messages listlengths 3 3 | source stringclasses 4 values |
|---|---|
[
{
"role": "system",
"content": "You are an expert in TOML format. Analyze the data structure and generate syntactically perfect TOML."
},
{
"role": "user",
"content": "Convert the following CSV to TOML:\ncompany,revenue,profit\r\nPatton-Hoffman,545077,48156\r\n"
},
{
"role": "assistant",
"content": "Approach:\n1. Analyze the input CSV structure: It represents a dictionary with 3 root fields (company, revenue, profit) and a nesting depth of 1.\n2. Identify the target format requirements: TOML requires strict syntax. Define hierarchy using dot-notation or [section] headers.\n3. Map the fields: The root keys will map directly, while nested objects need careful formatting.\n4. Handle data types: Ensure strings, numbers, and booleans are formatted correctly for the target.\n5. Construct the output: serialized the data strictly following the target schema.\n\nOutput:\ncompany = \"Patton-Hoffman\"\nrevenue = 545077\nprofit = 48156\n"
}
] | U10BEI_BASE |
[
{
"role": "system",
"content": "You are an expert in YAML format. Analyze the data structure and generate syntactically perfect YAML."
},
{
"role": "user",
"content": "Generate lab result data in YAML format."
},
{
"role": "assistant",
"content": "Approach:\n1. Define the schema: The lab_result should contain a dictionary with 3 root fields (id, test, value) and a nesting depth of 1.\n2. Select the format: YAML is chosen. Use indentation to represent nesting. lists are denoted by hyphens.\n3. Generate realistic content: Use plausible names, dates, and values appropriate for the domain.\n4. Structure the data: Organize the fields logically with appropriate nesting and hierarchies.\n5. Serialize: Output the final string making sure to escape special characters if necessary.\n\nOutput:\nid: 83a952aa-2b86-467d-80f4-4202ac02a3b2\ntest: sell\nvalue: 59.8729956334333\n"
}
] | U10BEI_BASE |
[
{
"role": "system",
"content": "You must NOT call tools.\nDo NOT include any extra text, code fences, or markdown.\nDo NOT output JSON or YAML.\nOutput MUST be valid TOML.\nThe first line MUST be: [result]\nThen include status and confidence keys under [result].\nThen include [result.fields].\nOutput ONLY TOML as plain text.\n"
},
{
"role": "user",
"content": "Please convert the following JSON code to TOML code.\n\n<code>{\"student\":\"Anna Jones\",\"score\":3,\"grade\":\"A\",\"meta_info\":{\"generated_at\":\"2026-01-23T05:23:24.080358\",\"version\":\"v1.0\",\"tags\":[\"full\",\"deal\",\"discuss\",\"my\",\"agree\"]},\"audit_log\":[{\"action\":\"created\",\"by\":\"robersonsteven\",\"timestamp\":\"2006-06-14T15:36:17.694508\"},{\"action\":\"created\",\"by\":\"allenian\",\"timestamp\":\"1997-06-03T02:28:35.524501\"}]}</code>"
},
{
"role": "assistant",
"content": "[result]\nstatus = \"ok\"\nconfidence = 0.800\n\n[result.fields]\n\"audit_log\" = \"[{\\\"action\\\": \\\"created\\\", \\\"by\\\": \\\"robersonsteven\\\", \\\"timestamp\\\": \\\"2006-06-14T15:36:17.694508\\\"}, {\\\"action\\\": \\\"created\\\", \\\"by\\\": \\\"allenian\\\", \\\"timestamp\\\": \\\"1997-06-03T02:28:35.524501\\\"}]\"\n\"grade\" = \"A\"\n\"meta_info\" = \"{\\\"generated_at\\\": \\\"2026-01-23T05:23:24.080358\\\", \\\"version\\\": \\\"v1.0\\\", \\\"tags\\\": [\\\"full\\\", \\\"deal\\\", \\\"discuss\\\", \\\"my\\\", \\\"agree\\\"]}\"\n\"score\" = 3\n\"student\" = \"Anna Jones\""
}
] | CONV_JSON2TOML_TPLFIX |
[
{
"role": "system",
"content": "You are an expert in JSON format. Analyze the data structure and generate syntactically perfect JSON."
},
{
"role": "user",
"content": "Convert the following CSV to JSON:\nrandom_content,meta_info.generated_at,meta_info.version,meta_info.tags,audit_log\r\ncollection,2026-01-23T05:23:24.956683,v3.0,\"[\"\"food\"\", \"\"wide\"\", \"\"blue\"\", \"\"detail\"\", \"\"against\"\"]\",\"[{\"\"action\"\": \"\"created\"\", \"\"by\"\": \"\"leonardjesse\"\", \"\"timestamp\"\": \"\"2016-03-15T01:52:11.626442\"\"}]\"\r\n"
},
{
"role": "assistant",
"content": "Approach:\n1. Analyze the input CSV structure: It represents a dictionary with 3 root fields (random_content, meta_info, audit_log) and a nesting depth of 3.\n2. Identify the target format requirements: JSON requires strict syntax. Ensure all keys are quoted and nested structures use curly braces.\n3. Map the fields: The root keys will map directly, while nested objects need careful formatting.\n4. Handle data types: Ensure strings, numbers, and booleans are formatted correctly for the target.\n5. Construct the output: serialized the data strictly following the target schema.\n\nOutput:\n{\n \"random_content\": \"collection\",\n \"meta_info\": {\n \"generated_at\": \"2026-01-23T05:23:24.956683\",\n \"version\": \"v3.0\",\n \"tags\": [\n \"food\",\n \"wide\",\n \"blue\",\n \"detail\",\n \"against\"\n ]\n },\n \"audit_log\": [\n {\n \"action\": \"created\",\n \"by\": \"leonardjesse\",\n \"timestamp\": \"2016-03-15T01:52:11.626442\"\n }\n ]\n}"
}
] | U10BEI_BASE |
[
{
"role": "system",
"content": "You must NOT call tools.\nDo NOT include any extra text, code fences, or markdown.\nDo NOT output JSON or YAML.\nOutput MUST be valid TOML.\nThe first line MUST be: [result]\nThen include status and confidence keys under [result].\nThen include [result.fields].\nOutput ONLY TOML as plain text.\n"
},
{
"role": "user",
"content": "Generate a valid TOML output using the fixed template.\nConstraints:\n- Output ONLY TOML, no extra text.\n- The first line must be [result].\n- Include [result.fields] with all required keys.\nRequired keys in fields: id,name,note,flag,score,category,empty\nValues:\n- id: 1220\n- name: item_1220\n- note: line1\\nline2\n- flag: True\n- score: 0.2\n- category: C\n- empty: \n"
},
{
"role": "assistant",
"content": "[result]\nstatus = \"ok\"\nconfidence = 0.800\n\n[result.fields]\n\"category\" = \"C\"\n\"empty\" = \"\"\n\"flag\" = true\n\"id\" = 1220\n\"name\" = \"item_1220\"\n\"note\" = \"line1\\\\nline2\"\n\"score\" = 0.2"
}
] | TOML_GEN_FIXED_VAR |
[
{
"role": "system",
"content": "You must NOT call tools.\nDo NOT include any extra text, code fences, or markdown.\nDo NOT output JSON or YAML.\nOutput MUST be valid TOML.\nThe first line MUST be: [result]\nThen include status and confidence keys under [result].\nThen include [result.fields].\nOutput ONLY TOML as plain text.\n"
},
{
"role": "user",
"content": "Please convert the following JSON code to TOML code.\n\n<code>{\n \"between\": [\n 2.44,\n 18,\n 20,\n 2.42\n ],\n \"over_1\": \"point\",\n \"source_2\": [\n 3.11\n ],\n \"statement_3\": {\n \"small\": \"they\",\n \"to_1\": 68,\n \"week_2\": 84,\n \"trouble_3\": 22,\n \"consumer_4\": 30\n }\n}</code>"
},
{
"role": "assistant",
"content": "[result]\nstatus = \"ok\"\nconfidence = 0.800\n\n[result.fields]\n\"between\" = \"[2.44, 18, 20, 2.42]\"\n\"over_1\" = \"point\"\n\"source_2\" = \"[3.11]\"\n\"statement_3\" = \"{\\\"small\\\": \\\"they\\\", \\\"to_1\\\": 68, \\\"week_2\\\": 84, \\\"trouble_3\\\": 22, \\\"consumer_4\\\": 30}\""
}
] | CONV_JSON2TOML_TPLFIX |
[
{
"role": "system",
"content": "You must NOT call tools.\nDo NOT include any extra text, code fences, or markdown.\nDo NOT output JSON or YAML.\nOutput MUST be valid TOML.\nThe first line MUST be: [result]\nThen include status and confidence keys under [result].\nThen include [result.fields].\nOutput ONLY TOML as plain text.\n"
},
{
"role": "user",
"content": "Please convert the following JSON code to TOML code.\n\n<code>{\n \"control\": {\n \"experience\": {\n \"raise\": null,\n \"thought_1\": 2.27\n },\n \"baby_1\": [\n true,\n null,\n null\n ],\n \"little_2\": [\n null,\n null,\n null,\n 4.24,\n 80\n ]\n },\n \"sport_1\": [\n [\n \"include\",\n 69,\n \"around\"\n ],\n {\n \"race\": 97,\n \"serve_1\": \"any\",\n \"yeah_2\": 100\n },\n [\n \"believe\",\n null\n ]\n ],\n \"meta_info\": {\n \"generated_at\": \"2026-01-23T05:23:24.235641\",\n \"version\": \"v2.0\",\n \"tags\": [\n \"church\",\n \"thus\",\n \"walk\",\n \"give\",\n \"same\"\n ]\n },\n \"audit_log\": [\n {\n \"action\": \"created\",\n \"by\": \"karasmith\",\n \"timestamp\": \"2025-08-19T11:51:33.330595\"\n },\n {\n \"action\": \"created\",\n \"by\": \"lucas98\",\n \"timestamp\": \"2018-04-12T03:57:55.136692\"\n },\n {\n \"action\": \"created\",\n \"by\": \"martineznathan\",\n \"timestamp\": \"2013-04-15T15:14:37.801682\"\n }\n ]\n}</code>"
},
{
"role": "assistant",
"content": "[result]\nstatus = \"ok\"\nconfidence = 0.800\n\n[result.fields]\n\"audit_log\" = \"[{\\\"action\\\": \\\"created\\\", \\\"by\\\": \\\"karasmith\\\", \\\"timestamp\\\": \\\"2025-08-19T11:51:33.330595\\\"}, {\\\"action\\\": \\\"created\\\", \\\"by\\\": \\\"lucas98\\\", \\\"timestamp\\\": \\\"2018-04-12T03:57:55.136692\\\"}, {\\\"action\\\": \\\"created\\\", \\\"by\\\": \\\"martineznathan\\\", \\\"timestamp\\\": \\\"2013-04-15T15:14:37.801682\\\"}]\"\n\"control\" = \"{\\\"experience\\\": {\\\"raise\\\": null, \\\"thought_1\\\": 2.27}, \\\"baby_1\\\": [true, null, null], \\\"little_2\\\": [null, null, null, 4.24, 80]}\"\n\"meta_info\" = \"{\\\"generated_at\\\": \\\"2026-01-23T05:23:24.235641\\\", \\\"version\\\": \\\"v2.0\\\", \\\"tags\\\": [\\\"church\\\", \\\"thus\\\", \\\"walk\\\", \\\"give\\\", \\\"same\\\"]}\"\n\"sport_1\" = \"[[\\\"include\\\", 69, \\\"around\\\"], {\\\"race\\\": 97, \\\"serve_1\\\": \\\"any\\\", \\\"yeah_2\\\": 100}, [\\\"believe\\\", null]]\""
}
] | CONV_JSON2TOML_TPLFIX |
[
{
"role": "system",
"content": "You are an expert in YAML format. Analyze the data structure and generate syntactically perfect YAML."
},
{
"role": "user",
"content": "Transform this data from XML to YAML:\n<?xml version=\"1.0\" encoding=\"UTF-8\"?>\n<random_structured_data>\n<random_content>[9.08, None]</random_content>\n<random_content>\n<boy>True</boy>\n<cost_1>2.98</cost_1>\n<during_2>False</during_2>\n</random_content>\n<random_content>\n<science>19</science>\n<against_1>8.38</against_1>\n<whose_2>True</whose_2>\n<foot_3 />\n<at_4 />\n</random_content>\n<random_content>into</random_content>\n</random_structured_data>"
},
{
"role": "assistant",
"content": "Approach:\n1. Analyze the input XML structure: It represents a dictionary with 1 root fields (random_content) and a nesting depth of 3.\n2. Identify the target format requirements: YAML requires strict syntax. Use indentation to represent nesting. lists are denoted by hyphens.\n3. Map the fields: The root keys will map directly, while nested objects need careful formatting.\n4. Handle data types: Ensure strings, numbers, and booleans are formatted correctly for the target.\n5. Construct the output: serialized the data strictly following the target schema.\n\nOutput:\nrandom_content:\n- - 9.08\n - null\n- boy: true\n cost_1: 2.98\n during_2: false\n- science: 19\n against_1: 8.38\n whose_2: true\n foot_3: null\n at_4: null\n- into\n"
}
] | U10BEI_BASE |
[
{
"role": "system",
"content": "You must NOT call tools.\nDo NOT include any extra text, code fences, or markdown.\nFollow the user instructions EXACTLY.\nOutput ONLY CSV content as plain text.\nStart with the header row.\n"
},
{
"role": "user",
"content": "Please output CSV code(Only output the CSV content as plain text, starting with the header row.).\n\nTask:\nGiven a recipe description, extract detailed structured information about the recipe into CSV format.\nFeature Requirements:\n1. Use the exact header fields and order.\n2. Output exactly one data row.\n3. Output exactly two lines: header line + one row.\n\nHeader:\nrecipe_id,recipe_name,chef_name,cuisine_type,preparation_time_minutes,cooking_time_minutes,servings,difficulty_level,main_ingredient,ingredient_list,ingredient_quantities,cooking_method,calories_per_serving,allergen_warnings,vegetarian_friendly,seasonality"
},
{
"role": "assistant",
"content": "recipe_id,recipe_name,chef_name,cuisine_type,preparation_time_minutes,cooking_time_minutes,servings,difficulty_level,main_ingredient,ingredient_list,ingredient_quantities,cooking_method,calories_per_serving,allergen_warnings,vegetarian_friendly,seasonality\nR00189,Stellar Stew 189,Mira Qor,Japanese,25,51,3,Medium,Nebula Beans,nebula_beans;star_salt;quantum_onion,200g;5g;1,fry,292,None,No,Summer"
}
] | CSV_TWO_LINE_FIXED |
[
{
"role": "system",
"content": "You are an expert in YAML format. Analyze the data structure and generate syntactically perfect YAML."
},
{
"role": "user",
"content": "Output structured data for api_specification using YAML."
},
{
"role": "assistant",
"content": "Approach:\n1. Define the schema: The api_specification should contain a dictionary with 3 root fields (endpoint, method, responses) and a nesting depth of 2.\n2. Select the format: YAML is chosen. Use indentation to represent nesting. lists are denoted by hyphens.\n3. Generate realistic content: Use plausible names, dates, and values appropriate for the domain.\n4. Structure the data: Organize the fields logically with appropriate nesting and hierarchies.\n5. Serialize: Output the final string making sure to escape special characters if necessary.\n\nOutput:\nendpoint: /api/v1/users\nmethod: GET\nresponses:\n '200': OK\n"
}
] | U10BEI_BASE |
[
{
"role": "system",
"content": "You must NOT call tools.\nDo NOT include any extra text, code fences, or markdown.\nDo NOT output JSON or YAML.\nOutput MUST be valid TOML.\nThe first line MUST be: [result]\nThen include status and confidence keys under [result].\nThen include [result.fields].\nOutput ONLY TOML as plain text.\n"
},
{
"role": "user",
"content": "Please convert the following JSON code to TOML code.\n\n<code>{\n \"experiment_id\": \"EXP-898577\",\n \"title\": \"Analysis of among\",\n \"metrics\": {\n \"accuracy\": 0.9221,\n \"loss\": 0.3625,\n \"epoch\": 97\n },\n \"config\": {\n \"batch_size\": 128,\n \"optimizer\": \"adam\",\n \"lr\": 0.001\n }\n}</code>"
},
{
"role": "assistant",
"content": "[result]\nstatus = \"ok\"\nconfidence = 0.800\n\n[result.fields]\n\"config\" = \"{\\\"batch_size\\\": 128, \\\"optimizer\\\": \\\"adam\\\", \\\"lr\\\": 0.001}\"\n\"experiment_id\" = \"EXP-898577\"\n\"metrics\" = \"{\\\"accuracy\\\": 0.9221, \\\"loss\\\": 0.3625, \\\"epoch\\\": 97}\"\n\"title\" = \"Analysis of among\""
}
] | CONV_JSON2TOML_TPLFIX |
[
{
"role": "system",
"content": "You must NOT call tools.\nDo NOT include any extra text, code fences, or markdown.\nDo NOT output JSON or YAML.\nOutput MUST be valid TOML.\nThe first line MUST be: [result]\nThen include status and confidence keys under [result].\nThen include [result.fields].\nOutput ONLY TOML as plain text.\n"
},
{
"role": "user",
"content": "Please convert the following JSON code to TOML code.\n\n<code>{\n \"audit_log\": [\n {\n \"action\": \"created\",\n \"by\": \"cheyennewashington\",\n \"timestamp\": \"1994-09-26T03:52:33.932107\"\n }\n ],\n \"endpoint\": \"/api/v1/users\",\n \"meta_info\": {\n \"generated_at\": \"2026-01-23T05:23:24.752822\",\n \"tags\": [\n \"feeling\",\n \"offer\",\n \"among\",\n \"but\",\n \"receive\"\n ],\n \"version\": \"v4.0\"\n },\n \"method\": \"GET\",\n \"responses\": {\n \"200\": \"OK\"\n }\n}</code>"
},
{
"role": "assistant",
"content": "[result]\nstatus = \"ok\"\nconfidence = 0.800\n\n[result.fields]\n\"audit_log\" = \"[{\\\"action\\\": \\\"created\\\", \\\"by\\\": \\\"cheyennewashington\\\", \\\"timestamp\\\": \\\"1994-09-26T03:52:33.932107\\\"}]\"\n\"endpoint\" = \"/api/v1/users\"\n\"meta_info\" = \"{\\\"generated_at\\\": \\\"2026-01-23T05:23:24.752822\\\", \\\"tags\\\": [\\\"feeling\\\", \\\"offer\\\", \\\"among\\\", \\\"but\\\", \\\"receive\\\"], \\\"version\\\": \\\"v4.0\\\"}\"\n\"method\" = \"GET\"\n\"responses\" = \"{\\\"200\\\": \\\"OK\\\"}\""
}
] | CONV_JSON2TOML_TPLFIX |
[
{
"role": "system",
"content": "You must NOT call tools.\nDo NOT include any extra text, code fences, or markdown.\nDo NOT output JSON or YAML.\nOutput MUST be valid TOML.\nThe first line MUST be: [result]\nThen include status and confidence keys under [result].\nThen include [result.fields].\nOutput ONLY TOML as plain text.\n"
},
{
"role": "user",
"content": "Please convert the following JSON code to TOML code.\n\n<code>{\n \"user\": \"jonathanhenry\",\n \"post\": \"Still light beyond high space probably. Represent speech certainly share.\",\n \"likes\": 399,\n \"meta_info\": {\n \"generated_at\": \"2026-01-23T05:23:25.079332\",\n \"version\": \"v4.0\",\n \"tags\": [\n \"run\",\n \"drug\",\n \"meet\",\n \"plant\",\n \"far\"\n ]\n },\n \"audit_log\": [\n {\n \"action\": \"created\",\n \"by\": \"underwoodgary\",\n \"timestamp\": \"1985-10-15T06:03:46.138891\"\n },\n {\n \"action\": \"created\",\n \"by\": \"brianbradley\",\n \"timestamp\": \"1978-01-12T23:26:56.795902\"\n }\n ]\n}</code>"
},
{
"role": "assistant",
"content": "[result]\nstatus = \"ok\"\nconfidence = 0.800\n\n[result.fields]\n\"audit_log\" = \"[{\\\"action\\\": \\\"created\\\", \\\"by\\\": \\\"underwoodgary\\\", \\\"timestamp\\\": \\\"1985-10-15T06:03:46.138891\\\"}, {\\\"action\\\": \\\"created\\\", \\\"by\\\": \\\"brianbradley\\\", \\\"timestamp\\\": \\\"1978-01-12T23:26:56.795902\\\"}]\"\n\"likes\" = 399\n\"meta_info\" = \"{\\\"generated_at\\\": \\\"2026-01-23T05:23:25.079332\\\", \\\"version\\\": \\\"v4.0\\\", \\\"tags\\\": [\\\"run\\\", \\\"drug\\\", \\\"meet\\\", \\\"plant\\\", \\\"far\\\"]}\"\n\"post\" = \"Still light beyond high space probably. Represent speech certainly share.\"\n\"user\" = \"jonathanhenry\""
}
] | CONV_JSON2TOML_TPLFIX |
[
{
"role": "system",
"content": "You must NOT call tools.\nDo NOT include any extra text, code fences, or markdown.\nDo NOT output JSON or YAML.\nOutput MUST be valid TOML.\nThe first line MUST be: [result]\nThen include status and confidence keys under [result].\nThen include [result.fields].\nOutput ONLY TOML as plain text.\n"
},
{
"role": "user",
"content": "Please convert the following JSON code to TOML code.\n\n<code>{\n \"experiment_id\": \"EXP-898577\",\n \"title\": \"Analysis of among\",\n \"metrics\": {\n \"accuracy\": 0.9221,\n \"loss\": 0.3625,\n \"epoch\": 97\n },\n \"config\": {\n \"batch_size\": 128,\n \"optimizer\": \"adam\",\n \"lr\": 0.001\n }\n}</code>"
},
{
"role": "assistant",
"content": "[result]\nstatus = \"ok\"\nconfidence = 0.800\n\n[result.fields]\n\"config\" = \"{\\\"batch_size\\\": 128, \\\"optimizer\\\": \\\"adam\\\", \\\"lr\\\": 0.001}\"\n\"experiment_id\" = \"EXP-898577\"\n\"metrics\" = \"{\\\"accuracy\\\": 0.9221, \\\"loss\\\": 0.3625, \\\"epoch\\\": 97}\"\n\"title\" = \"Analysis of among\""
}
] | CONV_JSON2TOML_TPLFIX |
[
{
"role": "system",
"content": "You must NOT call tools.\nDo NOT include any extra text, code fences, or markdown.\nFollow the user instructions EXACTLY.\nOutput ONLY CSV content as plain text.\nStart with the header row.\n"
},
{
"role": "user",
"content": "Please output CSV code(Only output the CSV content as plain text, starting with the header row.).\n\nTask:\nGiven a recipe description, extract detailed structured information about the recipe into CSV format.\nFeature Requirements:\n1. Use the exact header fields and order.\n2. Output exactly one data row.\n3. Output exactly two lines: header line + one row.\n\nHeader:\nrecipe_id,recipe_name,chef_name,cuisine_type,preparation_time_minutes,cooking_time_minutes,servings,difficulty_level,main_ingredient,ingredient_list,ingredient_quantities,cooking_method,calories_per_serving,allergen_warnings,vegetarian_friendly,seasonality"
},
{
"role": "assistant",
"content": "recipe_id,recipe_name,chef_name,cuisine_type,preparation_time_minutes,cooking_time_minutes,servings,difficulty_level,main_ingredient,ingredient_list,ingredient_quantities,cooking_method,calories_per_serving,allergen_warnings,vegetarian_friendly,seasonality\nR00361,Stellar Stew 361,Mira Qor,Galactic,16,17,4,Medium,Quantum Onion,nebula_beans;star_salt;quantum_onion,200g;5g;1,fry,864,dairy,Yes,Winter"
}
] | CSV_TWO_LINE_FIXED |
[
{
"role": "system",
"content": "You are an expert in JSON format. Analyze the data structure and generate syntactically perfect JSON."
},
{
"role": "user",
"content": "Generate transaction record data in JSON format."
},
{
"role": "assistant",
"content": "Approach:\n1. Define the schema: The transaction_record should contain a dictionary with 3 root fields (txn, amount, status) and a nesting depth of 1.\n2. Select the format: JSON is chosen. Ensure all keys are quoted and nested structures use curly braces.\n3. Generate realistic content: Use plausible names, dates, and values appropriate for the domain.\n4. Structure the data: Organize the fields logically with appropriate nesting and hierarchies.\n5. Serialize: Output the final string making sure to escape special characters if necessary.\n\nOutput:\n{\n \"txn\": \"c4317618-4baa-4293-87c0-6199aecbcef4\",\n \"amount\": 132.21,\n \"status\": \"completed\"\n}"
}
] | U10BEI_BASE |
[
{
"role": "system",
"content": "You must NOT call tools.\nDo NOT include any extra text, code fences, or markdown.\nDo NOT output JSON or YAML.\nOutput MUST be valid TOML.\nThe first line MUST be: [result]\nThen include status and confidence keys under [result].\nThen include [result.fields].\nOutput ONLY TOML as plain text.\n"
},
{
"role": "user",
"content": "Please convert the following JSON code to TOML code.\n\n<code>{\n \"contract\": \"deploy proactive methodologies\",\n \"parties\": [\n \"Robinson-Matthews\",\n \"Charles, Adams and Williams\"\n ],\n \"signed\": true,\n \"meta_info\": {\n \"generated_at\": \"2026-01-23T05:23:24.367974\",\n \"version\": \"v3.0\",\n \"tags\": [\n \"clear\",\n \"grow\",\n \"money\",\n \"street\",\n \"better\"\n ]\n },\n \"audit_log\": [\n {\n \"action\": \"created\",\n \"by\": \"cgonzalez\",\n \"timestamp\": \"1987-01-23T06:44:31.855869\"\n }\n ]\n}</code>"
},
{
"role": "assistant",
"content": "[result]\nstatus = \"ok\"\nconfidence = 0.800\n\n[result.fields]\n\"audit_log\" = \"[{\\\"action\\\": \\\"created\\\", \\\"by\\\": \\\"cgonzalez\\\", \\\"timestamp\\\": \\\"1987-01-23T06:44:31.855869\\\"}]\"\n\"contract\" = \"deploy proactive methodologies\"\n\"meta_info\" = \"{\\\"generated_at\\\": \\\"2026-01-23T05:23:24.367974\\\", \\\"version\\\": \\\"v3.0\\\", \\\"tags\\\": [\\\"clear\\\", \\\"grow\\\", \\\"money\\\", \\\"street\\\", \\\"better\\\"]}\"\n\"parties\" = \"[\\\"Robinson-Matthews\\\", \\\"Charles, Adams and Williams\\\"]\"\n\"signed\" = true"
}
] | CONV_JSON2TOML_TPLFIX |
[
{
"role": "system",
"content": "You must NOT call tools.\nDo NOT include any extra text, code fences, or markdown.\nDo NOT output JSON or YAML.\nOutput MUST be valid TOML.\nThe first line MUST be: [result]\nThen include status and confidence keys under [result].\nThen include [result.fields].\nOutput ONLY TOML as plain text.\n"
},
{
"role": "user",
"content": "Generate a valid TOML output using the fixed template.\nConstraints:\n- Output ONLY TOML, no extra text.\n- The first line must be [result].\n- Include [result.fields] with all required keys.\nRequired keys in fields: id,name,note,flag,score,category,empty\nValues:\n- id: 648\n- name: item_648\n- note: ok\n- flag: True\n- score: 0.48\n- category: A\n- empty: \n"
},
{
"role": "assistant",
"content": "[result]\nstatus = \"ok\"\nconfidence = 0.800\n\n[result.fields]\n\"category\" = \"A\"\n\"empty\" = \"\"\n\"flag\" = true\n\"id\" = 648\n\"name\" = \"item_648\"\n\"note\" = \"ok\"\n\"score\" = 0.48"
}
] | TOML_GEN_FIXED_VAR |
[
{
"role": "system",
"content": "You must NOT call tools.\nDo NOT include any extra text, code fences, or markdown.\nDo NOT output JSON or YAML.\nOutput MUST be valid TOML.\nThe first line MUST be: [result]\nThen include status and confidence keys under [result].\nThen include [result.fields].\nOutput ONLY TOML as plain text.\n"
},
{
"role": "user",
"content": "Please convert the following JSON code to TOML code.\n\n<code>{\n \"experiment_id\": \"EXP-186833\",\n \"title\": \"Analysis of table\",\n \"metrics\": {\n \"accuracy\": 0.9336,\n \"loss\": 0.3688,\n \"epoch\": 47\n },\n \"config\": {\n \"batch_size\": 32,\n \"optimizer\": \"adam\",\n \"lr\": 0.001\n }\n}</code>"
},
{
"role": "assistant",
"content": "[result]\nstatus = \"ok\"\nconfidence = 0.800\n\n[result.fields]\n\"config\" = \"{\\\"batch_size\\\": 32, \\\"optimizer\\\": \\\"adam\\\", \\\"lr\\\": 0.001}\"\n\"experiment_id\" = \"EXP-186833\"\n\"metrics\" = \"{\\\"accuracy\\\": 0.9336, \\\"loss\\\": 0.3688, \\\"epoch\\\": 47}\"\n\"title\" = \"Analysis of table\""
}
] | CONV_JSON2TOML_TPLFIX |
[
{
"role": "system",
"content": "You must NOT call tools.\nDo NOT include any extra text, code fences, or markdown.\nDo NOT output JSON or YAML.\nOutput MUST be valid TOML.\nThe first line MUST be: [result]\nThen include status and confidence keys under [result].\nThen include [result.fields].\nOutput ONLY TOML as plain text.\n"
},
{
"role": "user",
"content": "Please convert the following JSON code to TOML code.\n\n<code>{\n \"course\": \"Right-sized context-sensitive superstructure\",\n \"weeks\": 11,\n \"credits\": 3,\n \"meta_info\": {\n \"generated_at\": \"2026-01-23T05:23:24.088213\",\n \"version\": \"v3.0\",\n \"tags\": [\n \"theory\",\n \"movement\",\n \"evening\",\n \"Congress\",\n \"reality\"\n ]\n },\n \"audit_log\": [\n {\n \"action\": \"created\",\n \"by\": \"michaelsmith\",\n \"timestamp\": \"1996-11-12T19:13:19.777724\"\n }\n ]\n}</code>"
},
{
"role": "assistant",
"content": "[result]\nstatus = \"ok\"\nconfidence = 0.800\n\n[result.fields]\n\"audit_log\" = \"[{\\\"action\\\": \\\"created\\\", \\\"by\\\": \\\"michaelsmith\\\", \\\"timestamp\\\": \\\"1996-11-12T19:13:19.777724\\\"}]\"\n\"course\" = \"Right-sized context-sensitive superstructure\"\n\"credits\" = 3\n\"meta_info\" = \"{\\\"generated_at\\\": \\\"2026-01-23T05:23:24.088213\\\", \\\"version\\\": \\\"v3.0\\\", \\\"tags\\\": [\\\"theory\\\", \\\"movement\\\", \\\"evening\\\", \\\"Congress\\\", \\\"reality\\\"]}\"\n\"weeks\" = 11"
}
] | CONV_JSON2TOML_TPLFIX |
[
{
"role": "system",
"content": "You are an expert in TOML format. Analyze the data structure and generate syntactically perfect TOML."
},
{
"role": "user",
"content": "Reformat the XML below into TOML:\n<?xml version=\"1.0\" encoding=\"UTF-8\"?>\n<sales_analytics>\n<period>Q1</period>\n<sales>1350</sales>\n<region>Lake Tracy</region>\n</sales_analytics>"
},
{
"role": "assistant",
"content": "Approach:\n1. Analyze the input XML structure: It represents a dictionary with 3 root fields (period, sales, region) and a nesting depth of 1.\n2. Identify the target format requirements: TOML requires strict syntax. Define hierarchy using dot-notation or [section] headers.\n3. Map the fields: The root keys will map directly, while nested objects need careful formatting.\n4. Handle data types: Ensure strings, numbers, and booleans are formatted correctly for the target.\n5. Construct the output: serialized the data strictly following the target schema.\n\nOutput:\nperiod = \"Q1\"\nsales = 1350\nregion = \"Lake Tracy\"\n"
}
] | U10BEI_BASE |
[
{
"role": "system",
"content": "You must NOT call tools.\nDo NOT include any extra text, code fences, or markdown.\nFollow the user instructions EXACTLY.\nOutput ONLY CSV content as plain text.\nStart with the header row.\n"
},
{
"role": "user",
"content": "Please output CSV code(Only output the CSV content as plain text, starting with the header row.).\n\nTask:\nGiven a recipe description, extract detailed structured information about the recipe into CSV format.\nFeature Requirements:\n1. Use the exact header fields and order.\n2. Output exactly one data row.\n3. Output exactly two lines: header line + one row.\n\nHeader:\nrecipe_id,recipe_name,chef_name,cuisine_type,preparation_time_minutes,cooking_time_minutes,servings,difficulty_level,main_ingredient,ingredient_list,ingredient_quantities,cooking_method,calories_per_serving,allergen_warnings,vegetarian_friendly,seasonality"
},
{
"role": "assistant",
"content": "recipe_id,recipe_name,chef_name,cuisine_type,preparation_time_minutes,cooking_time_minutes,servings,difficulty_level,main_ingredient,ingredient_list,ingredient_quantities,cooking_method,calories_per_serving,allergen_warnings,vegetarian_friendly,seasonality\nR00659,Stellar Stew 659,Mira Qor,Thai,8,40,3,Easy,Nebula Beans,nebula_beans;star_salt;quantum_onion,200g;5g;1,fry,490,None,No,All Year"
}
] | CSV_TWO_LINE_FIXED |
[
{
"role": "system",
"content": "You must NOT call tools.\nDo NOT include any extra text, code fences, or markdown.\nDo NOT output JSON or YAML.\nOutput MUST be valid TOML.\nThe first line MUST be: [result]\nThen include status and confidence keys under [result].\nThen include [result.fields].\nOutput ONLY TOML as plain text.\n"
},
{
"role": "user",
"content": "Please convert the following JSON code to TOML code.\n\n<code>{\n \"txn\": \"42327eaf-8038-44fa-9c06-2285ed8de51b\",\n \"amount\": 40.99,\n \"status\": \"completed\",\n \"meta_info\": {\n \"generated_at\": \"2026-01-23T05:23:24.465239\",\n \"version\": \"v2.0\",\n \"tags\": [\n \"south\",\n \"push\",\n \"address\",\n \"business\",\n \"me\"\n ]\n },\n \"audit_log\": [\n {\n \"action\": \"created\",\n \"by\": \"showard\",\n \"timestamp\": \"2012-07-27T22:03:46.538443\"\n }\n ]\n}</code>"
},
{
"role": "assistant",
"content": "[result]\nstatus = \"completed\"\nconfidence = 0.800\n\n[result.fields]\n\"amount\" = 40.99\n\"audit_log\" = \"[{\\\"action\\\": \\\"created\\\", \\\"by\\\": \\\"showard\\\", \\\"timestamp\\\": \\\"2012-07-27T22:03:46.538443\\\"}]\"\n\"meta_info\" = \"{\\\"generated_at\\\": \\\"2026-01-23T05:23:24.465239\\\", \\\"version\\\": \\\"v2.0\\\", \\\"tags\\\": [\\\"south\\\", \\\"push\\\", \\\"address\\\", \\\"business\\\", \\\"me\\\"]}\"\n\"txn\" = \"42327eaf-8038-44fa-9c06-2285ed8de51b\""
}
] | CONV_JSON2TOML_TPLFIX |
[
{
"role": "system",
"content": "You must NOT call tools.\nDo NOT include any extra text, code fences, or markdown.\nDo NOT output JSON or YAML.\nOutput MUST be valid TOML.\nThe first line MUST be: [result]\nThen include status and confidence keys under [result].\nThen include [result.fields].\nOutput ONLY TOML as plain text.\n"
},
{
"role": "user",
"content": "Generate a valid TOML output using the fixed template.\nConstraints:\n- Output ONLY TOML, no extra text.\n- The first line must be [result].\n- Include [result.fields] with all required keys.\nRequired keys in fields: id,name,note,flag,score,category,empty\nValues:\n- id: 1452\n- name: item_1452\n- note: ok\n- flag: True\n- score: 0.52\n- category: A\n- empty: \n"
},
{
"role": "assistant",
"content": "[result]\nstatus = \"ok\"\nconfidence = 0.800\n\n[result.fields]\n\"category\" = \"A\"\n\"empty\" = \"\"\n\"flag\" = true\n\"id\" = 1452\n\"name\" = \"item_1452\"\n\"note\" = \"ok\"\n\"score\" = 0.52"
}
] | TOML_GEN_FIXED_VAR |
[
{
"role": "system",
"content": "You are an expert in YAML format. Analyze the data structure and generate syntactically perfect YAML."
},
{
"role": "user",
"content": "Reformat the XML below into YAML:\n<?xml version=\"1.0\" encoding=\"UTF-8\"?>\n<random_structured_data>\n<random_content>themselves</random_content>\n</random_structured_data>"
},
{
"role": "assistant",
"content": "Approach:\n1. Analyze the input XML structure: It represents a dictionary with 1 root fields (random_content) and a nesting depth of 1.\n2. Identify the target format requirements: YAML requires strict syntax. Use indentation to represent nesting. lists are denoted by hyphens.\n3. Map the fields: The root keys will map directly, while nested objects need careful formatting.\n4. Handle data types: Ensure strings, numbers, and booleans are formatted correctly for the target.\n5. Construct the output: serialized the data strictly following the target schema.\n\nOutput:\nrandom_content: themselves\n"
}
] | U10BEI_BASE |
[
{
"role": "system",
"content": "You are an expert in XML format. Analyze the data structure and generate syntactically perfect XML."
},
{
"role": "user",
"content": "Create a XML representation of product listing."
},
{
"role": "assistant",
"content": "Approach:\n1. Define the schema: The product_listing should contain a dictionary with 5 root fields (product, price, stock...) and a nesting depth of 3.\n2. Select the format: XML is chosen. Map dictionary keys to tags. Handle list items by repeating child tags.\n3. Generate realistic content: Use plausible names, dates, and values appropriate for the domain.\n4. Structure the data: Organize the fields logically with appropriate nesting and hierarchies.\n5. Serialize: Output the final string making sure to escape special characters if necessary.\n\nOutput:\n<?xml version=\"1.0\" encoding=\"UTF-8\"?>\n<product_listing>\n<product>company</product>\n<price>19.06</price>\n<stock>22</stock>\n<meta_info>\n<generated_at>2026-01-23T05:23:24.388067</generated_at>\n<version>v1.0</version>\n<tags>public</tags>\n<tags>sort</tags>\n<tags>maybe</tags>\n<tags>side</tags>\n<tags>PM</tags>\n</meta_info>\n<audit_log>\n<action>created</action>\n<by>leonliu</by>\n<timestamp>1980-07-08T11:08:17.939494</timestamp>\n</audit_log>\n</product_listing>"
}
] | U10BEI_BASE |
[
{
"role": "system",
"content": "You must NOT call tools.\nDo NOT include any extra text, code fences, or markdown.\nDo NOT output JSON or YAML.\nOutput MUST be valid TOML.\nThe first line MUST be: [result]\nThen include status and confidence keys under [result].\nThen include [result.fields].\nOutput ONLY TOML as plain text.\n"
},
{
"role": "user",
"content": "Generate a valid TOML output using the fixed template.\nConstraints:\n- Output ONLY TOML, no extra text.\n- The first line must be [result].\n- Include [result.fields] with all required keys.\nRequired keys in fields: id,name,note,flag,score,category,empty\nValues:\n- id: 1068\n- name: item_1068\n- note: ok\n- flag: True\n- score: 0.68\n- category: A\n- empty: \n"
},
{
"role": "assistant",
"content": "[result]\nstatus = \"ok\"\nconfidence = 0.800\n\n[result.fields]\n\"category\" = \"A\"\n\"empty\" = \"\"\n\"flag\" = true\n\"id\" = 1068\n\"name\" = \"item_1068\"\n\"note\" = \"ok\"\n\"score\" = 0.68"
}
] | TOML_GEN_FIXED_VAR |
[
{
"role": "system",
"content": "You must NOT call tools.\nDo NOT include any extra text, code fences, or markdown.\nDo NOT output JSON or YAML.\nOutput MUST be valid TOML.\nThe first line MUST be: [result]\nThen include status and confidence keys under [result].\nThen include [result.fields].\nOutput ONLY TOML as plain text.\n"
},
{
"role": "user",
"content": "Please convert the following JSON code to TOML code.\n\n<code>{\n \"contract\": \"orchestrate front-end schemas\",\n \"parties\": [\n \"Nguyen, Gilbert and Rodriguez\",\n \"Brown, Manning and Glass\"\n ],\n \"signed\": false,\n \"meta_info\": {\n \"generated_at\": \"2026-01-23T05:23:24.090610\",\n \"version\": \"v4.0\",\n \"tags\": [\n \"away\",\n \"benefit\",\n \"many\",\n \"from\",\n \"sure\"\n ]\n },\n \"audit_log\": [\n {\n \"action\": \"created\",\n \"by\": \"zkennedy\",\n \"timestamp\": \"1987-04-06T14:27:02.808851\"\n },\n {\n \"action\": \"created\",\n \"by\": \"sramos\",\n \"timestamp\": \"1985-09-29T11:43:59.919553\"\n }\n ]\n}</code>"
},
{
"role": "assistant",
"content": "[result]\nstatus = \"ok\"\nconfidence = 0.800\n\n[result.fields]\n\"audit_log\" = \"[{\\\"action\\\": \\\"created\\\", \\\"by\\\": \\\"zkennedy\\\", \\\"timestamp\\\": \\\"1987-04-06T14:27:02.808851\\\"}, {\\\"action\\\": \\\"created\\\", \\\"by\\\": \\\"sramos\\\", \\\"timestamp\\\": \\\"1985-09-29T11:43:59.919553\\\"}]\"\n\"contract\" = \"orchestrate front-end schemas\"\n\"meta_info\" = \"{\\\"generated_at\\\": \\\"2026-01-23T05:23:24.090610\\\", \\\"version\\\": \\\"v4.0\\\", \\\"tags\\\": [\\\"away\\\", \\\"benefit\\\", \\\"many\\\", \\\"from\\\", \\\"sure\\\"]}\"\n\"parties\" = \"[\\\"Nguyen, Gilbert and Rodriguez\\\", \\\"Brown, Manning and Glass\\\"]\"\n\"signed\" = false"
}
] | CONV_JSON2TOML_TPLFIX |
[
{
"role": "system",
"content": "You must NOT call tools.\nDo NOT include any extra text, code fences, or markdown.\nDo NOT output JSON or YAML.\nOutput MUST be valid TOML.\nThe first line MUST be: [result]\nThen include status and confidence keys under [result].\nThen include [result.fields].\nOutput ONLY TOML as plain text.\n"
},
{
"role": "user",
"content": "Please convert the following JSON code to TOML code.\n\n<code>{\n \"contract\": \"re-contextualize B2B channels\",\n \"parties\": [\n \"Tanner, Durham and Stokes\",\n \"Fuller LLC\"\n ],\n \"signed\": false,\n \"meta_info\": {\n \"generated_at\": \"2026-01-23T05:23:24.678156\",\n \"version\": \"v4.0\",\n \"tags\": [\n \"question\",\n \"determine\",\n \"election\",\n \"white\",\n \"wish\"\n ]\n },\n \"audit_log\": [\n {\n \"action\": \"created\",\n \"by\": \"dawn53\",\n \"timestamp\": \"2003-01-22T03:53:43.594201\"\n },\n {\n \"action\": \"created\",\n \"by\": \"jonathan57\",\n \"timestamp\": \"1987-06-24T22:35:20.626592\"\n },\n {\n \"action\": \"created\",\n \"by\": \"odaugherty\",\n \"timestamp\": \"1999-03-03T22:26:11.425318\"\n }\n ]\n}</code>"
},
{
"role": "assistant",
"content": "[result]\nstatus = \"ok\"\nconfidence = 0.800\n\n[result.fields]\n\"audit_log\" = \"[{\\\"action\\\": \\\"created\\\", \\\"by\\\": \\\"dawn53\\\", \\\"timestamp\\\": \\\"2003-01-22T03:53:43.594201\\\"}, {\\\"action\\\": \\\"created\\\", \\\"by\\\": \\\"jonathan57\\\", \\\"timestamp\\\": \\\"1987-06-24T22:35:20.626592\\\"}, {\\\"action\\\": \\\"created\\\", \\\"by\\\": \\\"odaugherty\\\", \\\"timestamp\\\": \\\"1999-03-03T22:26:11.425318\\\"}]\"\n\"contract\" = \"re-contextualize B2B channels\"\n\"meta_info\" = \"{\\\"generated_at\\\": \\\"2026-01-23T05:23:24.678156\\\", \\\"version\\\": \\\"v4.0\\\", \\\"tags\\\": [\\\"question\\\", \\\"determine\\\", \\\"election\\\", \\\"white\\\", \\\"wish\\\"]}\"\n\"parties\" = \"[\\\"Tanner, Durham and Stokes\\\", \\\"Fuller LLC\\\"]\"\n\"signed\" = false"
}
] | CONV_JSON2TOML_TPLFIX |
[
{
"role": "system",
"content": "You must NOT call tools.\nDo NOT include any extra text, code fences, or markdown.\nDo NOT output JSON or YAML.\nOutput MUST be valid TOML.\nThe first line MUST be: [result]\nThen include status and confidence keys under [result].\nThen include [result.fields].\nOutput ONLY TOML as plain text.\n"
},
{
"role": "user",
"content": "Please convert the following JSON code to TOML code.\n\n<code>{\n \"total\": {\n \"particularly\": false,\n \"wind_1\": 1.02,\n \"summer_2\": 46,\n \"evening_3\": 26,\n \"about_4\": 2\n },\n \"data_1\": [\n 97\n ],\n \"party_2\": {\n \"economy\": \"someone\",\n \"crime_1\": null\n },\n \"beautiful_3\": [\n 92,\n \"idea\",\n 9\n ],\n \"direction_4\": [\n 30,\n true\n ],\n \"meta_info\": {\n \"generated_at\": \"2026-01-23T05:23:24.788801\",\n \"version\": \"v2.0\",\n \"tags\": [\n \"talk\",\n \"crime\",\n \"never\",\n \"water\",\n \"section\"\n ]\n },\n \"audit_log\": [\n {\n \"action\": \"created\",\n \"by\": \"jdouglas\",\n \"timestamp\": \"2003-03-09T15:38:06.701850\"\n },\n {\n \"action\": \"created\",\n \"by\": \"kristy85\",\n \"timestamp\": \"1971-06-29T19:40:11.877869\"\n },\n {\n \"action\": \"created\",\n \"by\": \"belljohn\",\n \"timestamp\": \"2009-02-24T04:20:41.539267\"\n }\n ]\n}</code>"
},
{
"role": "assistant",
"content": "[result]\nstatus = \"ok\"\nconfidence = 0.800\n\n[result.fields]\n\"audit_log\" = \"[{\\\"action\\\": \\\"created\\\", \\\"by\\\": \\\"jdouglas\\\", \\\"timestamp\\\": \\\"2003-03-09T15:38:06.701850\\\"}, {\\\"action\\\": \\\"created\\\", \\\"by\\\": \\\"kristy85\\\", \\\"timestamp\\\": \\\"1971-06-29T19:40:11.877869\\\"}, {\\\"action\\\": \\\"created\\\", \\\"by\\\": \\\"belljohn\\\", \\\"timestamp\\\": \\\"2009-02-24T04:20:41.539267\\\"}]\"\n\"beautiful_3\" = \"[92, \\\"idea\\\", 9]\"\n\"data_1\" = \"[97]\"\n\"direction_4\" = \"[30, true]\"\n\"meta_info\" = \"{\\\"generated_at\\\": \\\"2026-01-23T05:23:24.788801\\\", \\\"version\\\": \\\"v2.0\\\", \\\"tags\\\": [\\\"talk\\\", \\\"crime\\\", \\\"never\\\", \\\"water\\\", \\\"section\\\"]}\"\n\"party_2\" = \"{\\\"economy\\\": \\\"someone\\\", \\\"crime_1\\\": null}\"\n\"total\" = \"{\\\"particularly\\\": false, \\\"wind_1\\\": 1.02, \\\"summer_2\\\": 46, \\\"evening_3\\\": 26, \\\"about_4\\\": 2}\""
}
] | CONV_JSON2TOML_TPLFIX |
[
{
"role": "system",
"content": "You must NOT call tools.\nDo NOT include any extra text, code fences, or markdown.\nDo NOT output JSON or YAML.\nOutput MUST be valid TOML.\nThe first line MUST be: [result]\nThen include status and confidence keys under [result].\nThen include [result.fields].\nOutput ONLY TOML as plain text.\n"
},
{
"role": "user",
"content": "Please convert the following JSON code to TOML code.\n\n<code>{\n \"student\": \"Mark Coleman\",\n \"score\": 16,\n \"grade\": \"B\",\n \"meta_info\": {\n \"generated_at\": \"2026-01-23T05:23:25.209787\",\n \"version\": \"v4.0\",\n \"tags\": [\n \"PM\",\n \"wall\",\n \"cup\",\n \"land\",\n \"face\"\n ]\n },\n \"audit_log\": [\n {\n \"action\": \"created\",\n \"by\": \"taylorwhite\",\n \"timestamp\": \"2009-02-18T00:21:02.288646\"\n },\n {\n \"action\": \"created\",\n \"by\": \"richard32\",\n \"timestamp\": \"2010-05-04T08:50:09.926227\"\n },\n {\n \"action\": \"created\",\n \"by\": \"michaeldodson\",\n \"timestamp\": \"1972-09-20T19:37:41.771338\"\n }\n ]\n}</code>"
},
{
"role": "assistant",
"content": "[result]\nstatus = \"ok\"\nconfidence = 0.800\n\n[result.fields]\n\"audit_log\" = \"[{\\\"action\\\": \\\"created\\\", \\\"by\\\": \\\"taylorwhite\\\", \\\"timestamp\\\": \\\"2009-02-18T00:21:02.288646\\\"}, {\\\"action\\\": \\\"created\\\", \\\"by\\\": \\\"richard32\\\", \\\"timestamp\\\": \\\"2010-05-04T08:50:09.926227\\\"}, {\\\"action\\\": \\\"created\\\", \\\"by\\\": \\\"michaeldodson\\\", \\\"timestamp\\\": \\\"1972-09-20T19:37:41.771338\\\"}]\"\n\"grade\" = \"B\"\n\"meta_info\" = \"{\\\"generated_at\\\": \\\"2026-01-23T05:23:25.209787\\\", \\\"version\\\": \\\"v4.0\\\", \\\"tags\\\": [\\\"PM\\\", \\\"wall\\\", \\\"cup\\\", \\\"land\\\", \\\"face\\\"]}\"\n\"score\" = 16\n\"student\" = \"Mark Coleman\""
}
] | CONV_JSON2TOML_TPLFIX |
[
{
"role": "system",
"content": "You are an expert in YAML format. Analyze the data structure and generate syntactically perfect YAML."
},
{
"role": "user",
"content": "Generate random structured data data in YAML format."
},
{
"role": "assistant",
"content": "Approach:\n1. Define the schema: The random_structured_data should contain a dictionary with 3 root fields (animal, rich_1, stand_2) and a nesting depth of 3.\n2. Select the format: YAML is chosen. Use indentation to represent nesting. lists are denoted by hyphens.\n3. Generate realistic content: Use plausible names, dates, and values appropriate for the domain.\n4. Structure the data: Organize the fields logically with appropriate nesting and hierarchies.\n5. Serialize: Output the final string making sure to escape special characters if necessary.\n\nOutput:\nanimal:\n create:\n - 3.0\nrich_1:\n listen:\n - false\n - 3.83\nstand_2:\n each: half\n great_1:\n why: 68\n song_2:\n building: false\n true_1: false\n development_2: null\n range_3: 1.51\n up_3: against\n"
}
] | U10BEI_BASE |
[
{
"role": "system",
"content": "You are an expert in JSON format. Analyze the data structure and generate syntactically perfect JSON."
},
{
"role": "user",
"content": "Create a JSON representation of contract."
},
{
"role": "assistant",
"content": "Approach:\n1. Define the schema: The contract should contain a dictionary with 3 root fields (contract, parties, signed) and a nesting depth of 2.\n2. Select the format: JSON is chosen. Ensure all keys are quoted and nested structures use curly braces.\n3. Generate realistic content: Use plausible names, dates, and values appropriate for the domain.\n4. Structure the data: Organize the fields logically with appropriate nesting and hierarchies.\n5. Serialize: Output the final string making sure to escape special characters if necessary.\n\nOutput:\n{\n \"contract\": \"re-intermediate virtual networks\",\n \"parties\": [\n \"Floyd Group\",\n \"Powell-Huff\"\n ],\n \"signed\": true\n}"
}
] | U10BEI_BASE |
[
{
"role": "system",
"content": "You must NOT call tools.\nDo NOT include any extra text, code fences, or markdown.\nDo NOT output JSON or YAML.\nOutput MUST be valid TOML.\nThe first line MUST be: [result]\nThen include status and confidence keys under [result].\nThen include [result.fields].\nOutput ONLY TOML as plain text.\n"
},
{
"role": "user",
"content": "Please convert the following JSON code to TOML code.\n\n<code>{\n \"student\": \"Timothy Sanchez\",\n \"score\": 80,\n \"grade\": \"A\",\n \"meta_info\": {\n \"generated_at\": \"2026-01-23T05:23:23.542922\",\n \"version\": \"v5.0\",\n \"tags\": [\n \"kitchen\",\n \"education\",\n \"form\",\n \"offer\",\n \"suggest\"\n ]\n },\n \"audit_log\": [\n {\n \"action\": \"created\",\n \"by\": \"mstephenson\",\n \"timestamp\": \"2003-03-19T02:15:49.572356\"\n },\n {\n \"action\": \"created\",\n \"by\": \"wclarke\",\n \"timestamp\": \"2008-04-07T10:46:49.583850\"\n },\n {\n \"action\": \"created\",\n \"by\": \"bryan25\",\n \"timestamp\": \"1996-11-20T06:37:43.971508\"\n }\n ]\n}</code>"
},
{
"role": "assistant",
"content": "[result]\nstatus = \"ok\"\nconfidence = 0.800\n\n[result.fields]\n\"audit_log\" = \"[{\\\"action\\\": \\\"created\\\", \\\"by\\\": \\\"mstephenson\\\", \\\"timestamp\\\": \\\"2003-03-19T02:15:49.572356\\\"}, {\\\"action\\\": \\\"created\\\", \\\"by\\\": \\\"wclarke\\\", \\\"timestamp\\\": \\\"2008-04-07T10:46:49.583850\\\"}, {\\\"action\\\": \\\"created\\\", \\\"by\\\": \\\"bryan25\\\", \\\"timestamp\\\": \\\"1996-11-20T06:37:43.971508\\\"}]\"\n\"grade\" = \"A\"\n\"meta_info\" = \"{\\\"generated_at\\\": \\\"2026-01-23T05:23:23.542922\\\", \\\"version\\\": \\\"v5.0\\\", \\\"tags\\\": [\\\"kitchen\\\", \\\"education\\\", \\\"form\\\", \\\"offer\\\", \\\"suggest\\\"]}\"\n\"score\" = 80\n\"student\" = \"Timothy Sanchez\""
}
] | CONV_JSON2TOML_TPLFIX |
[
{
"role": "system",
"content": "You must NOT call tools.\nDo NOT include any extra text, code fences, or markdown.\nDo NOT output JSON or YAML.\nOutput MUST be valid TOML.\nThe first line MUST be: [result]\nThen include status and confidence keys under [result].\nThen include [result.fields].\nOutput ONLY TOML as plain text.\n"
},
{
"role": "user",
"content": "Please convert the following JSON code to TOML code.\n\n<code>{\n \"endpoint\": \"/api/v1/users\",\n \"method\": \"GET\",\n \"responses\": {\n \"200\": \"OK\"\n },\n \"meta_info\": {\n \"generated_at\": \"2026-01-23T05:23:23.804235\",\n \"version\": \"v4.0\",\n \"tags\": [\n \"building\",\n \"meeting\",\n \"half\",\n \"high\",\n \"leader\"\n ]\n },\n \"audit_log\": [\n {\n \"action\": \"created\",\n \"by\": \"wrightricky\",\n \"timestamp\": \"1986-01-26T08:37:37.622665\"\n }\n ]\n}</code>"
},
{
"role": "assistant",
"content": "[result]\nstatus = \"ok\"\nconfidence = 0.800\n\n[result.fields]\n\"audit_log\" = \"[{\\\"action\\\": \\\"created\\\", \\\"by\\\": \\\"wrightricky\\\", \\\"timestamp\\\": \\\"1986-01-26T08:37:37.622665\\\"}]\"\n\"endpoint\" = \"/api/v1/users\"\n\"meta_info\" = \"{\\\"generated_at\\\": \\\"2026-01-23T05:23:23.804235\\\", \\\"version\\\": \\\"v4.0\\\", \\\"tags\\\": [\\\"building\\\", \\\"meeting\\\", \\\"half\\\", \\\"high\\\", \\\"leader\\\"]}\"\n\"method\" = \"GET\"\n\"responses\" = \"{\\\"200\\\": \\\"OK\\\"}\""
}
] | CONV_JSON2TOML_TPLFIX |
[
{
"role": "system",
"content": "You are an expert in TOML format. Analyze the data structure and generate syntactically perfect TOML."
},
{
"role": "user",
"content": "Convert the following XML to TOML:\n<?xml version=\"1.0\" encoding=\"UTF-8\"?>\n<financial_report>\n<company>Brooks LLC</company>\n<revenue>176100</revenue>\n<profit>99846</profit>\n</financial_report>"
},
{
"role": "assistant",
"content": "Approach:\n1. Analyze the input XML structure: It represents a dictionary with 3 root fields (company, revenue, profit) and a nesting depth of 1.\n2. Identify the target format requirements: TOML requires strict syntax. Define hierarchy using dot-notation or [section] headers.\n3. Map the fields: The root keys will map directly, while nested objects need careful formatting.\n4. Handle data types: Ensure strings, numbers, and booleans are formatted correctly for the target.\n5. Construct the output: serialized the data strictly following the target schema.\n\nOutput:\ncompany = \"Brooks LLC\"\nrevenue = 176100\nprofit = 99846\n"
}
] | U10BEI_BASE |
[
{
"role": "system",
"content": "You must NOT call tools.\nDo NOT include any extra text, code fences, or markdown.\nDo NOT output JSON or YAML.\nOutput MUST be valid TOML.\nThe first line MUST be: [result]\nThen include status and confidence keys under [result].\nThen include [result.fields].\nOutput ONLY TOML as plain text.\n"
},
{
"role": "user",
"content": "Please convert the following JSON code to TOML code.\n\n<code>{\n \"and\": \"sign\",\n \"scene_1\": {\n \"way\": {\n \"religious\": 97,\n \"onto_1\": null,\n \"perform_2\": false,\n \"hundred_3\": 13\n },\n \"down_1\": [\n \"deep\",\n false,\n true\n ]\n },\n \"drug_2\": {\n \"smile\": \"just\",\n \"take_1\": \"song\"\n },\n \"product_3\": \"fine\",\n \"against_4\": \"well\"\n}</code>"
},
{
"role": "assistant",
"content": "[result]\nstatus = \"ok\"\nconfidence = 0.800\n\n[result.fields]\n\"against_4\" = \"well\"\n\"and\" = \"sign\"\n\"drug_2\" = \"{\\\"smile\\\": \\\"just\\\", \\\"take_1\\\": \\\"song\\\"}\"\n\"product_3\" = \"fine\"\n\"scene_1\" = \"{\\\"way\\\": {\\\"religious\\\": 97, \\\"onto_1\\\": null, \\\"perform_2\\\": false, \\\"hundred_3\\\": 13}, \\\"down_1\\\": [\\\"deep\\\", false, true]}\""
}
] | CONV_JSON2TOML_TPLFIX |
[
{
"role": "system",
"content": "You are an expert in XML format. Analyze the data structure and generate syntactically perfect XML."
},
{
"role": "user",
"content": "Generate product listing data in XML format."
},
{
"role": "assistant",
"content": "Approach:\n1. Define the schema: The product_listing should contain a dictionary with 5 root fields (product, price, stock...) and a nesting depth of 3.\n2. Select the format: XML is chosen. Map dictionary keys to tags. Handle list items by repeating child tags.\n3. Generate realistic content: Use plausible names, dates, and values appropriate for the domain.\n4. Structure the data: Organize the fields logically with appropriate nesting and hierarchies.\n5. Serialize: Output the final string making sure to escape special characters if necessary.\n\nOutput:\n<?xml version=\"1.0\" encoding=\"UTF-8\"?>\n<product_listing>\n<product>blue</product>\n<price>98.17</price>\n<stock>24</stock>\n<meta_info>\n<generated_at>2026-01-23T05:23:23.475727</generated_at>\n<version>v1.0</version>\n<tags>rate</tags>\n<tags>then</tags>\n<tags>item</tags>\n<tags>there</tags>\n<tags>play</tags>\n</meta_info>\n<audit_log>\n<action>created</action>\n<by>jason87</by>\n<timestamp>1975-10-26T14:02:52.131626</timestamp>\n</audit_log>\n</product_listing>"
}
] | U10BEI_BASE |
[
{
"role": "system",
"content": "You must NOT call tools.\nDo NOT include any extra text, code fences, or markdown.\nDo NOT output JSON or YAML.\nOutput MUST be valid TOML.\nThe first line MUST be: [result]\nThen include status and confidence keys under [result].\nThen include [result.fields].\nOutput ONLY TOML as plain text.\n"
},
{
"role": "user",
"content": "Please convert the following JSON code to TOML code.\n\n<code>{\n \"user\": \"robertoliver\",\n \"rating\": 5,\n \"comment\": \"Player dog ground share.\",\n \"meta_info\": {\n \"generated_at\": \"2026-01-23T05:23:24.483041\",\n \"version\": \"v1.0\",\n \"tags\": [\n \"food\",\n \"discuss\",\n \"on\",\n \"but\",\n \"machine\"\n ]\n },\n \"audit_log\": [\n {\n \"action\": \"created\",\n \"by\": \"ngray\",\n \"timestamp\": \"2009-03-14T04:10:42.701861\"\n },\n {\n \"action\": \"created\",\n \"by\": \"anthony18\",\n \"timestamp\": \"2003-06-25T07:17:43.018347\"\n },\n {\n \"action\": \"created\",\n \"by\": \"michael93\",\n \"timestamp\": \"2023-03-02T23:12:59.776379\"\n }\n ]\n}</code>"
},
{
"role": "assistant",
"content": "[result]\nstatus = \"ok\"\nconfidence = 0.800\n\n[result.fields]\n\"audit_log\" = \"[{\\\"action\\\": \\\"created\\\", \\\"by\\\": \\\"ngray\\\", \\\"timestamp\\\": \\\"2009-03-14T04:10:42.701861\\\"}, {\\\"action\\\": \\\"created\\\", \\\"by\\\": \\\"anthony18\\\", \\\"timestamp\\\": \\\"2003-06-25T07:17:43.018347\\\"}, {\\\"action\\\": \\\"created\\\", \\\"by\\\": \\\"michael93\\\", \\\"timestamp\\\": \\\"2023-03-02T23:12:59.776379\\\"}]\"\n\"comment\" = \"Player dog ground share.\"\n\"meta_info\" = \"{\\\"generated_at\\\": \\\"2026-01-23T05:23:24.483041\\\", \\\"version\\\": \\\"v1.0\\\", \\\"tags\\\": [\\\"food\\\", \\\"discuss\\\", \\\"on\\\", \\\"but\\\", \\\"machine\\\"]}\"\n\"rating\" = 5\n\"user\" = \"robertoliver\""
}
] | CONV_JSON2TOML_TPLFIX |
[
{
"role": "system",
"content": "You must NOT call tools.\nDo NOT include any extra text, code fences, or markdown.\nDo NOT output JSON or YAML.\nOutput MUST be valid TOML.\nThe first line MUST be: [result]\nThen include status and confidence keys under [result].\nThen include [result.fields].\nOutput ONLY TOML as plain text.\n"
},
{
"role": "user",
"content": "Please convert the following JSON code to TOML code.\n\n<code>{\n \"course\": \"Optional static paradigm\",\n \"weeks\": 12,\n \"credits\": 3,\n \"meta_info\": {\n \"generated_at\": \"2026-01-23T05:23:24.635785\",\n \"version\": \"v4.0\",\n \"tags\": [\n \"identify\",\n \"toward\",\n \"several\",\n \"town\",\n \"much\"\n ]\n },\n \"audit_log\": [\n {\n \"action\": \"created\",\n \"by\": \"mcooper\",\n \"timestamp\": \"1983-12-19T02:11:40.611403\"\n },\n {\n \"action\": \"created\",\n \"by\": \"kristinapowell\",\n \"timestamp\": \"1994-09-09T08:55:23.146594\"\n },\n {\n \"action\": \"created\",\n \"by\": \"griffinjustin\",\n \"timestamp\": \"1976-07-14T00:27:49.479468\"\n }\n ]\n}</code>"
},
{
"role": "assistant",
"content": "[result]\nstatus = \"ok\"\nconfidence = 0.800\n\n[result.fields]\n\"audit_log\" = \"[{\\\"action\\\": \\\"created\\\", \\\"by\\\": \\\"mcooper\\\", \\\"timestamp\\\": \\\"1983-12-19T02:11:40.611403\\\"}, {\\\"action\\\": \\\"created\\\", \\\"by\\\": \\\"kristinapowell\\\", \\\"timestamp\\\": \\\"1994-09-09T08:55:23.146594\\\"}, {\\\"action\\\": \\\"created\\\", \\\"by\\\": \\\"griffinjustin\\\", \\\"timestamp\\\": \\\"1976-07-14T00:27:49.479468\\\"}]\"\n\"course\" = \"Optional static paradigm\"\n\"credits\" = 3\n\"meta_info\" = \"{\\\"generated_at\\\": \\\"2026-01-23T05:23:24.635785\\\", \\\"version\\\": \\\"v4.0\\\", \\\"tags\\\": [\\\"identify\\\", \\\"toward\\\", \\\"several\\\", \\\"town\\\", \\\"much\\\"]}\"\n\"weeks\" = 12"
}
] | CONV_JSON2TOML_TPLFIX |
[
{
"role": "system",
"content": "You are an expert in XML format. Analyze the data structure and generate syntactically perfect XML."
},
{
"role": "user",
"content": "Convert the following YAML to XML:\nproduct: language\nprice: 62.14\nstock: 16\n"
},
{
"role": "assistant",
"content": "Approach:\n1. Analyze the input YAML structure: It represents a dictionary with 3 root fields (product, price, stock) and a nesting depth of 1.\n2. Identify the target format requirements: XML requires strict syntax. Map dictionary keys to tags. Handle list items by repeating child tags.\n3. Map the fields: The root keys will map directly, while nested objects need careful formatting.\n4. Handle data types: Ensure strings, numbers, and booleans are formatted correctly for the target.\n5. Construct the output: serialized the data strictly following the target schema.\n\nOutput:\n<?xml version=\"1.0\" encoding=\"UTF-8\"?>\n<product_listing>\n<product>language</product>\n<price>62.14</price>\n<stock>16</stock>\n</product_listing>"
}
] | U10BEI_BASE |
[
{
"role": "system",
"content": "You are an expert in CSV format. Analyze the data structure and generate syntactically perfect CSV."
},
{
"role": "user",
"content": "Generate financial report data in CSV format."
},
{
"role": "assistant",
"content": "Approach:\n1. Define the schema: The financial_report should contain a dictionary with 3 root fields (company, revenue, profit) and a nesting depth of 1.\n2. Select the format: CSV is chosen. Flatten the hierarchical structure. Use dot notation for nested keys in the header.\n3. Generate realistic content: Use plausible names, dates, and values appropriate for the domain.\n4. Structure the data: Organize the fields logically with appropriate nesting and hierarchies.\n5. Serialize: Output the final string making sure to escape special characters if necessary.\n\nOutput:\ncompany,revenue,profit\r\nMckee PLC,821695,2449\r\n"
}
] | U10BEI_BASE |
[
{
"role": "system",
"content": "You are an expert in JSON format. Analyze the data structure and generate syntactically perfect JSON."
},
{
"role": "user",
"content": "Convert the following CSV to JSON:\ntxn,amount,status\r\nbad3f73f-d4f6-4ef7-bb5a-b151c15e2023,428.34,completed\r\n"
},
{
"role": "assistant",
"content": "Approach:\n1. Analyze the input CSV structure: It represents a dictionary with 3 root fields (txn, amount, status) and a nesting depth of 1.\n2. Identify the target format requirements: JSON requires strict syntax. Ensure all keys are quoted and nested structures use curly braces.\n3. Map the fields: The root keys will map directly, while nested objects need careful formatting.\n4. Handle data types: Ensure strings, numbers, and booleans are formatted correctly for the target.\n5. Construct the output: serialized the data strictly following the target schema.\n\nOutput:\n{\n \"txn\": \"bad3f73f-d4f6-4ef7-bb5a-b151c15e2023\",\n \"amount\": 428.34,\n \"status\": \"completed\"\n}"
}
] | U10BEI_BASE |
[
{
"role": "system",
"content": "You are an expert in JSON format. Analyze the data structure and generate syntactically perfect JSON."
},
{
"role": "user",
"content": "Convert the following YAML to JSON:\nid: c2c86644-17e8-4cc3-833a-ba185b9dccb0\npatient: Ann Howard\ndiagnosis: in\nnotes: Relationship kind citizen former account ago. Campaign level him dream control.\n"
},
{
"role": "assistant",
"content": "Approach:\n1. Analyze the input YAML structure: It represents a dictionary with 4 root fields (id, patient, diagnosis...) and a nesting depth of 1.\n2. Identify the target format requirements: JSON requires strict syntax. Ensure all keys are quoted and nested structures use curly braces.\n3. Map the fields: The root keys will map directly, while nested objects need careful formatting.\n4. Handle data types: Ensure strings, numbers, and booleans are formatted correctly for the target.\n5. Construct the output: serialized the data strictly following the target schema.\n\nOutput:\n{\n \"id\": \"c2c86644-17e8-4cc3-833a-ba185b9dccb0\",\n \"patient\": \"Ann Howard\",\n \"diagnosis\": \"in\",\n \"notes\": \"Relationship kind citizen former account ago. Campaign level him dream control.\"\n}"
}
] | U10BEI_BASE |
[
{
"role": "system",
"content": "You must NOT call tools.\nDo NOT include any extra text, code fences, or markdown.\nFollow the user instructions EXACTLY.\nOutput ONLY CSV content as plain text.\nStart with the header row.\n"
},
{
"role": "user",
"content": "Please output CSV code(Only output the CSV content as plain text, starting with the header row.).\n\nTask:\nGiven a recipe description, extract detailed structured information about the recipe into CSV format.\nFeature Requirements:\n1. Use the exact header fields and order.\n2. Output exactly one data row.\n3. Output exactly two lines: header line + one row.\n\nHeader:\nrecipe_id,recipe_name,chef_name,cuisine_type,preparation_time_minutes,cooking_time_minutes,servings,difficulty_level,main_ingredient,ingredient_list,ingredient_quantities,cooking_method,calories_per_serving,allergen_warnings,vegetarian_friendly,seasonality"
},
{
"role": "assistant",
"content": "recipe_id,recipe_name,chef_name,cuisine_type,preparation_time_minutes,cooking_time_minutes,servings,difficulty_level,main_ingredient,ingredient_list,ingredient_quantities,cooking_method,calories_per_serving,allergen_warnings,vegetarian_friendly,seasonality\nR00857,Stellar Stew 857,Mira Qor,Galactic,5,47,2,Easy,Star Salt,nebula_beans;star_salt;quantum_onion,200g;5g;1,simmer,699,nuts,No,Winter"
}
] | CSV_TWO_LINE_FIXED |
[
{
"role": "system",
"content": "You must NOT call tools.\nDo NOT include any extra text, code fences, or markdown.\nDo NOT output JSON or YAML.\nOutput MUST be valid TOML.\nThe first line MUST be: [result]\nThen include status and confidence keys under [result].\nThen include [result.fields].\nOutput ONLY TOML as plain text.\n"
},
{
"role": "user",
"content": "Please convert the following JSON code to TOML code.\n\n<code>{\n \"company\": \"Morgan PLC\",\n \"revenue\": 472910,\n \"profit\": 69461,\n \"meta_info\": {\n \"generated_at\": \"2026-01-23T05:23:23.822824\",\n \"version\": \"v1.0\",\n \"tags\": [\n \"Democrat\",\n \"anyone\",\n \"plan\",\n \"series\",\n \"accept\"\n ]\n },\n \"audit_log\": [\n {\n \"action\": \"created\",\n \"by\": \"robertlopez\",\n \"timestamp\": \"1990-08-05T18:35:53.366661\"\n }\n ]\n}</code>"
},
{
"role": "assistant",
"content": "[result]\nstatus = \"ok\"\nconfidence = 0.800\n\n[result.fields]\n\"audit_log\" = \"[{\\\"action\\\": \\\"created\\\", \\\"by\\\": \\\"robertlopez\\\", \\\"timestamp\\\": \\\"1990-08-05T18:35:53.366661\\\"}]\"\n\"company\" = \"Morgan PLC\"\n\"meta_info\" = \"{\\\"generated_at\\\": \\\"2026-01-23T05:23:23.822824\\\", \\\"version\\\": \\\"v1.0\\\", \\\"tags\\\": [\\\"Democrat\\\", \\\"anyone\\\", \\\"plan\\\", \\\"series\\\", \\\"accept\\\"]}\"\n\"profit\" = 69461\n\"revenue\" = 472910"
}
] | CONV_JSON2TOML_TPLFIX |
[
{
"role": "system",
"content": "You are an expert in JSON format. Analyze the data structure and generate syntactically perfect JSON."
},
{
"role": "user",
"content": "Generate experiment result data in JSON format."
},
{
"role": "assistant",
"content": "Approach:\n1. Define the schema: The experiment_result should contain a dictionary with 4 root fields (experiment_id, title, metrics...) and a nesting depth of 2.\n2. Select the format: JSON is chosen. Ensure all keys are quoted and nested structures use curly braces.\n3. Generate realistic content: Use plausible names, dates, and values appropriate for the domain.\n4. Structure the data: Organize the fields logically with appropriate nesting and hierarchies.\n5. Serialize: Output the final string making sure to escape special characters if necessary.\n\nOutput:\n{\n \"experiment_id\": \"EXP-603598\",\n \"title\": \"Analysis of effort\",\n \"metrics\": {\n \"accuracy\": 0.7761,\n \"loss\": 0.0694,\n \"epoch\": 36\n },\n \"config\": {\n \"batch_size\": 128,\n \"optimizer\": \"sgd\",\n \"lr\": 0.001\n }\n}"
}
] | U10BEI_BASE |
[
{
"role": "system",
"content": "You must NOT call tools.\nDo NOT include any extra text, code fences, or markdown.\nDo NOT output JSON or YAML.\nOutput MUST be valid TOML.\nThe first line MUST be: [result]\nThen include status and confidence keys under [result].\nThen include [result.fields].\nOutput ONLY TOML as plain text.\n"
},
{
"role": "user",
"content": "Generate a valid TOML output using the fixed template.\nConstraints:\n- Output ONLY TOML, no extra text.\n- The first line must be [result].\n- Include [result.fields] with all required keys.\nRequired keys in fields: id,name,note,flag,score,category,empty\nValues:\n- id: 1370\n- name: item_1370\n- note: line1\\nline2\n- flag: True\n- score: 0.7\n- category: C\n- empty: filled\n"
},
{
"role": "assistant",
"content": "[result]\nstatus = \"ok\"\nconfidence = 0.800\n\n[result.fields]\n\"category\" = \"C\"\n\"empty\" = \"filled\"\n\"flag\" = true\n\"id\" = 1370\n\"name\" = \"item_1370\"\n\"note\" = \"line1\\\\nline2\"\n\"score\" = 0.7"
}
] | TOML_GEN_FIXED_VAR |
[
{
"role": "system",
"content": "You must NOT call tools.\nDo NOT include any extra text, code fences, or markdown.\nDo NOT output JSON or YAML.\nOutput MUST be valid TOML.\nThe first line MUST be: [result]\nThen include status and confidence keys under [result].\nThen include [result.fields].\nOutput ONLY TOML as plain text.\n"
},
{
"role": "user",
"content": "Please convert the following JSON code to TOML code.\n\n<code>{\n \"company\": \"Huang LLC\",\n \"revenue\": 445146,\n \"profit\": 44757,\n \"meta_info\": {\n \"generated_at\": \"2026-01-23T05:23:23.967735\",\n \"version\": \"v3.0\",\n \"tags\": [\n \"pattern\",\n \"break\",\n \"form\",\n \"course\",\n \"pressure\"\n ]\n },\n \"audit_log\": [\n {\n \"action\": \"created\",\n \"by\": \"lynnedwards\",\n \"timestamp\": \"2022-09-23T21:58:41.642551\"\n },\n {\n \"action\": \"created\",\n \"by\": \"gharris\",\n \"timestamp\": \"2013-10-05T14:09:55.559145\"\n }\n ]\n}</code>"
},
{
"role": "assistant",
"content": "[result]\nstatus = \"ok\"\nconfidence = 0.800\n\n[result.fields]\n\"audit_log\" = \"[{\\\"action\\\": \\\"created\\\", \\\"by\\\": \\\"lynnedwards\\\", \\\"timestamp\\\": \\\"2022-09-23T21:58:41.642551\\\"}, {\\\"action\\\": \\\"created\\\", \\\"by\\\": \\\"gharris\\\", \\\"timestamp\\\": \\\"2013-10-05T14:09:55.559145\\\"}]\"\n\"company\" = \"Huang LLC\"\n\"meta_info\" = \"{\\\"generated_at\\\": \\\"2026-01-23T05:23:23.967735\\\", \\\"version\\\": \\\"v3.0\\\", \\\"tags\\\": [\\\"pattern\\\", \\\"break\\\", \\\"form\\\", \\\"course\\\", \\\"pressure\\\"]}\"\n\"profit\" = 44757\n\"revenue\" = 445146"
}
] | CONV_JSON2TOML_TPLFIX |
[
{
"role": "system",
"content": "You are an expert in JSON format. Analyze the data structure and generate syntactically perfect JSON."
},
{
"role": "user",
"content": "Generate contract data in JSON format."
},
{
"role": "assistant",
"content": "Approach:\n1. Define the schema: The contract should contain a dictionary with 5 root fields (contract, parties, signed...) and a nesting depth of 3.\n2. Select the format: JSON is chosen. Ensure all keys are quoted and nested structures use curly braces.\n3. Generate realistic content: Use plausible names, dates, and values appropriate for the domain.\n4. Structure the data: Organize the fields logically with appropriate nesting and hierarchies.\n5. Serialize: Output the final string making sure to escape special characters if necessary.\n\nOutput:\n{\n \"contract\": \"scale front-end experiences\",\n \"parties\": [\n \"Holloway Group\",\n \"Johnson, Bryant and White\"\n ],\n \"signed\": true,\n \"meta_info\": {\n \"generated_at\": \"2026-01-23T05:23:24.292856\",\n \"version\": \"v3.0\",\n \"tags\": [\n \"moment\",\n \"production\",\n \"education\",\n \"control\",\n \"leader\"\n ]\n },\n \"audit_log\": [\n {\n \"action\": \"created\",\n \"by\": \"hlong\",\n \"timestamp\": \"2018-10-11T10:02:42.738596\"\n },\n {\n \"action\": \"created\",\n \"by\": \"matthew69\",\n \"timestamp\": \"2023-11-06T17:36:50.842047\"\n }\n ]\n}"
}
] | U10BEI_BASE |
[
{
"role": "system",
"content": "You must NOT call tools.\nDo NOT include any extra text, code fences, or markdown.\nFollow the user instructions EXACTLY.\nOutput ONLY CSV content as plain text.\nStart with the header row.\n"
},
{
"role": "user",
"content": "Please output CSV code(Only output the CSV content as plain text, starting with the header row.).\n\nTask:\nGiven a recipe description, extract detailed structured information about the recipe into CSV format.\nFeature Requirements:\n1. Use the exact header fields and order.\n2. Output exactly one data row.\n3. Output exactly two lines: header line + one row.\n\nHeader:\nrecipe_id,recipe_name,chef_name,cuisine_type,preparation_time_minutes,cooking_time_minutes,servings,difficulty_level,main_ingredient,ingredient_list,ingredient_quantities,cooking_method,calories_per_serving,allergen_warnings,vegetarian_friendly,seasonality"
},
{
"role": "assistant",
"content": "recipe_id,recipe_name,chef_name,cuisine_type,preparation_time_minutes,cooking_time_minutes,servings,difficulty_level,main_ingredient,ingredient_list,ingredient_quantities,cooking_method,calories_per_serving,allergen_warnings,vegetarian_friendly,seasonality\nR00971,Stellar Stew 971,Mira Qor,Galactic,17,37,5,Medium,Nebula Beans,nebula_beans;star_salt;quantum_onion,200g;5g;1,fry,684,dairy,Yes,Winter"
}
] | CSV_TWO_LINE_FIXED |
[
{
"role": "system",
"content": "You must NOT call tools.\nDo NOT include any extra text, code fences, or markdown.\nDo NOT output JSON or YAML.\nOutput MUST be valid TOML.\nThe first line MUST be: [result]\nThen include status and confidence keys under [result].\nThen include [result.fields].\nOutput ONLY TOML as plain text.\n"
},
{
"role": "user",
"content": "Please convert the following JSON code to TOML code.\n\n<code>{\n \"paper_id\": \"arXiv:2020.64913\",\n \"title\": \"Upgradable interactive function\",\n \"authors\": [\n {\n \"name\": \"Victor Hernandez\",\n \"affiliation\": \"Richard, Rodriguez and Allen\",\n \"email\": \"brownallison@example.com\"\n },\n {\n \"name\": \"Angela Osborn\",\n \"affiliation\": \"Gonzalez-Lopez\",\n \"email\": \"qdavis@example.org\"\n }\n ],\n \"abstract\": \"Radio international total ago. Man speak professor. Girl green night human board. Agency soldier for finish.\",\n \"keywords\": [\n \"court\",\n \"computer\",\n \"possible\"\n ],\n \"publication\": {\n \"venue\": \"Science\",\n \"year\": 2024,\n \"doi\": \"10.1246/KUZy\"\n }\n}</code>"
},
{
"role": "assistant",
"content": "[result]\nstatus = \"ok\"\nconfidence = 0.800\n\n[result.fields]\n\"abstract\" = \"Radio international total ago. Man speak professor. Girl green night human board. Agency soldier for finish.\"\n\"authors\" = \"[{\\\"name\\\": \\\"Victor Hernandez\\\", \\\"affiliation\\\": \\\"Richard, Rodriguez and Allen\\\", \\\"email\\\": \\\"brownallison@example.com\\\"}, {\\\"name\\\": \\\"Angela Osborn\\\", \\\"affiliation\\\": \\\"Gonzalez-Lopez\\\", \\\"email\\\": \\\"qdavis@example.org\\\"}]\"\n\"keywords\" = \"[\\\"court\\\", \\\"computer\\\", \\\"possible\\\"]\"\n\"paper_id\" = \"arXiv:2020.64913\"\n\"publication\" = \"{\\\"venue\\\": \\\"Science\\\", \\\"year\\\": 2024, \\\"doi\\\": \\\"10.1246/KUZy\\\"}\"\n\"title\" = \"Upgradable interactive function\""
}
] | CONV_JSON2TOML_TPLFIX |
[
{
"role": "system",
"content": "You must NOT call tools.\nDo NOT include any extra text, code fences, or markdown.\nDo NOT output JSON or YAML.\nOutput MUST be valid TOML.\nThe first line MUST be: [result]\nThen include status and confidence keys under [result].\nThen include [result.fields].\nOutput ONLY TOML as plain text.\n"
},
{
"role": "user",
"content": "Please convert the following JSON code to TOML code.\n\n<code>{\n \"course\": \"Universal stable project\",\n \"weeks\": 11,\n \"credits\": 3,\n \"meta_info\": {\n \"generated_at\": \"2026-01-23T05:23:24.363558\",\n \"version\": \"v3.0\",\n \"tags\": [\n \"bar\",\n \"time\",\n \"yes\",\n \"yourself\",\n \"experience\"\n ]\n },\n \"audit_log\": [\n {\n \"action\": \"created\",\n \"by\": \"elizabethdaniels\",\n \"timestamp\": \"2011-12-29T01:41:46.979012\"\n }\n ]\n}</code>"
},
{
"role": "assistant",
"content": "[result]\nstatus = \"ok\"\nconfidence = 0.800\n\n[result.fields]\n\"audit_log\" = \"[{\\\"action\\\": \\\"created\\\", \\\"by\\\": \\\"elizabethdaniels\\\", \\\"timestamp\\\": \\\"2011-12-29T01:41:46.979012\\\"}]\"\n\"course\" = \"Universal stable project\"\n\"credits\" = 3\n\"meta_info\" = \"{\\\"generated_at\\\": \\\"2026-01-23T05:23:24.363558\\\", \\\"version\\\": \\\"v3.0\\\", \\\"tags\\\": [\\\"bar\\\", \\\"time\\\", \\\"yes\\\", \\\"yourself\\\", \\\"experience\\\"]}\"\n\"weeks\" = 11"
}
] | CONV_JSON2TOML_TPLFIX |
[
{
"role": "system",
"content": "You must NOT call tools.\nDo NOT include any extra text, code fences, or markdown.\nDo NOT output JSON or YAML.\nOutput MUST be valid TOML.\nThe first line MUST be: [result]\nThen include status and confidence keys under [result].\nThen include [result.fields].\nOutput ONLY TOML as plain text.\n"
},
{
"role": "user",
"content": "Please convert the following JSON code to TOML code.\n\n<code>{\n \"product\": \"do\",\n \"price\": 70.43,\n \"stock\": 0,\n \"meta_info\": {\n \"generated_at\": \"2026-01-23T05:23:23.441531\",\n \"version\": \"v5.0\",\n \"tags\": [\n \"book\",\n \"suddenly\",\n \"stay\",\n \"economic\",\n \"standard\"\n ]\n },\n \"audit_log\": [\n {\n \"action\": \"created\",\n \"by\": \"wilsonyvonne\",\n \"timestamp\": \"1991-11-07T15:38:28.707919\"\n }\n ]\n}</code>"
},
{
"role": "assistant",
"content": "[result]\nstatus = \"ok\"\nconfidence = 0.800\n\n[result.fields]\n\"audit_log\" = \"[{\\\"action\\\": \\\"created\\\", \\\"by\\\": \\\"wilsonyvonne\\\", \\\"timestamp\\\": \\\"1991-11-07T15:38:28.707919\\\"}]\"\n\"meta_info\" = \"{\\\"generated_at\\\": \\\"2026-01-23T05:23:23.441531\\\", \\\"version\\\": \\\"v5.0\\\", \\\"tags\\\": [\\\"book\\\", \\\"suddenly\\\", \\\"stay\\\", \\\"economic\\\", \\\"standard\\\"]}\"\n\"price\" = 70.43\n\"product\" = \"do\"\n\"stock\" = 0"
}
] | CONV_JSON2TOML_TPLFIX |
[
{
"role": "system",
"content": "You are an expert in XML format. Analyze the data structure and generate syntactically perfect XML."
},
{
"role": "user",
"content": "Convert the following CSV to XML:\nwriter,baby_1.serve,baby_1.soon_1,baby_1.worry_2,baby_1.every_3\r\n\"[\"\"season\"\", false, \"\"interesting\"\", null]\",concern,section,False,\r\n"
},
{
"role": "assistant",
"content": "Approach:\n1. Analyze the input CSV structure: It represents a dictionary with 2 root fields (writer, baby_1) and a nesting depth of 2.\n2. Identify the target format requirements: XML requires strict syntax. Map dictionary keys to tags. Handle list items by repeating child tags.\n3. Map the fields: The root keys will map directly, while nested objects need careful formatting.\n4. Handle data types: Ensure strings, numbers, and booleans are formatted correctly for the target.\n5. Construct the output: serialized the data strictly following the target schema.\n\nOutput:\n<?xml version=\"1.0\" encoding=\"UTF-8\"?>\n<random_structured_data>\n<writer>season</writer>\n<writer>False</writer>\n<writer>interesting</writer>\n<writer>None</writer>\n<baby_1>\n<serve>concern</serve>\n<soon_1>section</soon_1>\n<worry_2>False</worry_2>\n<every_3 />\n</baby_1>\n</random_structured_data>"
}
] | U10BEI_BASE |
[
{
"role": "system",
"content": "You must NOT call tools.\nDo NOT include any extra text, code fences, or markdown.\nDo NOT output JSON or YAML.\nOutput MUST be valid TOML.\nThe first line MUST be: [result]\nThen include status and confidence keys under [result].\nThen include [result.fields].\nOutput ONLY TOML as plain text.\n"
},
{
"role": "user",
"content": "Generate a valid TOML output using the fixed template.\nConstraints:\n- Output ONLY TOML, no extra text.\n- The first line must be [result].\n- Include [result.fields] with all required keys.\nRequired keys in fields: id,name,note,flag,score,category,empty\nValues:\n- id: 783\n- name: item_783\n- note: ok\n- flag: False\n- score: 0.83\n- category: A\n- empty: filled\n"
},
{
"role": "assistant",
"content": "[result]\nstatus = \"ok\"\nconfidence = 0.800\n\n[result.fields]\n\"category\" = \"A\"\n\"empty\" = \"filled\"\n\"flag\" = false\n\"id\" = 783\n\"name\" = \"item_783\"\n\"note\" = \"ok\"\n\"score\" = 0.83"
}
] | TOML_GEN_FIXED_VAR |
[
{
"role": "system",
"content": "You are an expert in JSON format. Analyze the data structure and generate syntactically perfect JSON."
},
{
"role": "user",
"content": "Reformat the CSV below into JSON:\ntxn,amount,status,meta_info.generated_at,meta_info.version,meta_info.tags,audit_log\r\n42327eaf-8038-44fa-9c06-2285ed8de51b,40.99,completed,2026-01-23T05:23:24.465239,v2.0,\"[\"\"south\"\", \"\"push\"\", \"\"address\"\", \"\"business\"\", \"\"me\"\"]\",\"[{\"\"action\"\": \"\"created\"\", \"\"by\"\": \"\"showard\"\", \"\"timestamp\"\": \"\"2012-07-27T22:03:46.538443\"\"}]\"\r\n"
},
{
"role": "assistant",
"content": "Approach:\n1. Analyze the input CSV structure: It represents a dictionary with 5 root fields (txn, amount, status...) and a nesting depth of 3.\n2. Identify the target format requirements: JSON requires strict syntax. Ensure all keys are quoted and nested structures use curly braces.\n3. Map the fields: The root keys will map directly, while nested objects need careful formatting.\n4. Handle data types: Ensure strings, numbers, and booleans are formatted correctly for the target.\n5. Construct the output: serialized the data strictly following the target schema.\n\nOutput:\n{\n \"txn\": \"42327eaf-8038-44fa-9c06-2285ed8de51b\",\n \"amount\": 40.99,\n \"status\": \"completed\",\n \"meta_info\": {\n \"generated_at\": \"2026-01-23T05:23:24.465239\",\n \"version\": \"v2.0\",\n \"tags\": [\n \"south\",\n \"push\",\n \"address\",\n \"business\",\n \"me\"\n ]\n },\n \"audit_log\": [\n {\n \"action\": \"created\",\n \"by\": \"showard\",\n \"timestamp\": \"2012-07-27T22:03:46.538443\"\n }\n ]\n}"
}
] | U10BEI_BASE |
[
{
"role": "system",
"content": "You must NOT call tools.\nDo NOT include any extra text, code fences, or markdown.\nDo NOT output JSON or YAML.\nOutput MUST be valid TOML.\nThe first line MUST be: [result]\nThen include status and confidence keys under [result].\nThen include [result.fields].\nOutput ONLY TOML as plain text.\n"
},
{
"role": "user",
"content": "Generate a valid TOML output using the fixed template.\nConstraints:\n- Output ONLY TOML, no extra text.\n- The first line must be [result].\n- Include [result.fields] with all required keys.\nRequired keys in fields: id,name,note,flag,score,category,empty\nValues:\n- id: 474\n- name: item_474\n- note: ok\n- flag: True\n- score: 0.74\n- category: A\n- empty: filled\n"
},
{
"role": "assistant",
"content": "[result]\nstatus = \"ok\"\nconfidence = 0.800\n\n[result.fields]\n\"category\" = \"A\"\n\"empty\" = \"filled\"\n\"flag\" = true\n\"id\" = 474\n\"name\" = \"item_474\"\n\"note\" = \"ok\"\n\"score\" = 0.74"
}
] | TOML_GEN_FIXED_VAR |
[
{
"role": "system",
"content": "You must NOT call tools.\nDo NOT include any extra text, code fences, or markdown.\nFollow the user instructions EXACTLY.\nOutput ONLY CSV content as plain text.\nStart with the header row.\n"
},
{
"role": "user",
"content": "Please output CSV code(Only output the CSV content as plain text, starting with the header row.).\n\nTask:\nGiven a recipe description, extract detailed structured information about the recipe into CSV format.\nFeature Requirements:\n1. Use the exact header fields and order.\n2. Output exactly one data row.\n3. Output exactly two lines: header line + one row.\n\nHeader:\nrecipe_id,recipe_name,chef_name,cuisine_type,preparation_time_minutes,cooking_time_minutes,servings,difficulty_level,main_ingredient,ingredient_list,ingredient_quantities,cooking_method,calories_per_serving,allergen_warnings,vegetarian_friendly,seasonality"
},
{
"role": "assistant",
"content": "recipe_id,recipe_name,chef_name,cuisine_type,preparation_time_minutes,cooking_time_minutes,servings,difficulty_level,main_ingredient,ingredient_list,ingredient_quantities,cooking_method,calories_per_serving,allergen_warnings,vegetarian_friendly,seasonality\nR00690,Stellar Stew 690,Mira Qor,Thai,30,55,3,Medium,Star Salt,nebula_beans;star_salt;quantum_onion,200g;5g;1,simmer,376,None,No,Winter"
}
] | CSV_TWO_LINE_FIXED |
[
{
"role": "system",
"content": "You must NOT call tools.\nDo NOT include any extra text, code fences, or markdown.\nDo NOT output JSON or YAML.\nOutput MUST be valid TOML.\nThe first line MUST be: [result]\nThen include status and confidence keys under [result].\nThen include [result.fields].\nOutput ONLY TOML as plain text.\n"
},
{
"role": "user",
"content": "Please convert the following JSON code to TOML code.\n\n<code>{\n \"headline\": \"Cultural development indeed.\",\n \"body\": \"Item fire theory. Indicate either reduce commercial certain woman. Ago remember collection author.\",\n \"author\": \"Jennifer Marquez\",\n \"meta_info\": {\n \"generated_at\": \"2026-01-23T05:23:23.519320\",\n \"version\": \"v2.0\",\n \"tags\": [\n \"ok\",\n \"sea\",\n \"try\",\n \"human\",\n \"water\"\n ]\n },\n \"audit_log\": [\n {\n \"action\": \"created\",\n \"by\": \"cowanlauren\",\n \"timestamp\": \"2018-06-01T01:52:24.754575\"\n }\n ]\n}</code>"
},
{
"role": "assistant",
"content": "[result]\nstatus = \"ok\"\nconfidence = 0.800\n\n[result.fields]\n\"audit_log\" = \"[{\\\"action\\\": \\\"created\\\", \\\"by\\\": \\\"cowanlauren\\\", \\\"timestamp\\\": \\\"2018-06-01T01:52:24.754575\\\"}]\"\n\"author\" = \"Jennifer Marquez\"\n\"body\" = \"Item fire theory. Indicate either reduce commercial certain woman. Ago remember collection author.\"\n\"headline\" = \"Cultural development indeed.\"\n\"meta_info\" = \"{\\\"generated_at\\\": \\\"2026-01-23T05:23:23.519320\\\", \\\"version\\\": \\\"v2.0\\\", \\\"tags\\\": [\\\"ok\\\", \\\"sea\\\", \\\"try\\\", \\\"human\\\", \\\"water\\\"]}\""
}
] | CONV_JSON2TOML_TPLFIX |
[
{
"role": "system",
"content": "You are an expert in TOML format. Analyze the data structure and generate syntactically perfect TOML."
},
{
"role": "user",
"content": "Generate minified TOML for error_log (no whitespace)."
},
{
"role": "assistant",
"content": "Approach:\n1. Define the schema: The error_log should contain a dictionary with 3 root fields (timestamp, level, message) and a nesting depth of 1.\n2. Select the format: TOML is chosen. Define hierarchy using dot-notation or [section] headers. Output must be minified (no extra whitespace/newlines).\n3. Generate realistic content: Use plausible names, dates, and values appropriate for the domain.\n4. Structure the data: Organize the fields logically with appropriate nesting and hierarchies.\n5. Serialize: Output the final string making sure to escape special characters if necessary and compress output.\n\nOutput:\ntimestamp = \"2026-01-23T05:23:25.199125\"\nlevel = \"ERROR\"\nmessage = \"Lay at between.\"\n"
}
] | U10BEI_BASE |
[
{
"role": "system",
"content": "You are an expert in TOML format. Analyze the data structure and generate syntactically perfect TOML."
},
{
"role": "user",
"content": "Transform this data from JSON to TOML:\n{\n \"user\": \"robertoliver\",\n \"rating\": 5,\n \"comment\": \"Player dog ground share.\",\n \"meta_info\": {\n \"generated_at\": \"2026-01-23T05:23:24.483041\",\n \"version\": \"v1.0\",\n \"tags\": [\n \"food\",\n \"discuss\",\n \"on\",\n \"but\",\n \"machine\"\n ]\n },\n \"audit_log\": [\n {\n \"action\": \"created\",\n \"by\": \"ngray\",\n \"timestamp\": \"2009-03-14T04:10:42.701861\"\n },\n {\n \"action\": \"created\",\n \"by\": \"anthony18\",\n \"timestamp\": \"2003-06-25T07:17:43.018347\"\n },\n {\n \"action\": \"created\",\n \"by\": \"michael93\",\n \"timestamp\": \"2023-03-02T23:12:59.776379\"\n }\n ]\n}"
},
{
"role": "assistant",
"content": "Approach:\n1. Analyze the input JSON structure: It represents a dictionary with 5 root fields (user, rating, comment...) and a nesting depth of 3.\n2. Identify the target format requirements: TOML requires strict syntax. Define hierarchy using dot-notation or [section] headers.\n3. Map the fields: The root keys will map directly, while nested objects need careful formatting.\n4. Handle data types: Ensure strings, numbers, and booleans are formatted correctly for the target.\n5. Construct the output: serialized the data strictly following the target schema.\n\nOutput:\nuser = \"robertoliver\"\nrating = 5\ncomment = \"Player dog ground share.\"\n[[audit_log]]\naction = \"created\"\nby = \"ngray\"\ntimestamp = \"2009-03-14T04:10:42.701861\"\n\n[[audit_log]]\naction = \"created\"\nby = \"anthony18\"\ntimestamp = \"2003-06-25T07:17:43.018347\"\n\n[[audit_log]]\naction = \"created\"\nby = \"michael93\"\ntimestamp = \"2023-03-02T23:12:59.776379\"\n\n[meta_info]\ngenerated_at = \"2026-01-23T05:23:24.483041\"\nversion = \"v1.0\"\ntags = [ \"food\", \"discuss\", \"on\", \"but\", \"machine\",]\n"
}
] | U10BEI_BASE |
[
{
"role": "system",
"content": "You must NOT call tools.\nDo NOT include any extra text, code fences, or markdown.\nDo NOT output JSON or YAML.\nOutput MUST be valid TOML.\nThe first line MUST be: [result]\nThen include status and confidence keys under [result].\nThen include [result.fields].\nOutput ONLY TOML as plain text.\n"
},
{
"role": "user",
"content": "Please convert the following JSON code to TOML code.\n\n<code>{\n \"experiment_id\": \"EXP-547570\",\n \"title\": \"Analysis of war\",\n \"metrics\": {\n \"accuracy\": 0.8235,\n \"loss\": 0.1439,\n \"epoch\": 18\n },\n \"config\": {\n \"batch_size\": 128,\n \"optimizer\": \"adam\",\n \"lr\": 0.001\n },\n \"meta_info\": {\n \"generated_at\": \"2026-01-23T05:23:23.504265\",\n \"version\": \"v1.0\",\n \"tags\": [\n \"along\",\n \"gas\",\n \"learn\",\n \"dinner\",\n \"director\"\n ]\n },\n \"audit_log\": [\n {\n \"action\": \"created\",\n \"by\": \"stephanie31\",\n \"timestamp\": \"1984-11-16T23:06:26.921884\"\n }\n ]\n}</code>"
},
{
"role": "assistant",
"content": "[result]\nstatus = \"ok\"\nconfidence = 0.800\n\n[result.fields]\n\"audit_log\" = \"[{\\\"action\\\": \\\"created\\\", \\\"by\\\": \\\"stephanie31\\\", \\\"timestamp\\\": \\\"1984-11-16T23:06:26.921884\\\"}]\"\n\"config\" = \"{\\\"batch_size\\\": 128, \\\"optimizer\\\": \\\"adam\\\", \\\"lr\\\": 0.001}\"\n\"experiment_id\" = \"EXP-547570\"\n\"meta_info\" = \"{\\\"generated_at\\\": \\\"2026-01-23T05:23:23.504265\\\", \\\"version\\\": \\\"v1.0\\\", \\\"tags\\\": [\\\"along\\\", \\\"gas\\\", \\\"learn\\\", \\\"dinner\\\", \\\"director\\\"]}\"\n\"metrics\" = \"{\\\"accuracy\\\": 0.8235, \\\"loss\\\": 0.1439, \\\"epoch\\\": 18}\"\n\"title\" = \"Analysis of war\""
}
] | CONV_JSON2TOML_TPLFIX |
[
{
"role": "system",
"content": "You must NOT call tools.\nDo NOT include any extra text, code fences, or markdown.\nDo NOT output JSON or YAML.\nOutput MUST be valid TOML.\nThe first line MUST be: [result]\nThen include status and confidence keys under [result].\nThen include [result.fields].\nOutput ONLY TOML as plain text.\n"
},
{
"role": "user",
"content": "Please convert the following JSON code to TOML code.\n\n<code>{\n \"user\": \"ugrant\",\n \"rating\": 2,\n \"comment\": \"Million feel whole put those.\",\n \"meta_info\": {\n \"generated_at\": \"2026-01-23T05:23:23.913687\",\n \"version\": \"v4.0\",\n \"tags\": [\n \"maintain\",\n \"near\",\n \"fund\",\n \"lead\",\n \"reach\"\n ]\n },\n \"audit_log\": [\n {\n \"action\": \"created\",\n \"by\": \"jonathan36\",\n \"timestamp\": \"2020-11-20T08:09:40.298188\"\n }\n ]\n}</code>"
},
{
"role": "assistant",
"content": "[result]\nstatus = \"ok\"\nconfidence = 0.800\n\n[result.fields]\n\"audit_log\" = \"[{\\\"action\\\": \\\"created\\\", \\\"by\\\": \\\"jonathan36\\\", \\\"timestamp\\\": \\\"2020-11-20T08:09:40.298188\\\"}]\"\n\"comment\" = \"Million feel whole put those.\"\n\"meta_info\" = \"{\\\"generated_at\\\": \\\"2026-01-23T05:23:23.913687\\\", \\\"version\\\": \\\"v4.0\\\", \\\"tags\\\": [\\\"maintain\\\", \\\"near\\\", \\\"fund\\\", \\\"lead\\\", \\\"reach\\\"]}\"\n\"rating\" = 2\n\"user\" = \"ugrant\""
}
] | CONV_JSON2TOML_TPLFIX |
[
{
"role": "system",
"content": "You must NOT call tools.\nDo NOT include any extra text, code fences, or markdown.\nDo NOT output JSON or YAML.\nOutput MUST be valid TOML.\nThe first line MUST be: [result]\nThen include status and confidence keys under [result].\nThen include [result.fields].\nOutput ONLY TOML as plain text.\n"
},
{
"role": "user",
"content": "Generate a valid TOML output using the fixed template.\nConstraints:\n- Output ONLY TOML, no extra text.\n- The first line must be [result].\n- Include [result.fields] with all required keys.\nRequired keys in fields: id,name,note,flag,score,category,empty\nValues:\n- id: 183\n- name: item_183\n- note: ok\n- flag: False\n- score: 0.83\n- category: A\n- empty: filled\n"
},
{
"role": "assistant",
"content": "[result]\nstatus = \"ok\"\nconfidence = 0.800\n\n[result.fields]\n\"category\" = \"A\"\n\"empty\" = \"filled\"\n\"flag\" = false\n\"id\" = 183\n\"name\" = \"item_183\"\n\"note\" = \"ok\"\n\"score\" = 0.83"
}
] | TOML_GEN_FIXED_VAR |
[
{
"role": "system",
"content": "You are an expert in YAML format. Analyze the data structure and generate syntactically perfect YAML."
},
{
"role": "user",
"content": "Reformat the CSV below into YAML:\ncourse,weeks,credits\r\nCustomer-focused asynchronous approach,9,3\r\n"
},
{
"role": "assistant",
"content": "Approach:\n1. Analyze the input CSV structure: It represents a dictionary with 3 root fields (course, weeks, credits) and a nesting depth of 1.\n2. Identify the target format requirements: YAML requires strict syntax. Use indentation to represent nesting. lists are denoted by hyphens.\n3. Map the fields: The root keys will map directly, while nested objects need careful formatting.\n4. Handle data types: Ensure strings, numbers, and booleans are formatted correctly for the target.\n5. Construct the output: serialized the data strictly following the target schema.\n\nOutput:\ncourse: Customer-focused asynchronous approach\nweeks: 9\ncredits: 3\n"
}
] | U10BEI_BASE |
[
{
"role": "system",
"content": "You must NOT call tools.\nDo NOT include any extra text, code fences, or markdown.\nDo NOT output JSON or YAML.\nOutput MUST be valid TOML.\nThe first line MUST be: [result]\nThen include status and confidence keys under [result].\nThen include [result.fields].\nOutput ONLY TOML as plain text.\n"
},
{
"role": "user",
"content": "Generate a valid TOML output using the fixed template.\nConstraints:\n- Output ONLY TOML, no extra text.\n- The first line must be [result].\n- Include [result.fields] with all required keys.\nRequired keys in fields: id,name,note,flag,score,category,empty\nValues:\n- id: 745\n- name: item_745\n- note: line1\\nline2\n- flag: False\n- score: 0.45\n- category: B\n- empty: filled\n"
},
{
"role": "assistant",
"content": "[result]\nstatus = \"ok\"\nconfidence = 0.800\n\n[result.fields]\n\"category\" = \"B\"\n\"empty\" = \"filled\"\n\"flag\" = false\n\"id\" = 745\n\"name\" = \"item_745\"\n\"note\" = \"line1\\\\nline2\"\n\"score\" = 0.45"
}
] | TOML_GEN_FIXED_VAR |
[
{
"role": "system",
"content": "You must NOT call tools.\nDo NOT include any extra text, code fences, or markdown.\nDo NOT output JSON or YAML.\nOutput MUST be valid TOML.\nThe first line MUST be: [result]\nThen include status and confidence keys under [result].\nThen include [result.fields].\nOutput ONLY TOML as plain text.\n"
},
{
"role": "user",
"content": "Please convert the following JSON code to TOML code.\n\n<code>{\n \"maybe\": {\n \"pull\": 14,\n \"speech_1\": \"end\",\n \"treatment_2\": \"school\"\n },\n \"compare_1\": {\n \"service\": \"market\",\n \"become_1\": null,\n \"mind_2\": 8.5\n },\n \"yes_2\": {\n \"add\": true,\n \"hand_1\": null,\n \"analysis_2\": \"chair\",\n \"south_3\": \"age\",\n \"force_4\": false\n },\n \"exist_3\": \"dinner\",\n \"model_4\": [\n true,\n null,\n \"find\",\n null,\n 0.91\n ]\n}</code>"
},
{
"role": "assistant",
"content": "[result]\nstatus = \"ok\"\nconfidence = 0.800\n\n[result.fields]\n\"compare_1\" = \"{\\\"service\\\": \\\"market\\\", \\\"become_1\\\": null, \\\"mind_2\\\": 8.5}\"\n\"exist_3\" = \"dinner\"\n\"maybe\" = \"{\\\"pull\\\": 14, \\\"speech_1\\\": \\\"end\\\", \\\"treatment_2\\\": \\\"school\\\"}\"\n\"model_4\" = \"[true, null, \\\"find\\\", null, 0.91]\"\n\"yes_2\" = \"{\\\"add\\\": true, \\\"hand_1\\\": null, \\\"analysis_2\\\": \\\"chair\\\", \\\"south_3\\\": \\\"age\\\", \\\"force_4\\\": false}\""
}
] | CONV_JSON2TOML_TPLFIX |
[
{
"role": "system",
"content": "You are an expert in TOML format. Analyze the data structure and generate syntactically perfect TOML."
},
{
"role": "user",
"content": "Convert to TOML with sorted keys:\ncomment: Opportunity yeah religious course he.\nrating: 2\nuser: fieldserin\n"
},
{
"role": "assistant",
"content": "Approach:\n1. Analyze the input YAML structure: It represents a dictionary with 3 root fields (user, rating, comment) and a nesting depth of 1.\n2. Identify the target format requirements: TOML requires strict syntax. Define hierarchy using dot-notation or [section] headers. Keys must be sorted alphabetically.\n3. Map the fields: The root keys will map directly, while nested objects need careful formatting.\n4. Handle data types: Ensure strings, numbers, and booleans are formatted correctly for the target.\n5. Construct the output: serialized the data strictly following the target schema and applying constraints.\n\nOutput:\ncomment = \"Opportunity yeah religious course he.\"\nrating = 2\nuser = \"fieldserin\"\n"
}
] | U10BEI_BASE |
[
{
"role": "system",
"content": "You must NOT call tools.\nDo NOT include any extra text, code fences, or markdown.\nDo NOT output JSON or YAML.\nOutput MUST be valid TOML.\nThe first line MUST be: [result]\nThen include status and confidence keys under [result].\nThen include [result.fields].\nOutput ONLY TOML as plain text.\n"
},
{
"role": "user",
"content": "Generate a valid TOML output using the fixed template.\nConstraints:\n- Output ONLY TOML, no extra text.\n- The first line must be [result].\n- Include [result.fields] with all required keys.\nRequired keys in fields: id,name,note,flag,score,category,empty\nValues:\n- id: 677\n- name: item_677\n- note: ok\n- flag: False\n- score: 0.77\n- category: C\n- empty: filled\n"
},
{
"role": "assistant",
"content": "[result]\nstatus = \"ok\"\nconfidence = 0.800\n\n[result.fields]\n\"category\" = \"C\"\n\"empty\" = \"filled\"\n\"flag\" = false\n\"id\" = 677\n\"name\" = \"item_677\"\n\"note\" = \"ok\"\n\"score\" = 0.77"
}
] | TOML_GEN_FIXED_VAR |
[
{
"role": "system",
"content": "You are an expert in TOML format. Analyze the data structure and generate syntactically perfect TOML."
},
{
"role": "user",
"content": "Convert to TOML with sorted keys:\ntrial: should\n"
},
{
"role": "assistant",
"content": "Approach:\n1. Analyze the input YAML structure: It represents a dictionary with 1 root fields (trial) and a nesting depth of 1.\n2. Identify the target format requirements: TOML requires strict syntax. Define hierarchy using dot-notation or [section] headers. Keys must be sorted alphabetically.\n3. Map the fields: The root keys will map directly, while nested objects need careful formatting.\n4. Handle data types: Ensure strings, numbers, and booleans are formatted correctly for the target.\n5. Construct the output: serialized the data strictly following the target schema and applying constraints.\n\nOutput:\ntrial = \"should\"\n"
}
] | U10BEI_BASE |
[
{
"role": "system",
"content": "You are an expert in TOML format. Analyze the data structure and generate syntactically perfect TOML."
},
{
"role": "user",
"content": "Output structured data for random_structured_data using TOML."
},
{
"role": "assistant",
"content": "Approach:\n1. Define the schema: The random_structured_data should contain a dictionary with 1 root fields (random_content) and a nesting depth of 1.\n2. Select the format: TOML is chosen. Define hierarchy using dot-notation or [section] headers.\n3. Generate realistic content: Use plausible names, dates, and values appropriate for the domain.\n4. Structure the data: Organize the fields logically with appropriate nesting and hierarchies.\n5. Serialize: Output the final string making sure to escape special characters if necessary.\n\nOutput:\nrandom_content = \"establish\"\n"
}
] | U10BEI_BASE |
[
{
"role": "system",
"content": "You must NOT call tools.\nDo NOT include any extra text, code fences, or markdown.\nDo NOT output JSON or YAML.\nOutput MUST be valid TOML.\nThe first line MUST be: [result]\nThen include status and confidence keys under [result].\nThen include [result.fields].\nOutput ONLY TOML as plain text.\n"
},
{
"role": "user",
"content": "Generate a valid TOML output using the fixed template.\nConstraints:\n- Output ONLY TOML, no extra text.\n- The first line must be [result].\n- Include [result.fields] with all required keys.\nRequired keys in fields: id,name,note,flag,score,category,empty\nValues:\n- id: 1300\n- name: item_1300\n- note: line1\\nline2\n- flag: True\n- score: 0.0\n- category: B\n- empty: \n"
},
{
"role": "assistant",
"content": "[result]\nstatus = \"ok\"\nconfidence = 0.800\n\n[result.fields]\n\"category\" = \"B\"\n\"empty\" = \"\"\n\"flag\" = true\n\"id\" = 1300\n\"name\" = \"item_1300\"\n\"note\" = \"line1\\\\nline2\"\n\"score\" = 0.0"
}
] | TOML_GEN_FIXED_VAR |
[
{
"role": "system",
"content": "You must NOT call tools.\nDo NOT include any extra text, code fences, or markdown.\nDo NOT output JSON or YAML.\nOutput MUST be valid TOML.\nThe first line MUST be: [result]\nThen include status and confidence keys under [result].\nThen include [result.fields].\nOutput ONLY TOML as plain text.\n"
},
{
"role": "user",
"content": "Please convert the following JSON code to TOML code.\n\n<code>{\n \"experiment_id\": \"EXP-270313\",\n \"title\": \"Analysis of order\",\n \"metrics\": {\n \"accuracy\": 0.9795,\n \"loss\": 0.4141,\n \"epoch\": 58\n },\n \"config\": {\n \"batch_size\": 128,\n \"optimizer\": \"adam\",\n \"lr\": 0.001\n },\n \"meta_info\": {\n \"generated_at\": \"2026-01-23T05:23:24.339861\",\n \"version\": \"v1.0\",\n \"tags\": [\n \"reflect\",\n \"tough\",\n \"run\",\n \"tell\",\n \"week\"\n ]\n },\n \"audit_log\": [\n {\n \"action\": \"created\",\n \"by\": \"jennifer11\",\n \"timestamp\": \"2004-05-16T22:06:48.489427\"\n },\n {\n \"action\": \"created\",\n \"by\": \"josephbush\",\n \"timestamp\": \"1990-03-04T11:26:48.314085\"\n },\n {\n \"action\": \"created\",\n \"by\": \"arodriguez\",\n \"timestamp\": \"2003-03-24T18:13:17.593186\"\n }\n ]\n}</code>"
},
{
"role": "assistant",
"content": "[result]\nstatus = \"ok\"\nconfidence = 0.800\n\n[result.fields]\n\"audit_log\" = \"[{\\\"action\\\": \\\"created\\\", \\\"by\\\": \\\"jennifer11\\\", \\\"timestamp\\\": \\\"2004-05-16T22:06:48.489427\\\"}, {\\\"action\\\": \\\"created\\\", \\\"by\\\": \\\"josephbush\\\", \\\"timestamp\\\": \\\"1990-03-04T11:26:48.314085\\\"}, {\\\"action\\\": \\\"created\\\", \\\"by\\\": \\\"arodriguez\\\", \\\"timestamp\\\": \\\"2003-03-24T18:13:17.593186\\\"}]\"\n\"config\" = \"{\\\"batch_size\\\": 128, \\\"optimizer\\\": \\\"adam\\\", \\\"lr\\\": 0.001}\"\n\"experiment_id\" = \"EXP-270313\"\n\"meta_info\" = \"{\\\"generated_at\\\": \\\"2026-01-23T05:23:24.339861\\\", \\\"version\\\": \\\"v1.0\\\", \\\"tags\\\": [\\\"reflect\\\", \\\"tough\\\", \\\"run\\\", \\\"tell\\\", \\\"week\\\"]}\"\n\"metrics\" = \"{\\\"accuracy\\\": 0.9795, \\\"loss\\\": 0.4141, \\\"epoch\\\": 58}\"\n\"title\" = \"Analysis of order\""
}
] | CONV_JSON2TOML_TPLFIX |
[
{
"role": "system",
"content": "You must NOT call tools.\nDo NOT include any extra text, code fences, or markdown.\nDo NOT output JSON or YAML.\nOutput MUST be valid TOML.\nThe first line MUST be: [result]\nThen include status and confidence keys under [result].\nThen include [result.fields].\nOutput ONLY TOML as plain text.\n"
},
{
"role": "user",
"content": "Generate a valid TOML output using the fixed template.\nConstraints:\n- Output ONLY TOML, no extra text.\n- The first line must be [result].\n- Include [result.fields] with all required keys.\nRequired keys in fields: id,name,note,flag,score,category,empty\nValues:\n- id: 1334\n- name: item_1334\n- note: ok\n- flag: True\n- score: 0.34\n- category: C\n- empty: filled\n"
},
{
"role": "assistant",
"content": "[result]\nstatus = \"ok\"\nconfidence = 0.800\n\n[result.fields]\n\"category\" = \"C\"\n\"empty\" = \"filled\"\n\"flag\" = true\n\"id\" = 1334\n\"name\" = \"item_1334\"\n\"note\" = \"ok\"\n\"score\" = 0.34"
}
] | TOML_GEN_FIXED_VAR |
[
{
"role": "system",
"content": "You must NOT call tools.\nDo NOT include any extra text, code fences, or markdown.\nDo NOT output JSON or YAML.\nOutput MUST be valid TOML.\nThe first line MUST be: [result]\nThen include status and confidence keys under [result].\nThen include [result.fields].\nOutput ONLY TOML as plain text.\n"
},
{
"role": "user",
"content": "Please convert the following JSON code to TOML code.\n\n<code>{\n \"txn\": \"f317ba19-1a6b-453a-96a5-c60cffc2297c\",\n \"amount\": 431.61,\n \"status\": \"completed\",\n \"meta_info\": {\n \"generated_at\": \"2026-01-23T05:23:25.173096\",\n \"version\": \"v4.0\",\n \"tags\": [\n \"range\",\n \"include\",\n \"make\",\n \"nation\",\n \"six\"\n ]\n },\n \"audit_log\": [\n {\n \"action\": \"created\",\n \"by\": \"tannerkatherine\",\n \"timestamp\": \"1982-06-23T20:51:09.230015\"\n },\n {\n \"action\": \"created\",\n \"by\": \"ljohnson\",\n \"timestamp\": \"2002-07-28T08:20:27.405076\"\n }\n ]\n}</code>"
},
{
"role": "assistant",
"content": "[result]\nstatus = \"completed\"\nconfidence = 0.800\n\n[result.fields]\n\"amount\" = 431.61\n\"audit_log\" = \"[{\\\"action\\\": \\\"created\\\", \\\"by\\\": \\\"tannerkatherine\\\", \\\"timestamp\\\": \\\"1982-06-23T20:51:09.230015\\\"}, {\\\"action\\\": \\\"created\\\", \\\"by\\\": \\\"ljohnson\\\", \\\"timestamp\\\": \\\"2002-07-28T08:20:27.405076\\\"}]\"\n\"meta_info\" = \"{\\\"generated_at\\\": \\\"2026-01-23T05:23:25.173096\\\", \\\"version\\\": \\\"v4.0\\\", \\\"tags\\\": [\\\"range\\\", \\\"include\\\", \\\"make\\\", \\\"nation\\\", \\\"six\\\"]}\"\n\"txn\" = \"f317ba19-1a6b-453a-96a5-c60cffc2297c\""
}
] | CONV_JSON2TOML_TPLFIX |
[
{
"role": "system",
"content": "You must NOT call tools.\nDo NOT include any extra text, code fences, or markdown.\nDo NOT output JSON or YAML.\nOutput MUST be valid TOML.\nThe first line MUST be: [result]\nThen include status and confidence keys under [result].\nThen include [result.fields].\nOutput ONLY TOML as plain text.\n"
},
{
"role": "user",
"content": "Please convert the following JSON code to TOML code.\n\n<code>{\n \"contract\": \"implement out-of-the-box networks\",\n \"parties\": [\n \"Garcia, Mcconnell and Tucker\",\n \"Wright, Miller and Jensen\"\n ],\n \"signed\": true,\n \"meta_info\": {\n \"generated_at\": \"2026-01-23T05:23:24.888898\",\n \"version\": \"v5.0\",\n \"tags\": [\n \"actually\",\n \"give\",\n \"whether\",\n \"interview\",\n \"air\"\n ]\n },\n \"audit_log\": [\n {\n \"action\": \"created\",\n \"by\": \"katie50\",\n \"timestamp\": \"2001-06-16T00:02:06.348273\"\n },\n {\n \"action\": \"created\",\n \"by\": \"aburgess\",\n \"timestamp\": \"2021-05-17T05:19:06.692184\"\n },\n {\n \"action\": \"created\",\n \"by\": \"sanchezjoshua\",\n \"timestamp\": \"1986-10-10T17:53:31.606275\"\n }\n ]\n}</code>"
},
{
"role": "assistant",
"content": "[result]\nstatus = \"ok\"\nconfidence = 0.800\n\n[result.fields]\n\"audit_log\" = \"[{\\\"action\\\": \\\"created\\\", \\\"by\\\": \\\"katie50\\\", \\\"timestamp\\\": \\\"2001-06-16T00:02:06.348273\\\"}, {\\\"action\\\": \\\"created\\\", \\\"by\\\": \\\"aburgess\\\", \\\"timestamp\\\": \\\"2021-05-17T05:19:06.692184\\\"}, {\\\"action\\\": \\\"created\\\", \\\"by\\\": \\\"sanchezjoshua\\\", \\\"timestamp\\\": \\\"1986-10-10T17:53:31.606275\\\"}]\"\n\"contract\" = \"implement out-of-the-box networks\"\n\"meta_info\" = \"{\\\"generated_at\\\": \\\"2026-01-23T05:23:24.888898\\\", \\\"version\\\": \\\"v5.0\\\", \\\"tags\\\": [\\\"actually\\\", \\\"give\\\", \\\"whether\\\", \\\"interview\\\", \\\"air\\\"]}\"\n\"parties\" = \"[\\\"Garcia, Mcconnell and Tucker\\\", \\\"Wright, Miller and Jensen\\\"]\"\n\"signed\" = true"
}
] | CONV_JSON2TOML_TPLFIX |
[
{
"role": "system",
"content": "You must NOT call tools.\nDo NOT include any extra text, code fences, or markdown.\nDo NOT output JSON or YAML.\nOutput MUST be valid TOML.\nThe first line MUST be: [result]\nThen include status and confidence keys under [result].\nThen include [result.fields].\nOutput ONLY TOML as plain text.\n"
},
{
"role": "user",
"content": "Please convert the following JSON code to TOML code.\n\n<code>{\n \"user\": \"ybrown\",\n \"rating\": 1,\n \"comment\": \"Political water capital skin behind.\",\n \"meta_info\": {\n \"generated_at\": \"2026-01-23T05:23:24.853397\",\n \"version\": \"v3.0\",\n \"tags\": [\n \"care\",\n \"bar\",\n \"rich\",\n \"attorney\",\n \"eight\"\n ]\n },\n \"audit_log\": [\n {\n \"action\": \"created\",\n \"by\": \"rodney28\",\n \"timestamp\": \"1979-05-25T18:38:37.273960\"\n },\n {\n \"action\": \"created\",\n \"by\": \"destiny21\",\n \"timestamp\": \"2006-06-02T16:47:23.651747\"\n }\n ]\n}</code>"
},
{
"role": "assistant",
"content": "[result]\nstatus = \"ok\"\nconfidence = 0.800\n\n[result.fields]\n\"audit_log\" = \"[{\\\"action\\\": \\\"created\\\", \\\"by\\\": \\\"rodney28\\\", \\\"timestamp\\\": \\\"1979-05-25T18:38:37.273960\\\"}, {\\\"action\\\": \\\"created\\\", \\\"by\\\": \\\"destiny21\\\", \\\"timestamp\\\": \\\"2006-06-02T16:47:23.651747\\\"}]\"\n\"comment\" = \"Political water capital skin behind.\"\n\"meta_info\" = \"{\\\"generated_at\\\": \\\"2026-01-23T05:23:24.853397\\\", \\\"version\\\": \\\"v3.0\\\", \\\"tags\\\": [\\\"care\\\", \\\"bar\\\", \\\"rich\\\", \\\"attorney\\\", \\\"eight\\\"]}\"\n\"rating\" = 1\n\"user\" = \"ybrown\""
}
] | CONV_JSON2TOML_TPLFIX |
[
{
"role": "system",
"content": "You must NOT call tools.\nDo NOT include any extra text, code fences, or markdown.\nDo NOT output JSON or YAML.\nOutput MUST be valid TOML.\nThe first line MUST be: [result]\nThen include status and confidence keys under [result].\nThen include [result.fields].\nOutput ONLY TOML as plain text.\n"
},
{
"role": "user",
"content": "Please convert the following JSON code to TOML code.\n\n<code>{\n \"course\": \"Synergized radical moderator\",\n \"weeks\": 11,\n \"credits\": 3,\n \"meta_info\": {\n \"generated_at\": \"2026-01-23T05:23:25.179761\",\n \"version\": \"v4.0\",\n \"tags\": [\n \"threat\",\n \"available\",\n \"house\",\n \"away\",\n \"civil\"\n ]\n },\n \"audit_log\": [\n {\n \"action\": \"created\",\n \"by\": \"hhurst\",\n \"timestamp\": \"1981-08-11T17:46:51.059611\"\n },\n {\n \"action\": \"created\",\n \"by\": \"jonathan70\",\n \"timestamp\": \"2011-07-26T06:10:08.993728\"\n },\n {\n \"action\": \"created\",\n \"by\": \"wprice\",\n \"timestamp\": \"1992-07-09T08:42:04.828467\"\n }\n ]\n}</code>"
},
{
"role": "assistant",
"content": "[result]\nstatus = \"ok\"\nconfidence = 0.800\n\n[result.fields]\n\"audit_log\" = \"[{\\\"action\\\": \\\"created\\\", \\\"by\\\": \\\"hhurst\\\", \\\"timestamp\\\": \\\"1981-08-11T17:46:51.059611\\\"}, {\\\"action\\\": \\\"created\\\", \\\"by\\\": \\\"jonathan70\\\", \\\"timestamp\\\": \\\"2011-07-26T06:10:08.993728\\\"}, {\\\"action\\\": \\\"created\\\", \\\"by\\\": \\\"wprice\\\", \\\"timestamp\\\": \\\"1992-07-09T08:42:04.828467\\\"}]\"\n\"course\" = \"Synergized radical moderator\"\n\"credits\" = 3\n\"meta_info\" = \"{\\\"generated_at\\\": \\\"2026-01-23T05:23:25.179761\\\", \\\"version\\\": \\\"v4.0\\\", \\\"tags\\\": [\\\"threat\\\", \\\"available\\\", \\\"house\\\", \\\"away\\\", \\\"civil\\\"]}\"\n\"weeks\" = 11"
}
] | CONV_JSON2TOML_TPLFIX |
[
{
"role": "system",
"content": "You are an expert in XML format. Analyze the data structure and generate syntactically perfect XML."
},
{
"role": "user",
"content": "Convert the following JSON to XML:\n{\n \"random_content\": \"friend\"\n}"
},
{
"role": "assistant",
"content": "Approach:\n1. Analyze the input JSON structure: It represents a dictionary with 1 root fields (random_content) and a nesting depth of 1.\n2. Identify the target format requirements: XML requires strict syntax. Map dictionary keys to tags. Handle list items by repeating child tags.\n3. Map the fields: The root keys will map directly, while nested objects need careful formatting.\n4. Handle data types: Ensure strings, numbers, and booleans are formatted correctly for the target.\n5. Construct the output: serialized the data strictly following the target schema.\n\nOutput:\n<?xml version=\"1.0\" encoding=\"UTF-8\"?>\n<random_structured_data>\n<random_content>friend</random_content>\n</random_structured_data>"
}
] | U10BEI_BASE |
[
{
"role": "system",
"content": "You are an expert in CSV format. Analyze the data structure and generate syntactically perfect CSV."
},
{
"role": "user",
"content": "Generate customer review data in CSV format."
},
{
"role": "assistant",
"content": "Approach:\n1. Define the schema: The customer_review should contain a dictionary with 5 root fields (user, rating, comment...) and a nesting depth of 3.\n2. Select the format: CSV is chosen. Flatten the hierarchical structure. Use dot notation for nested keys in the header.\n3. Generate realistic content: Use plausible names, dates, and values appropriate for the domain.\n4. Structure the data: Organize the fields logically with appropriate nesting and hierarchies.\n5. Serialize: Output the final string making sure to escape special characters if necessary.\n\nOutput:\nuser,rating,comment,meta_info.generated_at,meta_info.version,meta_info.tags,audit_log\r\nruizcrystal,2,Series event military success action not.,2026-01-23T05:23:23.578178,v5.0,\"[\"\"benefit\"\", \"\"edge\"\", \"\"care\"\", \"\"discuss\"\", \"\"while\"\"]\",\"[{\"\"action\"\": \"\"created\"\", \"\"by\"\": \"\"lgarrett\"\", \"\"timestamp\"\": \"\"2018-01-29T21:15:54.636125\"\"}, {\"\"action\"\": \"\"created\"\", \"\"by\"\": \"\"pbarron\"\", \"\"timestamp\"\": \"\"2020-11-02T01:27:42.639087\"\"}, {\"\"action\"\": \"\"created\"\", \"\"by\"\": \"\"gabriella22\"\", \"\"timestamp\"\": \"\"1977-10-04T14:51:40.157754\"\"}]\"\r\n"
}
] | U10BEI_BASE |
[
{
"role": "system",
"content": "You must NOT call tools.\nDo NOT include any extra text, code fences, or markdown.\nDo NOT output JSON or YAML.\nOutput MUST be valid TOML.\nThe first line MUST be: [result]\nThen include status and confidence keys under [result].\nThen include [result.fields].\nOutput ONLY TOML as plain text.\n"
},
{
"role": "user",
"content": "Please convert the following JSON code to TOML code.\n\n<code>{\n \"you\": {\n \"watch\": {\n \"must\": 6.9,\n \"air_1\": null,\n \"individual_2\": \"the\"\n },\n \"million_1\": \"care\",\n \"edge_2\": [\n \"security\",\n \"attack\",\n false\n ],\n \"conference_3\": [\n 1.41,\n false\n ]\n },\n \"red_1\": [\n [\n null,\n true,\n null\n ],\n {\n \"blood\": \"continue\"\n },\n [\n null,\n true\n ],\n \"war\"\n ],\n \"majority_2\": {\n \"although\": \"animal\"\n },\n \"several_3\": \"read\",\n \"meta_info\": {\n \"generated_at\": \"2026-01-23T05:23:23.742496\",\n \"version\": \"v5.0\",\n \"tags\": [\n \"program\",\n \"current\",\n \"physical\",\n \"send\",\n \"popular\"\n ]\n },\n \"audit_log\": [\n {\n \"action\": \"created\",\n \"by\": \"jenniferbradford\",\n \"timestamp\": \"1981-04-07T17:14:45.762019\"\n }\n ]\n}</code>"
},
{
"role": "assistant",
"content": "[result]\nstatus = \"ok\"\nconfidence = 0.800\n\n[result.fields]\n\"audit_log\" = \"[{\\\"action\\\": \\\"created\\\", \\\"by\\\": \\\"jenniferbradford\\\", \\\"timestamp\\\": \\\"1981-04-07T17:14:45.762019\\\"}]\"\n\"majority_2\" = \"{\\\"although\\\": \\\"animal\\\"}\"\n\"meta_info\" = \"{\\\"generated_at\\\": \\\"2026-01-23T05:23:23.742496\\\", \\\"version\\\": \\\"v5.0\\\", \\\"tags\\\": [\\\"program\\\", \\\"current\\\", \\\"physical\\\", \\\"send\\\", \\\"popular\\\"]}\"\n\"red_1\" = \"[[null, true, null], {\\\"blood\\\": \\\"continue\\\"}, [null, true], \\\"war\\\"]\"\n\"several_3\" = \"read\"\n\"you\" = \"{\\\"watch\\\": {\\\"must\\\": 6.9, \\\"air_1\\\": null, \\\"individual_2\\\": \\\"the\\\"}, \\\"million_1\\\": \\\"care\\\", \\\"edge_2\\\": [\\\"security\\\", \\\"attack\\\", false], \\\"conference_3\\\": [1.41, false]}\""
}
] | CONV_JSON2TOML_TPLFIX |
[
{
"role": "system",
"content": "You must NOT call tools.\nDo NOT include any extra text, code fences, or markdown.\nDo NOT output JSON or YAML.\nOutput MUST be valid TOML.\nThe first line MUST be: [result]\nThen include status and confidence keys under [result].\nThen include [result.fields].\nOutput ONLY TOML as plain text.\n"
},
{
"role": "user",
"content": "Generate a valid TOML output using the fixed template.\nConstraints:\n- Output ONLY TOML, no extra text.\n- The first line must be [result].\n- Include [result.fields] with all required keys.\nRequired keys in fields: id,name,note,flag,score,category,empty\nValues:\n- id: 1223\n- name: item_1223\n- note: ok\n- flag: False\n- score: 0.23\n- category: C\n- empty: filled\n"
},
{
"role": "assistant",
"content": "[result]\nstatus = \"ok\"\nconfidence = 0.800\n\n[result.fields]\n\"category\" = \"C\"\n\"empty\" = \"filled\"\n\"flag\" = false\n\"id\" = 1223\n\"name\" = \"item_1223\"\n\"note\" = \"ok\"\n\"score\" = 0.23"
}
] | TOML_GEN_FIXED_VAR |
[
{
"role": "system",
"content": "You are an expert in CSV format. Analyze the data structure and generate syntactically perfect CSV."
},
{
"role": "user",
"content": "Output structured data for random_structured_data using CSV."
},
{
"role": "assistant",
"content": "Approach:\n1. Define the schema: The random_structured_data should contain a dictionary with 3 root fields (random_content, meta_info, audit_log) and a nesting depth of 3.\n2. Select the format: CSV is chosen. Flatten the hierarchical structure. Use dot notation for nested keys in the header.\n3. Generate realistic content: Use plausible names, dates, and values appropriate for the domain.\n4. Structure the data: Organize the fields logically with appropriate nesting and hierarchies.\n5. Serialize: Output the final string making sure to escape special characters if necessary.\n\nOutput:\nrandom_content,meta_info.generated_at,meta_info.version,meta_info.tags,audit_log\r\nhot,2026-01-23T05:23:24.194578,v4.0,\"[\"\"cell\"\", \"\"left\"\", \"\"significant\"\", \"\"entire\"\", \"\"treatment\"\"]\",\"[{\"\"action\"\": \"\"created\"\", \"\"by\"\": \"\"kyle45\"\", \"\"timestamp\"\": \"\"1993-06-16T16:54:40.020800\"\"}, {\"\"action\"\": \"\"created\"\", \"\"by\"\": \"\"cantusarah\"\", \"\"timestamp\"\": \"\"2013-07-31T10:01:41.881583\"\"}]\"\r\n"
}
] | U10BEI_BASE |
[
{
"role": "system",
"content": "You must NOT call tools.\nDo NOT include any extra text, code fences, or markdown.\nDo NOT output JSON or YAML.\nOutput MUST be valid TOML.\nThe first line MUST be: [result]\nThen include status and confidence keys under [result].\nThen include [result.fields].\nOutput ONLY TOML as plain text.\n"
},
{
"role": "user",
"content": "Generate a valid TOML output using the fixed template.\nConstraints:\n- Output ONLY TOML, no extra text.\n- The first line must be [result].\n- Include [result.fields] with all required keys.\nRequired keys in fields: id,name,note,flag,score,category,empty\nValues:\n- id: 1421\n- name: item_1421\"\\\n- note: ok\n- flag: False\n- score: 0.21\n- category: C\n- empty: filled\n"
},
{
"role": "assistant",
"content": "[result]\nstatus = \"ok\"\nconfidence = 0.800\n\n[result.fields]\n\"category\" = \"C\"\n\"empty\" = \"filled\"\n\"flag\" = false\n\"id\" = 1421\n\"name\" = \"item_1421\\\"\\\\\"\n\"note\" = \"ok\"\n\"score\" = 0.21"
}
] | TOML_GEN_FIXED_VAR |
[
{
"role": "system",
"content": "You are an expert in TOML format. Analyze the data structure and generate syntactically perfect TOML."
},
{
"role": "user",
"content": "Create a TOML representation of research paper."
},
{
"role": "assistant",
"content": "Approach:\n1. Define the schema: The research_paper should contain a dictionary with 6 root fields (paper_id, title, authors...) and a nesting depth of 3.\n2. Select the format: TOML is chosen. Define hierarchy using dot-notation or [section] headers.\n3. Generate realistic content: Use plausible names, dates, and values appropriate for the domain.\n4. Structure the data: Organize the fields logically with appropriate nesting and hierarchies.\n5. Serialize: Output the final string making sure to escape special characters if necessary.\n\nOutput:\npaper_id = \"arXiv:2022.98366\"\ntitle = \"Extended 6thgeneration Graphical User Interface\"\nabstract = \"Memory stage seek coach whether between. Analysis star health.\\nTree she country risk day old. Ball check pretty couple possible street.\"\nkeywords = [ \"trip\", \"now\", \"five\", \"will\",]\n[[authors]]\nname = \"John Hensley\"\naffiliation = \"Schroeder-Robinson\"\nemail = \"xmartin@example.com\"\n\n[[authors]]\nname = \"Michael Melendez\"\naffiliation = \"Marshall Ltd\"\nemail = \"williamsrebecca@example.com\"\n\n[[authors]]\nname = \"Andrew Roberts\"\naffiliation = \"Jackson-Powers\"\nemail = \"charlenesandoval@example.org\"\n\n[publication]\nvenue = \"Nature\"\nyear = 2023\ndoi = \"10.9119/YjXW\"\n"
}
] | U10BEI_BASE |
[
{
"role": "system",
"content": "You are an expert in CSV format. Analyze the data structure and generate syntactically perfect CSV."
},
{
"role": "user",
"content": "Reformat the XML below into CSV:\n<?xml version=\"1.0\" encoding=\"UTF-8\"?>\n<contract>\n<contract>orchestrate distributed solutions</contract>\n<parties>Mckay, Tyler and Gray</parties>\n<parties>Davis Inc</parties>\n<signed>True</signed>\n<meta_info>\n<generated_at>2026-01-23T05:23:25.346467</generated_at>\n<version>v5.0</version>\n<tags>election</tags>\n<tags>concern</tags>\n<tags>social</tags>\n<tags>series</tags>\n<tags>old</tags>\n</meta_info>\n<audit_log>\n<action>created</action>\n<by>michaeltorres</by>\n<timestamp>2020-07-26T02:01:15.646398</timestamp>\n</audit_log>\n<audit_log>\n<action>created</action>\n<by>kimberly60</by>\n<timestamp>1976-05-19T04:38:55.037965</timestamp>\n</audit_log>\n<audit_log>\n<action>created</action>\n<by>carlaosborne</by>\n<timestamp>2006-08-23T08:19:25.058486</timestamp>\n</audit_log>\n</contract>"
},
{
"role": "assistant",
"content": "Approach:\n1. Analyze the input XML structure: It represents a dictionary with 5 root fields (contract, parties, signed...) and a nesting depth of 3.\n2. Identify the target format requirements: CSV requires strict syntax. Flatten the hierarchical structure. Use dot notation for nested keys in the header.\n3. Map the fields: The root keys will map directly, while nested objects need careful formatting.\n4. Handle data types: Ensure strings, numbers, and booleans are formatted correctly for the target.\n5. Construct the output: serialized the data strictly following the target schema.\n\nOutput:\ncontract,parties,signed,meta_info.generated_at,meta_info.version,meta_info.tags,audit_log\r\norchestrate distributed solutions,\"[\"\"Mckay, Tyler and Gray\"\", \"\"Davis Inc\"\"]\",True,2026-01-23T05:23:25.346467,v5.0,\"[\"\"election\"\", \"\"concern\"\", \"\"social\"\", \"\"series\"\", \"\"old\"\"]\",\"[{\"\"action\"\": \"\"created\"\", \"\"by\"\": \"\"michaeltorres\"\", \"\"timestamp\"\": \"\"2020-07-26T02:01:15.646398\"\"}, {\"\"action\"\": \"\"created\"\", \"\"by\"\": \"\"kimberly60\"\", \"\"timestamp\"\": \"\"1976-05-19T04:38:55.037965\"\"}, {\"\"action\"\": \"\"created\"\", \"\"by\"\": \"\"carlaosborne\"\", \"\"timestamp\"\": \"\"2006-08-23T08:19:25.058486\"\"}]\"\r\n"
}
] | U10BEI_BASE |
[
{
"role": "system",
"content": "You are an expert in TOML format. Analyze the data structure and generate syntactically perfect TOML."
},
{
"role": "user",
"content": "Transform this data from CSV to TOML:\ncompany,revenue,profit,meta_info.generated_at,meta_info.version,meta_info.tags,audit_log\r\nAdkins Inc,58522,71087,2026-01-23T05:23:23.466877,v2.0,\"[\"\"television\"\", \"\"where\"\", \"\"interview\"\", \"\"high\"\", \"\"possible\"\"]\",\"[{\"\"action\"\": \"\"created\"\", \"\"by\"\": \"\"jsmith\"\", \"\"timestamp\"\": \"\"1979-10-28T11:22:35.300270\"\"}]\"\r\n"
},
{
"role": "assistant",
"content": "Approach:\n1. Analyze the input CSV structure: It represents a dictionary with 5 root fields (company, revenue, profit...) and a nesting depth of 3.\n2. Identify the target format requirements: TOML requires strict syntax. Define hierarchy using dot-notation or [section] headers.\n3. Map the fields: The root keys will map directly, while nested objects need careful formatting.\n4. Handle data types: Ensure strings, numbers, and booleans are formatted correctly for the target.\n5. Construct the output: serialized the data strictly following the target schema.\n\nOutput:\ncompany = \"Adkins Inc\"\nrevenue = 58522\nprofit = 71087\n[[audit_log]]\naction = \"created\"\nby = \"jsmith\"\ntimestamp = \"1979-10-28T11:22:35.300270\"\n\n[meta_info]\ngenerated_at = \"2026-01-23T05:23:23.466877\"\nversion = \"v2.0\"\ntags = [ \"television\", \"where\", \"interview\", \"high\", \"possible\",]\n"
}
] | U10BEI_BASE |
[
{
"role": "system",
"content": "You must NOT call tools.\nDo NOT include any extra text, code fences, or markdown.\nFollow the user instructions EXACTLY.\nOutput ONLY CSV content as plain text.\nStart with the header row.\n"
},
{
"role": "user",
"content": "Please output CSV code(Only output the CSV content as plain text, starting with the header row.).\n\nTask:\nGiven a recipe description, extract detailed structured information about the recipe into CSV format.\nFeature Requirements:\n1. Use the exact header fields and order.\n2. Output exactly one data row.\n3. Output exactly two lines: header line + one row.\n\nHeader:\nrecipe_id,recipe_name,chef_name,cuisine_type,preparation_time_minutes,cooking_time_minutes,servings,difficulty_level,main_ingredient,ingredient_list,ingredient_quantities,cooking_method,calories_per_serving,allergen_warnings,vegetarian_friendly,seasonality"
},
{
"role": "assistant",
"content": "recipe_id,recipe_name,chef_name,cuisine_type,preparation_time_minutes,cooking_time_minutes,servings,difficulty_level,main_ingredient,ingredient_list,ingredient_quantities,cooking_method,calories_per_serving,allergen_warnings,vegetarian_friendly,seasonality\nR00685,Stellar Stew 685,Mira Qor,Italian,27,39,1,Easy,Nebula Beans,nebula_beans;star_salt;quantum_onion,200g;5g;1,simmer,893,dairy,No,Winter"
}
] | CSV_TWO_LINE_FIXED |
[
{
"role": "system",
"content": "You must NOT call tools.\nDo NOT include any extra text, code fences, or markdown.\nDo NOT output JSON or YAML.\nOutput MUST be valid TOML.\nThe first line MUST be: [result]\nThen include status and confidence keys under [result].\nThen include [result.fields].\nOutput ONLY TOML as plain text.\n"
},
{
"role": "user",
"content": "Generate a valid TOML output using the fixed template.\nConstraints:\n- Output ONLY TOML, no extra text.\n- The first line must be [result].\n- Include [result.fields] with all required keys.\nRequired keys in fields: id,name,note,flag,score,category,empty\nValues:\n- id: 1186\n- name: item_1186\n- note: ok\n- flag: True\n- score: 0.86\n- category: B\n- empty: filled\n"
},
{
"role": "assistant",
"content": "[result]\nstatus = \"ok\"\nconfidence = 0.800\n\n[result.fields]\n\"category\" = \"B\"\n\"empty\" = \"filled\"\n\"flag\" = true\n\"id\" = 1186\n\"name\" = \"item_1186\"\n\"note\" = \"ok\"\n\"score\" = 0.86"
}
] | TOML_GEN_FIXED_VAR |
[
{
"role": "system",
"content": "You must NOT call tools.\nDo NOT include any extra text, code fences, or markdown.\nFollow the user instructions EXACTLY.\nOutput ONLY CSV content as plain text.\nStart with the header row.\n"
},
{
"role": "user",
"content": "Please output CSV code(Only output the CSV content as plain text, starting with the header row.).\n\nTask:\nGiven a recipe description, extract detailed structured information about the recipe into CSV format.\nFeature Requirements:\n1. Use the exact header fields and order.\n2. Output exactly one data row.\n3. Output exactly two lines: header line + one row.\n\nHeader:\nrecipe_id,recipe_name,chef_name,cuisine_type,preparation_time_minutes,cooking_time_minutes,servings,difficulty_level,main_ingredient,ingredient_list,ingredient_quantities,cooking_method,calories_per_serving,allergen_warnings,vegetarian_friendly,seasonality"
},
{
"role": "assistant",
"content": "recipe_id,recipe_name,chef_name,cuisine_type,preparation_time_minutes,cooking_time_minutes,servings,difficulty_level,main_ingredient,ingredient_list,ingredient_quantities,cooking_method,calories_per_serving,allergen_warnings,vegetarian_friendly,seasonality\nR00929,Stellar Stew 929,Mira Qor,Japanese,13,72,7,Hard,Quantum Onion,nebula_beans;star_salt;quantum_onion,200g;5g;1,simmer,511,dairy,Yes,All Year"
}
] | CSV_TWO_LINE_FIXED |
[
{
"role": "system",
"content": "You must NOT call tools.\nDo NOT include any extra text, code fences, or markdown.\nDo NOT output JSON or YAML.\nOutput MUST be valid TOML.\nThe first line MUST be: [result]\nThen include status and confidence keys under [result].\nThen include [result.fields].\nOutput ONLY TOML as plain text.\n"
},
{
"role": "user",
"content": "Please convert the following JSON code to TOML code.\n\n<code>{\n \"period\": \"Q1\",\n \"sales\": 3775,\n \"region\": \"Deborahbury\",\n \"meta_info\": {\n \"generated_at\": \"2026-01-23T05:23:23.963341\",\n \"version\": \"v5.0\",\n \"tags\": [\n \"lose\",\n \"community\",\n \"since\",\n \"head\",\n \"serve\"\n ]\n },\n \"audit_log\": [\n {\n \"action\": \"created\",\n \"by\": \"griffinmark\",\n \"timestamp\": \"1974-02-10T08:08:11.089686\"\n },\n {\n \"action\": \"created\",\n \"by\": \"moralesstacy\",\n \"timestamp\": \"1972-02-20T17:10:06.489182\"\n },\n {\n \"action\": \"created\",\n \"by\": \"veronica08\",\n \"timestamp\": \"1993-01-30T03:31:20.401491\"\n }\n ]\n}</code>"
},
{
"role": "assistant",
"content": "[result]\nstatus = \"ok\"\nconfidence = 0.800\n\n[result.fields]\n\"audit_log\" = \"[{\\\"action\\\": \\\"created\\\", \\\"by\\\": \\\"griffinmark\\\", \\\"timestamp\\\": \\\"1974-02-10T08:08:11.089686\\\"}, {\\\"action\\\": \\\"created\\\", \\\"by\\\": \\\"moralesstacy\\\", \\\"timestamp\\\": \\\"1972-02-20T17:10:06.489182\\\"}, {\\\"action\\\": \\\"created\\\", \\\"by\\\": \\\"veronica08\\\", \\\"timestamp\\\": \\\"1993-01-30T03:31:20.401491\\\"}]\"\n\"meta_info\" = \"{\\\"generated_at\\\": \\\"2026-01-23T05:23:23.963341\\\", \\\"version\\\": \\\"v5.0\\\", \\\"tags\\\": [\\\"lose\\\", \\\"community\\\", \\\"since\\\", \\\"head\\\", \\\"serve\\\"]}\"\n\"period\" = \"Q1\"\n\"region\" = \"Deborahbury\"\n\"sales\" = 3775"
}
] | CONV_JSON2TOML_TPLFIX |
[
{
"role": "system",
"content": "You are an expert in JSON format. Analyze the data structure and generate syntactically perfect JSON."
},
{
"role": "user",
"content": "Transform this data from TOML to JSON:\nperiod = \"Q1\"\nsales = 2719\nregion = \"Jasonchester\"\n"
},
{
"role": "assistant",
"content": "Approach:\n1. Analyze the input TOML structure: It represents a dictionary with 3 root fields (period, sales, region) and a nesting depth of 1.\n2. Identify the target format requirements: JSON requires strict syntax. Ensure all keys are quoted and nested structures use curly braces.\n3. Map the fields: The root keys will map directly, while nested objects need careful formatting.\n4. Handle data types: Ensure strings, numbers, and booleans are formatted correctly for the target.\n5. Construct the output: serialized the data strictly following the target schema.\n\nOutput:\n{\n \"period\": \"Q1\",\n \"sales\": 2719,\n \"region\": \"Jasonchester\"\n}"
}
] | U10BEI_BASE |
[
{
"role": "system",
"content": "You must NOT call tools.\nDo NOT include any extra text, code fences, or markdown.\nDo NOT output JSON or YAML.\nOutput MUST be valid TOML.\nThe first line MUST be: [result]\nThen include status and confidence keys under [result].\nThen include [result.fields].\nOutput ONLY TOML as plain text.\n"
},
{
"role": "user",
"content": "Generate a valid TOML output using the fixed template.\nConstraints:\n- Output ONLY TOML, no extra text.\n- The first line must be [result].\n- Include [result.fields] with all required keys.\nRequired keys in fields: id,name,note,flag,score,category,empty\nValues:\n- id: 834\n- name: item_834\n- note: ok\n- flag: True\n- score: 0.34\n- category: A\n- empty: filled\n"
},
{
"role": "assistant",
"content": "[result]\nstatus = \"ok\"\nconfidence = 0.800\n\n[result.fields]\n\"category\" = \"A\"\n\"empty\" = \"filled\"\n\"flag\" = true\n\"id\" = 834\n\"name\" = \"item_834\"\n\"note\" = \"ok\"\n\"score\" = 0.34"
}
] | TOML_GEN_FIXED_VAR |
[
{
"role": "system",
"content": "You must NOT call tools.\nDo NOT include any extra text, code fences, or markdown.\nDo NOT output JSON or YAML.\nOutput MUST be valid TOML.\nThe first line MUST be: [result]\nThen include status and confidence keys under [result].\nThen include [result.fields].\nOutput ONLY TOML as plain text.\n"
},
{
"role": "user",
"content": "Please convert the following JSON code to TOML code.\n\n<code>{\n \"company\": \"Green Group\",\n \"revenue\": 169862,\n \"profit\": 60179,\n \"meta_info\": {\n \"generated_at\": \"2026-01-23T05:23:25.360733\",\n \"version\": \"v3.0\",\n \"tags\": [\n \"up\",\n \"cover\",\n \"strong\",\n \"sister\",\n \"hundred\"\n ]\n },\n \"audit_log\": [\n {\n \"action\": \"created\",\n \"by\": \"curtisjohn\",\n \"timestamp\": \"1992-06-25T21:08:25.104142\"\n },\n {\n \"action\": \"created\",\n \"by\": \"kelly44\",\n \"timestamp\": \"1977-05-29T12:19:25.421845\"\n }\n ]\n}</code>"
},
{
"role": "assistant",
"content": "[result]\nstatus = \"ok\"\nconfidence = 0.800\n\n[result.fields]\n\"audit_log\" = \"[{\\\"action\\\": \\\"created\\\", \\\"by\\\": \\\"curtisjohn\\\", \\\"timestamp\\\": \\\"1992-06-25T21:08:25.104142\\\"}, {\\\"action\\\": \\\"created\\\", \\\"by\\\": \\\"kelly44\\\", \\\"timestamp\\\": \\\"1977-05-29T12:19:25.421845\\\"}]\"\n\"company\" = \"Green Group\"\n\"meta_info\" = \"{\\\"generated_at\\\": \\\"2026-01-23T05:23:25.360733\\\", \\\"version\\\": \\\"v3.0\\\", \\\"tags\\\": [\\\"up\\\", \\\"cover\\\", \\\"strong\\\", \\\"sister\\\", \\\"hundred\\\"]}\"\n\"profit\" = 60179\n\"revenue\" = 169862"
}
] | CONV_JSON2TOML_TPLFIX |
[
{
"role": "system",
"content": "You are an expert in CSV format. Analyze the data structure and generate syntactically perfect CSV."
},
{
"role": "user",
"content": "Generate minified CSV for experiment_result (no whitespace)."
},
{
"role": "assistant",
"content": "Approach:\n1. Define the schema: The experiment_result should contain a dictionary with 4 root fields (experiment_id, title, metrics...) and a nesting depth of 2.\n2. Select the format: CSV is chosen. Flatten the hierarchical structure. Use dot notation for nested keys in the header. Output must be minified (no extra whitespace/newlines).\n3. Generate realistic content: Use plausible names, dates, and values appropriate for the domain.\n4. Structure the data: Organize the fields logically with appropriate nesting and hierarchies.\n5. Serialize: Output the final string making sure to escape special characters if necessary and compress output.\n\nOutput:\nexperiment_id,title,metrics.accuracy,metrics.loss,metrics.epoch,config.batch_size,config.optimizer,config.lr\r\nEXP-412677,Analysis of director,0.9772,0.389,15,128,sgd,0.001\r\n"
}
] | U10BEI_BASE |
[
{
"role": "system",
"content": "You must NOT call tools.\nDo NOT include any extra text, code fences, or markdown.\nDo NOT output JSON or YAML.\nOutput MUST be valid TOML.\nThe first line MUST be: [result]\nThen include status and confidence keys under [result].\nThen include [result.fields].\nOutput ONLY TOML as plain text.\n"
},
{
"role": "user",
"content": "Generate a valid TOML output using the fixed template.\nConstraints:\n- Output ONLY TOML, no extra text.\n- The first line must be [result].\n- Include [result.fields] with all required keys.\nRequired keys in fields: id,name,note,flag,score,category,empty\nValues:\n- id: 815\n- name: item_815\n- note: line1\\nline2\n- flag: False\n- score: 0.15\n- category: C\n- empty: filled\n"
},
{
"role": "assistant",
"content": "[result]\nstatus = \"ok\"\nconfidence = 0.800\n\n[result.fields]\n\"category\" = \"C\"\n\"empty\" = \"filled\"\n\"flag\" = false\n\"id\" = 815\n\"name\" = \"item_815\"\n\"note\" = \"line1\\\\nline2\"\n\"score\" = 0.15"
}
] | TOML_GEN_FIXED_VAR |
[
{
"role": "system",
"content": "You are an expert in JSON format. Analyze the data structure and generate syntactically perfect JSON."
},
{
"role": "user",
"content": "Create a JSON representation of error log."
},
{
"role": "assistant",
"content": "Approach:\n1. Define the schema: The error_log should contain a dictionary with 3 root fields (timestamp, level, message) and a nesting depth of 1.\n2. Select the format: JSON is chosen. Ensure all keys are quoted and nested structures use curly braces.\n3. Generate realistic content: Use plausible names, dates, and values appropriate for the domain.\n4. Structure the data: Organize the fields logically with appropriate nesting and hierarchies.\n5. Serialize: Output the final string making sure to escape special characters if necessary.\n\nOutput:\n{\n \"timestamp\": \"2026-01-23T05:23:24.128136\",\n \"level\": \"ERROR\",\n \"message\": \"Information agent wall very reach I.\"\n}"
}
] | U10BEI_BASE |
[
{
"role": "system",
"content": "You must NOT call tools.\nDo NOT include any extra text, code fences, or markdown.\nDo NOT output JSON or YAML.\nOutput MUST be valid TOML.\nThe first line MUST be: [result]\nThen include status and confidence keys under [result].\nThen include [result.fields].\nOutput ONLY TOML as plain text.\n"
},
{
"role": "user",
"content": "Please convert the following JSON code to TOML code.\n\n<code>{\n \"product\": \"rule\",\n \"price\": 81.46,\n \"stock\": 50,\n \"meta_info\": {\n \"generated_at\": \"2026-01-23T05:23:24.140606\",\n \"version\": \"v3.0\",\n \"tags\": [\n \"dream\",\n \"exactly\",\n \"budget\",\n \"hit\",\n \"turn\"\n ]\n },\n \"audit_log\": [\n {\n \"action\": \"created\",\n \"by\": \"hwebster\",\n \"timestamp\": \"2008-07-26T02:20:46.095720\"\n },\n {\n \"action\": \"created\",\n \"by\": \"michael54\",\n \"timestamp\": \"1990-08-12T06:31:11.332181\"\n }\n ]\n}</code>"
},
{
"role": "assistant",
"content": "[result]\nstatus = \"ok\"\nconfidence = 0.800\n\n[result.fields]\n\"audit_log\" = \"[{\\\"action\\\": \\\"created\\\", \\\"by\\\": \\\"hwebster\\\", \\\"timestamp\\\": \\\"2008-07-26T02:20:46.095720\\\"}, {\\\"action\\\": \\\"created\\\", \\\"by\\\": \\\"michael54\\\", \\\"timestamp\\\": \\\"1990-08-12T06:31:11.332181\\\"}]\"\n\"meta_info\" = \"{\\\"generated_at\\\": \\\"2026-01-23T05:23:24.140606\\\", \\\"version\\\": \\\"v3.0\\\", \\\"tags\\\": [\\\"dream\\\", \\\"exactly\\\", \\\"budget\\\", \\\"hit\\\", \\\"turn\\\"]}\"\n\"price\" = 81.46\n\"product\" = \"rule\"\n\"stock\" = 50"
}
] | CONV_JSON2TOML_TPLFIX |
[
{
"role": "system",
"content": "You must NOT call tools.\nDo NOT include any extra text, code fences, or markdown.\nDo NOT output JSON or YAML.\nOutput MUST be valid TOML.\nThe first line MUST be: [result]\nThen include status and confidence keys under [result].\nThen include [result.fields].\nOutput ONLY TOML as plain text.\n"
},
{
"role": "user",
"content": "Please convert the following JSON code to TOML code.\n\n<code>{\n \"user\": \"blakeanderson\",\n \"post\": \"Cause fight rock test if more. Hair sport arm usually member TV. Store town something.\",\n \"likes\": 496,\n \"meta_info\": {\n \"generated_at\": \"2026-01-23T05:23:24.383161\",\n \"version\": \"v5.0\",\n \"tags\": [\n \"affect\",\n \"forget\",\n \"worry\",\n \"from\",\n \"there\"\n ]\n },\n \"audit_log\": [\n {\n \"action\": \"created\",\n \"by\": \"melissawest\",\n \"timestamp\": \"1980-01-02T21:39:34.500222\"\n },\n {\n \"action\": \"created\",\n \"by\": \"garyjones\",\n \"timestamp\": \"2009-02-05T11:46:49.233577\"\n }\n ]\n}</code>"
},
{
"role": "assistant",
"content": "[result]\nstatus = \"ok\"\nconfidence = 0.800\n\n[result.fields]\n\"audit_log\" = \"[{\\\"action\\\": \\\"created\\\", \\\"by\\\": \\\"melissawest\\\", \\\"timestamp\\\": \\\"1980-01-02T21:39:34.500222\\\"}, {\\\"action\\\": \\\"created\\\", \\\"by\\\": \\\"garyjones\\\", \\\"timestamp\\\": \\\"2009-02-05T11:46:49.233577\\\"}]\"\n\"likes\" = 496\n\"meta_info\" = \"{\\\"generated_at\\\": \\\"2026-01-23T05:23:24.383161\\\", \\\"version\\\": \\\"v5.0\\\", \\\"tags\\\": [\\\"affect\\\", \\\"forget\\\", \\\"worry\\\", \\\"from\\\", \\\"there\\\"]}\"\n\"post\" = \"Cause fight rock test if more. Hair sport arm usually member TV. Store town something.\"\n\"user\" = \"blakeanderson\""
}
] | CONV_JSON2TOML_TPLFIX |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.