Dataset Viewer
Auto-converted to Parquet Duplicate
id
stringclasses
15 values
input
stringlengths
198
1.41k
chosen
stringlengths
39
1.29k
rejected
stringlengths
21
1.31k
meta
dict
#X_TAG_RENAME
Generate financial report data in XML format. Required constraints: - Return only XML without any Markdown fences or prose. - Root tag must be <financial_report>; keep allowed tags and attributes only. - Do not change tag names or add unknown elements.
<?xml version="1.0" encoding="UTF-8"?> <financial_report> <company>Miller, Strickland and Lee</company> <revenue>907554</revenue> <profit>79706</profit> </financial_report>
<?xml version="1.0" encoding="UTF-8"?> <financial_report> <company>Miller, Strickland and Lee</company> <revenue_data>907554</revenue_data> <profit>79706</profit> </financial_report>
{ "format": "xml", "rule_id": "X_TAG_RENAME", "severity": "moderate", "seed_id": "", "parseable_full": true, "parseable_inner": false, "markdown_wrapped": false, "margin_score": 2, "margin_reasons": [ "missing_xml_tags" ], "violation_category": "severe" }
#J_STRING_TRUNCATE
Convert the following YAML to JSON: period: Q1 sales: 4903 region: Port Blakeborough Required constraints: - Return only JSON without any Markdown fences or prose. - Top-level keys must be exactly: period, sales, region (no extra or missing keys).
{ "period": "Q1", "sales": 4903, "region": "Port Blakeborough" }
{ "period": "Q1", "sales": 4903, "region": "Port Bl" }
{ "format": "json", "rule_id": "J_STRING_TRUNCATE", "severity": "moderate", "seed_id": "", "parseable_full": true, "parseable_inner": false, "markdown_wrapped": false, "margin_score": 1, "margin_reasons": [ "value_changes" ], "violation_category": "moderate" }
#X_TAG_RENAME
Convert the following YAML to XML: paper_id: arXiv:2024.31103 title: Quality-focused fresh-thinking customer loyalty authors: - name: Darlene Mckinney affiliation: White PLC email: ssmith@example.org - name: Henry Gonzalez affiliation: Acosta-Reynolds email: kyleanderson@example.com abstract: 'Heavy miss exactl...
<?xml version="1.0" encoding="UTF-8"?> <research_paper> <paper_id>arXiv:2024.31103</paper_id> <title>Quality-focused fresh-thinking customer loyalty</title> <authors> <name>Darlene Mckinney</name> <affiliation>White PLC</affiliation> <email>ssmith@example.org</email> </authors> <authors> <name>Henry Gonzalez</name> <af...
<?xml version="1.0" encoding="UTF-8"?> <research_paper> <paper_id>arXiv:2024.31103</paper_id> <title>Quality-focused fresh-thinking customer loyalty</title> <authors> <name>Darlene Mckinney</name> <affiliation>White PLC</affiliation> <email_node>ssmith@example.org</email_node> </authors> <authors> <name>Henry Gonzalez<...
{ "format": "xml", "rule_id": "X_TAG_RENAME", "severity": "moderate", "seed_id": "", "parseable_full": true, "parseable_inner": false, "markdown_wrapped": false, "margin_score": 2, "margin_reasons": [ "missing_xml_tags" ], "violation_category": "severe" }
#X_TAG_RENAME
Convert the following TOML to XML: student = "Denise Chandler" score = 40 grade = "C" Required constraints: - Return only XML without any Markdown fences or prose. - Root tag must be <student_assessment>; keep allowed tags and attributes only. - Do not change tag names or add unknown elements.
<?xml version="1.0" encoding="UTF-8"?> <student_assessment> <student>Denise Chandler</student> <score>40</score> <grade>C</grade> </student_assessment>
<?xml version="1.0" encoding="UTF-8"?> <student_assessment> <student>Denise Chandler</student> <score>40</score> <grade_data>C</grade_data> </student_assessment>
{ "format": "xml", "rule_id": "X_TAG_RENAME", "severity": "moderate", "seed_id": "", "parseable_full": true, "parseable_inner": false, "markdown_wrapped": false, "margin_score": 2, "margin_reasons": [ "missing_xml_tags" ], "violation_category": "severe" }
#C_CODE_FENCE_WRAP
Output structured data for social_media_post using YAML. Required constraints: - Return only YAML without any Markdown fences or prose. - Top-level keys must be exactly: Approach, 1. Define the schema, 2. Select the format, 3. Generate realistic content, 4. Structure the data, 5. Serialize, Output, user, post, likes (...
Approach: 1. Define the schema: The social_media_post should contain a dictionary with 3 root fields (user, post, likes) and a nesting depth of 1. 2. Select the format: YAML is chosen. Use indentation to represent nesting. lists are denoted by hyphens. 3. Generate realistic content: Use plausible names, dates, and valu...
## Result ```yaml Approach: 1. Define the schema: The social_media_post should contain a dictionary with 3 root fields (user, post, likes) and a nesting depth of 1. 2. Select the format: YAML is chosen. Use indentation to represent nesting. lists are denoted by hyphens. 3. Generate realistic content: Use plausible name...
{ "format": "yaml", "rule_id": "C_CODE_FENCE_WRAP", "severity": "minor", "seed_id": "", "parseable_full": false, "parseable_inner": true, "markdown_wrapped": true, "margin_score": 1, "margin_reasons": [ "prohibited_markdown" ], "violation_category": "minor" }
#T_STRING_TRUNCATE
Output structured data for student_assessment using TOML. Required constraints: - Return only TOML without any Markdown fences or prose. - Top-level keys must be exactly: student, score, grade (no extra or missing keys).
student = "Edward Dixon" score = 22 grade = "C"
student = "Edwar" score = 22 grade = "C"
{ "format": "toml", "rule_id": "T_STRING_TRUNCATE", "severity": "moderate", "seed_id": "", "parseable_full": true, "parseable_inner": false, "markdown_wrapped": false, "margin_score": 1, "margin_reasons": [ "value_changes" ], "violation_category": "moderate" }
#J_LIST_EXTRA_ITEM
Convert the following YAML to JSON: id: 52d5492f-14d5-4e12-bb4f-27ab05134a37 patient: Sabrina Dean diagnosis: commercial notes: 'Board argue color ago season its there. Number Mr however enjoy night. Cell set fine their. Smile forget simply possible hotel respond do. Great country dinner student card enter prote...
{ "id": "52d5492f-14d5-4e12-bb4f-27ab05134a37", "patient": "Sabrina Dean", "diagnosis": "commercial", "notes": "Board argue color ago season its there. Number Mr however enjoy night. Cell set fine their.\nSmile forget simply possible hotel respond do. Great country dinner student card enter protect to.", "met...
{ "id": "52d5492f-14d5-4e12-bb4f-27ab05134a37", "patient": "Sabrina Dean", "diagnosis": "commercial", "notes": "Board argue color ago season its there. Number Mr however enjoy night. Cell set fine their.\nSmile forget simply possible hotel respond do. Great country dinner student card enter protect to.", "met...
{ "format": "json", "rule_id": "J_LIST_EXTRA_ITEM", "severity": "moderate", "seed_id": "", "parseable_full": true, "parseable_inner": false, "markdown_wrapped": false, "margin_score": 1, "margin_reasons": [ "value_changes" ], "violation_category": "moderate" }
#T_KEY_DROP
Convert the following XML to TOML: <?xml version="1.0" encoding="UTF-8"?> <experiment_result> <experiment_id>EXP-100053</experiment_id> <title>Analysis of hour</title> <metrics> <accuracy>0.791</accuracy> <loss>0.43</loss> <epoch>30</epoch> </metrics> <config> <batch_size>32</batch_size> <optimizer>sgd</optimizer> <lr>...
experiment_id = "EXP-100053" title = "Analysis of hour" [metrics] accuracy = 0.791 loss = 0.43 epoch = 30 [config] batch_size = 32 optimizer = "sgd" lr = 0.001
title = "Analysis of hour" [metrics] accuracy = 0.791 loss = 0.43 epoch = 30 [config] batch_size = 32 optimizer = "sgd" lr = 0.001
{ "format": "toml", "rule_id": "T_KEY_DROP", "severity": "moderate", "seed_id": "", "parseable_full": true, "parseable_inner": false, "markdown_wrapped": false, "margin_score": 2, "margin_reasons": [ "missing_required_keys" ], "violation_category": "severe" }
#C_CODE_FENCE_WRAP
Create a YAML representation of product listing. Required constraints: - Return only YAML without any Markdown fences or prose. - Top-level keys must be exactly: Approach, 1. Define the schema, 2. Select the format, 3. Generate realistic content, 4. Structure the data, 5. Serialize, Output, product, price, stock, meta...
Approach: 1. Define the schema: The product_listing should contain a dictionary with 5 root fields (product, price, stock...) and a nesting depth of 3. 2. Select the format: YAML is chosen. Use indentation to represent nesting. lists are denoted by hyphens. 3. Generate realistic content: Use plausible names, dates, and...
## Result ```yaml Approach: 1. Define the schema: The product_listing should contain a dictionary with 5 root fields (product, price, stock...) and a nesting depth of 3. 2. Select the format: YAML is chosen. Use indentation to represent nesting. lists are denoted by hyphens. 3. Generate realistic content: Use plausible...
{ "format": "yaml", "rule_id": "C_CODE_FENCE_WRAP", "severity": "minor", "seed_id": "", "parseable_full": false, "parseable_inner": true, "markdown_wrapped": true, "margin_score": 1, "margin_reasons": [ "prohibited_markdown" ], "violation_category": "minor" }
#T_KEY_DROP
Output structured data for error_log using TOML. Required constraints: - Return only TOML without any Markdown fences or prose. - Top-level keys must be exactly: timestamp, level, message (no extra or missing keys).
timestamp = "2026-01-22T04:03:07.550510" level = "ERROR" message = "The report contain ten."
timestamp = "2026-01-22T04:03:07.550510" level = "ERROR"
{ "format": "toml", "rule_id": "T_KEY_DROP", "severity": "moderate", "seed_id": "", "parseable_full": true, "parseable_inner": false, "markdown_wrapped": false, "margin_score": 2, "margin_reasons": [ "missing_required_keys" ], "violation_category": "severe" }
#T_KEY_DROP
Output structured data for customer_review using TOML. Required constraints: - Return only TOML without any Markdown fences or prose. - Top-level keys must be exactly: user, rating, comment (no extra or missing keys).
user = "mcbridedennis" rating = 4 comment = "Now through hospital one."
rating = 4 comment = "Now through hospital one."
{ "format": "toml", "rule_id": "T_KEY_DROP", "severity": "moderate", "seed_id": "", "parseable_full": true, "parseable_inner": false, "markdown_wrapped": false, "margin_score": 2, "margin_reasons": [ "missing_required_keys" ], "violation_category": "severe" }
#J_STRING_TRUNCATE
Generate research paper data in JSON format. Required constraints: - Return only JSON without any Markdown fences or prose. - Top-level keys must be exactly: paper_id, title, authors, abstract, keywords, publication (no extra or missing keys). - All list items must be unique; do not duplicate entries.
{ "paper_id": "arXiv:2021.91678", "title": "Right-sized composite synergy", "authors": [ { "name": "Christopher Paul", "affiliation": "Gonzalez-Woods", "email": "bernard88@example.com" }, { "name": "Jennifer Williams", "affiliation": "Chase, Moyer and Perez", "email...
{ "paper_id": "arXiv:", "title": "Right-sized composite synergy", "authors": [ { "name": "Christopher Paul", "affiliation": "Gonzalez-Woods", "email": "bernard88@example.com" }, { "name": "Jennifer Williams", "affiliation": "Chase, Moyer and Perez", "email": "perryc...
{ "format": "json", "rule_id": "J_STRING_TRUNCATE", "severity": "moderate", "seed_id": "", "parseable_full": true, "parseable_inner": false, "markdown_wrapped": false, "margin_score": 1, "margin_reasons": [ "value_changes" ], "violation_category": "moderate" }
#X_TAG_RENAME
Convert the following CSV to XML: headline,body,author,meta_info.generated_at,meta_info.version,meta_info.tags,audit_log Pm meet type.,"Left family agency tend necessary floor. Shake budget care lose. Newspaper choice never perhaps rather. Boy where source condition. Their opportunity lay standard in their clear mothe...
<?xml version="1.0" encoding="UTF-8"?> <news_article> <headline>Pm meet type.</headline> <body>Left family agency tend necessary floor. Shake budget care lose. Newspaper choice never perhaps rather. Boy where source condition. Their opportunity lay standard in their clear mother.</body> <author>Lauren Robbins</author> ...
<?xml version="1.0" encoding="UTF-8"?> <news_article> <headline>Pm meet type.</headline> <body>Left family agency tend necessary floor. Shake budget care lose. Newspaper choice never perhaps rather. Boy where source condition. Their opportunity lay standard in their clear mother.</body> <author>Lauren Robbins</author> ...
{ "format": "xml", "rule_id": "X_TAG_RENAME", "severity": "moderate", "seed_id": "", "parseable_full": true, "parseable_inner": false, "markdown_wrapped": false, "margin_score": 2, "margin_reasons": [ "missing_xml_tags" ], "violation_category": "severe" }
#X_TAG_RENAME
Reformat the JSON below into XML: { "rx": "7d4b6095-90a3-4ea8-8a9d-ff58756d8ff5", "drug": "energy", "dosage": "10mg" } Required constraints: - Return only XML without any Markdown fences or prose. - Root tag must be <prescription>; keep allowed tags and attributes only. - Do not change tag names or add unknown e...
<?xml version="1.0" encoding="UTF-8"?> <prescription> <rx>7d4b6095-90a3-4ea8-8a9d-ff58756d8ff5</rx> <drug>energy</drug> <dosage>10mg</dosage> </prescription>
<?xml version="1.0" encoding="UTF-8"?> <prescription_node> <rx>7d4b6095-90a3-4ea8-8a9d-ff58756d8ff5</rx> <drug>energy</drug> <dosage>10mg</dosage> </prescription_node>
{ "format": "xml", "rule_id": "X_TAG_RENAME", "severity": "moderate", "seed_id": "", "parseable_full": true, "parseable_inner": false, "markdown_wrapped": false, "margin_score": 2, "margin_reasons": [ "missing_xml_tags" ], "violation_category": "severe" }
#Y_LIST_ITEM_DROP
Convert the following CSV to YAML: user,rating,comment,meta_info.generated_at,meta_info.version,meta_info.tags,audit_log rdean,3,Which always national southern notice rise hotel.,2026-01-22T04:03:07.452608,v3.0,"[""keep"", ""to"", ""national"", ""century"", ""happy""]","[{""action"": ""created"", ""by"": ""theresasmit...
Approach: 1. Analyze the input CSV structure: It represents a dictionary with 5 root fields (user, rating, comment...) and a nesting depth of 3. 2. Identify the target format requirements: YAML requires strict syntax. Use indentation to represent nesting. lists are denoted by hyphens. 3. Map the fields: The root keys w...
1. Analyze the input CSV structure: It represents a dictionary with 5 root fields (user, rating, comment...) and a nesting depth of 3. 2. Identify the target format requirements: YAML requires strict syntax. Use indentation to represent nesting. lists are denoted by hyphens. 3. Map the fields: The root keys will ma...
{ "format": "yaml", "rule_id": "Y_LIST_ITEM_DROP", "severity": "minor", "seed_id": "", "parseable_full": true, "parseable_inner": false, "markdown_wrapped": false, "margin_score": 1, "margin_reasons": [ "value_changes" ], "violation_category": "moderate" }
#T_KEY_DROP
Convert the following JSON to TOML: { "rx": "fa8df1e7-11bd-487d-bd7d-6f5104299197", "drug": "others", "dosage": "10mg" } Required constraints: - Return only TOML without any Markdown fences or prose. - Top-level keys must be exactly: rx, drug, dosage (no extra or missing keys).
rx = "fa8df1e7-11bd-487d-bd7d-6f5104299197" drug = "others" dosage = "10mg"
rx = "fa8df1e7-11bd-487d-bd7d-6f5104299197" dosage = "10mg"
{ "format": "toml", "rule_id": "T_KEY_DROP", "severity": "moderate", "seed_id": "", "parseable_full": true, "parseable_inner": false, "markdown_wrapped": false, "margin_score": 2, "margin_reasons": [ "missing_required_keys" ], "violation_category": "severe" }
#S_ROWCOUNT
Output structured data for student_assessment using CSV. Required constraints: - Return only CSV without any Markdown fences or prose. - CSV header columns must be exactly this order: 1. Define the schema: The student_assessment should contain a dictionary with 3 root fields (student, score, grade) and a nesting dep...
1. Define the schema: The student_assessment should contain a dictionary with 3 root fields (student, score, grade) and a nesting depth of 1. 2. Select the format: CSV is chosen. Flatten the hierarchical structure. Use dot notation for nested keys in the header. 3. Generate realistic content: Use plausible names, dates...
1. Define the schema: The student_assessment should contain a dictionary with 3 root fields (student, score, grade) and a nesting depth of 1. 2. Select the format: CSV is chosen. Flatten the hierarchical structure. Use dot notation for nested keys in the header. 3. Generate realistic content: Use plausible names, dates...
{ "format": "csv", "rule_id": "S_ROWCOUNT", "severity": "minor", "seed_id": "", "parseable_full": true, "parseable_inner": false, "markdown_wrapped": false, "margin_score": 1, "margin_reasons": [ "csv_rowcount_diff" ], "violation_category": "minor" }
#J_LIST_EXTRA_ITEM
Output structured data for error_log using JSON. Required constraints: - Return only JSON without any Markdown fences or prose. - Top-level keys must be exactly: timestamp, level, message, meta_info, audit_log (no extra or missing keys). - Field meta_info.generated_at is required. - All list items must be unique; do n...
{ "timestamp": "2026-01-22T04:03:06.699723", "level": "ERROR", "message": "Research recently college street imagine memory discuss.", "meta_info": { "generated_at": "2026-01-22T04:03:06.699816", "version": "v2.0", "tags": [ "push", "cultural", "establish", "pull", "hit"...
{ "timestamp": "2026-01-22T04:03:06.699723", "level": "ERROR", "message": "Research recently college street imagine memory discuss.", "meta_info": { "generated_at": "2026-01-22T04:03:06.699816", "version": "v2.0", "tags": [ "push", "cultural", "establish", "pull", "hit"...
{ "format": "json", "rule_id": "J_LIST_EXTRA_ITEM", "severity": "moderate", "seed_id": "", "parseable_full": true, "parseable_inner": false, "markdown_wrapped": false, "margin_score": 1, "margin_reasons": [ "value_changes" ], "violation_category": "moderate" }
#X_TAG_RENAME
Generate financial report data in XML format. Required constraints: - Return only XML without any Markdown fences or prose. - Root tag must be <financial_report>; keep allowed tags and attributes only. - Do not change tag names or add unknown elements.
<?xml version="1.0" encoding="UTF-8"?> <financial_report> <company>Silva Inc</company> <revenue>702036</revenue> <profit>68525</profit> <meta_info> <generated_at>2026-01-22T04:03:05.223463</generated_at> <version>v5.0</version> <tags>consider</tags> <tags>name</tags> <tags>run</tags> <tags>training</tags> <tags>dark</t...
<?xml version="1.0" encoding="UTF-8"?> <financial_report> <company_node>Silva Inc</company_node> <revenue>702036</revenue> <profit>68525</profit> <meta_info> <generated_at>2026-01-22T04:03:05.223463</generated_at> <version>v5.0</version> <tags>consider</tags> <tags>name</tags> <tags>run</tags> <tags>training</tags> <ta...
{ "format": "xml", "rule_id": "X_TAG_RENAME", "severity": "moderate", "seed_id": "", "parseable_full": true, "parseable_inner": false, "markdown_wrapped": false, "margin_score": 2, "margin_reasons": [ "missing_xml_tags" ], "violation_category": "severe" }
#X_TAG_RENAME
Reformat the CSV below into XML: paper_id,title,authors,abstract,keywords,publication.venue,publication.year,publication.doi arXiv:2022.39534,Cross-group 4thgeneration knowledge user,"[{""name"": ""Eric Thomas"", ""affiliation"": ""Hutchinson-Smith"", ""email"": ""lorisanchez@example.net""}, {""name"": ""Alexandra Dix...
<?xml version="1.0" encoding="UTF-8"?> <research_paper> <paper_id>arXiv:2022.39534</paper_id> <title>Cross-group 4thgeneration knowledge user</title> <authors> <name>Eric Thomas</name> <affiliation>Hutchinson-Smith</affiliation> <email>lorisanchez@example.net</email> </authors> <authors> <name>Alexandra Dixon</name> <a...
<?xml version="1.0" encoding="UTF-8"?> <research_paper> <paper_id>arXiv:2022.39534</paper_id> <title>Cross-group 4thgeneration knowledge user</title> <authors> <name>Eric Thomas</name> <affiliation>Hutchinson-Smith</affiliation> <email_node>lorisanchez@example.net</email_node> </authors> <authors> <name>Alexandra Dixon...
{ "format": "xml", "rule_id": "X_TAG_RENAME", "severity": "moderate", "seed_id": "", "parseable_full": true, "parseable_inner": false, "markdown_wrapped": false, "margin_score": 2, "margin_reasons": [ "missing_xml_tags" ], "violation_category": "severe" }
#S_ROWCOUNT
Convert the following TOML to CSV: id = "6ff47b77-a436-4180-bc41-36a6fe6835b9" test = "between" value = 7.159897593981157 [[audit_log]] action = "created" by = "michaelcurry" timestamp = "2021-12-06T10:41:36.796956" [[audit_log]] action = "created" by = "brenda92" timestamp = "2014-11-08T13:17:03.460189" [meta_info] ...
1. Analyze the input TOML structure: It represents a dictionary with 5 root fields (id, test, value...) and a nesting depth of 3. 2. Identify the target format requirements: CSV requires strict syntax. Flatten the hierarchical structure. Use dot notation for nested keys in the header. 3. Map the fields: The root keys w...
1. Analyze the input TOML structure: It represents a dictionary with 5 root fields (id, test, value...) and a nesting depth of 3. 2. Identify the target format requirements: CSV requires strict syntax. Flatten the hierarchical structure. Use dot notation for nested keys in the header. 3. Map the fields: The root keys w...
{ "format": "csv", "rule_id": "S_ROWCOUNT", "severity": "minor", "seed_id": "", "parseable_full": true, "parseable_inner": false, "markdown_wrapped": false, "margin_score": 1, "margin_reasons": [ "csv_rowcount_diff" ], "violation_category": "minor" }
#T_KEY_DROP
Create a TOML representation of contract. Required constraints: - Return only TOML without any Markdown fences or prose. - Top-level keys must be exactly: contract, parties, signed (no extra or missing keys). - All list items must be unique; do not duplicate entries.
contract = "leverage vertical e-commerce" parties = [ "Wilson Group", "Ross-Irwin",] signed = true
contract = "leverage vertical e-commerce" parties = [ "Wilson Group", "Ross-Irwin",]
{ "format": "toml", "rule_id": "T_KEY_DROP", "severity": "moderate", "seed_id": "", "parseable_full": true, "parseable_inner": false, "markdown_wrapped": false, "margin_score": 2, "margin_reasons": [ "missing_required_keys" ], "violation_category": "severe" }
#X_TAG_RENAME
Generate course syllabus data in XML format. Required constraints: - Return only XML without any Markdown fences or prose. - Root tag must be <course_syllabus>; keep allowed tags and attributes only. - Do not change tag names or add unknown elements.
<?xml version="1.0" encoding="UTF-8"?> <course_syllabus> <course>Face-to-face directional info-mediaries</course> <weeks>9</weeks> <credits>3</credits> </course_syllabus>
<?xml version="1.0" encoding="UTF-8"?> <course_syllabus> <course>Face-to-face directional info-mediaries</course> <weeks_data>9</weeks_data> <credits>3</credits> </course_syllabus>
{ "format": "xml", "rule_id": "X_TAG_RENAME", "severity": "moderate", "seed_id": "", "parseable_full": true, "parseable_inner": false, "markdown_wrapped": false, "margin_score": 2, "margin_reasons": [ "missing_xml_tags" ], "violation_category": "severe" }
#Y_VALUE_PERTURB
Output structured data for social_media_post using YAML. Required constraints: - Return only YAML without any Markdown fences or prose. - Top-level keys must be exactly: Approach, 1. Define the schema, 2. Select the format, 3. Generate realistic content, 4. Structure the data, 5. Serialize, Output, user, post, likes (...
Approach: 1. Define the schema: The social_media_post should contain a dictionary with 3 root fields (user, post, likes) and a nesting depth of 1. 2. Select the format: YAML is chosen. Use indentation to represent nesting. lists are denoted by hyphens. 3. Generate realistic content: Use plausible names, dates, and valu...
1. Define the schema: The social_media_post should contain a dictionary with 3 root fields (user, post, likes) and a nesting depth of 1. 2. Select the format: YAML is chosen. Use indentation to represent nesting. lists are denoted by hyphens. 3. Generate realistic content: Use plausible names, dates, and values app...
{ "format": "yaml", "rule_id": "Y_VALUE_PERTURB", "severity": "minor", "seed_id": "", "parseable_full": true, "parseable_inner": false, "markdown_wrapped": false, "margin_score": 1, "margin_reasons": [ "value_changes" ], "violation_category": "moderate" }
#J_STRING_TRUNCATE
Create a JSON representation of student assessment. Required constraints: - Return only JSON without any Markdown fences or prose. - Top-level keys must be exactly: student, score, grade, meta_info, audit_log (no extra or missing keys). - Field meta_info.generated_at is required. - All list items must be unique; do no...
{ "student": "Tracy Armstrong", "score": 17, "grade": "B", "meta_info": { "generated_at": "2026-01-22T04:03:05.432409", "version": "v1.0", "tags": [ "through", "other", "message", "fall", "behavior" ] }, "audit_log": [ { "action": "created", "by"...
{ "student": "Tracy ", "score": 17, "grade": "B", "meta_info": { "generated_at": "2026-01-22T04:03:05.432409", "version": "v1.0", "tags": [ "through", "other", "message", "fall", "behavior" ] }, "audit_log": [ { "action": "created", "by": "huffma...
{ "format": "json", "rule_id": "J_STRING_TRUNCATE", "severity": "moderate", "seed_id": "", "parseable_full": true, "parseable_inner": false, "markdown_wrapped": false, "margin_score": 1, "margin_reasons": [ "value_changes" ], "violation_category": "moderate" }
#C_CODE_FENCE_WRAP
Reformat the CSV below into XML: user,post,likes boyersylvia,Interview knowledge ready hold establish hotel several. Money subject away without special. Oil whether short air become.,450 Required constraints: - Return only XML without any Markdown fences or prose. - Root tag must be <social_media_post>; keep allowed ...
<?xml version="1.0" encoding="UTF-8"?> <social_media_post> <user>boyersylvia</user> <post>Interview knowledge ready hold establish hotel several. Money subject away without special. Oil whether short air become.</post> <likes>450</likes> </social_media_post>
## Result ```xml <?xml version="1.0" encoding="UTF-8"?> <social_media_post> <user>boyersylvia</user> <post>Interview knowledge ready hold establish hotel several. Money subject away without special. Oil whether short air become.</post> <likes>450</likes> </social_media_post> ```
{ "format": "xml", "rule_id": "C_CODE_FENCE_WRAP", "severity": "minor", "seed_id": "", "parseable_full": false, "parseable_inner": true, "markdown_wrapped": true, "margin_score": 1, "margin_reasons": [ "prohibited_markdown" ], "violation_category": "minor" }
#C_CODE_FENCE_WRAP
Generate lab result data in XML format. Required constraints: - Return only XML without any Markdown fences or prose. - Root tag must be <lab_result>; keep allowed tags and attributes only. - Do not change tag names or add unknown elements.
<?xml version="1.0" encoding="UTF-8"?> <lab_result> <id>ce6ab4d4-4e57-435d-bf44-68b0ddda8b05</id> <test>necessary</test> <value>36.569220334710764</value> <meta_info> <generated_at>2026-01-22T04:03:06.974558</generated_at> <version>v5.0</version> <tags>feeling</tags> <tags>ask</tags> <tags>ready</tags> <tags>interestin...
## Result ```xml <?xml version="1.0" encoding="UTF-8"?> <lab_result> <id>ce6ab4d4-4e57-435d-bf44-68b0ddda8b05</id> <test>necessary</test> <value>36.569220334710764</value> <meta_info> <generated_at>2026-01-22T04:03:06.974558</generated_at> <version>v5.0</version> <tags>feeling</tags> <tags>ask</tags> <tags>ready</tags>...
{ "format": "xml", "rule_id": "C_CODE_FENCE_WRAP", "severity": "minor", "seed_id": "", "parseable_full": false, "parseable_inner": true, "markdown_wrapped": true, "margin_score": 1, "margin_reasons": [ "prohibited_markdown" ], "violation_category": "minor" }
#X_TAG_RENAME
Output structured data for lab_result using XML. Required constraints: - Return only XML without any Markdown fences or prose. - Root tag must be <lab_result>; keep allowed tags and attributes only. - Do not change tag names or add unknown elements.
<?xml version="1.0" encoding="UTF-8"?> <lab_result> <id>0e5c588b-7c37-481c-8f39-8514d2cb626f</id> <test>me</test> <value>73.86088081333102</value> </lab_result>
<?xml version="1.0" encoding="UTF-8"?> <lab_result> <id>0e5c588b-7c37-481c-8f39-8514d2cb626f</id> <test_node>me</test_node> <value>73.86088081333102</value> </lab_result>
{ "format": "xml", "rule_id": "X_TAG_RENAME", "severity": "moderate", "seed_id": "", "parseable_full": true, "parseable_inner": false, "markdown_wrapped": false, "margin_score": 2, "margin_reasons": [ "missing_xml_tags" ], "violation_category": "severe" }
#X_TAG_RENAME
Create a XML representation of product listing. Required constraints: - Return only XML without any Markdown fences or prose. - Root tag must be <product_listing>; keep allowed tags and attributes only. - Do not change tag names or add unknown elements.
<?xml version="1.0" encoding="UTF-8"?> <product_listing> <product>style</product> <price>54.2</price> <stock>9</stock> <meta_info> <generated_at>2026-01-22T04:03:07.393083</generated_at> <version>v5.0</version> <tags>season</tags> <tags>goal</tags> <tags>continue</tags> <tags>value</tags> <tags>other</tags> </meta_info...
<?xml version="1.0" encoding="UTF-8"?> <product_listing> <product>style</product> <price>54.2</price> <stock>9</stock> <meta_info> <generated_at>2026-01-22T04:03:07.393083</generated_at> <version>v5.0</version> <tags_node>season</tags_node> <tags_node>goal</tags_node> <tags_node>continue</tags_node> <tags_node>value</t...
{ "format": "xml", "rule_id": "X_TAG_RENAME", "severity": "moderate", "seed_id": "", "parseable_full": true, "parseable_inner": false, "markdown_wrapped": false, "margin_score": 2, "margin_reasons": [ "missing_xml_tags" ], "violation_category": "severe" }
#X_TAG_RENAME
Transform this data from JSON to XML: { "period": "Q1", "sales": 3795, "region": "East Travishaven", "meta_info": { "generated_at": "2026-01-22T04:03:05.381605", "version": "v2.0", "tags": [ "eight", "her", "mean", "nation", "theory" ] }, "audit_log": [ { ...
<?xml version="1.0" encoding="UTF-8"?> <sales_analytics> <period>Q1</period> <sales>3795</sales> <region>East Travishaven</region> <meta_info> <generated_at>2026-01-22T04:03:05.381605</generated_at> <version>v2.0</version> <tags>eight</tags> <tags>her</tags> <tags>mean</tags> <tags>nation</tags> <tags>theory</tags> </m...
<?xml version="1.0" encoding="UTF-8"?> <sales_analytics> <period>Q1</period> <sales_node>3795</sales_node> <region>East Travishaven</region> <meta_info> <generated_at>2026-01-22T04:03:05.381605</generated_at> <version>v2.0</version> <tags>eight</tags> <tags>her</tags> <tags>mean</tags> <tags>nation</tags> <tags>theory<...
{ "format": "xml", "rule_id": "X_TAG_RENAME", "severity": "moderate", "seed_id": "", "parseable_full": true, "parseable_inner": false, "markdown_wrapped": false, "margin_score": 2, "margin_reasons": [ "missing_xml_tags" ], "violation_category": "severe" }
#X_TAG_RENAME
Generate contract data in XML format. Required constraints: - Return only XML without any Markdown fences or prose. - Root tag must be <contract>; keep allowed tags and attributes only. - Do not change tag names or add unknown elements.
<?xml version="1.0" encoding="UTF-8"?> <contract> <contract>deploy integrated ROI</contract> <parties>Nelson-Hamilton</parties> <parties>Jackson-Burton</parties> <signed>True</signed> </contract>
<?xml version="1.0" encoding="UTF-8"?> <contract> <contract>deploy integrated ROI</contract> <parties>Nelson-Hamilton</parties> <parties>Jackson-Burton</parties> <signed_node>True</signed_node> </contract>
{ "format": "xml", "rule_id": "X_TAG_RENAME", "severity": "moderate", "seed_id": "", "parseable_full": true, "parseable_inner": false, "markdown_wrapped": false, "margin_score": 2, "margin_reasons": [ "missing_xml_tags" ], "violation_category": "severe" }
#Y_LIST_ITEM_DROP
Create a YAML representation of clinical note. Required constraints: - Return only YAML without any Markdown fences or prose. - Top-level keys must be exactly: Approach, 1. Define the schema, 2. Select the format, 3. Generate realistic content, 4. Structure the data, 5. Serialize, Output, id, patient, diagnosis, notes...
Approach: 1. Define the schema: The clinical_note should contain a dictionary with 6 root fields (id, patient, diagnosis...) and a nesting depth of 3. 2. Select the format: YAML is chosen. Use indentation to represent nesting. lists are denoted by hyphens. 3. Generate realistic content: Use plausible names, dates, and ...
1. Define the schema: The clinical_note should contain a dictionary with 6 root fields (id, patient, diagnosis...) and a nesting depth of 3. 2. Select the format: YAML is chosen. Use indentation to represent nesting. lists are denoted by hyphens. 3. Generate realistic content: Use plausible names, dates, and values...
{ "format": "yaml", "rule_id": "Y_LIST_ITEM_DROP", "severity": "minor", "seed_id": "", "parseable_full": true, "parseable_inner": false, "markdown_wrapped": false, "margin_score": 1, "margin_reasons": [ "value_changes" ], "violation_category": "moderate" }
#T_KEY_DROP
Create a TOML representation of clinical note. Required constraints: - Return only TOML without any Markdown fences or prose. - Top-level keys must be exactly: id, patient, diagnosis, notes (no extra or missing keys).
id = "67e1b420-549d-4744-8cb3-627e50fd14ec" patient = "Brandon Herrera" diagnosis = "arrive" notes = "Poor story computer single. Need similar consumer a. Special new company across want PM right of."
patient = "Brandon Herrera" diagnosis = "arrive" notes = "Poor story computer single. Need similar consumer a. Special new company across want PM right of."
{ "format": "toml", "rule_id": "T_KEY_DROP", "severity": "moderate", "seed_id": "", "parseable_full": true, "parseable_inner": false, "markdown_wrapped": false, "margin_score": 2, "margin_reasons": [ "missing_required_keys" ], "violation_category": "severe" }
#X_TAG_RENAME
Transform this data from CSV to XML: timestamp,level,message 2026-01-22T04:03:06.143016,ERROR,My condition happy enjoy hear. Required constraints: - Return only XML without any Markdown fences or prose. - Root tag must be <error_log>; keep allowed tags and attributes only. - Do not change tag names or add unknown ele...
<?xml version="1.0" encoding="UTF-8"?> <error_log> <timestamp>2026-01-22T04:03:06.143016</timestamp> <level>ERROR</level> <message>My condition happy enjoy hear.</message> </error_log>
<?xml version="1.0" encoding="UTF-8"?> <error_log> <timestamp_node>2026-01-22T04:03:06.143016</timestamp_node> <level>ERROR</level> <message>My condition happy enjoy hear.</message> </error_log>
{ "format": "xml", "rule_id": "X_TAG_RENAME", "severity": "moderate", "seed_id": "", "parseable_full": true, "parseable_inner": false, "markdown_wrapped": false, "margin_score": 2, "margin_reasons": [ "missing_xml_tags" ], "violation_category": "severe" }
#X_TAG_RENAME
Reformat the TOML below into XML: student = "Kelly Tucker" score = 82 grade = "B" Required constraints: - Return only XML without any Markdown fences or prose. - Root tag must be <student_assessment>; keep allowed tags and attributes only. - Do not change tag names or add unknown elements.
<?xml version="1.0" encoding="UTF-8"?> <student_assessment> <student>Kelly Tucker</student> <score>82</score> <grade>B</grade> </student_assessment>
<?xml version="1.0" encoding="UTF-8"?> <student_assessment> <student>Kelly Tucker</student> <score>82</score> <grade_data>B</grade_data> </student_assessment>
{ "format": "xml", "rule_id": "X_TAG_RENAME", "severity": "moderate", "seed_id": "", "parseable_full": true, "parseable_inner": false, "markdown_wrapped": false, "margin_score": 2, "margin_reasons": [ "missing_xml_tags" ], "violation_category": "severe" }
#Y_VALUE_PERTURB
Create a YAML representation of student assessment. Required constraints: - Return only YAML without any Markdown fences or prose. - Top-level keys must be exactly: Approach, 1. Define the schema, 2. Select the format, 3. Generate realistic content, 4. Structure the data, 5. Serialize, Output, student, score, grade (n...
Approach: 1. Define the schema: The student_assessment should contain a dictionary with 3 root fields (student, score, grade) and a nesting depth of 1. 2. Select the format: YAML is chosen. Use indentation to represent nesting. lists are denoted by hyphens. 3. Generate realistic content: Use plausible names, dates, and...
1. Define the schema: The student_assessment should contain a dictionary with 3 root fields (student, score, grade) and a nesting depth of 1. 2. Select the format: YAML is chosen. Use indentation to represent nesting. lists are denoted by hyphens. 3. Generate realistic content: Use plausible names, dates, and value...
{ "format": "yaml", "rule_id": "Y_VALUE_PERTURB", "severity": "minor", "seed_id": "", "parseable_full": true, "parseable_inner": false, "markdown_wrapped": false, "margin_score": 1, "margin_reasons": [ "value_changes" ], "violation_category": "moderate" }
#J_STRING_TRUNCATE
Reformat the CSV below into JSON: id,patient,diagnosis,notes,meta_info.generated_at,meta_info.version,meta_info.tags,audit_log 0a9e4c0d-c2da-4338-9e23-30b43df699e9,Robert Simmons,student,Interview particularly treatment hospital join attack race among. Style safe level forward need once through. Every good over return...
{ "id": "0a9e4c0d-c2da-4338-9e23-30b43df699e9", "patient": "Robert Simmons", "diagnosis": "student", "notes": "Interview particularly treatment hospital join attack race among. Style safe level forward need once through. Every good over return.", "meta_info": { "generated_at": "2026-01-22T04:03:07.587322"...
{ "id": "0a9e4c0d-c2da-43", "patient": "Robert Simmons", "diagnosis": "student", "notes": "Interview particularly treatment hospital join attack race among. Style safe level forward need once through. Every good over return.", "meta_info": { "generated_at": "2026-01-22T04:03:07.587322", "version": "v2...
{ "format": "json", "rule_id": "J_STRING_TRUNCATE", "severity": "moderate", "seed_id": "", "parseable_full": true, "parseable_inner": false, "markdown_wrapped": false, "margin_score": 1, "margin_reasons": [ "value_changes" ], "violation_category": "moderate" }
#X_TAG_RENAME
Reformat the CSV below into XML: contract,parties,signed deploy integrated ROI,"[""Nelson-Hamilton"", ""Jackson-Burton""]",True Required constraints: - Return only XML without any Markdown fences or prose. - Root tag must be <contract>; keep allowed tags and attributes only. - Do not change tag names or add unknown e...
<?xml version="1.0" encoding="UTF-8"?> <contract> <contract>deploy integrated ROI</contract> <parties>Nelson-Hamilton</parties> <parties>Jackson-Burton</parties> <signed>True</signed> </contract>
<?xml version="1.0" encoding="UTF-8"?> <contract> <contract>deploy integrated ROI</contract> <parties>Nelson-Hamilton</parties> <parties>Jackson-Burton</parties> <signed_node>True</signed_node> </contract>
{ "format": "xml", "rule_id": "X_TAG_RENAME", "severity": "moderate", "seed_id": "", "parseable_full": true, "parseable_inner": false, "markdown_wrapped": false, "margin_score": 2, "margin_reasons": [ "missing_xml_tags" ], "violation_category": "severe" }
#X_TAG_RENAME
Reformat the TOML below into XML: txn = "84ee87a6-9fc0-406b-aaa9-04fb2e2835df" amount = 364.43 status = "completed" Required constraints: - Return only XML without any Markdown fences or prose. - Root tag must be <transaction_record>; keep allowed tags and attributes only. - Do not change tag names or add unknown elem...
<?xml version="1.0" encoding="UTF-8"?> <transaction_record> <txn>84ee87a6-9fc0-406b-aaa9-04fb2e2835df</txn> <amount>364.43</amount> <status>completed</status> </transaction_record>
<?xml version="1.0" encoding="UTF-8"?> <transaction_record> <txn>84ee87a6-9fc0-406b-aaa9-04fb2e2835df</txn> <amount_data>364.43</amount_data> <status>completed</status> </transaction_record>
{ "format": "xml", "rule_id": "X_TAG_RENAME", "severity": "moderate", "seed_id": "", "parseable_full": true, "parseable_inner": false, "markdown_wrapped": false, "margin_score": 2, "margin_reasons": [ "missing_xml_tags" ], "violation_category": "severe" }
#X_TAG_RENAME
Generate error log data in XML format. Required constraints: - Return only XML without any Markdown fences or prose. - Root tag must be <error_log>; keep allowed tags and attributes only. - Do not change tag names or add unknown elements.
<?xml version="1.0" encoding="UTF-8"?> <error_log> <timestamp>2026-01-22T04:03:07.375258</timestamp> <level>ERROR</level> <message>Which later bag public impact.</message> <meta_info> <generated_at>2026-01-22T04:03:07.375423</generated_at> <version>v1.0</version> <tags>truth</tags> <tags>finally</tags> <tags>whole</tag...
<?xml version="1.0" encoding="UTF-8"?> <error_log_node> <timestamp>2026-01-22T04:03:07.375258</timestamp> <level>ERROR</level> <message>Which later bag public impact.</message> <meta_info> <generated_at>2026-01-22T04:03:07.375423</generated_at> <version>v1.0</version> <tags>truth</tags> <tags>finally</tags> <tags>whole...
{ "format": "xml", "rule_id": "X_TAG_RENAME", "severity": "moderate", "seed_id": "", "parseable_full": true, "parseable_inner": false, "markdown_wrapped": false, "margin_score": 2, "margin_reasons": [ "missing_xml_tags" ], "violation_category": "severe" }
#T_KEY_DROP
Generate clinical note data in TOML format. Required constraints: - Return only TOML without any Markdown fences or prose. - Top-level keys must be exactly: id, patient, diagnosis, notes (no extra or missing keys).
id = "408dd72f-cc90-4b95-a360-eb4dcccd011c" patient = "Peter Black" diagnosis = "involve" notes = "Country drop senior business. Here but south tough return result.\nPage Mrs perhaps trouble value. Return capital newspaper hour sign."
patient = "Peter Black" diagnosis = "involve" notes = "Country drop senior business. Here but south tough return result.\nPage Mrs perhaps trouble value. Return capital newspaper hour sign."
{ "format": "toml", "rule_id": "T_KEY_DROP", "severity": "moderate", "seed_id": "", "parseable_full": true, "parseable_inner": false, "markdown_wrapped": false, "margin_score": 2, "margin_reasons": [ "missing_required_keys" ], "violation_category": "severe" }
#C_CODE_FENCE_WRAP
Convert the following JSON to XML: { "timestamp": "2026-01-22T04:03:05.358571", "level": "ERROR", "message": "Himself Republican world." } Required constraints: - Return only XML without any Markdown fences or prose. - Root tag must be <error_log>; keep allowed tags and attributes only. - Do not change tag names...
<?xml version="1.0" encoding="UTF-8"?> <error_log> <timestamp>2026-01-22T04:03:05.358571</timestamp> <level>ERROR</level> <message>Himself Republican world.</message> </error_log>
## Result ```xml <?xml version="1.0" encoding="UTF-8"?> <error_log> <timestamp>2026-01-22T04:03:05.358571</timestamp> <level>ERROR</level> <message>Himself Republican world.</message> </error_log> ```
{ "format": "xml", "rule_id": "C_CODE_FENCE_WRAP", "severity": "minor", "seed_id": "", "parseable_full": false, "parseable_inner": true, "markdown_wrapped": true, "margin_score": 1, "margin_reasons": [ "prohibited_markdown" ], "violation_category": "minor" }
#J_DUPLICATE_LIST_ITEM
Create a JSON representation of experiment result. Required constraints: - Return only JSON without any Markdown fences or prose. - Top-level keys must be exactly: experiment_id, title, metrics, config, meta_info, audit_log (no extra or missing keys). - Field meta_info.generated_at is required. - All list items must b...
{ "experiment_id": "EXP-385056", "title": "Analysis of cover", "metrics": { "accuracy": 0.9136, "loss": 0.0837, "epoch": 10 }, "config": { "batch_size": 32, "optimizer": "adam", "lr": 0.001 }, "meta_info": { "generated_at": "2026-01-22T04:03:05.211272", "version": "v4.0", ...
{ "experiment_id": "EXP-385056", "title": "Analysis of cover", "metrics": { "accuracy": 0.9136, "loss": 0.0837, "epoch": 10 }, "config": { "batch_size": 32, "optimizer": "adam", "lr": 0.001 }, "meta_info": { "generated_at": "2026-01-22T04:03:05.211272", "version": "v4.0", ...
{ "format": "json", "rule_id": "J_DUPLICATE_LIST_ITEM", "severity": "moderate", "seed_id": "", "parseable_full": true, "parseable_inner": false, "markdown_wrapped": false, "margin_score": 1, "margin_reasons": [ "value_changes" ], "violation_category": "moderate" }
#J_DUPLICATE_LIST_ITEM
Output structured data for prescription using JSON. Required constraints: - Return only JSON without any Markdown fences or prose. - Top-level keys must be exactly: rx, drug, dosage, meta_info, audit_log (no extra or missing keys). - Field meta_info.generated_at is required. - All list items must be unique; do not dup...
{ "rx": "cc11ba92-4294-42c2-b903-2a38469292dd", "drug": "television", "dosage": "10mg", "meta_info": { "generated_at": "2026-01-22T04:03:06.734304", "version": "v1.0", "tags": [ "lead", "factor", "quickly", "task", "piece" ] }, "audit_log": [ { "action...
{ "rx": "cc11ba92-4294-42c2-b903-2a38469292dd", "drug": "television", "dosage": "10mg", "meta_info": { "generated_at": "2026-01-22T04:03:06.734304", "version": "v1.0", "tags": [ "lead", "factor", "quickly", "task", "piece" ] }, "audit_log": [ { "action...
{ "format": "json", "rule_id": "J_DUPLICATE_LIST_ITEM", "severity": "moderate", "seed_id": "", "parseable_full": true, "parseable_inner": false, "markdown_wrapped": false, "margin_score": 1, "margin_reasons": [ "value_changes" ], "violation_category": "moderate" }
#C_CODE_FENCE_WRAP
Generate financial report data in YAML format. Required constraints: - Return only YAML without any Markdown fences or prose. - Top-level keys must be exactly: Approach, 1. Define the schema, 2. Select the format, 3. Generate realistic content, 4. Structure the data, 5. Serialize, Output, company, revenue, profit, met...
Approach: 1. Define the schema: The financial_report should contain a dictionary with 5 root fields (company, revenue, profit...) and a nesting depth of 3. 2. Select the format: YAML is chosen. Use indentation to represent nesting. lists are denoted by hyphens. 3. Generate realistic content: Use plausible names, dates,...
## Result ```yaml Approach: 1. Define the schema: The financial_report should contain a dictionary with 5 root fields (company, revenue, profit...) and a nesting depth of 3. 2. Select the format: YAML is chosen. Use indentation to represent nesting. lists are denoted by hyphens. 3. Generate realistic content: Use plaus...
{ "format": "yaml", "rule_id": "C_CODE_FENCE_WRAP", "severity": "minor", "seed_id": "", "parseable_full": false, "parseable_inner": true, "markdown_wrapped": true, "margin_score": 1, "margin_reasons": [ "prohibited_markdown" ], "violation_category": "minor" }
#T_KEY_DROP
Reformat the YAML below into TOML: timestamp: '2026-01-22T04:03:05.062367' level: ERROR message: Buy site face type middle voice. Required constraints: - Return only TOML without any Markdown fences or prose. - Top-level keys must be exactly: timestamp, level, message (no extra or missing keys).
timestamp = "2026-01-22T04:03:05.062367" level = "ERROR" message = "Buy site face type middle voice."
timestamp = "2026-01-22T04:03:05.062367" level = "ERROR"
{ "format": "toml", "rule_id": "T_KEY_DROP", "severity": "moderate", "seed_id": "", "parseable_full": true, "parseable_inner": false, "markdown_wrapped": false, "margin_score": 2, "margin_reasons": [ "missing_required_keys" ], "violation_category": "severe" }
#C_CODE_FENCE_WRAP
Output structured data for error_log using CSV. Required constraints: - Return only CSV without any Markdown fences or prose. - CSV header columns must be exactly this order: 1. Define the schema: The error_log should contain a dictionary with 3 root fields (timestamp, level, message) and a nesting depth of 1.. - Do...
1. Define the schema: The error_log should contain a dictionary with 3 root fields (timestamp, level, message) and a nesting depth of 1. 2. Select the format: CSV is chosen. Flatten the hierarchical structure. Use dot notation for nested keys in the header. 3. Generate realistic content: Use plausible names, dates, and...
## Result ```csv 1. Define the schema: The error_log should contain a dictionary with 3 root fields (timestamp, level, message) and a nesting depth of 1. 2. Select the format: CSV is chosen. Flatten the hierarchical structure. Use dot notation for nested keys in the header. 3. Generate realistic content: Use plausible ...
{ "format": "csv", "rule_id": "C_CODE_FENCE_WRAP", "severity": "minor", "seed_id": "", "parseable_full": true, "parseable_inner": true, "markdown_wrapped": true, "margin_score": 1, "margin_reasons": [ "prohibited_markdown" ], "violation_category": "minor" }
#T_KEY_DROP
Create a TOML representation of social media post. Required constraints: - Return only TOML without any Markdown fences or prose. - Top-level keys must be exactly: user, post, likes (no extra or missing keys).
user = "janetmartinez" post = "Large group turn small. Expert heavy despite detail political building. Tv right cell report management dream." likes = 358
post = "Large group turn small. Expert heavy despite detail political building. Tv right cell report management dream." likes = 358
{ "format": "toml", "rule_id": "T_KEY_DROP", "severity": "moderate", "seed_id": "", "parseable_full": true, "parseable_inner": false, "markdown_wrapped": false, "margin_score": 2, "margin_reasons": [ "missing_required_keys" ], "violation_category": "severe" }
#T_STRING_TRUNCATE
Output structured data for student_assessment using TOML. Required constraints: - Return only TOML without any Markdown fences or prose. - Top-level keys must be exactly: student, score, grade (no extra or missing keys).
student = "Leah Palmer" score = 69 grade = "C"
student = "Leah" score = 69 grade = "C"
{ "format": "toml", "rule_id": "T_STRING_TRUNCATE", "severity": "moderate", "seed_id": "", "parseable_full": true, "parseable_inner": false, "markdown_wrapped": false, "margin_score": 1, "margin_reasons": [ "value_changes" ], "violation_category": "moderate" }
#T_KEY_DROP
Create a TOML representation of contract. Required constraints: - Return only TOML without any Markdown fences or prose. - Top-level keys must be exactly: contract, parties, signed, audit_log, meta_info (no extra or missing keys). - Field meta_info.generated_at is required. - All list items must be unique; do not dupl...
contract = "utilize impactful ROI" parties = [ "Barton and Sons", "Cain-Smith",] signed = true [[audit_log]] action = "created" by = "christopher37" timestamp = "1973-12-18T00:59:16.652501" [[audit_log]] action = "created" by = "rebeccagilbert" timestamp = "1998-01-20T17:55:49.323027" [meta_info] generated_at = "2026...
contract = "utilize impactful ROI" parties = [ "Barton and Sons", "Cain-Smith",] signed = true [[audit_log]] action = "created" by = "christopher37" timestamp = "1973-12-18T00:59:16.652501" [[audit_log]] action = "created" by = "rebeccagilbert" timestamp = "1998-01-20T17:55:49.323027" [meta_info] generated_at = "2026...
{ "format": "toml", "rule_id": "T_KEY_DROP", "severity": "moderate", "seed_id": "", "parseable_full": true, "parseable_inner": false, "markdown_wrapped": false, "margin_score": 2, "margin_reasons": [ "missing_required_keys" ], "violation_category": "severe" }
#T_KEY_DROP
Convert the following XML to TOML: <?xml version="1.0" encoding="UTF-8"?> <customer_review> <user>katherine10</user> <rating>2</rating> <comment>Word hotel save on finish relate drop because.</comment> </customer_review> Required constraints: - Return only TOML without any Markdown fences or prose. - Top-level keys mu...
user = "katherine10" rating = 2 comment = "Word hotel save on finish relate drop because."
rating = 2 comment = "Word hotel save on finish relate drop because."
{ "format": "toml", "rule_id": "T_KEY_DROP", "severity": "moderate", "seed_id": "", "parseable_full": true, "parseable_inner": false, "markdown_wrapped": false, "margin_score": 2, "margin_reasons": [ "missing_required_keys" ], "violation_category": "severe" }
#X_TAG_RENAME
Generate api specification data in XML format. Required constraints: - Return only XML without any Markdown fences or prose. - Root tag must be <api_specification>; keep allowed tags and attributes only. - Do not change tag names or add unknown elements.
<?xml version="1.0" encoding="UTF-8"?> <api_specification> <endpoint>/api/v1/users</endpoint> <method>GET</method> <responses> <_200>OK</_200> </responses> <meta_info> <generated_at>2026-01-22T04:03:05.184771</generated_at> <version>v2.0</version> <tags>teach</tags> <tags>next</tags> <tags>degree</tags> <tags>debate</t...
<?xml version="1.0" encoding="UTF-8"?> <api_specification> <endpoint_node>/api/v1/users</endpoint_node> <method>GET</method> <responses> <_200>OK</_200> </responses> <meta_info> <generated_at>2026-01-22T04:03:05.184771</generated_at> <version>v2.0</version> <tags>teach</tags> <tags>next</tags> <tags>degree</tags> <tags...
{ "format": "xml", "rule_id": "X_TAG_RENAME", "severity": "moderate", "seed_id": "", "parseable_full": true, "parseable_inner": false, "markdown_wrapped": false, "margin_score": 2, "margin_reasons": [ "missing_xml_tags" ], "violation_category": "severe" }
#J_VALUE_PERTURB
Convert the following CSV to JSON: id,test,value,meta_info.generated_at,meta_info.version,meta_info.tags,audit_log c540bd9d-4cd8-44b3-8173-3113630fa509,tonight,82.54530668576918,2026-01-22T04:03:07.041771,v3.0,"[""catch"", ""animal"", ""federal"", ""gun"", ""task""]","[{""action"": ""created"", ""by"": ""ahaas"", ""ti...
{ "id": "c540bd9d-4cd8-44b3-8173-3113630fa509", "test": "tonight", "value": 82.54530668576918, "meta_info": { "generated_at": "2026-01-22T04:03:07.041771", "version": "v3.0", "tags": [ "catch", "animal", "federal", "gun", "task" ] }, "audit_log": [ { "...
{ "id": "c540bd9d-4cd8-44b3-8173-3113630fa509", "test": "tonight", "value": 66.04, "meta_info": { "generated_at": "2026-01-22T04:03:07.041771", "version": "v3.0", "tags": [ "catch", "animal", "federal", "gun", "task" ] }, "audit_log": [ { "action": "cr...
{ "format": "json", "rule_id": "J_VALUE_PERTURB", "severity": "minor", "seed_id": "", "parseable_full": true, "parseable_inner": false, "markdown_wrapped": false, "margin_score": 1, "margin_reasons": [ "value_changes" ], "violation_category": "moderate" }
#X_TAG_RENAME
Output structured data for transaction_record using XML. Required constraints: - Return only XML without any Markdown fences or prose. - Root tag must be <transaction_record>; keep allowed tags and attributes only. - Do not change tag names or add unknown elements.
<?xml version="1.0" encoding="UTF-8"?> <transaction_record> <txn>3ed32e60-9522-4eaf-9732-08552512ba97</txn> <amount>246.17</amount> <status>completed</status> </transaction_record>
<?xml version="1.0" encoding="UTF-8"?> <transaction_record> <txn>3ed32e60-9522-4eaf-9732-08552512ba97</txn> <amount_data>246.17</amount_data> <status>completed</status> </transaction_record>
{ "format": "xml", "rule_id": "X_TAG_RENAME", "severity": "moderate", "seed_id": "", "parseable_full": true, "parseable_inner": false, "markdown_wrapped": false, "margin_score": 2, "margin_reasons": [ "missing_xml_tags" ], "violation_category": "severe" }
#Y_LIST_ITEM_DROP
Transform this data from XML to YAML: <?xml version="1.0" encoding="UTF-8"?> <product_listing> <product>speech</product> <price>22.23</price> <stock>5</stock> <meta_info> <generated_at>2026-01-22T04:03:07.429400</generated_at> <version>v4.0</version> <tags>attack</tags> <tags>speech</tags> <tags>recently</tags> <tags>a...
Approach: 1. Analyze the input XML structure: It represents a dictionary with 5 root fields (product, price, stock...) and a nesting depth of 3. 2. Identify the target format requirements: YAML requires strict syntax. Use indentation to represent nesting. lists are denoted by hyphens. 3. Map the fields: The root keys w...
1. Analyze the input XML structure: It represents a dictionary with 5 root fields (product, price, stock...) and a nesting depth of 3. 2. Identify the target format requirements: YAML requires strict syntax. Use indentation to represent nesting. lists are denoted by hyphens. 3. Map the fields: The root keys will ma...
{ "format": "yaml", "rule_id": "Y_LIST_ITEM_DROP", "severity": "minor", "seed_id": "", "parseable_full": true, "parseable_inner": false, "markdown_wrapped": false, "margin_score": 1, "margin_reasons": [ "value_changes" ], "violation_category": "moderate" }
#T_KEY_DROP
Transform this data from JSON to TOML: { "id": "7301a4df-9a3e-4620-b997-fd62584c98fb", "test": "hold", "value": 45.46268797137946 } Required constraints: - Return only TOML without any Markdown fences or prose. - Top-level keys must be exactly: id, test, value (no extra or missing keys).
id = "7301a4df-9a3e-4620-b997-fd62584c98fb" test = "hold" value = 45.46268797137946
test = "hold" value = 45.46268797137946
{ "format": "toml", "rule_id": "T_KEY_DROP", "severity": "moderate", "seed_id": "", "parseable_full": true, "parseable_inner": false, "markdown_wrapped": false, "margin_score": 2, "margin_reasons": [ "missing_required_keys" ], "violation_category": "severe" }
#C_CODE_FENCE_WRAP
Transform this data from XML to YAML: <?xml version="1.0" encoding="UTF-8"?> <lab_result> <id>5901af85-0032-4d6d-9e78-60a5ec3114bb</id> <test>action</test> <value>27.646140856629607</value> <meta_info> <generated_at>2026-01-22T04:03:06.179957</generated_at> <version>v1.0</version> <tags>attorney</tags> <tags>yard</tags...
Approach: 1. Analyze the input XML structure: It represents a dictionary with 5 root fields (id, test, value...) and a nesting depth of 3. 2. Identify the target format requirements: YAML requires strict syntax. Use indentation to represent nesting. lists are denoted by hyphens. 3. Map the fields: The root keys will ma...
## Result ```yaml Approach: 1. Analyze the input XML structure: It represents a dictionary with 5 root fields (id, test, value...) and a nesting depth of 3. 2. Identify the target format requirements: YAML requires strict syntax. Use indentation to represent nesting. lists are denoted by hyphens. 3. Map the fields: The...
{ "format": "yaml", "rule_id": "C_CODE_FENCE_WRAP", "severity": "minor", "seed_id": "", "parseable_full": false, "parseable_inner": true, "markdown_wrapped": true, "margin_score": 1, "margin_reasons": [ "prohibited_markdown" ], "violation_category": "minor" }
#C_CODE_FENCE_WRAP
Convert the following YAML to XML: user: james92 post: Two final modern open image room. Arrive live about hear front all. Available effort alone. likes: 864 Required constraints: - Return only XML without any Markdown fences or prose. - Root tag must be <social_media_post>; keep allowed tags and attributes only. - ...
<?xml version="1.0" encoding="UTF-8"?> <social_media_post> <user>james92</user> <post>Two final modern open image room. Arrive live about hear front all. Available effort alone.</post> <likes>864</likes> </social_media_post>
## Result ```xml <?xml version="1.0" encoding="UTF-8"?> <social_media_post> <user>james92</user> <post>Two final modern open image room. Arrive live about hear front all. Available effort alone.</post> <likes>864</likes> </social_media_post> ```
{ "format": "xml", "rule_id": "C_CODE_FENCE_WRAP", "severity": "minor", "seed_id": "", "parseable_full": false, "parseable_inner": true, "markdown_wrapped": true, "margin_score": 1, "margin_reasons": [ "prohibited_markdown" ], "violation_category": "minor" }
#X_TAG_RENAME
Reformat the TOML below into XML: timestamp = "2026-01-22T04:03:07.576869" level = "ERROR" message = "Total enjoy article century charge." [[audit_log]] action = "created" by = "ronaldberry" timestamp = "1972-10-05T06:28:45.163277" [[audit_log]] action = "created" by = "erikacunningham" timestamp = "2016-03-31T08:39:2...
<?xml version="1.0" encoding="UTF-8"?> <error_log> <timestamp>2026-01-22T04:03:07.576869</timestamp> <level>ERROR</level> <message>Total enjoy article century charge.</message> <meta_info> <generated_at>2026-01-22T04:03:07.576980</generated_at> <version>v1.0</version> <tags>job</tags> <tags>former</tags> <tags>will</ta...
<?xml version="1.0" encoding="UTF-8"?> <error_log_node> <timestamp>2026-01-22T04:03:07.576869</timestamp> <level>ERROR</level> <message>Total enjoy article century charge.</message> <meta_info> <generated_at>2026-01-22T04:03:07.576980</generated_at> <version>v1.0</version> <tags>job</tags> <tags>former</tags> <tags>wil...
{ "format": "xml", "rule_id": "X_TAG_RENAME", "severity": "moderate", "seed_id": "", "parseable_full": true, "parseable_inner": false, "markdown_wrapped": false, "margin_score": 2, "margin_reasons": [ "missing_xml_tags" ], "violation_category": "severe" }
#C_CODE_FENCE_WRAP
Transform this data from TOML to YAML: txn = "6e1f092f-d57b-4a36-842d-93e63cba9405" amount = 425.58 status = "completed" [[audit_log]] action = "created" by = "youngamanda" timestamp = "2016-04-18T16:32:04.400302" [[audit_log]] action = "created" by = "amckee" timestamp = "1986-08-27T00:14:30.888935" [meta_info] gene...
Approach: 1. Analyze the input TOML structure: It represents a dictionary with 5 root fields (txn, amount, status...) and a nesting depth of 3. 2. Identify the target format requirements: YAML requires strict syntax. Use indentation to represent nesting. lists are denoted by hyphens. 3. Map the fields: The root keys wi...
## Result ```yaml Approach: 1. Analyze the input TOML structure: It represents a dictionary with 5 root fields (txn, amount, status...) and a nesting depth of 3. 2. Identify the target format requirements: YAML requires strict syntax. Use indentation to represent nesting. lists are denoted by hyphens. 3. Map the fields...
{ "format": "yaml", "rule_id": "C_CODE_FENCE_WRAP", "severity": "minor", "seed_id": "", "parseable_full": false, "parseable_inner": true, "markdown_wrapped": true, "margin_score": 1, "margin_reasons": [ "prohibited_markdown" ], "violation_category": "minor" }
#T_KEY_DROP
Convert the following JSON to TOML: { "id": "3f4ecc21-764e-4f31-bee1-0ed33cc0b1c3", "patient": "Marissa Diaz", "diagnosis": "everything", "notes": "Such despite education game happy. Sure near present bad blood film. President artist maybe the poor majority." } Required constraints: - Return only TOML without ...
id = "3f4ecc21-764e-4f31-bee1-0ed33cc0b1c3" patient = "Marissa Diaz" diagnosis = "everything" notes = "Such despite education game happy. Sure near present bad blood film. President artist maybe the poor majority."
patient = "Marissa Diaz" diagnosis = "everything" notes = "Such despite education game happy. Sure near present bad blood film. President artist maybe the poor majority."
{ "format": "toml", "rule_id": "T_KEY_DROP", "severity": "moderate", "seed_id": "", "parseable_full": true, "parseable_inner": false, "markdown_wrapped": false, "margin_score": 2, "margin_reasons": [ "missing_required_keys" ], "violation_category": "severe" }
#X_TAG_RENAME
Create a XML representation of clinical note. Required constraints: - Return only XML without any Markdown fences or prose. - Root tag must be <clinical_note>; keep allowed tags and attributes only. - Do not change tag names or add unknown elements.
<?xml version="1.0" encoding="UTF-8"?> <clinical_note> <id>408dd72f-cc90-4b95-a360-eb4dcccd011c</id> <patient>Peter Black</patient> <diagnosis>involve</diagnosis> <notes>Country drop senior business. Here but south tough return result. Page Mrs perhaps trouble value. Return capital newspaper hour sign.</notes> </clinic...
<?xml version="1.0" encoding="UTF-8"?> <clinical_note> <id>408dd72f-cc90-4b95-a360-eb4dcccd011c</id> <patient>Peter Black</patient> <diagnosis_node>involve</diagnosis_node> <notes>Country drop senior business. Here but south tough return result. Page Mrs perhaps trouble value. Return capital newspaper hour sign.</notes...
{ "format": "xml", "rule_id": "X_TAG_RENAME", "severity": "moderate", "seed_id": "", "parseable_full": true, "parseable_inner": false, "markdown_wrapped": false, "margin_score": 2, "margin_reasons": [ "missing_xml_tags" ], "violation_category": "severe" }
#S_STRING_MODIFY
Output structured data for news_article using CSV. Required constraints: - Return only CSV without any Markdown fences or prose. - CSV header columns must be exactly this order: 1. Define the schema: The news_article should contain a dictionary with 3 root fields (headline, body, author) and a nesting depth of 1.. -...
1. Define the schema: The news_article should contain a dictionary with 3 root fields (headline, body, author) and a nesting depth of 1. 2. Select the format: CSV is chosen. Flatten the hierarchical structure. Use dot notation for nested keys in the header. 3. Generate realistic content: Use plausible names, dates, and...
1. Define the schema: The news_article should contain a dictionary with 3 root fields (headline, body, author) and a nesting depth of 1. 2. Select the format: CSV is chosen. Flatten the hierarchical structure. Use dot notation for nested keys in the header. 3. Generate realistic content: Use plausible names, dates, and...
{ "format": "csv", "rule_id": "S_STRING_MODIFY", "severity": "minor", "seed_id": "", "parseable_full": true, "parseable_inner": false, "markdown_wrapped": false, "margin_score": 1, "margin_reasons": [ "csv_value_diff" ], "violation_category": "minor" }
#S_STRING_MODIFY
Reformat the JSON below into CSV: { "id": "4a54db7d-1dec-49c6-b3d8-1608e75b010b", "patient": "Julie Young", "diagnosis": "training", "notes": "Wind situation expert a. Food financial mean use decide including technology. Western increase significant whatever help chair kid." } Required constraints: - Return on...
1. Analyze the input JSON structure: It represents a dictionary with 4 root fields (id, patient, diagnosis...) and a nesting depth of 1. 2. Identify the target format requirements: CSV requires strict syntax. Flatten the hierarchical structure. Use dot notation for nested keys in the header. 3. Map the fields: The root...
1. Analyze the input JSON structure: It represents a dictionary with 4 root fields (id, patient, diagnosis...) and a nesting depth of 1. 2. Identify the target format requirements: CSV requires strict syntax. Flatten the hierarchical structure. Use dot notation for nested keys in the header. 3. Map the fields: The root...
{ "format": "csv", "rule_id": "S_STRING_MODIFY", "severity": "minor", "seed_id": "", "parseable_full": true, "parseable_inner": false, "markdown_wrapped": false, "margin_score": 1, "margin_reasons": [ "csv_value_diff" ], "violation_category": "minor" }
#J_DUPLICATE_LIST_ITEM
Reformat the CSV below into JSON: id,patient,diagnosis,notes,meta_info.generated_at,meta_info.version,meta_info.tags,audit_log e5b99474-5346-4b5d-ad9f-f47ebf3796c3,Heather Kent,near,"Real character over take defense. Speech young similar. Here where about discover finish apply. Program employee or skill beautiful duri...
{ "id": "e5b99474-5346-4b5d-ad9f-f47ebf3796c3", "patient": "Heather Kent", "diagnosis": "near", "notes": "Real character over take defense. Speech young similar. Here where about discover finish apply.\nProgram employee or skill beautiful during. Church woman board so skin reach.", "meta_info": { "generat...
{ "id": "e5b99474-5346-4b5d-ad9f-f47ebf3796c3", "patient": "Heather Kent", "diagnosis": "near", "notes": "Real character over take defense. Speech young similar. Here where about discover finish apply.\nProgram employee or skill beautiful during. Church woman board so skin reach.", "meta_info": { "generat...
{ "format": "json", "rule_id": "J_DUPLICATE_LIST_ITEM", "severity": "moderate", "seed_id": "", "parseable_full": true, "parseable_inner": false, "markdown_wrapped": false, "margin_score": 1, "margin_reasons": [ "value_changes" ], "violation_category": "moderate" }
#C_CODE_FENCE_WRAP
Convert the following YAML to CSV: txn: f775a27f-d20f-4777-a898-02a324473f67 amount: 180.81 status: completed Required constraints: - Return only CSV without any Markdown fences or prose. - CSV header columns must be exactly this order: 1. Analyze the input YAML structure: It represents a dictionary with 3 root fields...
1. Analyze the input YAML structure: It represents a dictionary with 3 root fields (txn, amount, status) and a nesting depth of 1. 2. Identify the target format requirements: CSV requires strict syntax. Flatten the hierarchical structure. Use dot notation for nested keys in the header. 3. Map the fields: The root keys ...
## Result ```csv 1. Analyze the input YAML structure: It represents a dictionary with 3 root fields (txn, amount, status) and a nesting depth of 1. 2. Identify the target format requirements: CSV requires strict syntax. Flatten the hierarchical structure. Use dot notation for nested keys in the header. 3. Map the field...
{ "format": "csv", "rule_id": "C_CODE_FENCE_WRAP", "severity": "minor", "seed_id": "", "parseable_full": true, "parseable_inner": true, "markdown_wrapped": true, "margin_score": 1, "margin_reasons": [ "prohibited_markdown" ], "violation_category": "minor" }
#J_DUPLICATE_LIST_ITEM
Output structured data for sales_analytics using JSON. Required constraints: - Return only JSON without any Markdown fences or prose. - Top-level keys must be exactly: period, sales, region, meta_info, audit_log (no extra or missing keys). - Field meta_info.generated_at is required. - All list items must be unique; do...
{ "period": "Q1", "sales": 936, "region": "Hayeston", "meta_info": { "generated_at": "2026-01-22T04:03:06.672988", "version": "v1.0", "tags": [ "hope", "old", "executive", "maintain", "result" ] }, "audit_log": [ { "action": "created", "by": "gon...
{ "period": "Q1", "sales": 936, "region": "Hayeston", "meta_info": { "generated_at": "2026-01-22T04:03:06.672988", "version": "v1.0", "tags": [ "hope", "old", "executive", "maintain", "result" ] }, "audit_log": [ { "action": "created", "by": "gon...
{ "format": "json", "rule_id": "J_DUPLICATE_LIST_ITEM", "severity": "moderate", "seed_id": "", "parseable_full": true, "parseable_inner": false, "markdown_wrapped": false, "margin_score": 1, "margin_reasons": [ "value_changes" ], "violation_category": "moderate" }
#J_LIST_EXTRA_ITEM
Generate product listing data in JSON format. Required constraints: - Return only JSON without any Markdown fences or prose. - Top-level keys must be exactly: product, price, stock, meta_info, audit_log (no extra or missing keys). - Field meta_info.generated_at is required. - All list items must be unique; do not dupl...
{ "product": "natural", "price": 17.08, "stock": 24, "meta_info": { "generated_at": "2026-01-22T04:03:06.521445", "version": "v4.0", "tags": [ "sense", "effect", "stock", "suggest", "order" ] }, "audit_log": [ { "action": "created", "by": "dpalme...
{ "product": "natural", "price": 17.08, "stock": 24, "meta_info": { "generated_at": "2026-01-22T04:03:06.521445", "version": "v4.0", "tags": [ "sense", "effect", "stock", "suggest", "order" ] }, "audit_log": [ { "action": "created", "by": "dpalme...
{ "format": "json", "rule_id": "J_LIST_EXTRA_ITEM", "severity": "moderate", "seed_id": "", "parseable_full": true, "parseable_inner": false, "markdown_wrapped": false, "margin_score": 1, "margin_reasons": [ "value_changes" ], "violation_category": "moderate" }
#T_KEY_DROP
Create a TOML representation of product listing. Required constraints: - Return only TOML without any Markdown fences or prose. - Top-level keys must be exactly: product, price, stock (no extra or missing keys).
product = "seek" price = 22.82 stock = 16
product = "seek" price = 22.82
{ "format": "toml", "rule_id": "T_KEY_DROP", "severity": "moderate", "seed_id": "", "parseable_full": true, "parseable_inner": false, "markdown_wrapped": false, "margin_score": 2, "margin_reasons": [ "missing_required_keys" ], "violation_category": "severe" }
#T_KEY_DROP
Create a TOML representation of product listing. Required constraints: - Return only TOML without any Markdown fences or prose. - Top-level keys must be exactly: product, price, stock (no extra or missing keys).
product = "fill" price = 38.15 stock = 30
product = "fill" price = 38.15
{ "format": "toml", "rule_id": "T_KEY_DROP", "severity": "moderate", "seed_id": "", "parseable_full": true, "parseable_inner": false, "markdown_wrapped": false, "margin_score": 2, "margin_reasons": [ "missing_required_keys" ], "violation_category": "severe" }
#C_CODE_FENCE_WRAP
Convert the following XML to CSV: <?xml version="1.0" encoding="UTF-8"?> <transaction_record> <txn>efa82ca6-ea4c-417c-80e7-b01530ea0616</txn> <amount>172.91</amount> <status>completed</status> <meta_info> <generated_at>2026-01-22T04:03:07.744438</generated_at> <version>v2.0</version> <tags>help</tags> <tags>forward</ta...
1. Analyze the input XML structure: It represents a dictionary with 5 root fields (txn, amount, status...) and a nesting depth of 3. 2. Identify the target format requirements: CSV requires strict syntax. Flatten the hierarchical structure. Use dot notation for nested keys in the header. 3. Map the fields: The root key...
## Result ```csv 1. Analyze the input XML structure: It represents a dictionary with 5 root fields (txn, amount, status...) and a nesting depth of 3. 2. Identify the target format requirements: CSV requires strict syntax. Flatten the hierarchical structure. Use dot notation for nested keys in the header. 3. Map the fie...
{ "format": "csv", "rule_id": "C_CODE_FENCE_WRAP", "severity": "minor", "seed_id": "", "parseable_full": true, "parseable_inner": true, "markdown_wrapped": true, "margin_score": 1, "margin_reasons": [ "prohibited_markdown" ], "violation_category": "minor" }
#J_KEY_RENAME_WRONG
Convert the following CSV to JSON: endpoint,method,responses.200,meta_info.generated_at,meta_info.version,meta_info.tags,audit_log /api/v1/users,GET,OK,2026-01-22T04:03:05.913823,v4.0,"[""drive"", ""game"", ""industry"", ""line"", ""attention""]","[{""action"": ""created"", ""by"": ""zcooper"", ""timestamp"": ""2005-0...
{ "endpoint": "/api/v1/users", "method": "GET", "responses": { "200": "OK" }, "meta_info": { "generated_at": "2026-01-22T04:03:05.913823", "version": "v4.0", "tags": [ "drive", "game", "industry", "line", "attention" ] }, "audit_log": [ { "action...
{ "endpoint": "/api/v1/users", "method": "GET", "meta_info": { "generated_at": "2026-01-22T04:03:05.913823", "version": "v4.0", "tags": [ "drive", "game", "industry", "line", "attention" ] }, "audit_log": [ { "action": "created", "by": "zcooper", ...
{ "format": "json", "rule_id": "J_KEY_RENAME_WRONG", "severity": "moderate", "seed_id": "", "parseable_full": true, "parseable_inner": false, "markdown_wrapped": false, "margin_score": 3, "margin_reasons": [ "missing_required_keys", "added_unknown_keys" ], "violation_category": "severe" }
#X_TAG_RENAME
Output structured data for lab_result using XML. Required constraints: - Return only XML without any Markdown fences or prose. - Root tag must be <lab_result>; keep allowed tags and attributes only. - Do not change tag names or add unknown elements.
<?xml version="1.0" encoding="UTF-8"?> <lab_result> <id>7716ced4-1e7f-4311-9bcb-41101453f974</id> <test>painting</test> <value>39.41475886335455</value> </lab_result>
<?xml version="1.0" encoding="UTF-8"?> <lab_result> <id>7716ced4-1e7f-4311-9bcb-41101453f974</id> <test_node>painting</test_node> <value>39.41475886335455</value> </lab_result>
{ "format": "xml", "rule_id": "X_TAG_RENAME", "severity": "moderate", "seed_id": "", "parseable_full": true, "parseable_inner": false, "markdown_wrapped": false, "margin_score": 2, "margin_reasons": [ "missing_xml_tags" ], "violation_category": "severe" }
#X_TAG_RENAME
Create a XML representation of prescription. Required constraints: - Return only XML without any Markdown fences or prose. - Root tag must be <prescription>; keep allowed tags and attributes only. - Do not change tag names or add unknown elements.
<?xml version="1.0" encoding="UTF-8"?> <prescription> <rx>e083656c-c363-43a0-ad8a-5a0367d00f19</rx> <drug>because</drug> <dosage>10mg</dosage> </prescription>
<?xml version="1.0" encoding="UTF-8"?> <prescription_node> <rx>e083656c-c363-43a0-ad8a-5a0367d00f19</rx> <drug>because</drug> <dosage>10mg</dosage> </prescription_node>
{ "format": "xml", "rule_id": "X_TAG_RENAME", "severity": "moderate", "seed_id": "", "parseable_full": true, "parseable_inner": false, "markdown_wrapped": false, "margin_score": 2, "margin_reasons": [ "missing_xml_tags" ], "violation_category": "severe" }
#X_TAG_RENAME
Reformat the TOML below into XML: contract = "synergize viral portals" parties = [ "Becker, Barrett and Neal", "Anderson LLC",] signed = false Required constraints: - Return only XML without any Markdown fences or prose. - Root tag must be <contract>; keep allowed tags and attributes only. - Do not change tag names or...
<?xml version="1.0" encoding="UTF-8"?> <contract> <contract>synergize viral portals</contract> <parties>Becker, Barrett and Neal</parties> <parties>Anderson LLC</parties> <signed>False</signed> </contract>
<?xml version="1.0" encoding="UTF-8"?> <contract> <contract>synergize viral portals</contract> <parties>Becker, Barrett and Neal</parties> <parties>Anderson LLC</parties> <signed_node>False</signed_node> </contract>
{ "format": "xml", "rule_id": "X_TAG_RENAME", "severity": "moderate", "seed_id": "", "parseable_full": true, "parseable_inner": false, "markdown_wrapped": false, "margin_score": 2, "margin_reasons": [ "missing_xml_tags" ], "violation_category": "severe" }
#T_KEY_DROP
Output structured data for sales_analytics using TOML. Required constraints: - Return only TOML without any Markdown fences or prose. - Top-level keys must be exactly: period, sales, region, audit_log, meta_info (no extra or missing keys). - Field meta_info.generated_at is required. - All list items must be unique; do...
period = "Q1" sales = 3795 region = "East Travishaven" [[audit_log]] action = "created" by = "blackstephen" timestamp = "1992-08-29T14:58:22.613839" [meta_info] generated_at = "2026-01-22T04:03:05.381605" version = "v2.0" tags = [ "eight", "her", "mean", "nation", "theory",]
period = "Q1" sales = 3795 region = "East Travishaven" [[audit_log]] by = "blackstephen" timestamp = "1992-08-29T14:58:22.613839" [meta_info] generated_at = "2026-01-22T04:03:05.381605" version = "v2.0" tags = [ "eight", "her", "mean", "nation", "theory",]
{ "format": "toml", "rule_id": "T_KEY_DROP", "severity": "moderate", "seed_id": "", "parseable_full": true, "parseable_inner": false, "markdown_wrapped": false, "margin_score": 2, "margin_reasons": [ "missing_required_keys" ], "violation_category": "severe" }
#X_TAG_RENAME
Transform this data from CSV to XML: user,post,likes,meta_info.generated_at,meta_info.version,meta_info.tags,audit_log villanuevamichael,Development popular town left whole news fight. Baby it I into both. Choice town whether that top sport son.,804,2026-01-22T04:03:06.892984,v2.0,"[""same"", ""rise"", ""center"", ""s...
<?xml version="1.0" encoding="UTF-8"?> <social_media_post> <user>villanuevamichael</user> <post>Development popular town left whole news fight. Baby it I into both. Choice town whether that top sport son.</post> <likes>804</likes> <meta_info> <generated_at>2026-01-22T04:03:06.892984</generated_at> <version>v2.0</versio...
<?xml version="1.0" encoding="UTF-8"?> <social_media_post_node> <user>villanuevamichael</user> <post>Development popular town left whole news fight. Baby it I into both. Choice town whether that top sport son.</post> <likes>804</likes> <meta_info> <generated_at>2026-01-22T04:03:06.892984</generated_at> <version>v2.0</v...
{ "format": "xml", "rule_id": "X_TAG_RENAME", "severity": "moderate", "seed_id": "", "parseable_full": true, "parseable_inner": false, "markdown_wrapped": false, "margin_score": 2, "margin_reasons": [ "missing_xml_tags" ], "violation_category": "severe" }
#X_TAG_RENAME
Generate experiment result data in XML format. Required constraints: - Return only XML without any Markdown fences or prose. - Root tag must be <experiment_result>; keep allowed tags and attributes only. - Do not change tag names or add unknown elements.
<?xml version="1.0" encoding="UTF-8"?> <experiment_result> <experiment_id>EXP-467634</experiment_id> <title>Analysis of blue</title> <metrics> <accuracy>0.8001</accuracy> <loss>0.4591</loss> <epoch>32</epoch> </metrics> <config> <batch_size>64</batch_size> <optimizer>sgd</optimizer> <lr>0.001</lr> </config> </experimen...
<?xml version="1.0" encoding="UTF-8"?> <experiment_result> <experiment_id>EXP-467634</experiment_id> <title_data>Analysis of blue</title_data> <metrics> <accuracy>0.8001</accuracy> <loss>0.4591</loss> <epoch>32</epoch> </metrics> <config> <batch_size>64</batch_size> <optimizer>sgd</optimizer> <lr>0.001</lr> </config> <...
{ "format": "xml", "rule_id": "X_TAG_RENAME", "severity": "moderate", "seed_id": "", "parseable_full": true, "parseable_inner": false, "markdown_wrapped": false, "margin_score": 2, "margin_reasons": [ "missing_xml_tags" ], "violation_category": "severe" }
#T_KEY_DROP
Output structured data for financial_report using TOML. Required constraints: - Return only TOML without any Markdown fences or prose. - Top-level keys must be exactly: company, revenue, profit (no extra or missing keys).
company = "Miller, Glenn and Becker" revenue = 266473 profit = 22717
revenue = 266473 profit = 22717
{ "format": "toml", "rule_id": "T_KEY_DROP", "severity": "moderate", "seed_id": "", "parseable_full": true, "parseable_inner": false, "markdown_wrapped": false, "margin_score": 2, "margin_reasons": [ "missing_required_keys" ], "violation_category": "severe" }
#S_STRING_MODIFY
Generate financial report data in CSV format. Required constraints: - Return only CSV without any Markdown fences or prose. - CSV header columns must be exactly this order: 1. Define the schema: The financial_report should contain a dictionary with 3 root fields (company, revenue, profit) and a nesting depth of 1.. ...
1. Define the schema: The financial_report should contain a dictionary with 3 root fields (company, revenue, profit) and a nesting depth of 1. 2. Select the format: CSV is chosen. Flatten the hierarchical structure. Use dot notation for nested keys in the header. 3. Generate realistic content: Use plausible names, date...
1. Define the schema: The financial_report should contain a dictionary with 3 root fields (company, revenue, profit) and a nesting depth of 1. 2. Select the format: CSV is chosen. Flatten the hierarchical structure. Use dot notation for nested keys in the header. 3. Generate realistic content: Use plausible names, date...
{ "format": "csv", "rule_id": "S_STRING_MODIFY", "severity": "minor", "seed_id": "", "parseable_full": true, "parseable_inner": false, "markdown_wrapped": false, "margin_score": 1, "margin_reasons": [ "csv_value_diff" ], "violation_category": "minor" }
#X_TAG_RENAME
Output structured data for news_article using XML. Required constraints: - Return only XML without any Markdown fences or prose. - Root tag must be <news_article>; keep allowed tags and attributes only. - Do not change tag names or add unknown elements.
<?xml version="1.0" encoding="UTF-8"?> <news_article> <headline>Yet poor evidence environment general.</headline> <body>Poor worker recently factor form sure argue. Long start according sense story. Share Congress sometimes tough green when image.</body> <author>James Morales</author> </news_article>
<?xml version="1.0" encoding="UTF-8"?> <news_article_node> <headline>Yet poor evidence environment general.</headline> <body>Poor worker recently factor form sure argue. Long start according sense story. Share Congress sometimes tough green when image.</body> <author>James Morales</author> </news_article_node>
{ "format": "xml", "rule_id": "X_TAG_RENAME", "severity": "moderate", "seed_id": "", "parseable_full": true, "parseable_inner": false, "markdown_wrapped": false, "margin_score": 2, "margin_reasons": [ "missing_xml_tags" ], "violation_category": "severe" }
#X_TAG_RENAME
Create a XML representation of prescription. Required constraints: - Return only XML without any Markdown fences or prose. - Root tag must be <prescription>; keep allowed tags and attributes only. - Do not change tag names or add unknown elements.
<?xml version="1.0" encoding="UTF-8"?> <prescription> <rx>d6f5cd79-d10b-4964-a659-9baa73a728dd</rx> <drug>ever</drug> <dosage>10mg</dosage> </prescription>
<?xml version="1.0" encoding="UTF-8"?> <prescription_node> <rx>d6f5cd79-d10b-4964-a659-9baa73a728dd</rx> <drug>ever</drug> <dosage>10mg</dosage> </prescription_node>
{ "format": "xml", "rule_id": "X_TAG_RENAME", "severity": "moderate", "seed_id": "", "parseable_full": true, "parseable_inner": false, "markdown_wrapped": false, "margin_score": 2, "margin_reasons": [ "missing_xml_tags" ], "violation_category": "severe" }
#J_STRING_TRUNCATE
Convert the following TOML to JSON: txn = "578a4db9-6f3f-4115-9252-9169becd3f07" amount = 444.96 status = "completed" Required constraints: - Return only JSON without any Markdown fences or prose. - Top-level keys must be exactly: txn, amount, status (no extra or missing keys).
{ "txn": "578a4db9-6f3f-4115-9252-9169becd3f07", "amount": 444.96, "status": "completed" }
{ "txn": "578a4db9-6f3f-41", "amount": 444.96, "status": "completed" }
{ "format": "json", "rule_id": "J_STRING_TRUNCATE", "severity": "moderate", "seed_id": "", "parseable_full": true, "parseable_inner": false, "markdown_wrapped": false, "margin_score": 1, "margin_reasons": [ "value_changes" ], "violation_category": "moderate" }
#T_STRING_TRUNCATE
Generate lab result data in TOML format. Required constraints: - Return only TOML without any Markdown fences or prose. - Top-level keys must be exactly: id, test, value (no extra or missing keys).
id = "0e5c588b-7c37-481c-8f39-8514d2cb626f" test = "me" value = 73.86088081333102
id = "0e5c588b-7c37-481c-8f3" test = "me" value = 73.86088081333102
{ "format": "toml", "rule_id": "T_STRING_TRUNCATE", "severity": "moderate", "seed_id": "", "parseable_full": true, "parseable_inner": false, "markdown_wrapped": false, "margin_score": 1, "margin_reasons": [ "value_changes" ], "violation_category": "moderate" }
#T_KEY_DROP
Generate clinical note data in TOML format. Required constraints: - Return only TOML without any Markdown fences or prose. - Top-level keys must be exactly: id, patient, diagnosis, notes (no extra or missing keys).
id = "709596e8-9c88-4c54-99d6-361f90141992" patient = "Kristi Smith" diagnosis = "pressure" notes = "Media situation stay team professional sea. According new seat produce table. Mind north give enjoy law. Field rich people prepare."
patient = "Kristi Smith" diagnosis = "pressure" notes = "Media situation stay team professional sea. According new seat produce table. Mind north give enjoy law. Field rich people prepare."
{ "format": "toml", "rule_id": "T_KEY_DROP", "severity": "moderate", "seed_id": "", "parseable_full": true, "parseable_inner": false, "markdown_wrapped": false, "margin_score": 2, "margin_reasons": [ "missing_required_keys" ], "violation_category": "severe" }
#T_KEY_DROP
Reformat the CSV below into TOML: txn,amount,status 260f9831-29af-467a-8739-aabd00406e01,483.09,completed Required constraints: - Return only TOML without any Markdown fences or prose. - Top-level keys must be exactly: txn, amount, status (no extra or missing keys).
txn = "260f9831-29af-467a-8739-aabd00406e01" amount = 483.09 status = "completed"
amount = 483.09 status = "completed"
{ "format": "toml", "rule_id": "T_KEY_DROP", "severity": "moderate", "seed_id": "", "parseable_full": true, "parseable_inner": false, "markdown_wrapped": false, "margin_score": 2, "margin_reasons": [ "missing_required_keys" ], "violation_category": "severe" }
#X_TAG_RENAME
Reformat the YAML below into XML: course: Total eco-centric complexity weeks: 11 credits: 3 meta_info: generated_at: '2026-01-22T04:03:05.033189' version: v4.0 tags: - position - wall - mean - fill - finish audit_log: - action: created by: steven00 timestamp: '2016-05-23T14:40:26.686413' Required c...
<?xml version="1.0" encoding="UTF-8"?> <course_syllabus> <course>Total eco-centric complexity</course> <weeks>11</weeks> <credits>3</credits> <meta_info> <generated_at>2026-01-22T04:03:05.033189</generated_at> <version>v4.0</version> <tags>position</tags> <tags>wall</tags> <tags>mean</tags> <tags>fill</tags> <tags>fini...
<?xml version="1.0" encoding="UTF-8"?> <course_syllabus> <course>Total eco-centric complexity</course> <weeks>11</weeks> <credits>3</credits> <meta_info> <generated_at>2026-01-22T04:03:05.033189</generated_at> <version>v4.0</version> <tags_node>position</tags_node> <tags_node>wall</tags_node> <tags_node>mean</tags_node...
{ "format": "xml", "rule_id": "X_TAG_RENAME", "severity": "moderate", "seed_id": "", "parseable_full": true, "parseable_inner": false, "markdown_wrapped": false, "margin_score": 2, "margin_reasons": [ "missing_xml_tags" ], "violation_category": "severe" }
#X_TAG_RENAME
Create a XML representation of transaction record. Required constraints: - Return only XML without any Markdown fences or prose. - Root tag must be <transaction_record>; keep allowed tags and attributes only. - Do not change tag names or add unknown elements.
<?xml version="1.0" encoding="UTF-8"?> <transaction_record> <txn>9e58c806-c5a4-4012-ae74-e12005bb13ca</txn> <amount>231.99</amount> <status>completed</status> </transaction_record>
<?xml version="1.0" encoding="UTF-8"?> <transaction_record> <txn>9e58c806-c5a4-4012-ae74-e12005bb13ca</txn> <amount_data>231.99</amount_data> <status>completed</status> </transaction_record>
{ "format": "xml", "rule_id": "X_TAG_RENAME", "severity": "moderate", "seed_id": "", "parseable_full": true, "parseable_inner": false, "markdown_wrapped": false, "margin_score": 2, "margin_reasons": [ "missing_xml_tags" ], "violation_category": "severe" }
#J_KEY_RENAME_WRONG
Output structured data for news_article using JSON. Required constraints: - Return only JSON without any Markdown fences or prose. - Top-level keys must be exactly: headline, body, author (no extra or missing keys).
{ "headline": "Become there think west wear wear.", "body": "One task machine. Want professional bar issue. Four plant responsibility.\nOne rich strategy. Gun Republican child. Year discussion land order respond entire peace.", "author": "Monica Wagner" }
{ "headline": "Become there think west wear wear.", "author": "Monica Wagner", "body_id": "One task machine. Want professional bar issue. Four plant responsibility.\nOne rich strategy. Gun Republican child. Year discussion land order respond entire peace." }
{ "format": "json", "rule_id": "J_KEY_RENAME_WRONG", "severity": "moderate", "seed_id": "", "parseable_full": true, "parseable_inner": false, "markdown_wrapped": false, "margin_score": 3, "margin_reasons": [ "missing_required_keys", "added_unknown_keys" ], "violation_category": "severe" }
#X_TAG_RENAME
Output structured data for course_syllabus using XML. Required constraints: - Return only XML without any Markdown fences or prose. - Root tag must be <course_syllabus>; keep allowed tags and attributes only. - Do not change tag names or add unknown elements.
<?xml version="1.0" encoding="UTF-8"?> <course_syllabus> <course>Adaptive transitional circuit</course> <weeks>11</weeks> <credits>3</credits> </course_syllabus>
<?xml version="1.0" encoding="UTF-8"?> <course_syllabus> <course>Adaptive transitional circuit</course> <weeks_data>11</weeks_data> <credits>3</credits> </course_syllabus>
{ "format": "xml", "rule_id": "X_TAG_RENAME", "severity": "moderate", "seed_id": "", "parseable_full": true, "parseable_inner": false, "markdown_wrapped": false, "margin_score": 2, "margin_reasons": [ "missing_xml_tags" ], "violation_category": "severe" }
#C_CODE_FENCE_WRAP
Create a YAML representation of contract. Required constraints: - Return only YAML without any Markdown fences or prose. - Top-level keys must be exactly: Approach, 1. Define the schema, 2. Select the format, 3. Generate realistic content, 4. Structure the data, 5. Serialize, Output, contract, parties, signed (no extr...
Approach: 1. Define the schema: The contract should contain a dictionary with 3 root fields (contract, parties, signed) and a nesting depth of 2. 2. Select the format: YAML is chosen. Use indentation to represent nesting. lists are denoted by hyphens. 3. Generate realistic content: Use plausible names, dates, and value...
## Result ```yaml Approach: 1. Define the schema: The contract should contain a dictionary with 3 root fields (contract, parties, signed) and a nesting depth of 2. 2. Select the format: YAML is chosen. Use indentation to represent nesting. lists are denoted by hyphens. 3. Generate realistic content: Use plausible names...
{ "format": "yaml", "rule_id": "C_CODE_FENCE_WRAP", "severity": "minor", "seed_id": "", "parseable_full": false, "parseable_inner": true, "markdown_wrapped": true, "margin_score": 1, "margin_reasons": [ "prohibited_markdown" ], "violation_category": "minor" }
#J_LIST_EXTRA_ITEM
Output structured data for lab_result using JSON. Required constraints: - Return only JSON without any Markdown fences or prose. - Top-level keys must be exactly: id, test, value, meta_info, audit_log (no extra or missing keys). - Field meta_info.generated_at is required. - All list items must be unique; do not duplic...
{ "id": "90f49f2d-5ee5-422c-b44b-def4fd362bc0", "test": "structure", "value": 14.241576373452492, "meta_info": { "generated_at": "2026-01-22T04:03:06.965516", "version": "v2.0", "tags": [ "conference", "level", "Mr", "benefit", "tree" ] }, "audit_log": [ { ...
{ "id": "90f49f2d-5ee5-422c-b44b-def4fd362bc0", "test": "structure", "value": 14.241576373452492, "meta_info": { "generated_at": "2026-01-22T04:03:06.965516", "version": "v2.0", "tags": [ "conference", "level", "Mr", "benefit", "tree" ] }, "audit_log": [ { ...
{ "format": "json", "rule_id": "J_LIST_EXTRA_ITEM", "severity": "moderate", "seed_id": "", "parseable_full": true, "parseable_inner": false, "markdown_wrapped": false, "margin_score": 1, "margin_reasons": [ "value_changes" ], "violation_category": "moderate" }
#X_TAG_RENAME
Reformat the JSON below into XML: { "experiment_id": "EXP-750258", "title": "Analysis of laugh", "metrics": { "accuracy": 0.9467, "loss": 0.1876, "epoch": 97 }, "config": { "batch_size": 128, "optimizer": "adam", "lr": 0.001 } } Required constraints: - Return only XML without any Ma...
<?xml version="1.0" encoding="UTF-8"?> <experiment_result> <experiment_id>EXP-750258</experiment_id> <title>Analysis of laugh</title> <metrics> <accuracy>0.9467</accuracy> <loss>0.1876</loss> <epoch>97</epoch> </metrics> <config> <batch_size>128</batch_size> <optimizer>adam</optimizer> <lr>0.001</lr> </config> </experi...
<?xml version="1.0" encoding="UTF-8"?> <experiment_result> <experiment_id>EXP-750258</experiment_id> <title_data>Analysis of laugh</title_data> <metrics> <accuracy>0.9467</accuracy> <loss>0.1876</loss> <epoch>97</epoch> </metrics> <config> <batch_size>128</batch_size> <optimizer>adam</optimizer> <lr>0.001</lr> </config...
{ "format": "xml", "rule_id": "X_TAG_RENAME", "severity": "moderate", "seed_id": "", "parseable_full": true, "parseable_inner": false, "markdown_wrapped": false, "margin_score": 2, "margin_reasons": [ "missing_xml_tags" ], "violation_category": "severe" }
#X_TAG_RENAME
Output structured data for news_article using XML. Required constraints: - Return only XML without any Markdown fences or prose. - Root tag must be <news_article>; keep allowed tags and attributes only. - Do not change tag names or add unknown elements.
<?xml version="1.0" encoding="UTF-8"?> <news_article> <headline>Hair yourself finally.</headline> <body>Send join too and. Let project affect. Huge black should degree. Sometimes answer past art play turn act. Son their listen occur player. Challenge first about line program may foot.</body> <author>Louis Allen</author...
<?xml version="1.0" encoding="UTF-8"?> <news_article_node> <headline>Hair yourself finally.</headline> <body>Send join too and. Let project affect. Huge black should degree. Sometimes answer past art play turn act. Son their listen occur player. Challenge first about line program may foot.</body> <author>Louis Allen</a...
{ "format": "xml", "rule_id": "X_TAG_RENAME", "severity": "moderate", "seed_id": "", "parseable_full": true, "parseable_inner": false, "markdown_wrapped": false, "margin_score": 2, "margin_reasons": [ "missing_xml_tags" ], "violation_category": "severe" }
#X_TAG_RENAME
Output structured data for research_paper using XML. Required constraints: - Return only XML without any Markdown fences or prose. - Root tag must be <research_paper>; keep allowed tags and attributes only. - Do not change tag names or add unknown elements.
<?xml version="1.0" encoding="UTF-8"?> <research_paper> <paper_id>arXiv:2024.84903</paper_id> <title>Stand-alone scalable encryption</title> <authors> <name>John Shields</name> <affiliation>Schaefer LLC</affiliation> <email>wigginsshawn@example.net</email> </authors> <authors> <name>Stephanie Wilson</name> <affiliation...
<?xml version="1.0" encoding="UTF-8"?> <research_paper> <paper_id>arXiv:2024.84903</paper_id> <title>Stand-alone scalable encryption</title> <authors> <name>John Shields</name> <affiliation>Schaefer LLC</affiliation> <email_node>wigginsshawn@example.net</email_node> </authors> <authors> <name>Stephanie Wilson</name> <a...
{ "format": "xml", "rule_id": "X_TAG_RENAME", "severity": "moderate", "seed_id": "", "parseable_full": true, "parseable_inner": false, "markdown_wrapped": false, "margin_score": 2, "margin_reasons": [ "missing_xml_tags" ], "violation_category": "severe" }
#T_KEY_DROP
Transform this data from JSON to TOML: { "company": "Salazar Inc", "revenue": 849808, "profit": 42226 } Required constraints: - Return only TOML without any Markdown fences or prose. - Top-level keys must be exactly: company, revenue, profit (no extra or missing keys).
company = "Salazar Inc" revenue = 849808 profit = 42226
revenue = 849808 profit = 42226
{ "format": "toml", "rule_id": "T_KEY_DROP", "severity": "moderate", "seed_id": "", "parseable_full": true, "parseable_inner": false, "markdown_wrapped": false, "margin_score": 2, "margin_reasons": [ "missing_required_keys" ], "violation_category": "severe" }
#X_TAG_RENAME
Generate product listing data in XML format. Required constraints: - Return only XML without any Markdown fences or prose. - Root tag must be <product_listing>; keep allowed tags and attributes only. - Do not change tag names or add unknown elements.
<?xml version="1.0" encoding="UTF-8"?> <product_listing> <product>live</product> <price>22.45</price> <stock>49</stock> </product_listing>
<?xml version="1.0" encoding="UTF-8"?> <product_listing> <product_data>live</product_data> <price>22.45</price> <stock>49</stock> </product_listing>
{ "format": "xml", "rule_id": "X_TAG_RENAME", "severity": "moderate", "seed_id": "", "parseable_full": true, "parseable_inner": false, "markdown_wrapped": false, "margin_score": 2, "margin_reasons": [ "missing_xml_tags" ], "violation_category": "severe" }
#J_STRING_TRUNCATE
Output structured data for prescription using JSON. Required constraints: - Return only JSON without any Markdown fences or prose. - Top-level keys must be exactly: rx, drug, dosage (no extra or missing keys).
{ "rx": "1d4c1c1b-177b-489f-b58f-53f77587035d", "drug": "president", "dosage": "10mg" }
{ "rx": "1d4c1c1b-177b-489f", "drug": "president", "dosage": "10mg" }
{ "format": "json", "rule_id": "J_STRING_TRUNCATE", "severity": "moderate", "seed_id": "", "parseable_full": true, "parseable_inner": false, "markdown_wrapped": false, "margin_score": 1, "margin_reasons": [ "value_changes" ], "violation_category": "moderate" }
#T_KEY_DROP
Output structured data for student_assessment using TOML. Required constraints: - Return only TOML without any Markdown fences or prose. - Top-level keys must be exactly: student, score, grade (no extra or missing keys).
student = "Renee Mosley" score = 35 grade = "A"
score = 35 grade = "A"
{ "format": "toml", "rule_id": "T_KEY_DROP", "severity": "moderate", "seed_id": "", "parseable_full": true, "parseable_inner": false, "markdown_wrapped": false, "margin_score": 2, "margin_reasons": [ "missing_required_keys" ], "violation_category": "severe" }
#Y_LIST_ITEM_DROP
Convert the following TOML to YAML: period = "Q1" sales = 3598 region = "Davisfurt" [[audit_log]] action = "created" by = "johnathankelly" timestamp = "2016-09-06T12:02:55.898012" [[audit_log]] action = "created" by = "ortizscott" timestamp = "1977-01-21T19:18:24.176833" [[audit_log]] action = "created" by = "cassidy...
Approach: 1. Analyze the input TOML structure: It represents a dictionary with 5 root fields (period, sales, region...) and a nesting depth of 3. 2. Identify the target format requirements: YAML requires strict syntax. Use indentation to represent nesting. lists are denoted by hyphens. 3. Map the fields: The root keys ...
1. Analyze the input TOML structure: It represents a dictionary with 5 root fields (period, sales, region...) and a nesting depth of 3. 2. Identify the target format requirements: YAML requires strict syntax. Use indentation to represent nesting. lists are denoted by hyphens. 3. Map the fields: The root keys will m...
{ "format": "yaml", "rule_id": "Y_LIST_ITEM_DROP", "severity": "minor", "seed_id": "", "parseable_full": true, "parseable_inner": false, "markdown_wrapped": false, "margin_score": 1, "margin_reasons": [ "value_changes" ], "violation_category": "moderate" }
End of preview. Expand in Data Studio

YAML Metadata Warning:empty or missing yaml metadata in repo card

Check out the documentation for more information.

Structured ORPO v3 (50/30/20 violation mix)

This dataset contains 2,450 ORPO preference pairs with a targeted violation mix: 50% severe, 30% moderate, 20% minor. All rejected outputs are syntax-valid (or parseable inside fences) and represent clear instruction violations.

Source & Key Differences vs. v1/v2

  • Seed: u-10bei/structured_data_with_cot_dataset_512_v4 (train split)
  • v2 added: prompt constraint augmentation + margin scoring.
  • v3 adds: violation categorization and global 50/30/20 balancing.

Prompt Constraint Augmentation

  • Inject explicit constraints based on the chosen block, e.g.:
    • Return only FORMAT without Markdown.
    • Exact top-level keys / CSV header order.
    • Uniqueness for list items (when applicable).

Margin & Categorization

  • Parsed structural diffs produce margin_score and margin_reasons:
    • Severe: missing_required_keys, type_mismatch, csv_header_mismatch/column_order_change, missing_xml_tags
    • Moderate: added_unknown_keys, value_changes, xml_attr_diff, csv_value_diff, csv_rowcount_diff
    • Minor: prohibited_markdown, light row/dup tweaks (when allowed)
  • Only pairs with margin_score ≥ 1 are kept.

Balancing & Filtering

  • Global target mix: 50% severe, 30% moderate, 20% minor (minor ≤ 20%).
  • Deduplication: SHA1 over normalized input+chosen+rejected.
  • Token limit: (input + chosen + rejected) ≤ 1536 tokens (cl100k_base).

Data Schema

  • jsonl with keys: id, input, chosen, rejected, meta.
  • meta: {format, rule_id, severity, parseable_full, parseable_inner, markdown_wrapped, margin_score, margin_reasons, violation_category}.

Intended Use

Train ORPO to emphasize spec violations that directly impact scoring while preserving syntax fidelity.

Downloads last month
18