - Question-based input given to a model for generating relevant answers or information.
To utilize prompts in a dataset, it must be registered within the Tensoract Studio prior to usage. Refer to Model Integration for more details..
Registered models are visible in the drop-down menus for each recipe.

Elaborate explanations for each recipe are outlined below:

1. Classification Recipe¶

The “Classification Prompt” in a dataset serves the function of categorizing or labeling data points based on predefined criteria.

Model Info - Choose registered model.
Prompt Design
- Prompt Name- Enter Name of prompt.
- Prompt Text -It is default customized prompt text refers to a concise set of instructions or a statement given to a system or model to guide its behavior or generate a specific response.
- Temperature(0-1) - Enter temperature value. The Temperature setting in an LLM controls the randomness of its output.A higher temperature makes the output more creative but less focused, while a lower temperature makes the output more predictable and controlled.
- Classification Classes - Enter Classification classes.
- Default Class - Enter default class. (Document is classified as default class label in case the document belongs to none of the classes)
Item Processing -
- If Run upon dataset item upload checkbox is enabled,classification is run on each document in the dataset.
- If Run upon dataset item upload is not enabled, you get an option to select Classify as an Action in the dataset items page.
Test Recipe - It is the test text payload which can be customized.
Click Test and validate the model output to ensure the integration works as expected.
Before Save, make sure you have provided all the necessary details.
Click Save and the prompt is added in prompt history for classification of documents.
Classification Test Recipe Payloads and Response

Classification Test Recipe Payload

{
Charles Darvin Beagle Street 1687

Objective: Highly skilled software engineer with a passion for developing innovative solutions and a strong background in various programming languages. Seeking a challenging position in a dynamic organization where I can apply my technical expertise and contribute to the development of cutting-edge software solutions.

Education: Bachelor of Science in Computer Science Oxford, 1999

Skills:

Programming Languages: Java, Python, C++, JavaScript Web Development: HTML, CSS, React, Node.js Database Management: SQL, MongoDB Version Control: Git Software Development Lifecycle (SDLC) Object-Oriented Design and Analysis (OOAD) Problem-solving and debugging skills Strong knowledge of algorithms and data structures Teamwork and collaboration Excellent verbal and written communication skills Experience: Software Engineer Intern Beagle INC, 2020-Present

Collaborated with a team of software engineers to develop and maintain web applications using Java and JavaScript. Participated in code reviews and provided feedback to ensure high-quality code. Implemented new features and resolved bugs in existing software applications. Assisted in database management and optimization for improved performance. Conducted unit testing and assisted in integration testing to ensure functionality and reliability. Software Developer Oxtech, 2005-1029

English (Fluent) French (Intermediate) References: Available upon request

}

Classification Test Recipe Response

CV

After saving the recipe, you will find the Classify option available in the Action menu.
Choose dataset items and classify them. This will initiate the prompt execution and display the results in the Prompt History section as shown below.
Prompts can be compared as illustrated below.
Classification Export

{
    "itemId": "25631d2d8688a56035222fa2",
    "type": "application/pdf",
    "source": "https://sandboxdocuments.tensoract.com/presigned/c0183e9d6cdc03119adb941df9947466.pdf?sig=4fb4f97b88a8493249d6a1d34d306ba367c2f6d7ab4cc4631b0b5cb0f171a8fab3f47c5665cc65dd6b8048884bf2cc32bb5f42e7df0cf9540353a742406021d2:3895574330ecba8ac1213e64be64b124:64d75643:f83a56600fe4a22a9347956d1e8d8824",
    "text": "Seyyed Meghdad Hasheminasab 1190 Windsor Dr., Apt 13, West Lafayette, IN 47906 Phone: 765-775-3983, Email: hashemin@purdue.edu Objective To obtain a full-time position focusing on LiDAR/image-based 3D reconstruction, starting summer 2021 Education PhD Student, Geomatics Engineering Aug 2017 - Present Purdue University, Indiana BS + MS, Geomatics Engineering Sep 2009 - Feb 2016 K.N. Toosi University of Technology, Tehran, Iran Experience Research Assistant in TERRA (Advisor: Prof. Ayman Habib): Aug 2017 - Present Purdue TERRA project funded by a $6.5 million grant from the U.S. Department of Energy for research aimed at producing superior strains of sorghum suitable for growing as a biofuel. My tasks include: - UAV and wheel-based mobile mapping system data acquisition, - Algorithm development for processing and analyzing of RGB images, Hyperspectral images, and LiDAR point clouds, and - UAV and wheel-based mobile mapping system calibration. Research Assistant in JTRP (Advisor: Prof. Ayman Habib): Aug 2017-Present - Purdue JTRP facilitates collaboration between the Indiana Department of Transportation, higher education institutions and industry to implement innovations in Indiana transportation infrastructure. My tasks include: - UAV and wheel-based mobile mapping system data acquisition, - LiDAR point cloud processing and classification for road lane markings extraction, potholes, etc., that can be useful for connected and autonomous vehicles, and - UAV-based image processing for crash scene documentation. Research Assistant (Advisor: Prof. Hamid Ebadi): Sep 2014-Feb - 2016 In the Digital Photogrammetry Research Group my task included: - Close-range photogrammetry data acquisition, and - Implementing robust Structure from Motion strategies for 3D reconstruction using close-range photogrammetry. Honors and Awards Purdue University Ross Fellowship 2017-2018 academic year Professor Edward M. Mikhail Endowment 2019-2020 academic year Peer Reviewed Journal Papers 1. Zhou, T., Hasheminasab, S. M., Ravi, R., & Habib, A. (2020). LiDAR-Aided Interior Orientation Parameters Refinement Strategy for Consumer-Grade Cameras Onboard UAV Remote Sensing Systems. Remote Sensing, 12(14), 2268. 2. Hasheminasab, S. M., Zhou, T., & Habib, A. (2020). GNSS/INS-Assisted Structure from Motion Strategies for UAV-Based Imagery over Mechanized Agricultural Fields. Remote Sensing, 12(3), 351. 3. Ravi, R., Cheng, Y. T., Lin, Y. C., Lin, Y. J., Hasheminasab, S. M., Zhou, T., & Habib, A. (2019). Lane Width Estimation in Work Zones Using LiDAR-Based Mobile Mapping Systems. IEEE Transactions on Intelligent Transportation Systems. 4. Lin, Y. C., Cheng, Y. T., Zhou, T., Ravi, R., Hasheminasab, S. M., Flatt, J. E., & Habib, A. (2019). Evaluation of UAV LiDAR for Mapping Coastal Environments. Remote Sensing, 11(24), 2893. 5. LaForest, L., Hasheminasab, S. M., Zhou, T., Flatt, J. E., & Habib, A. (2019). New Strategies for Time Delay Estimation during System Calibration for UAV-Based GNSS/INS-Assisted Imaging Systems. Remote Sensing, 11(15), 1811. 6. He, F., Zhou, T., Xiong, W., Hasheminasab, S.M., & Habib, A. (2018). Automated Aerial Triangulation for UAV-Based Mapping. Remote Sensing, 10(12), 1952. 7. LaForest, L., Zhou, T., Hasheminasab, S. M., & Habib, A. (2020). \"Time Delay Estimation for GNSS/INS-assisted, UAV-based Push-broom Scanners \". Photogrammetric Engineering & Remote Sensing, (submitted). 8. Hasheminasab, S. M., Zhou, T., LaForest, L., & Habib, A. (2020). System Calibration with Time Delay Estimation for UAV-based Frame and Line Cameras Using Automatically-derived Tie Points (under preparation) Peer Reviewed Conference Proceedings 1. Lin, Y. C., Hasheminasab, S. M., Bullock, J. L., Horton, D., Baxmeyer, A., Habib, A., & Bullock, D. M. (2020, September). UAS Based Methodology for Measuring Glide Slope Angles of Airport Precision Approach Path Indicators (PAPI). In 2020 International Conference on Unmanned Aircraft Systems (ICUAS) (pp. 1468-1474). IEEE. 2. T., Zhou, Hasheminasab, S. M., Lin, Y. C. & Habib, A. \"Comparative Evaluation of Derived Image and Lidar Point Clouds from Uav-based Mobile Mapping Systems\". The International Archives of Photogrammetry, Remote Sensing and Spatial Information Sciences, 43, 169-175. 3. Ravi, R., Hasheminasab, S. M., Zhou, T., Masjedi, A., Quijano, K., Flatt, J. E., & Habib, A. (2019, May). UAV-based multi-sensor multi-platform integration for high throughput phenotyping. In Autonomous Air and Ground Sensing Systems for Agricultural Optimization and Phenotyping IV (Vol. 11008, p. 110080E). International Society for Optics and Photonics. 4. Hasheminasab, S. M., H. Ebadi, A. Sedaghat.\" An Integrated RANSAC and Graph-based Mismatch Elimination Approach for Wide-Baseline Image Matching.\" The International Archives of Photogrammetry, Remote Sensing and Spatial Information Sciences 40.1 (May 2015): 297 Technical Skills Programming C++, Python, MATLAB Libraries/Frameworks OpenCV, Boost, libLAS, Pytorch, Numpy Image/LiDAR Processing Tools Pix4D, PhotoScan, Trimble Inpho, ArcGIS, ENVI",
    "results": [
        {
            "recipeId": "0423cb04a48dac6c1fe9b935",
            "result": "cv",
            "recipeName": "Prompt Classification"
        }
    ]
}
{
    "itemId": "6ccfd758a2340efe076acfe6",
    "type": "application/pdf",
    "source": "https://sandboxdocuments.tensoract.com/presigned/08fc737f6645b08ceb7b4f11873cdc35.pdf?sig=5915332c01d5323546d421236a1aee5d605573392125ddca479ea922a42f09974a15d5db17aba9cfae426b806779a35e223cd0257376a124f0f23443e1a29080:673919c1675684a9c135d46e859b30a0:64d75643:a5a095243b1955415279c5f45d8960b6",
    "text": "Table of Contents North America The North America segment primarily consists of amounts earned from retail sales of consumer products (including from sellers) and subscriptions through North America-focused online and physical stores This segment includes export sales from these online stores. International The International segment primarily consists of amounts earned from retail sales of consumer products (including from sellers) and subscriptions through internationally-focused online stores. This segment includes export sales from these internationally-focused online stores (including export sales from these online stores to customers in the U.S., Mexico, and Canada), but excludes export sales from our North America- focused online stores AWS The AWS segment consists of amounts earned from global sales of compute, storage, database, and other service offerings for start-ups, enterprises government agencies, and academic institutions. Information on reportable segments and reconciliation to consolidated net income is as follows (in millions): Three Months Ended March 31, 2019 2020 North America Net sales $ 35,812 $ 46,127 Operating expenses 33,525 44,815 Operating income $ 2,287 $ 1,312 International Net sales $ 16,192 $ 19,106 Operating expenses 16,282 19,504 Operating income (loss) $ (90) $ (398) AWS Net sales $ 7,696 $ 10,219 Operating expenses 5,473 7,144 Operating income $ 2,223 $ 3,075 Consolidated Net sales $ 59,700 $ 75,452 Operating expenses 55,280 71,463 Operating income 4,420 3,989 Total non-operating income (expense) (19) (606) Provision for income taxes (836) (744) Equity-method investment activity net of tax (4) (104) Net income $ 3,561 $ 2,535 17",
    "results": [
        {
            "recipeId": "0423cb04a48dac6c1fe9b935",
            "result": "generic",
            "recipeName": "Prompt Classification"
        }
    ]
}

Key	Value
itemId	The item id of dataset item
type	The type of document
source	Presigned URL or S3 source of document
text	The text associated with document
results	List of dictionary containing results
recipeId	The id of the recipe
result	Classification result
recipeName	The name of the recipe

2. Summarization Recipe¶

Model Info - Choose registered model.
Prompt Design
- Prompt Name- Enter Name of prompt.
- Prompt Text -It is default customized prompt text refers to a concise set of instructions or a statement given to a system or model to guide its behavior or generate a specific response.
- Max words- Configure Max words to summarize the documents.
- Temperature(0-1) - Enter temperature value. The Temperature setting in an LLM controls the randomness of its output.A higher temperature makes the output more creative but less focused, while a lower temperature makes the output more predictable and controlled.
Item Processing -
- If Run upon dataset item upload checkbox is enabled,summarization is run on each document in the dataset.
- If Run upon dataset item upload is not enabled, you get an option to select Summarize as an Action in the dataset items page.
Test Recipe - It is the test text payload which can be customized.
Click Test and validate the model output to ensure the integration works as expected.
Before Save, make sure you have provided all the necessary details.
Click Save and the prompt is added in prompt history for summarization of documents.
Summarization Request Payload and Response

Summarization Test Payload

{
Charles Darvin Beagle Street 1687

Objective: Highly skilled software engineer with a passion for developing innovative solutions and a strong background in various programming languages. Seeking a challenging position in a dynamic organization where I can apply my technical expertise and contribute to the development of cutting-edge software solutions.

Education: Bachelor of Science in Computer Science Oxford, 1999

Skills:

Programming Languages: Java, Python, C++, JavaScript Web Development: HTML, CSS, React, Node.js Database Management: SQL, MongoDB Version Control: Git Software Development Lifecycle (SDLC) Object-Oriented Design and Analysis (OOAD) Problem-solving and debugging skills Strong knowledge of algorithms and data structures Teamwork and collaboration Excellent verbal and written communication skills Experience: Software Engineer Intern Beagle INC, 2020-Present

Collaborated with a team of software engineers to develop and maintain web applications using Java and JavaScript. Participated in code reviews and provided feedback to ensure high-quality code. Implemented new features and resolved bugs in existing software applications. Assisted in database management and optimization for improved performance. Conducted unit testing and assisted in integration testing to ensure functionality and reliability. Software Developer Oxtech, 2005-1029

English (Fluent) French (Intermediate) References: Available upon request

}

Summarization Test Recipe Response

Charles Darwin is a highly skilled software engineer with a BSc in Computer Science from Oxford. He has expertise in various programming languages, web development, database management, version control, SDLC, OOAD, problem-solving, debugging, algorithms, data structures, teamwork and communication. He has experience as a software engineer intern and software developer. He is fluent in English and has intermediate French.

After saving the recipe, you will find the Summarize option available in the Action menu.
Choose dataset items and summarize them. This will initiate the prompt execution and display the results in the Prompt History section as shown below.
Prompts can be compared as illustrated below.
Summarization Export

{
    "itemId": "29b6b78d997f0249909e7ee1",
    "type": "application/pdf",
    "source": "https://sandboxdocuments.tensoract.com/presigned/2187046aa26093d57a813adda8039b7f.pdf?sig=7a88acba6d7542e862e963786c5319987d513cb4af1cdd3a2ee2d893a796602d980ccf8a20f3ac16f545bf934540c6826edea5bd14af189d2f153eacb2788d24:cc1e5b3488c934391a19e81652a4b511:64d762ac:08d0ce58b154d43b6596563e6c5e9659",
    "text": "John Doe 14 Fourteen Street, Town, CA 99999 Cell: 000.000.0000 - email@email.com Professional Profile High performing, customer-oriented manager with 10 years' experience driving operations, production and goal achievement in financial industry. Advanced leadership skills with success in managing groups to obtain results and bridging the gap between organizational objectives and operational capabilities. Leading contributor to strategic programs with the ability to successfully deliver multiple concurrent projects under tight deadlines and success parameters. Adept in the design and execution of successful operational processes, programs and solutions to support and optimize team performance. Dynamic team building and training abilities; exceptional interpersonal, communication and problem resolution skills. Accomplishments Increased sales by 67% and achieved client retention of 78%, compared to 20% standard, by implementing effective strategies in 2013. Exceeded sales goals by 50% for the last 3 quarters. Played key role in branch maintaining 110% cross-sell ratio, one of highest in region. Excelled in quarterly sales with 160% and maintained a retention rate of 90% for Discover Financial Services. Pinnacle Performance Award for a 180% closed loan rate for 2012. Employee of the Quarter for first quarter of 2012 and third quarter of 2010. Experience Assistant Branch Manager October 2008 to Current University Federal Credit Union Salt Lake City, UT Manage operations and production at 2 branches, including high volume university campus branch. Hire, train, develop, motivate, coach and discipline branch personnel. Achieve and maintain high level of member service through sales and team meetings. Run daily branch activities and ensure compliance with established credit union policies and procedures. Mentor employees in effective sales methods and techniques. Determine feasibility of loans by analyzing financial status, credit and property evaluation of applicants. Provide expert financial advice on mortgage, educational and personal loans. Deliver informational sales presentations to potential business members and build new relationships. Perform marketing by participating in and sponsoring community events. Quality Coach January 2004 to September 2008 Discover Financial Services - Salt Lake City, UT Leveraged expertise in resolving issues to satisfy customers for Retention Department. Coached a team of 10-15 employees by monitoring calls in fast-paced call center environment. Progressive advancement from account manager, senior account manager to quality coach. Education BBA Program, University of Utah, Salt Lake City, JT-Enrollment Aug 2014 Associate of Science in Accounting, Salt Lake Community College, Salt Lake City, UT-2009 Certifications Mortgage Loan Originator (MLO) License, 2011 Advanced Mortgage/ Home Equity Line of Credit Loan Training Mortgage Underwriter Training Investment Banking Seminar",
    "results": [
        {
            "recipeId": "5728c102b696a9ddabfac9fa",
            "result": "John Doe is a customer-oriented manager with 10 years' experience in the financial industry. He has advanced leadership skills and has achieved impressive results such as increasing sales by 67%, exceeding sales goals by 50%, and maintaining a retention rate of 90%. He has also received awards such as Pinnacle Performance Award and Employee of the Quarter. He has a BBA Program enrollment and certifications in Mortgage Loan Originator, Mortgage/ Home Equity Line of Credit Loan Training, and Investment Banking Seminar.",
            "recipeName": "Prompt Summarization"
        }
    ]
}
{
    "itemId": "34d1c366e0666f77c4e6114a",
    "type": "application/pdf",
    "source": "https://sandboxdocuments.tensoract.com/presigned/82da49876fbfd709d464220de2d3aef5.pdf?sig=3eec609ac8cfdcfdb294c7d2f5a185ca8cc1c6ae6f33cb2c831448405db94c263742f6a6ba1495f1bc11bdd876b56314ca33abece599d4a0471fe0adfb405f5e:0942174326e67ab5c4a6bbf719168105:64d762ac:2deb65c55280af0706e03f625222aa84",
    "text": "2/27/2021 Amazon com: Customer reviews: OPI Nail Lacquer, A Great Opera Tunity Shera B ***** Should be know that this color isn't very glossy. Reviewed in the United States on May 19 2017 Color Freedom of Peach Verified Purchase Should be know that this color isn't very glossy. Even with top coat it is the least shiny polish I've wom Maybe it's meant to be but don't recall reading that in description 16 people found this helpful Helpful Report abuse MK Love it Reviewed in the United States on June 13 2016 Color Get Myself into . Jam-balaya Verified Purchase I got the color \"show us your tips\". Very unique blue Requires multiple coats for full coverage Formula was easy to work with 23 people found this helpful Helpful I Report abuse Sharon Beverly VINE VOICE Great Summer Color Reviewed in the United States on June 20. 2018 Color Her Spicy I Verified Purchase Nothing says 'Summer' like a hot coral-colored nail polish I change polish colors seasonally and OPI has some terrific shades This Hot and Spicy version is less orangey leaning towards that in between shade with a hint of peach In the past, enjoyed OPI polish that was between peach and pink This one isn't IT'S definitely coral Despite being a darker color you will find that it streaks with a single application Two applications are a must A third is even better For a seasonal change, shift into a color that makes you think of beaches surt, and pools; Hot and Spicy 4 people found this helpful Helpful I Report abuse Amazon Customer * * Great consistency and durability. Reviewed in the United States on May 6, 2018 Color Yank My Doodle I Verified Purchase I have used this particular color polish before but couldn't find this locally My purchase is the same polish I've used for over a year. One coat covers good. but still use two for best coverage and durability Then finish with a top coat of your favorite brand and . lasts over a week depending how many times your hands are in soapy water with out gloves. The consistency is very fluid not slow moving like an older bottle 8 people found this helpful Helpful I Report abuse Angela Top Contributor Makeup o https://www.amazon.com/gp/customer-reviews/B00014353Y/ 3/5",
    "results": [
        {
            "recipeId": "5728c102b696a9ddabfac9fa",
            "result": "OPI Nail Lacquer is a popular nail polish with a variety of colors. Customers have found it to be less glossy than expected, but with a top coat it can be made more shiny. It requires multiple coats for full coverage and is durable with a top coat. It is a great summer color and is a fluid consistency.",
            "recipeName": "Prompt Summarization"
        }
    ]
}

Key	Value
itemId	The item id of dataset item
type	The type of document
source	Presigned URl or S3 source of document
text	The text associated with document
results	List of dictionary containing results
recipeId	The id of the recipe
result	Summarization result
recipeName	The name of the recipe

3. Redaction Recipe¶

Model Info - Choose registered model.
Redaction options
- Prompt Name- Enter Name of prompt.
- Choose the PII redaction options from the list.
Item Processing -
- If Run upon dataset item upload checkbox is enabled,redaction is run on each document in the dataset.
- If Run upon dataset item upload is not enabled, you get an option to select Redact as an Action in the dataset items page.
Test Recipe - It is the test text payload which can be customized.
Click Test and validate the model output to ensure the integration works as expected.
Before Save, make sure you have provided all the necessary details.
Click Save and the prompt is added in prompt history for redaction of documents.
Redaction Request Payload and Response

Redaction Test Payload

{
Charles Darvin Beagle Street 1687

Objective: Highly skilled software engineer with a passion for developing innovative solutions and a strong background in various programming languages. Seeking a challenging position in a dynamic organization where I can apply my technical expertise and contribute to the development of cutting-edge software solutions.

Education: Bachelor of Science in Computer Science Oxford, 1999

Skills:

Programming Languages: Java, Python, C++, JavaScript Web Development: HTML, CSS, React, Node.js Database Management: SQL, MongoDB Version Control: Git Software Development Lifecycle (SDLC) Object-Oriented Design and Analysis (OOAD) Problem-solving and debugging skills Strong knowledge of algorithms and data structures Teamwork and collaboration Excellent verbal and written communication skills Experience: Software Engineer Intern Beagle INC, 2020-Present

Collaborated with a team of software engineers to develop and maintain web applications using Java and JavaScript. Participated in code reviews and provided feedback to ensure high-quality code. Implemented new features and resolved bugs in existing software applications. Assisted in database management and optimization for improved performance. Conducted unit testing and assisted in integration testing to ensure functionality and reliability. Software Developer Oxtech, 2005-1029

English (Fluent) French (Intermediate) References: Available upon request

}

Redaction Test Recipe Response

***************************** Objective: Highly skilled software engineer with a passion for developing innovative solutions and a strong background in various programming languages. Seeking a challenging position in a dynamic organization where I can apply my technical expertise and contribute to the development of cutting-edge software solutions. Education: Bachelor of Science in Computer Science Oxford, 1999 Skills: Programming Languages: Java, Python, C++, JavaScript Web Development: HTML, CSS, React, Node.js Database Management: SQL, MongoDB Version Control: Git Software Development Lifecycle (SDLC) Object-Oriented Design and Analysis (OOAD) Problem-solving and debugging skills Strong knowledge of algorithms and data structures Teamwork and collaboration Excellent verbal and written communication skills Experience: Software Engineer Intern ** INC, ****-Present Collaborated with a team of software engineers to develop and maintain web applications using Java and JavaScript. Participated in code reviews and provided feedback to ensure high-quality code. Implemented new features and resolved bugs in existing software applications. Assisted in database management and optimization for improved performance. Conducted unit testing and assisted in integration testing to ensure functionality and reliability. Software Developer Oxtech, 2005-1029 English (Fluent) French (Intermediate) References: Available upon request

After saving the recipe, you will find the Redact option available in the Action menu.
Choose dataset items and classify them. This will initiate the prompt execution and display the results in the Prompt History section as shown below.
Prompts can be compared as illustrated below.
- Prompts can be compared as illustrated below.
Redaction Export

{
    "itemId": "0383ffc5de7d25862c77b22a",
    "type": "application/pdf",
    "source": "s3://objectways-ergo-poc/input_documents/AmbA_J3Y61NDT0021050L2.tif",
    "text": "2 Medizinisch-Diagnostisches-Labor allino Kempten Dr. med. Juan Perez allgäuLab Gemeinscheftspraxis für Laborstoriumsmedizin Med Milerobiologie TBC Diagnost$ Mytologie Porositologie Hygione Infeksionsserologie Immunologie Hamanologie Hinischo Charmie Endokrinologie Molekularbiologie Virologie Hausanschrift Heiko Schuben- Gasse & 12345 Berlin Dr mail Am Pares Hals Dahas Class : 12345 Pario Herr Telefon: 05766/43907-0 Telefax 05765/43907-10 John Doe Musterstraße 17 Steuer-Nr. DE 012 345 07 12345 Samplehausen IK): 123456 780 BSNR 987 054321 Rechnung Nr. PP1912-1833 (DUPLIKAT) 17.01.2020 Bitte bei Überwaisung unbacingt angeben0 Patient Doe John geb. 31.3.1997 Katalog: GOA Datus Preis inkl ANR Einsender Greadpreis Steig.Satz Geb.-Ziff. Text der Behandlung Anzahl Betrag (EUR) Steig. Betrag CEURO 05.12.2019 22685 Dr. mod. Juan Perce Berlin 4525 CampyTobacter-AG 1 14.57 1.15 16.76 4530 Bakteriennachweris durch einfache Anzüchtung a 2 9.32 1.15 10.72 4538 Bakteriennachweis durch Anzüchtung auf Selekt 4 27.95 1.15 32.16 4539 Bakterisenachweis durch besenders aufwendige 1 14.57 1.15 16.76 3744 Calprotectin in Stuhl 1 26.23 1.15 30.16 4069 Pankreas-Elastase in St. 1 43.72 1.15 $0.28 3741 CRP quantitativ 1 11.66 1.15 13.41 3897 Gliadin-AK IOG (deamidierte Peptide) 1 29.73 1.15 34.19 3897 Gliadin-Ak ICA (deamidierte Peptide) 1 29.73 1.15 34.19 P10/3 Ersatz von Auslagen (GOA (10) 1 3.50 1.0 3.50 Endsume: 242.13 ber 24. 1.2020 Dis Untersuchungen sind nach der Gobührencednung für Armt (GOA) Stand 01.01.2009-berechnet Material- und Versandkosten werden entsprechend $ 10 GOA in Höhe der entstandanon Kosten erhoben Gem53 1 12 GOA is dia Rechnung sofort filig A-Ziffern gleich Leistungen gem&ß I 6 Abs 2 der GOA. Bitte überweison Sig innerhalb der nächsten 14 Tage die Rechnangssumme< auf dos unter angegebene Konto. Banieverbindung: PssudobankBortin BIC: ABCDEFGHI IBAN: DE01 2345 5789 101112 Konto Nr. 123 456 789",
    "results": [
        {
            "recipeId": "c92ab378dff3f872efc9e717",
            "result": "2 Medizinisch-Diagnostisches-Labor ************** Dr. med. ********** allgäuLab Gemeinscheftspraxis für Laborstoriumsmedizin Med Milerobiologie TBC Diagnost$ Mytologie Porositologie ******* Infeksionsserologie Immunologie Hamanologie **************** Endokrinologie Molekularbiologie Virologie Hausanschrift Heiko *******- Gasse & 12345 ****** Dr mail Am Pares Hals Dahas Class : 12345 ********** Telefon: ************* Telefax ************** ******** ************ 17 Steuer-Nr. DE ********** 12345 ************ IK): 123456 780 BSNR ********** Rechnung Nr. PP1912-1833 (DUPLIKAT) ********** Bitte bei Überwaisung unbacingt angeben0 Patient Doe **** geb. ********* Katalog: GOA Datus Preis inkl ANR Einsender Greadpreis Steig.Satz Geb.-Ziff. Text der Behandlung Anzahl Betrag (EUR) Steig. Betrag CEURO ********** 22685 Dr. mod. Juan Perce ****** ********************* 1 14.57 1.15 16.76 4530 Bakteriennachweris durch einfache Anzüchtung a 2 9.32 1.15 10.72 4538 Bakteriennachweis durch Anzüchtung auf Selekt 4 27.95 1.15 32.16 4539 Bakterisenachweis durch besenders aufwendige 1 14.57 1.15 16.76 3744 Calprotectin in Stuhl 1 26.23 1.15 30.16 4069 Pankreas-Elastase in St. 1 43.72 1.15 $0.28 3741 CRP quantitativ 1 11.66 1.15 13.41 3897 Gliadin-AK IOG (deamidierte Peptide) 1 29.73 1.15 34.19 3897 Gliadin-Ak ICA (deamidierte Peptide) 1 29.73 1.15 34.19 P10/3 Ersatz von Auslagen (GOA (10) 1 3.50 1.0 3.50 Endsume: 242.13 ber 24. 1.2020 Dis Untersuchungen sind nach der Gobührencednung für Armt (GOA) Stand 01.01.2009-berechnet Material- und Versandkosten werden entsprechend $ 10 GOA in Höhe der entstandanon Kosten erhoben Gem53 1 12 GOA is dia Rechnung sofort filig A-Ziffern gleich Leistungen gem&ß I 6 Abs 2 der GOA. Bitte überweison Sig innerhalb der nächsten 14 Tage die Rechnangssumme< auf dos unter angegebene Konto. Banieverbindung: PssudobankBortin BIC: ABCDEFGHI IBAN: DE01 2345 5789 101112 Konto Nr. ***********",
            "recipeName": "Prompt Redact All"
        }
    ]
}
{
    "itemId": "206318418e777d418699b427",
    "type": "application/pdf",
    "source": "s3://objectways-ergo-poc/input_documents/AmbA_J3Y61NDT0021050L2.tif",
    "text": "2 Medizinisch-Diagnostisches-Labor allino Kempten Dr. med. Juan Perez allgäuLab Gemeinscheftspraxis für Laborstoriumsmedizin Med Milerobiologie TBC Diagnost$ Mytologie Porositologie Hygione Infeksionsserologie Immunologie Hamanologie Hinischo Charmie Endokrinologie Molekularbiologie Virologie Hausanschrift Heiko Schuben- Gasse & 12345 Berlin Dr mail Am Pares Hals Dahas Class : 12345 Pario Herr Telefon: 05766/43907-0 Telefax 05765/43907-10 John Doe Musterstraße 17 Steuer-Nr. DE 012 345 07 12345 Samplehausen IK): 123456 780 BSNR 987 054321 Rechnung Nr. PP1912-1833 (DUPLIKAT) 17.01.2020 Bitte bei Überwaisung unbacingt angeben0 Patient Doe John geb. 31.3.1997 Katalog: GOA Datus Preis inkl ANR Einsender Greadpreis Steig.Satz Geb.-Ziff. Text der Behandlung Anzahl Betrag (EUR) Steig. Betrag CEURO 05.12.2019 22685 Dr. mod. Juan Perce Berlin 4525 CampyTobacter-AG 1 14.57 1.15 16.76 4530 Bakteriennachweris durch einfache Anzüchtung a 2 9.32 1.15 10.72 4538 Bakteriennachweis durch Anzüchtung auf Selekt 4 27.95 1.15 32.16 4539 Bakterisenachweis durch besenders aufwendige 1 14.57 1.15 16.76 3744 Calprotectin in Stuhl 1 26.23 1.15 30.16 4069 Pankreas-Elastase in St. 1 43.72 1.15 $0.28 3741 CRP quantitativ 1 11.66 1.15 13.41 3897 Gliadin-AK IOG (deamidierte Peptide) 1 29.73 1.15 34.19 3897 Gliadin-Ak ICA (deamidierte Peptide) 1 29.73 1.15 34.19 P10/3 Ersatz von Auslagen (GOA (10) 1 3.50 1.0 3.50 Endsume: 242.13 ber 24. 1.2020 Dis Untersuchungen sind nach der Gobührencednung für Armt (GOA) Stand 01.01.2009-berechnet Material- und Versandkosten werden entsprechend $ 10 GOA in Höhe der entstandanon Kosten erhoben Gem53 1 12 GOA is dia Rechnung sofort filig A-Ziffern gleich Leistungen gem&ß I 6 Abs 2 der GOA. Bitte überweison Sig innerhalb der nächsten 14 Tage die Rechnangssumme< auf dos unter angegebene Konto. Banieverbindung: PssudobankBortin BIC: ABCDEFGHI IBAN: DE01 2345 5789 101112 Konto Nr. 123 456 789",
    "results": [
        {
            "recipeId": "c92ab378dff3f872efc9e717",
            "result": "2 Medizinisch-Diagnostisches-Labor ************** Dr. med. ********** allgäuLab Gemeinscheftspraxis für Laborstoriumsmedizin Med Milerobiologie TBC Diagnost$ Mytologie Porositologie ******* Infeksionsserologie Immunologie Hamanologie **************** Endokrinologie Molekularbiologie Virologie Hausanschrift Heiko *******- Gasse & 12345 ****** Dr mail Am Pares Hals Dahas Class : 12345 ********** Telefon: ************* Telefax ************** ******** ************ 17 Steuer-Nr. DE ********** 12345 ************ IK): 123456 780 BSNR ********** Rechnung Nr. PP1912-1833 (DUPLIKAT) ********** Bitte bei Überwaisung unbacingt angeben0 Patient Doe **** geb. ********* Katalog: GOA Datus Preis inkl ANR Einsender Greadpreis Steig.Satz Geb.-Ziff. Text der Behandlung Anzahl Betrag (EUR) Steig. Betrag CEURO ********** 22685 Dr. mod. Juan Perce ****** ********************* 1 14.57 1.15 16.76 4530 Bakteriennachweris durch einfache Anzüchtung a 2 9.32 1.15 10.72 4538 Bakteriennachweis durch Anzüchtung auf Selekt 4 27.95 1.15 32.16 4539 Bakterisenachweis durch besenders aufwendige 1 14.57 1.15 16.76 3744 Calprotectin in Stuhl 1 26.23 1.15 30.16 4069 Pankreas-Elastase in St. 1 43.72 1.15 $0.28 3741 CRP quantitativ 1 11.66 1.15 13.41 3897 Gliadin-AK IOG (deamidierte Peptide) 1 29.73 1.15 34.19 3897 Gliadin-Ak ICA (deamidierte Peptide) 1 29.73 1.15 34.19 P10/3 Ersatz von Auslagen (GOA (10) 1 3.50 1.0 3.50 Endsume: 242.13 ber 24. 1.2020 Dis Untersuchungen sind nach der Gobührencednung für Armt (GOA) Stand 01.01.2009-berechnet Material- und Versandkosten werden entsprechend $ 10 GOA in Höhe der entstandanon Kosten erhoben Gem53 1 12 GOA is dia Rechnung sofort filig A-Ziffern gleich Leistungen gem&ß I 6 Abs 2 der GOA. Bitte überweison Sig innerhalb der nächsten 14 Tage die Rechnangssumme< auf dos unter angegebene Konto. Banieverbindung: PssudobankBortin BIC: ABCDEFGHI IBAN: DE01 2345 5789 101112 Konto Nr. ***********",
            "recipeName": "Prompt Redact All"
        }
    ]
}

Key	Value
itemId	The item id of dataset item
type	The type of document
source	Presigned URl or S3 source of document
text	The text associated with document
results	List of dictionary containing results
recipeId	The id of the recipe
result	Redaction result
recipeName	The name of the recipe

4. Q-A-Recipe¶

Model Info - Choose registered model.
Model Parameters
- Prompt Name- Enter the prompt name
- Summarize document text -
  Max words -Configure Max words to summarize the documents.
- Prompt Head - It refers to the initial portion of a prompt or text that provides context,sets the topic or outlines the task for generating a response.
- Add prompt example
  - New example - You can add a new example using the following configurations.
  - Documents to use - Paste the text.It can be utilized as reference material.
  - User question - Question related to above text.
  - Chatbot Answer - This signifies the response generated by the chatbot in reaction to the user’s question.
- Max number of similar documents (Top K)- this setting limits the number of most similar documents that will be presented or analyzed in relation to a particular reference document.
- Temperature(0-1) - The Temperature setting in an LLM controls the randomness of its output.A higher temperature makes the output more creative but less focused, while a lower temperature makes the output more predictable and controlled.
- Test Text - It is the test text payload which can be customized.
- Test question - Enter test question
- Test Model - You can test the model integration with the Test Text payload. Click Test Model and validate the model output to ensure the integration works as expected.
- Compiled Prompt - Once prompt head and prompt examples are added on the left of window, it is visible in the constructed prompt on the right.
Q-A Request Payload and Response

Redaction Test Recipe Response

{
Charles Darvin Beagle Street 1687

Objective: Highly skilled software engineer with a passion for developing innovative solutions and a strong background in various programming languages. Seeking a challenging position in a dynamic organization where I can apply my technical expertise and contribute to the development of cutting-edge software solutions.

Education: Bachelor of Science in Computer Science Oxford, 1999

Skills:

Programming Languages: Java, Python, C++, JavaScript Web Development: HTML, CSS, React, Node.js Database Management: SQL, MongoDB Version Control: Git Software Development Lifecycle (SDLC) Object-Oriented Design and Analysis (OOAD) Problem-solving and debugging skills Strong knowledge of algorithms and data structures Teamwork and collaboration Excellent verbal and written communication skills Experience: Software Engineer Intern Beagle INC, 2020-Present

Collaborated with a team of software engineers to develop and maintain web applications using Java and JavaScript. Participated in code reviews and provided feedback to ensure high-quality code. Implemented new features and resolved bugs in existing software applications. Assisted in database management and optimization for improved performance. Conducted unit testing and assisted in integration testing to ensure functionality and reliability. Software Developer Oxtech, 2005-1029

English (Fluent) French (Intermediate) References: Available upon request

}

Redaction Recipe Response

This is a resume for a software engineer.