{"id": "bank_statement_fin_002_p1_classify", "task": "classify", "class": "bank_statement", "doc": "bank_statement_fin_002.pdf", "images": ["images/train/bank_statement/bank_statement_fin_002_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"bank_statement\", \"confidence\": 1.0, \"issuer\": \"HSBC UK\", \"holder\": \"Miss London Andersson\", \"page_marker\": \"\"}]}"}]} {"id": "bank_statement_fin_002_extract", "task": "extract", "class": "bank_statement", "doc": "bank_statement_fin_002.pdf", "images": ["images/train/bank_statement/bank_statement_fin_002_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('bank_statement', 'The document class.')\nfields = (['account_holder_name', 'bank_name', 'statement_period_start', 'statement_period_end', 'ending_balance', 'address', 'bank_account_type', 'bank_account_number', 'transaction_amount', 'transaction_date', 'transaction_type', 'transaction_sender'], 'The exact field names to extract.')\nhints = ({'account_holder_name': 'Name on the account', 'bank_name': 'Name of the bank', 'statement_period_start': 'Start of statement period', 'statement_period_end': 'End of statement period', 'ending_balance': 'Closing balance for the period', 'address': 'Address of the account holder', 'bank_account_type': 'Type of bank account - Checking/Saving', 'bank_account_number': 'Bank account number', 'transaction_amount': 'Transaction amount', 'transaction_date': 'Transaction date', 'transaction_type': 'Transaction type', 'transaction_sender': 'Transaction sender'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"account_holder_name\", \"value\": \"Miss London Andersson\"}, {\"field\": \"bank_name\", \"value\": \"HSBC UK\"}, {\"field\": \"statement_period_start\", \"value\": \"2020-02-26\"}, {\"field\": \"statement_period_end\", \"value\": \"2020-03-25\"}, {\"field\": \"ending_balance\", \"value\": \"35.0\"}, {\"field\": \"address\", \"value\": \"77b/77d Shirland Road, London, W9 2EL\"}, {\"field\": \"bank_account_type\", \"value\": \"Checking\"}, {\"field\": \"bank_account_number\", \"value\": \"82677997\"}, {\"field\": \"transaction_amount\", \"value\": \"5.0\"}, {\"field\": \"transaction_date\", \"value\": \"2020-03-04\"}, {\"field\": \"transaction_type\", \"value\": \"TFR INTERNET TRANSFER\"}, {\"field\": \"transaction_sender\", \"value\": \"\"}]}"}]} {"id": "bank_statement_fin_003_p1_classify", "task": "classify", "class": "bank_statement", "doc": "bank_statement_fin_003.pdf", "images": ["images/train/bank_statement/bank_statement_fin_003_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"bank_statement\", \"confidence\": 1.0, \"issuer\": \"UOB\", \"holder\": \"Mr. NICHOLAS TAN\", \"page_marker\": \"\"}]}"}]} {"id": "bank_statement_fin_003_extract", "task": "extract", "class": "bank_statement", "doc": "bank_statement_fin_003.pdf", "images": ["images/train/bank_statement/bank_statement_fin_003_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('bank_statement', 'The document class.')\nfields = (['account_holder_name', 'bank_name', 'statement_period_start', 'statement_period_end', 'ending_balance', 'address', 'bank_account_type', 'bank_account_number', 'transaction_amount', 'transaction_date', 'transaction_type', 'transaction_sender'], 'The exact field names to extract.')\nhints = ({'account_holder_name': 'Name on the account', 'bank_name': 'Name of the bank', 'statement_period_start': 'Start of statement period', 'statement_period_end': 'End of statement period', 'ending_balance': 'Closing balance for the period', 'address': 'Address of the account holder', 'bank_account_type': 'Type of bank account - Checking/Saving', 'bank_account_number': 'Bank account number', 'transaction_amount': 'Transaction amount', 'transaction_date': 'Transaction date', 'transaction_type': 'Transaction type', 'transaction_sender': 'Transaction sender'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"account_holder_name\", \"value\": \"Mr. NICHOLAS TAN\"}, {\"field\": \"bank_name\", \"value\": \"UOB\"}, {\"field\": \"statement_period_start\", \"value\": \"\"}, {\"field\": \"statement_period_end\", \"value\": \"\"}, {\"field\": \"ending_balance\", \"value\": \"\"}, {\"field\": \"address\", \"value\": \"123 Orchard Plaza #08-25 Singapore 238888\"}, {\"field\": \"bank_account_type\", \"value\": \"\"}, {\"field\": \"bank_account_number\", \"value\": \"5431-9180-1141-1141\"}, {\"field\": \"transaction_amount\", \"value\": \"6.29\"}, {\"field\": \"transaction_date\", \"value\": \"17 Nov\"}, {\"field\": \"transaction_type\", \"value\": \"\"}, {\"field\": \"transaction_sender\", \"value\": \"Lazada SG SINGAPORE\"}]}"}]} {"id": "bank_statement_fin_004_p1_classify", "task": "classify", "class": "bank_statement", "doc": "bank_statement_fin_004.pdf", "images": ["images/train/bank_statement/bank_statement_fin_004_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"bank_statement\", \"confidence\": 1.0, \"issuer\": \"HSBC\", \"holder\": \"Mr Santa Claus\", \"page_marker\": \"\"}]}"}]} {"id": "bank_statement_fin_004_extract", "task": "extract", "class": "bank_statement", "doc": "bank_statement_fin_004.pdf", "images": ["images/train/bank_statement/bank_statement_fin_004_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('bank_statement', 'The document class.')\nfields = (['account_holder_name', 'bank_name', 'statement_period_start', 'statement_period_end', 'ending_balance', 'address', 'bank_account_type', 'bank_account_number', 'transaction_amount', 'transaction_date', 'transaction_type', 'transaction_sender'], 'The exact field names to extract.')\nhints = ({'account_holder_name': 'Name on the account', 'bank_name': 'Name of the bank', 'statement_period_start': 'Start of statement period', 'statement_period_end': 'End of statement period', 'ending_balance': 'Closing balance for the period', 'address': 'Address of the account holder', 'bank_account_type': 'Type of bank account - Checking/Saving', 'bank_account_number': 'Bank account number', 'transaction_amount': 'Transaction amount', 'transaction_date': 'Transaction date', 'transaction_type': 'Transaction type', 'transaction_sender': 'Transaction sender'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"account_holder_name\", \"value\": \"Mr Santa Claus\"}, {\"field\": \"bank_name\", \"value\": \"HSBC\"}, {\"field\": \"statement_period_start\", \"value\": \"\"}, {\"field\": \"statement_period_end\", \"value\": \"\"}, {\"field\": \"ending_balance\", \"value\": \"826.1\"}, {\"field\": \"address\", \"value\": \"Mr S Clause, 00 Lapand Street, Lapand, AA0 DAA\"}, {\"field\": \"bank_account_type\", \"value\": \"Checking\"}, {\"field\": \"bank_account_number\", \"value\": \"00000000\"}, {\"field\": \"transaction_amount\", \"value\": \"3617.12\"}, {\"field\": \"transaction_date\", \"value\": \"26-Jan-0000\"}, {\"field\": \"transaction_type\", \"value\": \"CR\"}, {\"field\": \"transaction_sender\", \"value\": \"LAR HOMECARE\"}]}"}]} {"id": "bank_statement_fin_005_p1_classify", "task": "classify", "class": "bank_statement", "doc": "bank_statement_fin_005.pdf", "images": ["images/train/bank_statement/bank_statement_fin_005_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"bank_statement\", \"confidence\": 1.0, \"issuer\": \"HSBC\", \"holder\": \"Mr CORTES ALVIN\", \"page_marker\": \"\"}]}"}]} {"id": "bank_statement_fin_005_extract", "task": "extract", "class": "bank_statement", "doc": "bank_statement_fin_005.pdf", "images": ["images/train/bank_statement/bank_statement_fin_005_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('bank_statement', 'The document class.')\nfields = (['account_holder_name', 'bank_name', 'statement_period_start', 'statement_period_end', 'ending_balance', 'address', 'bank_account_type', 'bank_account_number', 'transaction_amount', 'transaction_date', 'transaction_type', 'transaction_sender'], 'The exact field names to extract.')\nhints = ({'account_holder_name': 'Name on the account', 'bank_name': 'Name of the bank', 'statement_period_start': 'Start of statement period', 'statement_period_end': 'End of statement period', 'ending_balance': 'Closing balance for the period', 'address': 'Address of the account holder', 'bank_account_type': 'Type of bank account - Checking/Saving', 'bank_account_number': 'Bank account number', 'transaction_amount': 'Transaction amount', 'transaction_date': 'Transaction date', 'transaction_type': 'Transaction type', 'transaction_sender': 'Transaction sender'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"account_holder_name\", \"value\": \"Mr CORTES ALVIN\"}, {\"field\": \"bank_name\", \"value\": \"HSBC\"}, {\"field\": \"statement_period_start\", \"value\": \"2020-08-01\"}, {\"field\": \"statement_period_end\", \"value\": \"2020-08-25\"}, {\"field\": \"ending_balance\", \"value\": \"682.74\"}, {\"field\": \"address\", \"value\": \"30 COLLINSON WALK, LONDON\"}, {\"field\": \"bank_account_type\", \"value\": \"\"}, {\"field\": \"bank_account_number\", \"value\": \"01540225\"}, {\"field\": \"transaction_amount\", \"value\": \"50.0\"}, {\"field\": \"transaction_date\", \"value\": \"2020-08-08\"}, {\"field\": \"transaction_type\", \"value\": \"CR\"}, {\"field\": \"transaction_sender\", \"value\": \"JAME U\"}]}"}]} {"id": "bank_statement_fin_007_p1_classify", "task": "classify", "class": "bank_statement", "doc": "bank_statement_fin_007.pdf", "images": ["images/train/bank_statement/bank_statement_fin_007_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"bank_statement\", \"confidence\": 1.0, \"issuer\": \"YES BANK\", \"holder\": \"MR.TEJAS SAMIR KADAKIA\", \"page_marker\": \"\"}]}"}]} {"id": "bank_statement_fin_007_extract", "task": "extract", "class": "bank_statement", "doc": "bank_statement_fin_007.pdf", "images": ["images/train/bank_statement/bank_statement_fin_007_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('bank_statement', 'The document class.')\nfields = (['account_holder_name', 'bank_name', 'statement_period_start', 'statement_period_end', 'ending_balance', 'address', 'bank_account_type', 'bank_account_number', 'transaction_amount', 'transaction_date', 'transaction_type', 'transaction_sender'], 'The exact field names to extract.')\nhints = ({'account_holder_name': 'Name on the account', 'bank_name': 'Name of the bank', 'statement_period_start': 'Start of statement period', 'statement_period_end': 'End of statement period', 'ending_balance': 'Closing balance for the period', 'address': 'Address of the account holder', 'bank_account_type': 'Type of bank account - Checking/Saving', 'bank_account_number': 'Bank account number', 'transaction_amount': 'Transaction amount', 'transaction_date': 'Transaction date', 'transaction_type': 'Transaction type', 'transaction_sender': 'Transaction sender'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"account_holder_name\", \"value\": \"MR.TEJAS SAMIR KADAKIA\"}, {\"field\": \"bank_name\", \"value\": \"YES BANK\"}, {\"field\": \"statement_period_start\", \"value\": \"2018-11-01\"}, {\"field\": \"statement_period_end\", \"value\": \"2019-05-20\"}, {\"field\": \"ending_balance\", \"value\": \"\"}, {\"field\": \"address\", \"value\": \"605 GANGA GANGA YAMUNA CHS LTD SD MARG NANCY COLONY NEAR NANCY BUS DEPOT GREEN VIEW COMPLEX BORIVALI EAST S O MUMBAI MUMBAI MAHARASHTRA 400066\"}, {\"field\": \"bank_account_type\", \"value\": \"SA - SMART SALARY ADVANTAGE\"}, {\"field\": \"bank_account_number\", \"value\": \"046991800006712\"}, {\"field\": \"transaction_amount\", \"value\": \"1250.0\"}, {\"field\": \"transaction_date\", \"value\": \"2018-11-10\"}, {\"field\": \"transaction_type\", \"value\": \"Withdrawal\"}, {\"field\": \"transaction_sender\", \"value\": \"PCA:4904202659:037022010410165:SOMANDA VLNEYARDS NASHIK IND-831410163315\"}]}"}]} {"id": "bank_statement_fin_008_p1_classify", "task": "classify", "class": "bank_statement", "doc": "bank_statement_fin_008.pdf", "images": ["images/train/bank_statement/bank_statement_fin_008_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"bank_statement\", \"confidence\": 1.0, \"issuer\": \"LLOYDS BANK\", \"holder\": \"Oleh Beshleu\", \"page_marker\": \"\"}]}"}]} {"id": "bank_statement_fin_008_extract", "task": "extract", "class": "bank_statement", "doc": "bank_statement_fin_008.pdf", "images": ["images/train/bank_statement/bank_statement_fin_008_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('bank_statement', 'The document class.')\nfields = (['account_holder_name', 'bank_name', 'statement_period_start', 'statement_period_end', 'ending_balance', 'address', 'bank_account_type', 'bank_account_number', 'transaction_amount', 'transaction_date', 'transaction_type', 'transaction_sender'], 'The exact field names to extract.')\nhints = ({'account_holder_name': 'Name on the account', 'bank_name': 'Name of the bank', 'statement_period_start': 'Start of statement period', 'statement_period_end': 'End of statement period', 'ending_balance': 'Closing balance for the period', 'address': 'Address of the account holder', 'bank_account_type': 'Type of bank account - Checking/Saving', 'bank_account_number': 'Bank account number', 'transaction_amount': 'Transaction amount', 'transaction_date': 'Transaction date', 'transaction_type': 'Transaction type', 'transaction_sender': 'Transaction sender'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"account_holder_name\", \"value\": \"Oleh Beshleu\"}, {\"field\": \"bank_name\", \"value\": \"LLOYDS BANK\"}, {\"field\": \"statement_period_start\", \"value\": \"2019-09-01\"}, {\"field\": \"statement_period_end\", \"value\": \"2019-11-30\"}, {\"field\": \"ending_balance\", \"value\": \"2062.05\"}, {\"field\": \"address\", \"value\": \"293 STRONE ROAD, MANOR PARK(BARKING AND DAGENHAM), LONDON, E12 6TR\"}, {\"field\": \"bank_account_type\", \"value\": \"\"}, {\"field\": \"bank_account_number\", \"value\": \"45201526\"}, {\"field\": \"transaction_amount\", \"value\": \"300.0\"}, {\"field\": \"transaction_date\", \"value\": \"2019-09-04\"}, {\"field\": \"transaction_type\", \"value\": \"CPT\"}, {\"field\": \"transaction_sender\", \"value\": \"LNK COOPERATIVE SW CD 4821 08DEC19\"}]}"}]} {"id": "bank_statement_fin_009_p1_classify", "task": "classify", "class": "bank_statement", "doc": "bank_statement_fin_009.pdf", "images": ["images/train/bank_statement/bank_statement_fin_009_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"bank_statement\", \"confidence\": 1.0, \"issuer\": \"HSBC\", \"holder\": \"JOHN CITIZEN\", \"page_marker\": \"\"}]}"}]} {"id": "bank_statement_fin_009_extract", "task": "extract", "class": "bank_statement", "doc": "bank_statement_fin_009.pdf", "images": ["images/train/bank_statement/bank_statement_fin_009_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('bank_statement', 'The document class.')\nfields = (['account_holder_name', 'bank_name', 'statement_period_start', 'statement_period_end', 'ending_balance', 'address', 'bank_account_type', 'bank_account_number', 'transaction_amount', 'transaction_date', 'transaction_type', 'transaction_sender'], 'The exact field names to extract.')\nhints = ({'account_holder_name': 'Name on the account', 'bank_name': 'Name of the bank', 'statement_period_start': 'Start of statement period', 'statement_period_end': 'End of statement period', 'ending_balance': 'Closing balance for the period', 'address': 'Address of the account holder', 'bank_account_type': 'Type of bank account - Checking/Saving', 'bank_account_number': 'Bank account number', 'transaction_amount': 'Transaction amount', 'transaction_date': 'Transaction date', 'transaction_type': 'Transaction type', 'transaction_sender': 'Transaction sender'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"account_holder_name\", \"value\": \"JOHN CITIZEN\"}, {\"field\": \"bank_name\", \"value\": \"HSBC\"}, {\"field\": \"statement_period_start\", \"value\": \"\"}, {\"field\": \"statement_period_end\", \"value\": \"2019-08-31\"}, {\"field\": \"ending_balance\", \"value\": \"13063611.0\"}, {\"field\": \"address\", \"value\": \"185 Giang Vo, Dong Da District\"}, {\"field\": \"bank_account_type\", \"value\": \"Saving\"}, {\"field\": \"bank_account_number\", \"value\": \"002-096451-056\"}, {\"field\": \"transaction_amount\", \"value\": \"8400000.0\"}, {\"field\": \"transaction_date\", \"value\": \"2019-08-04\"}, {\"field\": \"transaction_type\", \"value\": \"Deposit\"}, {\"field\": \"transaction_sender\", \"value\": \"CHI TRANG NOT TIEH\"}]}"}]} {"id": "bank_statement_fin_010_p1_classify", "task": "classify", "class": "bank_statement", "doc": "bank_statement_fin_010.pdf", "images": ["images/train/bank_statement/bank_statement_fin_010_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"bank_statement\", \"confidence\": 1.0, \"issuer\": \"U.S. Bank National Association\", \"holder\": \"AFTAB AHMED\", \"page_marker\": \"\"}]}"}]} {"id": "bank_statement_fin_010_extract", "task": "extract", "class": "bank_statement", "doc": "bank_statement_fin_010.pdf", "images": ["images/train/bank_statement/bank_statement_fin_010_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('bank_statement', 'The document class.')\nfields = (['account_holder_name', 'bank_name', 'statement_period_start', 'statement_period_end', 'ending_balance', 'address', 'bank_account_type', 'bank_account_number', 'transaction_amount', 'transaction_date', 'transaction_type', 'transaction_sender'], 'The exact field names to extract.')\nhints = ({'account_holder_name': 'Name on the account', 'bank_name': 'Name of the bank', 'statement_period_start': 'Start of statement period', 'statement_period_end': 'End of statement period', 'ending_balance': 'Closing balance for the period', 'address': 'Address of the account holder', 'bank_account_type': 'Type of bank account - Checking/Saving', 'bank_account_number': 'Bank account number', 'transaction_amount': 'Transaction amount', 'transaction_date': 'Transaction date', 'transaction_type': 'Transaction type', 'transaction_sender': 'Transaction sender'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"account_holder_name\", \"value\": \"AFTAB AHMED\"}, {\"field\": \"bank_name\", \"value\": \"U.S. Bank National Association\"}, {\"field\": \"statement_period_start\", \"value\": \"\"}, {\"field\": \"statement_period_end\", \"value\": \"\"}, {\"field\": \"ending_balance\", \"value\": \"1263.46\"}, {\"field\": \"address\", \"value\": \"441 Glen Falls Rd, Highlands, NC, 28741\"}, {\"field\": \"bank_account_type\", \"value\": \"Checking\"}, {\"field\": \"bank_account_number\", \"value\": \"1-575-1553-5135\"}, {\"field\": \"transaction_amount\", \"value\": \"0.01\"}, {\"field\": \"transaction_date\", \"value\": \"\"}, {\"field\": \"transaction_type\", \"value\": \"Electronic Deposit\"}, {\"field\": \"transaction_sender\", \"value\": \"From Turo\"}]}"}]} {"id": "bank_statement_fin_011_p1_classify", "task": "classify", "class": "bank_statement", "doc": "bank_statement_fin_011.pdf", "images": ["images/train/bank_statement/bank_statement_fin_011_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"bank_statement\", \"confidence\": 1.0, \"issuer\": \"HNB\", \"holder\": \"MR. VELLIYAMPALAM SITHAMPARANATHAN\", \"page_marker\": \"\"}]}"}]} {"id": "bank_statement_fin_011_extract", "task": "extract", "class": "bank_statement", "doc": "bank_statement_fin_011.pdf", "images": ["images/train/bank_statement/bank_statement_fin_011_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('bank_statement', 'The document class.')\nfields = (['account_holder_name', 'bank_name', 'statement_period_start', 'statement_period_end', 'ending_balance', 'address', 'bank_account_type', 'bank_account_number', 'transaction_amount', 'transaction_date', 'transaction_type', 'transaction_sender'], 'The exact field names to extract.')\nhints = ({'account_holder_name': 'Name on the account', 'bank_name': 'Name of the bank', 'statement_period_start': 'Start of statement period', 'statement_period_end': 'End of statement period', 'ending_balance': 'Closing balance for the period', 'address': 'Address of the account holder', 'bank_account_type': 'Type of bank account - Checking/Saving', 'bank_account_number': 'Bank account number', 'transaction_amount': 'Transaction amount', 'transaction_date': 'Transaction date', 'transaction_type': 'Transaction type', 'transaction_sender': 'Transaction sender'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"account_holder_name\", \"value\": \"MR. VELLIYAMPALAM SITHAMPARANATHAN\"}, {\"field\": \"bank_name\", \"value\": \"HNB\"}, {\"field\": \"statement_period_start\", \"value\": \"2024-09-01\"}, {\"field\": \"statement_period_end\", \"value\": \"2024-12-29\"}, {\"field\": \"ending_balance\", \"value\": \"3245237.88\"}, {\"field\": \"address\", \"value\": \"ARALYSOUTH, VADDUKODDAI, KILINOCHCHI, SRI LANKA\"}, {\"field\": \"bank_account_type\", \"value\": \"\"}, {\"field\": \"bank_account_number\", \"value\": \"012020170975\"}, {\"field\": \"transaction_amount\", \"value\": \"199.0\"}, {\"field\": \"transaction_date\", \"value\": \"2024-09-01\"}, {\"field\": \"transaction_type\", \"value\": \"B/F\"}, {\"field\": \"transaction_sender\", \"value\": \"COM/DIALOG AXIATA PLC\"}]}"}]} {"id": "bank_statement_fin_012_p1_classify", "task": "classify", "class": "bank_statement", "doc": "bank_statement_fin_012.pdf", "images": ["images/train/bank_statement/bank_statement_fin_012_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"bank_statement\", \"confidence\": 1.0, \"issuer\": \"DBS Bank Ltd\", \"holder\": \"LEW TZE CHI\", \"page_marker\": \"\"}]}"}]} {"id": "bank_statement_fin_012_extract", "task": "extract", "class": "bank_statement", "doc": "bank_statement_fin_012.pdf", "images": ["images/train/bank_statement/bank_statement_fin_012_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('bank_statement', 'The document class.')\nfields = (['account_holder_name', 'bank_name', 'statement_period_start', 'statement_period_end', 'ending_balance', 'address', 'bank_account_type', 'bank_account_number', 'transaction_amount', 'transaction_date', 'transaction_type', 'transaction_sender'], 'The exact field names to extract.')\nhints = ({'account_holder_name': 'Name on the account', 'bank_name': 'Name of the bank', 'statement_period_start': 'Start of statement period', 'statement_period_end': 'End of statement period', 'ending_balance': 'Closing balance for the period', 'address': 'Address of the account holder', 'bank_account_type': 'Type of bank account - Checking/Saving', 'bank_account_number': 'Bank account number', 'transaction_amount': 'Transaction amount', 'transaction_date': 'Transaction date', 'transaction_type': 'Transaction type', 'transaction_sender': 'Transaction sender'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"account_holder_name\", \"value\": \"LEW TZE CHI\"}, {\"field\": \"bank_name\", \"value\": \"DBS Bank Ltd\"}, {\"field\": \"statement_period_start\", \"value\": \"\"}, {\"field\": \"statement_period_end\", \"value\": \"2017-05-31\"}, {\"field\": \"ending_balance\", \"value\": \"23388.28\"}, {\"field\": \"address\", \"value\": \"78A, LORONG MELAYU, SINGAPORE 416983\"}, {\"field\": \"bank_account_type\", \"value\": \"Saving\"}, {\"field\": \"bank_account_number\", \"value\": \"007-65076-0\"}, {\"field\": \"transaction_amount\", \"value\": \"\"}, {\"field\": \"transaction_date\", \"value\": \"\"}, {\"field\": \"transaction_type\", \"value\": \"\"}, {\"field\": \"transaction_sender\", \"value\": \"\"}]}"}]} {"id": "bank_statement_fin_013_p1_classify", "task": "classify", "class": "bank_statement", "doc": "bank_statement_fin_013.pdf", "images": ["images/train/bank_statement/bank_statement_fin_013_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"bank_statement\", \"confidence\": 1.0, \"issuer\": \"DBS Bank Ltd\", \"holder\": \"JOHN CITIZEN\", \"page_marker\": \"\"}]}"}]} {"id": "bank_statement_fin_013_extract", "task": "extract", "class": "bank_statement", "doc": "bank_statement_fin_013.pdf", "images": ["images/train/bank_statement/bank_statement_fin_013_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('bank_statement', 'The document class.')\nfields = (['account_holder_name', 'bank_name', 'statement_period_start', 'statement_period_end', 'ending_balance', 'address', 'bank_account_type', 'bank_account_number', 'transaction_amount', 'transaction_date', 'transaction_type', 'transaction_sender'], 'The exact field names to extract.')\nhints = ({'account_holder_name': 'Name on the account', 'bank_name': 'Name of the bank', 'statement_period_start': 'Start of statement period', 'statement_period_end': 'End of statement period', 'ending_balance': 'Closing balance for the period', 'address': 'Address of the account holder', 'bank_account_type': 'Type of bank account - Checking/Saving', 'bank_account_number': 'Bank account number', 'transaction_amount': 'Transaction amount', 'transaction_date': 'Transaction date', 'transaction_type': 'Transaction type', 'transaction_sender': 'Transaction sender'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"account_holder_name\", \"value\": \"JOHN CITIZEN\"}, {\"field\": \"bank_name\", \"value\": \"DBS Bank Ltd\"}, {\"field\": \"statement_period_start\", \"value\": \"\"}, {\"field\": \"statement_period_end\", \"value\": \"2021-10-10\"}, {\"field\": \"ending_balance\", \"value\": \"87.0\"}, {\"field\": \"address\", \"value\": \"XXXXXXXXXXX #1204 SINGAPORE 5484\"}, {\"field\": \"bank_account_type\", \"value\": \"\"}, {\"field\": \"bank_account_number\", \"value\": \"315-69697-7\"}, {\"field\": \"transaction_amount\", \"value\": \"6.0\"}, {\"field\": \"transaction_date\", \"value\": \"2021-09-12\"}, {\"field\": \"transaction_type\", \"value\": \"\"}, {\"field\": \"transaction_sender\", \"value\": \"GRAB*1505046607484863\"}]}"}]} {"id": "bank_statement_fin_016_p1_classify", "task": "classify", "class": "bank_statement", "doc": "bank_statement_fin_016.pdf", "images": ["images/train/bank_statement/bank_statement_fin_016_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"bank_statement\", \"confidence\": 1.0, \"issuer\": \"DBS/POSB\", \"holder\": \"Customer Name 1 Customer Name 2 Customer Name 3\", \"page_marker\": \"\"}]}"}]} {"id": "bank_statement_fin_016_extract", "task": "extract", "class": "bank_statement", "doc": "bank_statement_fin_016.pdf", "images": ["images/train/bank_statement/bank_statement_fin_016_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('bank_statement', 'The document class.')\nfields = (['account_holder_name', 'bank_name', 'statement_period_start', 'statement_period_end', 'ending_balance', 'address', 'bank_account_type', 'bank_account_number', 'transaction_amount', 'transaction_date', 'transaction_type', 'transaction_sender'], 'The exact field names to extract.')\nhints = ({'account_holder_name': 'Name on the account', 'bank_name': 'Name of the bank', 'statement_period_start': 'Start of statement period', 'statement_period_end': 'End of statement period', 'ending_balance': 'Closing balance for the period', 'address': 'Address of the account holder', 'bank_account_type': 'Type of bank account - Checking/Saving', 'bank_account_number': 'Bank account number', 'transaction_amount': 'Transaction amount', 'transaction_date': 'Transaction date', 'transaction_type': 'Transaction type', 'transaction_sender': 'Transaction sender'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"account_holder_name\", \"value\": \"Customer Name 1 Customer Name 2 Customer Name 3\"}, {\"field\": \"bank_name\", \"value\": \"DBS/POSB\"}, {\"field\": \"statement_period_start\", \"value\": \"\"}, {\"field\": \"statement_period_end\", \"value\": \"2020-12-31\"}, {\"field\": \"ending_balance\", \"value\": \"\"}, {\"field\": \"address\", \"value\": \"Address Line 1 Address Line 2 Address Line 3 Address Line 4\"}, {\"field\": \"bank_account_type\", \"value\": \"Saving\"}, {\"field\": \"bank_account_number\", \"value\": \"\"}, {\"field\": \"transaction_amount\", \"value\": \"\"}, {\"field\": \"transaction_date\", \"value\": \"\"}, {\"field\": \"transaction_type\", \"value\": \"\"}, {\"field\": \"transaction_sender\", \"value\": \"\"}]}"}]} {"id": "bank_statement_fin_017_p1_classify", "task": "classify", "class": "bank_statement", "doc": "bank_statement_fin_017.pdf", "images": ["images/train/bank_statement/bank_statement_fin_017_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"bank_statement\", \"confidence\": 1.0, \"issuer\": \"DBS Bank India Limited\", \"holder\": \"MOHAMMED EKRAM\", \"page_marker\": \"\"}]}"}]} {"id": "bank_statement_fin_017_extract", "task": "extract", "class": "bank_statement", "doc": "bank_statement_fin_017.pdf", "images": ["images/train/bank_statement/bank_statement_fin_017_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('bank_statement', 'The document class.')\nfields = (['account_holder_name', 'bank_name', 'statement_period_start', 'statement_period_end', 'ending_balance', 'address', 'bank_account_type', 'bank_account_number', 'transaction_amount', 'transaction_date', 'transaction_type', 'transaction_sender'], 'The exact field names to extract.')\nhints = ({'account_holder_name': 'Name on the account', 'bank_name': 'Name of the bank', 'statement_period_start': 'Start of statement period', 'statement_period_end': 'End of statement period', 'ending_balance': 'Closing balance for the period', 'address': 'Address of the account holder', 'bank_account_type': 'Type of bank account - Checking/Saving', 'bank_account_number': 'Bank account number', 'transaction_amount': 'Transaction amount', 'transaction_date': 'Transaction date', 'transaction_type': 'Transaction type', 'transaction_sender': 'Transaction sender'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"account_holder_name\", \"value\": \"MOHAMMED EKRAM\"}, {\"field\": \"bank_name\", \"value\": \"DBS Bank India Limited\"}, {\"field\": \"statement_period_start\", \"value\": \"2021-06-01\"}, {\"field\": \"statement_period_end\", \"value\": \"2021-06-30\"}, {\"field\": \"ending_balance\", \"value\": \"2.31\"}, {\"field\": \"address\", \"value\": \"H NO 3-1-295 TO 296 NIMBOLIADDA HIMAYATHNAGAR HYDERABAD-500027 TELANGANA, INDIA\"}, {\"field\": \"bank_account_type\", \"value\": \"Saving\"}, {\"field\": \"bank_account_number\", \"value\": \"881025899771\"}, {\"field\": \"transaction_amount\", \"value\": \"\"}, {\"field\": \"transaction_date\", \"value\": \"\"}, {\"field\": \"transaction_type\", \"value\": \"\"}, {\"field\": \"transaction_sender\", \"value\": \"\"}]}"}]} {"id": "bank_statement_fin_018_p1_classify", "task": "classify", "class": "bank_statement", "doc": "bank_statement_fin_018.pdf", "images": ["images/train/bank_statement/bank_statement_fin_018_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"bank_statement\", \"confidence\": 1.0, \"issuer\": \"IDFC FIRST Bank\", \"holder\": \"Mr. Chandrakant Yuvraj Bansode\", \"page_marker\": \"\"}]}"}]} {"id": "bank_statement_fin_018_extract", "task": "extract", "class": "bank_statement", "doc": "bank_statement_fin_018.pdf", "images": ["images/train/bank_statement/bank_statement_fin_018_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('bank_statement', 'The document class.')\nfields = (['account_holder_name', 'bank_name', 'statement_period_start', 'statement_period_end', 'ending_balance', 'address', 'bank_account_type', 'bank_account_number', 'transaction_amount', 'transaction_date', 'transaction_type', 'transaction_sender'], 'The exact field names to extract.')\nhints = ({'account_holder_name': 'Name on the account', 'bank_name': 'Name of the bank', 'statement_period_start': 'Start of statement period', 'statement_period_end': 'End of statement period', 'ending_balance': 'Closing balance for the period', 'address': 'Address of the account holder', 'bank_account_type': 'Type of bank account - Checking/Saving', 'bank_account_number': 'Bank account number', 'transaction_amount': 'Transaction amount', 'transaction_date': 'Transaction date', 'transaction_type': 'Transaction type', 'transaction_sender': 'Transaction sender'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"account_holder_name\", \"value\": \"Mr. Chandrakant Yuvraj Bansode\"}, {\"field\": \"bank_name\", \"value\": \"IDFC FIRST Bank\"}, {\"field\": \"statement_period_start\", \"value\": \"2018-04-01\"}, {\"field\": \"statement_period_end\", \"value\": \"2019-03-31\"}, {\"field\": \"ending_balance\", \"value\": \"45.12\"}, {\"field\": \"address\", \"value\": \"S/O: Yuvraj Bansode, Room NO-307 Bldg No-8 Pancharatan CHS Chandrabhaga, Lallubhai Compound, Mankhurd, Mumbai, Mumbai, Mumbai - 400043\"}, {\"field\": \"bank_account_type\", \"value\": \"Saving\"}, {\"field\": \"bank_account_number\", \"value\": \"10009455592\"}, {\"field\": \"transaction_amount\", \"value\": \"\"}, {\"field\": \"transaction_date\", \"value\": \"\"}, {\"field\": \"transaction_type\", \"value\": \"\"}, {\"field\": \"transaction_sender\", \"value\": \"\"}]}"}]} {"id": "bank_statement_fin_019_p1_classify", "task": "classify", "class": "bank_statement", "doc": "bank_statement_fin_019.pdf", "images": ["images/train/bank_statement/bank_statement_fin_019_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"bank_statement\", \"confidence\": 1.0, \"issuer\": \"Development Bank of Singapore Ltd\", \"holder\": \"MRS MARY FUENTES\", \"page_marker\": \"\"}]}"}]} {"id": "bank_statement_fin_019_extract", "task": "extract", "class": "bank_statement", "doc": "bank_statement_fin_019.pdf", "images": ["images/train/bank_statement/bank_statement_fin_019_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('bank_statement', 'The document class.')\nfields = (['account_holder_name', 'bank_name', 'statement_period_start', 'statement_period_end', 'ending_balance', 'address', 'bank_account_type', 'bank_account_number', 'transaction_amount', 'transaction_date', 'transaction_type', 'transaction_sender'], 'The exact field names to extract.')\nhints = ({'account_holder_name': 'Name on the account', 'bank_name': 'Name of the bank', 'statement_period_start': 'Start of statement period', 'statement_period_end': 'End of statement period', 'ending_balance': 'Closing balance for the period', 'address': 'Address of the account holder', 'bank_account_type': 'Type of bank account - Checking/Saving', 'bank_account_number': 'Bank account number', 'transaction_amount': 'Transaction amount', 'transaction_date': 'Transaction date', 'transaction_type': 'Transaction type', 'transaction_sender': 'Transaction sender'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"account_holder_name\", \"value\": \"MRS MARY FUENTES\"}, {\"field\": \"bank_name\", \"value\": \"Development Bank of Singapore Ltd\"}, {\"field\": \"statement_period_start\", \"value\": \"2024-11-23\"}, {\"field\": \"statement_period_end\", \"value\": \"2024-11-27\"}, {\"field\": \"ending_balance\", \"value\": \"306.0\"}, {\"field\": \"address\", \"value\": \"12 Marina Boulevard, DBS Asia Central, Marina Bay Financial Centre Tower 3, Singapore 018982\"}, {\"field\": \"bank_account_type\", \"value\": \"Checking\"}, {\"field\": \"bank_account_number\", \"value\": \"271-918298-0\"}, {\"field\": \"transaction_amount\", \"value\": \"306.0\"}, {\"field\": \"transaction_date\", \"value\": \"2024-11-23\"}, {\"field\": \"transaction_type\", \"value\": \"Debit\"}, {\"field\": \"transaction_sender\", \"value\": \"James morgan\"}]}"}]} {"id": "bank_statement_fin_023_p1_classify", "task": "classify", "class": "bank_statement", "doc": "bank_statement_fin_023.pdf", "images": ["images/train/bank_statement/bank_statement_fin_023_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"bank_statement\", \"confidence\": 1.0, \"issuer\": \"Commerce Bank\", \"holder\": \"Jane Customer\", \"page_marker\": \"\"}]}"}]} {"id": "bank_statement_fin_023_extract", "task": "extract", "class": "bank_statement", "doc": "bank_statement_fin_023.pdf", "images": ["images/train/bank_statement/bank_statement_fin_023_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('bank_statement', 'The document class.')\nfields = (['account_holder_name', 'bank_name', 'statement_period_start', 'statement_period_end', 'ending_balance', 'address', 'bank_account_type', 'bank_account_number', 'transaction_amount', 'transaction_date', 'transaction_type', 'transaction_sender'], 'The exact field names to extract.')\nhints = ({'account_holder_name': 'Name on the account', 'bank_name': 'Name of the bank', 'statement_period_start': 'Start of statement period', 'statement_period_end': 'End of statement period', 'ending_balance': 'Closing balance for the period', 'address': 'Address of the account holder', 'bank_account_type': 'Type of bank account - Checking/Saving', 'bank_account_number': 'Bank account number', 'transaction_amount': 'Transaction amount', 'transaction_date': 'Transaction date', 'transaction_type': 'Transaction type', 'transaction_sender': 'Transaction sender'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"account_holder_name\", \"value\": \"Jane Customer\"}, {\"field\": \"bank_name\", \"value\": \"Commerce Bank\"}, {\"field\": \"statement_period_start\", \"value\": \"2003-05-03\"}, {\"field\": \"statement_period_end\", \"value\": \"2003-06-05\"}, {\"field\": \"ending_balance\", \"value\": \"10521.19\"}, {\"field\": \"address\", \"value\": \"1234 Anywhere Dr., Small Town, MO 12345-6789\"}, {\"field\": \"bank_account_type\", \"value\": \"Checking\"}, {\"field\": \"bank_account_number\", \"value\": \"000009752\"}, {\"field\": \"transaction_amount\", \"value\": \"3615.08\"}, {\"field\": \"transaction_date\", \"value\": \"2003-05-15\"}, {\"field\": \"transaction_type\", \"value\": \"Deposit\"}, {\"field\": \"transaction_sender\", \"value\": \"\"}]}"}]} {"id": "bank_statement_fin_024_p1_classify", "task": "classify", "class": "bank_statement", "doc": "bank_statement_fin_024.pdf", "images": ["images/train/bank_statement/bank_statement_fin_024_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"bank_statement\", \"confidence\": 1.0, \"issuer\": \"TD Bank\", \"holder\": \"DONNELL BUSH\", \"page_marker\": \"\"}]}"}]} {"id": "bank_statement_fin_024_extract", "task": "extract", "class": "bank_statement", "doc": "bank_statement_fin_024.pdf", "images": ["images/train/bank_statement/bank_statement_fin_024_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('bank_statement', 'The document class.')\nfields = (['account_holder_name', 'bank_name', 'statement_period_start', 'statement_period_end', 'ending_balance', 'address', 'bank_account_type', 'bank_account_number', 'transaction_amount', 'transaction_date', 'transaction_type', 'transaction_sender'], 'The exact field names to extract.')\nhints = ({'account_holder_name': 'Name on the account', 'bank_name': 'Name of the bank', 'statement_period_start': 'Start of statement period', 'statement_period_end': 'End of statement period', 'ending_balance': 'Closing balance for the period', 'address': 'Address of the account holder', 'bank_account_type': 'Type of bank account - Checking/Saving', 'bank_account_number': 'Bank account number', 'transaction_amount': 'Transaction amount', 'transaction_date': 'Transaction date', 'transaction_type': 'Transaction type', 'transaction_sender': 'Transaction sender'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"account_holder_name\", \"value\": \"DONNELL BUSH\"}, {\"field\": \"bank_name\", \"value\": \"TD Bank\"}, {\"field\": \"statement_period_start\", \"value\": \"2021-03-01\"}, {\"field\": \"statement_period_end\", \"value\": \"2021-04-01\"}, {\"field\": \"ending_balance\", \"value\": \"13669.36\"}, {\"field\": \"address\", \"value\": \"8471 AVALON DR, RIVERDALE, GA 30274-4405\"}, {\"field\": \"bank_account_type\", \"value\": \"Checking\"}, {\"field\": \"bank_account_number\", \"value\": \"237-7323560\"}, {\"field\": \"transaction_amount\", \"value\": \"220.0\"}, {\"field\": \"transaction_date\", \"value\": \"2021-03-19\"}, {\"field\": \"transaction_type\", \"value\": \"ATM CHECK DEPOSIT\"}, {\"field\": \"transaction_sender\", \"value\": \"230 FORSGATE DR JAMESBURG NJ\"}]}"}]} {"id": "bank_statement_fin_026_p1_classify", "task": "classify", "class": "bank_statement", "doc": "bank_statement_fin_026.pdf", "images": ["images/train/bank_statement/bank_statement_fin_026_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"bank_statement\", \"confidence\": 1.0, \"issuer\": \"DBS Bank Ltd\", \"holder\": \"CHIN SING WEI\", \"page_marker\": \"\"}]}"}]} {"id": "bank_statement_fin_026_extract", "task": "extract", "class": "bank_statement", "doc": "bank_statement_fin_026.pdf", "images": ["images/train/bank_statement/bank_statement_fin_026_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('bank_statement', 'The document class.')\nfields = (['account_holder_name', 'bank_name', 'statement_period_start', 'statement_period_end', 'ending_balance', 'address', 'bank_account_type', 'bank_account_number', 'transaction_amount', 'transaction_date', 'transaction_type', 'transaction_sender'], 'The exact field names to extract.')\nhints = ({'account_holder_name': 'Name on the account', 'bank_name': 'Name of the bank', 'statement_period_start': 'Start of statement period', 'statement_period_end': 'End of statement period', 'ending_balance': 'Closing balance for the period', 'address': 'Address of the account holder', 'bank_account_type': 'Type of bank account - Checking/Saving', 'bank_account_number': 'Bank account number', 'transaction_amount': 'Transaction amount', 'transaction_date': 'Transaction date', 'transaction_type': 'Transaction type', 'transaction_sender': 'Transaction sender'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"account_holder_name\", \"value\": \"CHIN SING WEI\"}, {\"field\": \"bank_name\", \"value\": \"DBS Bank Ltd\"}, {\"field\": \"statement_period_start\", \"value\": \"\"}, {\"field\": \"statement_period_end\", \"value\": \"2014-10-09\"}, {\"field\": \"ending_balance\", \"value\": \"192.68\"}, {\"field\": \"address\", \"value\": \"BLK 546 HOUGANG STREET 51 #15-200 SINGAPORE 530546\"}, {\"field\": \"bank_account_type\", \"value\": \"\"}, {\"field\": \"bank_account_number\", \"value\": \"186-12669-6\"}, {\"field\": \"transaction_amount\", \"value\": \"39.6\"}, {\"field\": \"transaction_date\", \"value\": \"2014-09-11\"}, {\"field\": \"transaction_type\", \"value\": \"\"}, {\"field\": \"transaction_sender\", \"value\": \"FAIRPRICE XTRA-NEX MAL SINGAPORE\"}]}"}]} {"id": "bank_statement_fin_028_p1_classify", "task": "classify", "class": "bank_statement", "doc": "bank_statement_fin_028.pdf", "images": ["images/train/bank_statement/bank_statement_fin_028_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"bank_statement\", \"confidence\": 1.0, \"issuer\": \"BNI\", \"holder\": \"John Smith\", \"page_marker\": \"\"}]}"}]} {"id": "bank_statement_fin_028_extract", "task": "extract", "class": "bank_statement", "doc": "bank_statement_fin_028.pdf", "images": ["images/train/bank_statement/bank_statement_fin_028_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('bank_statement', 'The document class.')\nfields = (['account_holder_name', 'bank_name', 'statement_period_start', 'statement_period_end', 'ending_balance', 'address', 'bank_account_type', 'bank_account_number', 'transaction_amount', 'transaction_date', 'transaction_type', 'transaction_sender'], 'The exact field names to extract.')\nhints = ({'account_holder_name': 'Name on the account', 'bank_name': 'Name of the bank', 'statement_period_start': 'Start of statement period', 'statement_period_end': 'End of statement period', 'ending_balance': 'Closing balance for the period', 'address': 'Address of the account holder', 'bank_account_type': 'Type of bank account - Checking/Saving', 'bank_account_number': 'Bank account number', 'transaction_amount': 'Transaction amount', 'transaction_date': 'Transaction date', 'transaction_type': 'Transaction type', 'transaction_sender': 'Transaction sender'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"account_holder_name\", \"value\": \"John Smith\"}, {\"field\": \"bank_name\", \"value\": \"BNI\"}, {\"field\": \"statement_period_start\", \"value\": \"\"}, {\"field\": \"statement_period_end\", \"value\": \"\"}, {\"field\": \"ending_balance\", \"value\": \"31678.69\"}, {\"field\": \"address\", \"value\": \"2450 Courage St, STE 108, Brownsville, TX 78521\"}, {\"field\": \"bank_account_type\", \"value\": \"Checking\"}, {\"field\": \"bank_account_number\", \"value\": \"1234567890\"}, {\"field\": \"transaction_amount\", \"value\": \"\"}, {\"field\": \"transaction_date\", \"value\": \"\"}, {\"field\": \"transaction_type\", \"value\": \"\"}, {\"field\": \"transaction_sender\", \"value\": \"\"}]}"}]} {"id": "bank_statement_fin_029_p1_classify", "task": "classify", "class": "bank_statement", "doc": "bank_statement_fin_029.pdf", "images": ["images/train/bank_statement/bank_statement_fin_029_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"bank_statement\", \"confidence\": 1.0, \"issuer\": \"SunTrust\", \"holder\": \"John Smith\", \"page_marker\": \"\"}]}"}]} {"id": "bank_statement_fin_029_extract", "task": "extract", "class": "bank_statement", "doc": "bank_statement_fin_029.pdf", "images": ["images/train/bank_statement/bank_statement_fin_029_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('bank_statement', 'The document class.')\nfields = (['account_holder_name', 'bank_name', 'statement_period_start', 'statement_period_end', 'ending_balance', 'address', 'bank_account_type', 'bank_account_number', 'transaction_amount', 'transaction_date', 'transaction_type', 'transaction_sender'], 'The exact field names to extract.')\nhints = ({'account_holder_name': 'Name on the account', 'bank_name': 'Name of the bank', 'statement_period_start': 'Start of statement period', 'statement_period_end': 'End of statement period', 'ending_balance': 'Closing balance for the period', 'address': 'Address of the account holder', 'bank_account_type': 'Type of bank account - Checking/Saving', 'bank_account_number': 'Bank account number', 'transaction_amount': 'Transaction amount', 'transaction_date': 'Transaction date', 'transaction_type': 'Transaction type', 'transaction_sender': 'Transaction sender'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"account_holder_name\", \"value\": \"John Smith\"}, {\"field\": \"bank_name\", \"value\": \"SunTrust\"}, {\"field\": \"statement_period_start\", \"value\": \"\"}, {\"field\": \"statement_period_end\", \"value\": \"\"}, {\"field\": \"ending_balance\", \"value\": \"3898.57\"}, {\"field\": \"address\", \"value\": \"2450 Courage St, STE 108, Brownsville, TX 78521\"}, {\"field\": \"bank_account_type\", \"value\": \"\"}, {\"field\": \"bank_account_number\", \"value\": \"23785-54-9674458\"}, {\"field\": \"transaction_amount\", \"value\": \"223.26\"}, {\"field\": \"transaction_date\", \"value\": \"\"}, {\"field\": \"transaction_type\", \"value\": \"\"}, {\"field\": \"transaction_sender\", \"value\": \"Petron - C5 Station\"}]}"}]} {"id": "bank_statement_fin_030_p1_classify", "task": "classify", "class": "bank_statement", "doc": "bank_statement_fin_030.pdf", "images": ["images/train/bank_statement/bank_statement_fin_030_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"bank_statement\", \"confidence\": 1.0, \"issuer\": \"THIS FIRST BANK\", \"holder\": \"JOHN Q. CUSTOMER\", \"page_marker\": \"\"}]}"}]} {"id": "bank_statement_fin_030_extract", "task": "extract", "class": "bank_statement", "doc": "bank_statement_fin_030.pdf", "images": ["images/train/bank_statement/bank_statement_fin_030_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('bank_statement', 'The document class.')\nfields = (['account_holder_name', 'bank_name', 'statement_period_start', 'statement_period_end', 'ending_balance', 'address', 'bank_account_type', 'bank_account_number', 'transaction_amount', 'transaction_date', 'transaction_type', 'transaction_sender'], 'The exact field names to extract.')\nhints = ({'account_holder_name': 'Name on the account', 'bank_name': 'Name of the bank', 'statement_period_start': 'Start of statement period', 'statement_period_end': 'End of statement period', 'ending_balance': 'Closing balance for the period', 'address': 'Address of the account holder', 'bank_account_type': 'Type of bank account - Checking/Saving', 'bank_account_number': 'Bank account number', 'transaction_amount': 'Transaction amount', 'transaction_date': 'Transaction date', 'transaction_type': 'Transaction type', 'transaction_sender': 'Transaction sender'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"account_holder_name\", \"value\": \"JOHN Q. CUSTOMER\"}, {\"field\": \"bank_name\", \"value\": \"THIS FIRST BANK\"}, {\"field\": \"statement_period_start\", \"value\": \"2000-01-01\"}, {\"field\": \"statement_period_end\", \"value\": \"2000-02-01\"}, {\"field\": \"ending_balance\", \"value\": \"794.57\"}, {\"field\": \"address\", \"value\": \"1234 MAIN STREET ANYTOWN, STATE 00000-0000\"}, {\"field\": \"bank_account_type\", \"value\": \"Checking\"}, {\"field\": \"bank_account_number\", \"value\": \"12345-678910\"}, {\"field\": \"transaction_amount\", \"value\": \"230.66\"}, {\"field\": \"transaction_date\", \"value\": \"2000-01-05\"}, {\"field\": \"transaction_type\", \"value\": \"DEPOSIT\"}, {\"field\": \"transaction_sender\", \"value\": \"\"}]}"}]} {"id": "bank_statement_fin_031_p1_classify", "task": "classify", "class": "bank_statement", "doc": "bank_statement_fin_031.pdf", "images": ["images/train/bank_statement/bank_statement_fin_031_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"bank_statement\", \"confidence\": 1.0, \"issuer\": \"1 THIS FIRST BANK\", \"holder\": \"JOHN Q. CUSTOMER\", \"page_marker\": \"\"}]}"}]} {"id": "bank_statement_fin_031_extract", "task": "extract", "class": "bank_statement", "doc": "bank_statement_fin_031.pdf", "images": ["images/train/bank_statement/bank_statement_fin_031_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('bank_statement', 'The document class.')\nfields = (['account_holder_name', 'bank_name', 'statement_period_start', 'statement_period_end', 'ending_balance', 'address', 'bank_account_type', 'bank_account_number', 'transaction_amount', 'transaction_date', 'transaction_type', 'transaction_sender'], 'The exact field names to extract.')\nhints = ({'account_holder_name': 'Name on the account', 'bank_name': 'Name of the bank', 'statement_period_start': 'Start of statement period', 'statement_period_end': 'End of statement period', 'ending_balance': 'Closing balance for the period', 'address': 'Address of the account holder', 'bank_account_type': 'Type of bank account - Checking/Saving', 'bank_account_number': 'Bank account number', 'transaction_amount': 'Transaction amount', 'transaction_date': 'Transaction date', 'transaction_type': 'Transaction type', 'transaction_sender': 'Transaction sender'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"account_holder_name\", \"value\": \"JOHN Q. CUSTOMER\"}, {\"field\": \"bank_name\", \"value\": \"1 THIS FIRST BANK\"}, {\"field\": \"statement_period_start\", \"value\": \"2000-01-01\"}, {\"field\": \"statement_period_end\", \"value\": \"2000-02-01\"}, {\"field\": \"ending_balance\", \"value\": \"794.57\"}, {\"field\": \"address\", \"value\": \"1234 MAIN STREET ANYTOWN, STATE 00000-0000\"}, {\"field\": \"bank_account_type\", \"value\": \"Checking\"}, {\"field\": \"bank_account_number\", \"value\": \"12345-678910\"}, {\"field\": \"transaction_amount\", \"value\": \"230.66\"}, {\"field\": \"transaction_date\", \"value\": \"2000-01-05\"}, {\"field\": \"transaction_type\", \"value\": \"DEPOSIT\"}, {\"field\": \"transaction_sender\", \"value\": \"\"}]}"}]} {"id": "bank_statement_fin_033_p1_classify", "task": "classify", "class": "bank_statement", "doc": "bank_statement_fin_033.pdf", "images": ["images/train/bank_statement/bank_statement_fin_033_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"bank_statement\", \"confidence\": 1.0, \"issuer\": \"digibank by DBS\", \"holder\": \"FRITZ OKTA NEHEMNYA\", \"page_marker\": \"\"}]}"}]} {"id": "bank_statement_fin_033_extract", "task": "extract", "class": "bank_statement", "doc": "bank_statement_fin_033.pdf", "images": ["images/train/bank_statement/bank_statement_fin_033_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('bank_statement', 'The document class.')\nfields = (['account_holder_name', 'bank_name', 'statement_period_start', 'statement_period_end', 'ending_balance', 'address', 'bank_account_type', 'bank_account_number', 'transaction_amount', 'transaction_date', 'transaction_type', 'transaction_sender'], 'The exact field names to extract.')\nhints = ({'account_holder_name': 'Name on the account', 'bank_name': 'Name of the bank', 'statement_period_start': 'Start of statement period', 'statement_period_end': 'End of statement period', 'ending_balance': 'Closing balance for the period', 'address': 'Address of the account holder', 'bank_account_type': 'Type of bank account - Checking/Saving', 'bank_account_number': 'Bank account number', 'transaction_amount': 'Transaction amount', 'transaction_date': 'Transaction date', 'transaction_type': 'Transaction type', 'transaction_sender': 'Transaction sender'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"account_holder_name\", \"value\": \"FRITZ OKTA NEHEMNYA\"}, {\"field\": \"bank_name\", \"value\": \"digibank by DBS\"}, {\"field\": \"statement_period_start\", \"value\": \"2022-04-01\"}, {\"field\": \"statement_period_end\", \"value\": \"2022-04-30\"}, {\"field\": \"ending_balance\", \"value\": \"570.0\"}, {\"field\": \"address\", \"value\": \"GEDUNG K LT 1 DAN 2 KOMPLEK PKN STAN JL.BINTA RO JAYA V LANTAI I TANGERANG SELATAN, BANTEN INDONESIA, 15222\"}, {\"field\": \"bank_account_type\", \"value\": \"Saving\"}, {\"field\": \"bank_account_number\", \"value\": \"1700762359\"}, {\"field\": \"transaction_amount\", \"value\": \"9999585.0\"}, {\"field\": \"transaction_date\", \"value\": \"2022-04-20\"}, {\"field\": \"transaction_type\", \"value\": \"Deposit\"}, {\"field\": \"transaction_sender\", \"value\": \"BANK DBS INDONESIA, PT\"}]}"}]} {"id": "bank_statement_fin_034_p1_classify", "task": "classify", "class": "bank_statement", "doc": "bank_statement_fin_034.pdf", "images": ["images/train/bank_statement/bank_statement_fin_034_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"bank_statement\", \"confidence\": 1.0, \"issuer\": \"HDFC BANK\", \"holder\": \"MR. SATYAJIT APPOT\", \"page_marker\": \"\"}]}"}]} {"id": "bank_statement_fin_034_extract", "task": "extract", "class": "bank_statement", "doc": "bank_statement_fin_034.pdf", "images": ["images/train/bank_statement/bank_statement_fin_034_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('bank_statement', 'The document class.')\nfields = (['account_holder_name', 'bank_name', 'statement_period_start', 'statement_period_end', 'ending_balance', 'address', 'bank_account_type', 'bank_account_number', 'transaction_amount', 'transaction_date', 'transaction_type', 'transaction_sender'], 'The exact field names to extract.')\nhints = ({'account_holder_name': 'Name on the account', 'bank_name': 'Name of the bank', 'statement_period_start': 'Start of statement period', 'statement_period_end': 'End of statement period', 'ending_balance': 'Closing balance for the period', 'address': 'Address of the account holder', 'bank_account_type': 'Type of bank account - Checking/Saving', 'bank_account_number': 'Bank account number', 'transaction_amount': 'Transaction amount', 'transaction_date': 'Transaction date', 'transaction_type': 'Transaction type', 'transaction_sender': 'Transaction sender'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"account_holder_name\", \"value\": \"MR. SATYAJIT APPOT\"}, {\"field\": \"bank_name\", \"value\": \"HDFC BANK\"}, {\"field\": \"statement_period_start\", \"value\": \"2017-02-01\"}, {\"field\": \"statement_period_end\", \"value\": \"2017-09-01\"}, {\"field\": \"ending_balance\", \"value\": \"122.41\"}, {\"field\": \"address\", \"value\": \"NO ANJALI 2ND FLOOR, 1ST CROSS THAVEREKERE, B/H ANJENEYA TEMPLE, BANGALORE 560029, KARNATAKA INDIA\"}, {\"field\": \"bank_account_type\", \"value\": \"\"}, {\"field\": \"bank_account_number\", \"value\": \"17581610012048\"}, {\"field\": \"transaction_amount\", \"value\": \"6500.0\"}, {\"field\": \"transaction_date\", \"value\": \"2017-02-01\"}, {\"field\": \"transaction_type\", \"value\": \"Withdrawal\"}, {\"field\": \"transaction_sender\", \"value\": \"EAW-421424XXXXXX3014-SWCW3332-BANGALORE\"}]}"}]} {"id": "bank_statement_fin_035_p1_classify", "task": "classify", "class": "bank_statement", "doc": "bank_statement_fin_035.pdf", "images": ["images/train/bank_statement/bank_statement_fin_035_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"bank_statement\", \"confidence\": 1.0, \"issuer\": \"HSBC\", \"holder\": \"Mr Hayden Lamb\", \"page_marker\": \"\"}]}"}]} {"id": "bank_statement_fin_035_extract", "task": "extract", "class": "bank_statement", "doc": "bank_statement_fin_035.pdf", "images": ["images/train/bank_statement/bank_statement_fin_035_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('bank_statement', 'The document class.')\nfields = (['account_holder_name', 'bank_name', 'statement_period_start', 'statement_period_end', 'ending_balance', 'address', 'bank_account_type', 'bank_account_number', 'transaction_amount', 'transaction_date', 'transaction_type', 'transaction_sender'], 'The exact field names to extract.')\nhints = ({'account_holder_name': 'Name on the account', 'bank_name': 'Name of the bank', 'statement_period_start': 'Start of statement period', 'statement_period_end': 'End of statement period', 'ending_balance': 'Closing balance for the period', 'address': 'Address of the account holder', 'bank_account_type': 'Type of bank account - Checking/Saving', 'bank_account_number': 'Bank account number', 'transaction_amount': 'Transaction amount', 'transaction_date': 'Transaction date', 'transaction_type': 'Transaction type', 'transaction_sender': 'Transaction sender'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"account_holder_name\", \"value\": \"Mr Hayden Lamb\"}, {\"field\": \"bank_name\", \"value\": \"HSBC\"}, {\"field\": \"statement_period_start\", \"value\": \"2019-04-01\"}, {\"field\": \"statement_period_end\", \"value\": \"2019-04-30\"}, {\"field\": \"ending_balance\", \"value\": \"3150.16\"}, {\"field\": \"address\", \"value\": \"22 Abington Street Northampton NN1 2An\"}, {\"field\": \"bank_account_type\", \"value\": \"\"}, {\"field\": \"bank_account_number\", \"value\": \"22820374\"}, {\"field\": \"transaction_amount\", \"value\": \"300.0\"}, {\"field\": \"transaction_date\", \"value\": \"\"}, {\"field\": \"transaction_type\", \"value\": \"CR\"}, {\"field\": \"transaction_sender\", \"value\": \"TRAN S\"}]}"}]} {"id": "bank_statement_fin_036_p1_classify", "task": "classify", "class": "bank_statement", "doc": "bank_statement_fin_036.pdf", "images": ["images/train/bank_statement/bank_statement_fin_036_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"bank_statement\", \"confidence\": 1.0, \"issuer\": \"Your Bank\", \"holder\": \"Mr John Smith\", \"page_marker\": \"\"}]}"}]} {"id": "bank_statement_fin_036_extract", "task": "extract", "class": "bank_statement", "doc": "bank_statement_fin_036.pdf", "images": ["images/train/bank_statement/bank_statement_fin_036_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('bank_statement', 'The document class.')\nfields = (['account_holder_name', 'bank_name', 'statement_period_start', 'statement_period_end', 'ending_balance', 'address', 'bank_account_type', 'bank_account_number', 'transaction_amount', 'transaction_date', 'transaction_type', 'transaction_sender'], 'The exact field names to extract.')\nhints = ({'account_holder_name': 'Name on the account', 'bank_name': 'Name of the bank', 'statement_period_start': 'Start of statement period', 'statement_period_end': 'End of statement period', 'ending_balance': 'Closing balance for the period', 'address': 'Address of the account holder', 'bank_account_type': 'Type of bank account - Checking/Saving', 'bank_account_number': 'Bank account number', 'transaction_amount': 'Transaction amount', 'transaction_date': 'Transaction date', 'transaction_type': 'Transaction type', 'transaction_sender': 'Transaction sender'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"account_holder_name\", \"value\": \"Mr John Smith\"}, {\"field\": \"bank_name\", \"value\": \"Your Bank\"}, {\"field\": \"statement_period_start\", \"value\": \"2019-02-01\"}, {\"field\": \"statement_period_end\", \"value\": \"2019-03-01\"}, {\"field\": \"ending_balance\", \"value\": \"44079.83\"}, {\"field\": \"address\", \"value\": \"5 Any Road, Randomford, Anyshire, YZ985XY\"}, {\"field\": \"bank_account_type\", \"value\": \"Checking\"}, {\"field\": \"bank_account_number\", \"value\": \"99988877\"}, {\"field\": \"transaction_amount\", \"value\": \"\"}, {\"field\": \"transaction_date\", \"value\": \"\"}, {\"field\": \"transaction_type\", \"value\": \"\"}, {\"field\": \"transaction_sender\", \"value\": \"\"}]}"}]} {"id": "bank_statement_fin_038_p1_classify", "task": "classify", "class": "bank_statement", "doc": "bank_statement_fin_038.pdf", "images": ["images/train/bank_statement/bank_statement_fin_038_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"bank_statement\", \"confidence\": 1.0, \"issuer\": \"Bank of America\", \"holder\": \"Stuardo sanchez\", \"page_marker\": \"\"}]}"}]} {"id": "bank_statement_fin_038_extract", "task": "extract", "class": "bank_statement", "doc": "bank_statement_fin_038.pdf", "images": ["images/train/bank_statement/bank_statement_fin_038_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('bank_statement', 'The document class.')\nfields = (['account_holder_name', 'bank_name', 'statement_period_start', 'statement_period_end', 'ending_balance', 'address', 'bank_account_type', 'bank_account_number', 'transaction_amount', 'transaction_date', 'transaction_type', 'transaction_sender'], 'The exact field names to extract.')\nhints = ({'account_holder_name': 'Name on the account', 'bank_name': 'Name of the bank', 'statement_period_start': 'Start of statement period', 'statement_period_end': 'End of statement period', 'ending_balance': 'Closing balance for the period', 'address': 'Address of the account holder', 'bank_account_type': 'Type of bank account - Checking/Saving', 'bank_account_number': 'Bank account number', 'transaction_amount': 'Transaction amount', 'transaction_date': 'Transaction date', 'transaction_type': 'Transaction type', 'transaction_sender': 'Transaction sender'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"account_holder_name\", \"value\": \"Stuardo sanchez\"}, {\"field\": \"bank_name\", \"value\": \"Bank of America\"}, {\"field\": \"statement_period_start\", \"value\": \"\"}, {\"field\": \"statement_period_end\", \"value\": \"2018-01-08\"}, {\"field\": \"ending_balance\", \"value\": \"4543.36\"}, {\"field\": \"address\", \"value\": \"956 jefferson st se #08 minnesota\"}, {\"field\": \"bank_account_type\", \"value\": \"\"}, {\"field\": \"bank_account_number\", \"value\": \"\"}, {\"field\": \"transaction_amount\", \"value\": \"\"}, {\"field\": \"transaction_date\", \"value\": \"\"}, {\"field\": \"transaction_type\", \"value\": \"\"}, {\"field\": \"transaction_sender\", \"value\": \"\"}]}"}]} {"id": "bank_statement_fin_039_p1_classify", "task": "classify", "class": "bank_statement", "doc": "bank_statement_fin_039.pdf", "images": ["images/train/bank_statement/bank_statement_fin_039_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"bank_statement\", \"confidence\": 1.0, \"issuer\": \"BANK OF AMERICA\", \"holder\": \"Ashley C Jones\", \"page_marker\": \"\"}]}"}]} {"id": "bank_statement_fin_039_extract", "task": "extract", "class": "bank_statement", "doc": "bank_statement_fin_039.pdf", "images": ["images/train/bank_statement/bank_statement_fin_039_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('bank_statement', 'The document class.')\nfields = (['account_holder_name', 'bank_name', 'statement_period_start', 'statement_period_end', 'ending_balance', 'address', 'bank_account_type', 'bank_account_number', 'transaction_amount', 'transaction_date', 'transaction_type', 'transaction_sender'], 'The exact field names to extract.')\nhints = ({'account_holder_name': 'Name on the account', 'bank_name': 'Name of the bank', 'statement_period_start': 'Start of statement period', 'statement_period_end': 'End of statement period', 'ending_balance': 'Closing balance for the period', 'address': 'Address of the account holder', 'bank_account_type': 'Type of bank account - Checking/Saving', 'bank_account_number': 'Bank account number', 'transaction_amount': 'Transaction amount', 'transaction_date': 'Transaction date', 'transaction_type': 'Transaction type', 'transaction_sender': 'Transaction sender'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"account_holder_name\", \"value\": \"Ashley C Jones\"}, {\"field\": \"bank_name\", \"value\": \"BANK OF AMERICA\"}, {\"field\": \"statement_period_start\", \"value\": \"2025-01-01\"}, {\"field\": \"statement_period_end\", \"value\": \"2025-01-31\"}, {\"field\": \"ending_balance\", \"value\": \"50587.57\"}, {\"field\": \"address\", \"value\": \"4751 Colfax Ave. North Hollywood, CA 91602\"}, {\"field\": \"bank_account_type\", \"value\": \"Checking\"}, {\"field\": \"bank_account_number\", \"value\": \"8981 1929 0355\"}, {\"field\": \"transaction_amount\", \"value\": \"\"}, {\"field\": \"transaction_date\", \"value\": \"\"}, {\"field\": \"transaction_type\", \"value\": \"\"}, {\"field\": \"transaction_sender\", \"value\": \"\"}]}"}]} {"id": "bank_statement_fin_040_p1_classify", "task": "classify", "class": "bank_statement", "doc": "bank_statement_fin_040.pdf", "images": ["images/train/bank_statement/bank_statement_fin_040_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"bank_statement\", \"confidence\": 1.0, \"issuer\": \"Bank of America\", \"holder\": \"VALENTINE AUTO 24/7 MOBILE LOCKSMITH LOUIE T. VALENTINE\", \"page_marker\": \"\"}]}"}]} {"id": "bank_statement_fin_040_extract", "task": "extract", "class": "bank_statement", "doc": "bank_statement_fin_040.pdf", "images": ["images/train/bank_statement/bank_statement_fin_040_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('bank_statement', 'The document class.')\nfields = (['account_holder_name', 'bank_name', 'statement_period_start', 'statement_period_end', 'ending_balance', 'address', 'bank_account_type', 'bank_account_number', 'transaction_amount', 'transaction_date', 'transaction_type', 'transaction_sender'], 'The exact field names to extract.')\nhints = ({'account_holder_name': 'Name on the account', 'bank_name': 'Name of the bank', 'statement_period_start': 'Start of statement period', 'statement_period_end': 'End of statement period', 'ending_balance': 'Closing balance for the period', 'address': 'Address of the account holder', 'bank_account_type': 'Type of bank account - Checking/Saving', 'bank_account_number': 'Bank account number', 'transaction_amount': 'Transaction amount', 'transaction_date': 'Transaction date', 'transaction_type': 'Transaction type', 'transaction_sender': 'Transaction sender'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"account_holder_name\", \"value\": \"VALENTINE AUTO 24/7 MOBILE LOCKSMITH LOUIE T. VALENTINE\"}, {\"field\": \"bank_name\", \"value\": \"Bank of America\"}, {\"field\": \"statement_period_start\", \"value\": \"2019-05-20\"}, {\"field\": \"statement_period_end\", \"value\": \"2019-06-19\"}, {\"field\": \"ending_balance\", \"value\": \"578.46\"}, {\"field\": \"address\", \"value\": \"3135 ROADRUNNER RD, SAN MARCOS, CA 92078-6141\"}, {\"field\": \"bank_account_type\", \"value\": \"Checking\"}, {\"field\": \"bank_account_number\", \"value\": \"4570 3064 5567\"}, {\"field\": \"transaction_amount\", \"value\": \"\"}, {\"field\": \"transaction_date\", \"value\": \"\"}, {\"field\": \"transaction_type\", \"value\": \"\"}, {\"field\": \"transaction_sender\", \"value\": \"\"}]}"}]} {"id": "bank_statement_fin_041_p1_classify", "task": "classify", "class": "bank_statement", "doc": "bank_statement_fin_041.pdf", "images": ["images/train/bank_statement/bank_statement_fin_041_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"bank_statement\", \"confidence\": 1.0, \"issuer\": \"BANK OF AMERICA\", \"holder\": \"DENNIS SIMPSON II\", \"page_marker\": \"\"}]}"}]} {"id": "bank_statement_fin_041_extract", "task": "extract", "class": "bank_statement", "doc": "bank_statement_fin_041.pdf", "images": ["images/train/bank_statement/bank_statement_fin_041_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('bank_statement', 'The document class.')\nfields = (['account_holder_name', 'bank_name', 'statement_period_start', 'statement_period_end', 'ending_balance', 'address', 'bank_account_type', 'bank_account_number', 'transaction_amount', 'transaction_date', 'transaction_type', 'transaction_sender'], 'The exact field names to extract.')\nhints = ({'account_holder_name': 'Name on the account', 'bank_name': 'Name of the bank', 'statement_period_start': 'Start of statement period', 'statement_period_end': 'End of statement period', 'ending_balance': 'Closing balance for the period', 'address': 'Address of the account holder', 'bank_account_type': 'Type of bank account - Checking/Saving', 'bank_account_number': 'Bank account number', 'transaction_amount': 'Transaction amount', 'transaction_date': 'Transaction date', 'transaction_type': 'Transaction type', 'transaction_sender': 'Transaction sender'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"account_holder_name\", \"value\": \"DENNIS SIMPSON II\"}, {\"field\": \"bank_name\", \"value\": \"BANK OF AMERICA\"}, {\"field\": \"statement_period_start\", \"value\": \"2022-11-14\"}, {\"field\": \"statement_period_end\", \"value\": \"2022-12-13\"}, {\"field\": \"ending_balance\", \"value\": \"-1.7\"}, {\"field\": \"address\", \"value\": \"1405 W MICHIGAN ST, ORLANDO, FL 32805-6123\"}, {\"field\": \"bank_account_type\", \"value\": \"\"}, {\"field\": \"bank_account_number\", \"value\": \"\"}, {\"field\": \"transaction_amount\", \"value\": \"-10.77\"}, {\"field\": \"transaction_date\", \"value\": \"2022-12-11\"}, {\"field\": \"transaction_type\", \"value\": \"Purchase of Goods or Services\"}, {\"field\": \"transaction_sender\", \"value\": \"GOODWILL S.0BT #1 ORLANDO, Florida 32809 United States\"}]}"}]} {"id": "bank_statement_fin_042_p1_classify", "task": "classify", "class": "bank_statement", "doc": "bank_statement_fin_042.pdf", "images": ["images/train/bank_statement/bank_statement_fin_042_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"bank_statement\", \"confidence\": 1.0, \"issuer\": \"Bank of America\", \"holder\": \"WILLIAM SMITH\", \"page_marker\": \"\"}]}"}]} {"id": "bank_statement_fin_042_extract", "task": "extract", "class": "bank_statement", "doc": "bank_statement_fin_042.pdf", "images": ["images/train/bank_statement/bank_statement_fin_042_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('bank_statement', 'The document class.')\nfields = (['account_holder_name', 'bank_name', 'statement_period_start', 'statement_period_end', 'ending_balance', 'address', 'bank_account_type', 'bank_account_number', 'transaction_amount', 'transaction_date', 'transaction_type', 'transaction_sender'], 'The exact field names to extract.')\nhints = ({'account_holder_name': 'Name on the account', 'bank_name': 'Name of the bank', 'statement_period_start': 'Start of statement period', 'statement_period_end': 'End of statement period', 'ending_balance': 'Closing balance for the period', 'address': 'Address of the account holder', 'bank_account_type': 'Type of bank account - Checking/Saving', 'bank_account_number': 'Bank account number', 'transaction_amount': 'Transaction amount', 'transaction_date': 'Transaction date', 'transaction_type': 'Transaction type', 'transaction_sender': 'Transaction sender'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"account_holder_name\", \"value\": \"WILLIAM SMITH\"}, {\"field\": \"bank_name\", \"value\": \"Bank of America\"}, {\"field\": \"statement_period_start\", \"value\": \"2016-12-20\"}, {\"field\": \"statement_period_end\", \"value\": \"2017-01-19\"}, {\"field\": \"ending_balance\", \"value\": \"578.46\"}, {\"field\": \"address\", \"value\": \"3314 E KELTON LN UNIT 105, PHOENIX, AZ 85032-2762\"}, {\"field\": \"bank_account_type\", \"value\": \"Checking\"}, {\"field\": \"bank_account_number\", \"value\": \"4570 3064 5567\"}, {\"field\": \"transaction_amount\", \"value\": \"\"}, {\"field\": \"transaction_date\", \"value\": \"\"}, {\"field\": \"transaction_type\", \"value\": \"\"}, {\"field\": \"transaction_sender\", \"value\": \"\"}]}"}]} {"id": "bank_statement_fin_043_p1_classify", "task": "classify", "class": "bank_statement", "doc": "bank_statement_fin_043.pdf", "images": ["images/train/bank_statement/bank_statement_fin_043_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"bank_statement\", \"confidence\": 1.0, \"issuer\": \"Bank of America\", \"holder\": \"JOHN CITIZEN\", \"page_marker\": \"\"}]}"}]} {"id": "bank_statement_fin_043_extract", "task": "extract", "class": "bank_statement", "doc": "bank_statement_fin_043.pdf", "images": ["images/train/bank_statement/bank_statement_fin_043_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('bank_statement', 'The document class.')\nfields = (['account_holder_name', 'bank_name', 'statement_period_start', 'statement_period_end', 'ending_balance', 'address', 'bank_account_type', 'bank_account_number', 'transaction_amount', 'transaction_date', 'transaction_type', 'transaction_sender'], 'The exact field names to extract.')\nhints = ({'account_holder_name': 'Name on the account', 'bank_name': 'Name of the bank', 'statement_period_start': 'Start of statement period', 'statement_period_end': 'End of statement period', 'ending_balance': 'Closing balance for the period', 'address': 'Address of the account holder', 'bank_account_type': 'Type of bank account - Checking/Saving', 'bank_account_number': 'Bank account number', 'transaction_amount': 'Transaction amount', 'transaction_date': 'Transaction date', 'transaction_type': 'Transaction type', 'transaction_sender': 'Transaction sender'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"account_holder_name\", \"value\": \"JOHN CITIZEN\"}, {\"field\": \"bank_name\", \"value\": \"Bank of America\"}, {\"field\": \"statement_period_start\", \"value\": \"2021-03-22\"}, {\"field\": \"statement_period_end\", \"value\": \"2021-04-22\"}, {\"field\": \"ending_balance\", \"value\": \"28220.24\"}, {\"field\": \"address\", \"value\": \"510 Atlantic Avenue, Waterfront, Boston, MA 02210, United States of America\"}, {\"field\": \"bank_account_type\", \"value\": \"Checking\"}, {\"field\": \"bank_account_number\", \"value\": \"452 788 0004\"}, {\"field\": \"transaction_amount\", \"value\": \"5950.0\"}, {\"field\": \"transaction_date\", \"value\": \"2021-03-23\"}, {\"field\": \"transaction_type\", \"value\": \"Deposit\"}, {\"field\": \"transaction_sender\", \"value\": \"VISTAPRINT DEPOSIT 03/23 ID NUMBER #82913\"}]}"}]} {"id": "bank_statement_fin_044_p1_classify", "task": "classify", "class": "bank_statement", "doc": "bank_statement_fin_044.pdf", "images": ["images/train/bank_statement/bank_statement_fin_044_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"bank_statement\", \"confidence\": 1.0, \"issuer\": \"Bank of America\", \"holder\": \"NEAL'S CAPITAL, INC\", \"page_marker\": \"\"}]}"}]} {"id": "bank_statement_fin_044_extract", "task": "extract", "class": "bank_statement", "doc": "bank_statement_fin_044.pdf", "images": ["images/train/bank_statement/bank_statement_fin_044_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('bank_statement', 'The document class.')\nfields = (['account_holder_name', 'bank_name', 'statement_period_start', 'statement_period_end', 'ending_balance', 'address', 'bank_account_type', 'bank_account_number', 'transaction_amount', 'transaction_date', 'transaction_type', 'transaction_sender'], 'The exact field names to extract.')\nhints = ({'account_holder_name': 'Name on the account', 'bank_name': 'Name of the bank', 'statement_period_start': 'Start of statement period', 'statement_period_end': 'End of statement period', 'ending_balance': 'Closing balance for the period', 'address': 'Address of the account holder', 'bank_account_type': 'Type of bank account - Checking/Saving', 'bank_account_number': 'Bank account number', 'transaction_amount': 'Transaction amount', 'transaction_date': 'Transaction date', 'transaction_type': 'Transaction type', 'transaction_sender': 'Transaction sender'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"account_holder_name\", \"value\": \"NEAL'S CAPITAL, INC\"}, {\"field\": \"bank_name\", \"value\": \"Bank of America\"}, {\"field\": \"statement_period_start\", \"value\": \"2022-07-01\"}, {\"field\": \"statement_period_end\", \"value\": \"2022-07-31\"}, {\"field\": \"ending_balance\", \"value\": \"24912.11\"}, {\"field\": \"address\", \"value\": \"P.O. Box 15284, Wilmington, DE 19850\"}, {\"field\": \"bank_account_type\", \"value\": \"\"}, {\"field\": \"bank_account_number\", \"value\": \"2910 3298 6383\"}, {\"field\": \"transaction_amount\", \"value\": \"\"}, {\"field\": \"transaction_date\", \"value\": \"\"}, {\"field\": \"transaction_type\", \"value\": \"\"}, {\"field\": \"transaction_sender\", \"value\": \"\"}]}"}]} {"id": "bank_statement_fin_045_p1_classify", "task": "classify", "class": "bank_statement", "doc": "bank_statement_fin_045.pdf", "images": ["images/train/bank_statement/bank_statement_fin_045_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"bank_statement\", \"confidence\": 1.0, \"issuer\": \"BANK OF AMERICA\", \"holder\": \"JULIO JACQUES\", \"page_marker\": \"\"}]}"}]} {"id": "bank_statement_fin_045_extract", "task": "extract", "class": "bank_statement", "doc": "bank_statement_fin_045.pdf", "images": ["images/train/bank_statement/bank_statement_fin_045_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('bank_statement', 'The document class.')\nfields = (['account_holder_name', 'bank_name', 'statement_period_start', 'statement_period_end', 'ending_balance', 'address', 'bank_account_type', 'bank_account_number', 'transaction_amount', 'transaction_date', 'transaction_type', 'transaction_sender'], 'The exact field names to extract.')\nhints = ({'account_holder_name': 'Name on the account', 'bank_name': 'Name of the bank', 'statement_period_start': 'Start of statement period', 'statement_period_end': 'End of statement period', 'ending_balance': 'Closing balance for the period', 'address': 'Address of the account holder', 'bank_account_type': 'Type of bank account - Checking/Saving', 'bank_account_number': 'Bank account number', 'transaction_amount': 'Transaction amount', 'transaction_date': 'Transaction date', 'transaction_type': 'Transaction type', 'transaction_sender': 'Transaction sender'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"account_holder_name\", \"value\": \"JULIO JACQUES\"}, {\"field\": \"bank_name\", \"value\": \"BANK OF AMERICA\"}, {\"field\": \"statement_period_start\", \"value\": \"2023-12-27\"}, {\"field\": \"statement_period_end\", \"value\": \"2024-01-26\"}, {\"field\": \"ending_balance\", \"value\": \"68.23\"}, {\"field\": \"address\", \"value\": \"75 CARNEGIE AVE APT 7, EAST ORANGE NJ 07018-2924\"}, {\"field\": \"bank_account_type\", \"value\": \"\"}, {\"field\": \"bank_account_number\", \"value\": \"4400 6646 2942 6850\"}, {\"field\": \"transaction_amount\", \"value\": \"\"}, {\"field\": \"transaction_date\", \"value\": \"\"}, {\"field\": \"transaction_type\", \"value\": \"\"}, {\"field\": \"transaction_sender\", \"value\": \"\"}]}"}]} {"id": "bank_statement_fin_046_p1_classify", "task": "classify", "class": "bank_statement", "doc": "bank_statement_fin_046.pdf", "images": ["images/train/bank_statement/bank_statement_fin_046_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"bank_statement\", \"confidence\": 1.0, \"issuer\": \"HSBC\", \"holder\": \"AHMAD MUSTAPHA DAFER\", \"page_marker\": \"\"}]}"}]} {"id": "bank_statement_fin_046_extract", "task": "extract", "class": "bank_statement", "doc": "bank_statement_fin_046.pdf", "images": ["images/train/bank_statement/bank_statement_fin_046_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('bank_statement', 'The document class.')\nfields = (['account_holder_name', 'bank_name', 'statement_period_start', 'statement_period_end', 'ending_balance', 'address', 'bank_account_type', 'bank_account_number', 'transaction_amount', 'transaction_date', 'transaction_type', 'transaction_sender'], 'The exact field names to extract.')\nhints = ({'account_holder_name': 'Name on the account', 'bank_name': 'Name of the bank', 'statement_period_start': 'Start of statement period', 'statement_period_end': 'End of statement period', 'ending_balance': 'Closing balance for the period', 'address': 'Address of the account holder', 'bank_account_type': 'Type of bank account - Checking/Saving', 'bank_account_number': 'Bank account number', 'transaction_amount': 'Transaction amount', 'transaction_date': 'Transaction date', 'transaction_type': 'Transaction type', 'transaction_sender': 'Transaction sender'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"account_holder_name\", \"value\": \"AHMAD MUSTAPHA DAFER\"}, {\"field\": \"bank_name\", \"value\": \"HSBC\"}, {\"field\": \"statement_period_start\", \"value\": \"2013-01-11\"}, {\"field\": \"statement_period_end\", \"value\": \"2014-02-07\"}, {\"field\": \"ending_balance\", \"value\": \"3100000000.0\"}, {\"field\": \"address\", \"value\": \"43 Great Sutton Street, London ECIV 2DB, United Kingdom\"}, {\"field\": \"bank_account_type\", \"value\": \"Current\"}, {\"field\": \"bank_account_number\", \"value\": \"00990275\"}, {\"field\": \"transaction_amount\", \"value\": \"1000000000.0\"}, {\"field\": \"transaction_date\", \"value\": \"2013-01-18\"}, {\"field\": \"transaction_type\", \"value\": \"Credit\"}, {\"field\": \"transaction_sender\", \"value\": \"IBM UK\"}]}"}]} {"id": "bank_statement_fin_047_p1_classify", "task": "classify", "class": "bank_statement", "doc": "bank_statement_fin_047.pdf", "images": ["images/train/bank_statement/bank_statement_fin_047_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"bank_statement\", \"confidence\": 1.0, \"issuer\": \"BANK OF AMERICA\", \"holder\": \"MR JOHN DOE\", \"page_marker\": \"\"}]}"}]} {"id": "bank_statement_fin_047_extract", "task": "extract", "class": "bank_statement", "doc": "bank_statement_fin_047.pdf", "images": ["images/train/bank_statement/bank_statement_fin_047_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('bank_statement', 'The document class.')\nfields = (['account_holder_name', 'bank_name', 'statement_period_start', 'statement_period_end', 'ending_balance', 'address', 'bank_account_type', 'bank_account_number', 'transaction_amount', 'transaction_date', 'transaction_type', 'transaction_sender'], 'The exact field names to extract.')\nhints = ({'account_holder_name': 'Name on the account', 'bank_name': 'Name of the bank', 'statement_period_start': 'Start of statement period', 'statement_period_end': 'End of statement period', 'ending_balance': 'Closing balance for the period', 'address': 'Address of the account holder', 'bank_account_type': 'Type of bank account - Checking/Saving', 'bank_account_number': 'Bank account number', 'transaction_amount': 'Transaction amount', 'transaction_date': 'Transaction date', 'transaction_type': 'Transaction type', 'transaction_sender': 'Transaction sender'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"account_holder_name\", \"value\": \"MR JOHN DOE\"}, {\"field\": \"bank_name\", \"value\": \"BANK OF AMERICA\"}, {\"field\": \"statement_period_start\", \"value\": \"2021-02-01\"}, {\"field\": \"statement_period_end\", \"value\": \"2021-02-28\"}, {\"field\": \"ending_balance\", \"value\": \"814.14\"}, {\"field\": \"address\", \"value\": \"2 POST ALLEY, SEATTLE, WA 98101\"}, {\"field\": \"bank_account_type\", \"value\": \"Checking\"}, {\"field\": \"bank_account_number\", \"value\": \"1 2345 6789\"}, {\"field\": \"transaction_amount\", \"value\": \"\"}, {\"field\": \"transaction_date\", \"value\": \"\"}, {\"field\": \"transaction_type\", \"value\": \"\"}, {\"field\": \"transaction_sender\", \"value\": \"\"}]}"}]} {"id": "bank_statement_fin_048_p1_classify", "task": "classify", "class": "bank_statement", "doc": "bank_statement_fin_048.pdf", "images": ["images/train/bank_statement/bank_statement_fin_048_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"bank_statement\", \"confidence\": 1.0, \"issuer\": \"BANK OF AMERICA\", \"holder\": \"WMA, INC.\", \"page_marker\": \"\"}]}"}]} {"id": "bank_statement_fin_048_extract", "task": "extract", "class": "bank_statement", "doc": "bank_statement_fin_048.pdf", "images": ["images/train/bank_statement/bank_statement_fin_048_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('bank_statement', 'The document class.')\nfields = (['account_holder_name', 'bank_name', 'statement_period_start', 'statement_period_end', 'ending_balance', 'address', 'bank_account_type', 'bank_account_number', 'transaction_amount', 'transaction_date', 'transaction_type', 'transaction_sender'], 'The exact field names to extract.')\nhints = ({'account_holder_name': 'Name on the account', 'bank_name': 'Name of the bank', 'statement_period_start': 'Start of statement period', 'statement_period_end': 'End of statement period', 'ending_balance': 'Closing balance for the period', 'address': 'Address of the account holder', 'bank_account_type': 'Type of bank account - Checking/Saving', 'bank_account_number': 'Bank account number', 'transaction_amount': 'Transaction amount', 'transaction_date': 'Transaction date', 'transaction_type': 'Transaction type', 'transaction_sender': 'Transaction sender'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"account_holder_name\", \"value\": \"WMA, INC.\"}, {\"field\": \"bank_name\", \"value\": \"BANK OF AMERICA\"}, {\"field\": \"statement_period_start\", \"value\": \"2022-07-01\"}, {\"field\": \"statement_period_end\", \"value\": \"2022-07-31\"}, {\"field\": \"ending_balance\", \"value\": \"38689.02\"}, {\"field\": \"address\", \"value\": \"20120 JUBILEE WAY, PORTER RANCH, CA 91326-2412\"}, {\"field\": \"bank_account_type\", \"value\": \"Checking\"}, {\"field\": \"bank_account_number\", \"value\": \"\"}, {\"field\": \"transaction_amount\", \"value\": \"\"}, {\"field\": \"transaction_date\", \"value\": \"\"}, {\"field\": \"transaction_type\", \"value\": \"\"}, {\"field\": \"transaction_sender\", \"value\": \"\"}]}"}]} {"id": "bank_statement_fin_049_p1_classify", "task": "classify", "class": "bank_statement", "doc": "bank_statement_fin_049.pdf", "images": ["images/train/bank_statement/bank_statement_fin_049_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"bank_statement\", \"confidence\": 1.0, \"issuer\": \"Bank of America\", \"holder\": \"JOHN CITIZEN\", \"page_marker\": \"\"}]}"}]} {"id": "bank_statement_fin_049_extract", "task": "extract", "class": "bank_statement", "doc": "bank_statement_fin_049.pdf", "images": ["images/train/bank_statement/bank_statement_fin_049_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('bank_statement', 'The document class.')\nfields = (['account_holder_name', 'bank_name', 'statement_period_start', 'statement_period_end', 'ending_balance', 'address', 'bank_account_type', 'bank_account_number', 'transaction_amount', 'transaction_date', 'transaction_type', 'transaction_sender'], 'The exact field names to extract.')\nhints = ({'account_holder_name': 'Name on the account', 'bank_name': 'Name of the bank', 'statement_period_start': 'Start of statement period', 'statement_period_end': 'End of statement period', 'ending_balance': 'Closing balance for the period', 'address': 'Address of the account holder', 'bank_account_type': 'Type of bank account - Checking/Saving', 'bank_account_number': 'Bank account number', 'transaction_amount': 'Transaction amount', 'transaction_date': 'Transaction date', 'transaction_type': 'Transaction type', 'transaction_sender': 'Transaction sender'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"account_holder_name\", \"value\": \"JOHN CITIZEN\"}, {\"field\": \"bank_name\", \"value\": \"Bank of America\"}, {\"field\": \"statement_period_start\", \"value\": \"\"}, {\"field\": \"statement_period_end\", \"value\": \"2021-08-01\"}, {\"field\": \"ending_balance\", \"value\": \"1665.99\"}, {\"field\": \"address\", \"value\": \"1068 WOODWARD AVENUE, SOUTH BEND, IN 46616\"}, {\"field\": \"bank_account_type\", \"value\": \"\"}, {\"field\": \"bank_account_number\", \"value\": \"4400 6632 2149 9471\"}, {\"field\": \"transaction_amount\", \"value\": \"91.6\"}, {\"field\": \"transaction_date\", \"value\": \"2021-07-01\"}, {\"field\": \"transaction_type\", \"value\": \"\"}, {\"field\": \"transaction_sender\", \"value\": \"CARMELA'S RESTAURANT, SOUTH BEND, IN\"}]}"}]} {"id": "bank_statement_fin_050_p1_classify", "task": "classify", "class": "bank_statement", "doc": "bank_statement_fin_050.pdf", "images": ["images/train/bank_statement/bank_statement_fin_050_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"bank_statement\", \"confidence\": 1.0, \"issuer\": \"BANK OF AMERICA\", \"holder\": \"JOHN CITIZEN\", \"page_marker\": \"\"}]}"}]} {"id": "bank_statement_fin_050_extract", "task": "extract", "class": "bank_statement", "doc": "bank_statement_fin_050.pdf", "images": ["images/train/bank_statement/bank_statement_fin_050_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('bank_statement', 'The document class.')\nfields = (['account_holder_name', 'bank_name', 'statement_period_start', 'statement_period_end', 'ending_balance', 'address', 'bank_account_type', 'bank_account_number', 'transaction_amount', 'transaction_date', 'transaction_type', 'transaction_sender'], 'The exact field names to extract.')\nhints = ({'account_holder_name': 'Name on the account', 'bank_name': 'Name of the bank', 'statement_period_start': 'Start of statement period', 'statement_period_end': 'End of statement period', 'ending_balance': 'Closing balance for the period', 'address': 'Address of the account holder', 'bank_account_type': 'Type of bank account - Checking/Saving', 'bank_account_number': 'Bank account number', 'transaction_amount': 'Transaction amount', 'transaction_date': 'Transaction date', 'transaction_type': 'Transaction type', 'transaction_sender': 'Transaction sender'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"account_holder_name\", \"value\": \"JOHN CITIZEN\"}, {\"field\": \"bank_name\", \"value\": \"BANK OF AMERICA\"}, {\"field\": \"statement_period_start\", \"value\": \"2022-07-01\"}, {\"field\": \"statement_period_end\", \"value\": \"2022-07-30\"}, {\"field\": \"ending_balance\", \"value\": \"\"}, {\"field\": \"address\", \"value\": \"\"}, {\"field\": \"bank_account_type\", \"value\": \"\"}, {\"field\": \"bank_account_number\", \"value\": \"3830 1549 6274\"}, {\"field\": \"transaction_amount\", \"value\": \"300.0\"}, {\"field\": \"transaction_date\", \"value\": \"2022-07-03\"}, {\"field\": \"transaction_type\", \"value\": \"Zelle Transfer\"}, {\"field\": \"transaction_sender\", \"value\": \"Conf# 580341df4\"}]}"}]} {"id": "bank_statement_fin_051_p1_classify", "task": "classify", "class": "bank_statement", "doc": "bank_statement_fin_051.pdf", "images": ["images/train/bank_statement/bank_statement_fin_051_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"bank_statement\", \"confidence\": 1.0, \"issuer\": \"BANK OF AMERICA\", \"holder\": \"JOHN CITIZEN\", \"page_marker\": \"\"}]}"}]} {"id": "bank_statement_fin_051_extract", "task": "extract", "class": "bank_statement", "doc": "bank_statement_fin_051.pdf", "images": ["images/train/bank_statement/bank_statement_fin_051_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('bank_statement', 'The document class.')\nfields = (['account_holder_name', 'bank_name', 'statement_period_start', 'statement_period_end', 'ending_balance', 'address', 'bank_account_type', 'bank_account_number', 'transaction_amount', 'transaction_date', 'transaction_type', 'transaction_sender'], 'The exact field names to extract.')\nhints = ({'account_holder_name': 'Name on the account', 'bank_name': 'Name of the bank', 'statement_period_start': 'Start of statement period', 'statement_period_end': 'End of statement period', 'ending_balance': 'Closing balance for the period', 'address': 'Address of the account holder', 'bank_account_type': 'Type of bank account - Checking/Saving', 'bank_account_number': 'Bank account number', 'transaction_amount': 'Transaction amount', 'transaction_date': 'Transaction date', 'transaction_type': 'Transaction type', 'transaction_sender': 'Transaction sender'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"account_holder_name\", \"value\": \"JOHN CITIZEN\"}, {\"field\": \"bank_name\", \"value\": \"BANK OF AMERICA\"}, {\"field\": \"statement_period_start\", \"value\": \"2022-07-01\"}, {\"field\": \"statement_period_end\", \"value\": \"2022-07-30\"}, {\"field\": \"ending_balance\", \"value\": \"\"}, {\"field\": \"address\", \"value\": \"\"}, {\"field\": \"bank_account_type\", \"value\": \"Checking\"}, {\"field\": \"bank_account_number\", \"value\": \"3830 1549 6274\"}, {\"field\": \"transaction_amount\", \"value\": \"-152.21\"}, {\"field\": \"transaction_date\", \"value\": \"2022-07-21\"}, {\"field\": \"transaction_type\", \"value\": \"CHECKCARD 0225 AUTOCODE NY. 718-404-9691 NY 55432861056200004165961 CKCD 8999\"}, {\"field\": \"transaction_sender\", \"value\": \"\"}]}"}]} {"id": "bank_statement_fin_052_p1_classify", "task": "classify", "class": "bank_statement", "doc": "bank_statement_fin_052.pdf", "images": ["images/train/bank_statement/bank_statement_fin_052_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"bank_statement\", \"confidence\": 1.0, \"issuer\": \"BANK OF AMERICA\", \"holder\": \"JOHN CITIZEN\", \"page_marker\": \"\"}]}"}]} {"id": "bank_statement_fin_052_extract", "task": "extract", "class": "bank_statement", "doc": "bank_statement_fin_052.pdf", "images": ["images/train/bank_statement/bank_statement_fin_052_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('bank_statement', 'The document class.')\nfields = (['account_holder_name', 'bank_name', 'statement_period_start', 'statement_period_end', 'ending_balance', 'address', 'bank_account_type', 'bank_account_number', 'transaction_amount', 'transaction_date', 'transaction_type', 'transaction_sender'], 'The exact field names to extract.')\nhints = ({'account_holder_name': 'Name on the account', 'bank_name': 'Name of the bank', 'statement_period_start': 'Start of statement period', 'statement_period_end': 'End of statement period', 'ending_balance': 'Closing balance for the period', 'address': 'Address of the account holder', 'bank_account_type': 'Type of bank account - Checking/Saving', 'bank_account_number': 'Bank account number', 'transaction_amount': 'Transaction amount', 'transaction_date': 'Transaction date', 'transaction_type': 'Transaction type', 'transaction_sender': 'Transaction sender'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"account_holder_name\", \"value\": \"JOHN CITIZEN\"}, {\"field\": \"bank_name\", \"value\": \"BANK OF AMERICA\"}, {\"field\": \"statement_period_start\", \"value\": \"2022-07-01\"}, {\"field\": \"statement_period_end\", \"value\": \"2022-07-30\"}, {\"field\": \"ending_balance\", \"value\": \"\"}, {\"field\": \"address\", \"value\": \"\"}, {\"field\": \"bank_account_type\", \"value\": \"\"}, {\"field\": \"bank_account_number\", \"value\": \"3830 1549 6274\"}, {\"field\": \"transaction_amount\", \"value\": \"-35.0\"}, {\"field\": \"transaction_date\", \"value\": \"2022-07-25\"}, {\"field\": \"transaction_type\", \"value\": \"OVERDRAFT ITEM FEE FOR ACTIVITY\"}, {\"field\": \"transaction_sender\", \"value\": \"\"}]}"}]} {"id": "bank_statement_fin_053_p1_classify", "task": "classify", "class": "bank_statement", "doc": "bank_statement_fin_053.pdf", "images": ["images/train/bank_statement/bank_statement_fin_053_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"bank_statement\", \"confidence\": 1.0, \"issuer\": \"Commonwealth Bank\", \"holder\": \"MR ROGERS KERRY\", \"page_marker\": \"\"}]}"}]} {"id": "bank_statement_fin_053_extract", "task": "extract", "class": "bank_statement", "doc": "bank_statement_fin_053.pdf", "images": ["images/train/bank_statement/bank_statement_fin_053_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('bank_statement', 'The document class.')\nfields = (['account_holder_name', 'bank_name', 'statement_period_start', 'statement_period_end', 'ending_balance', 'address', 'bank_account_type', 'bank_account_number', 'transaction_amount', 'transaction_date', 'transaction_type', 'transaction_sender'], 'The exact field names to extract.')\nhints = ({'account_holder_name': 'Name on the account', 'bank_name': 'Name of the bank', 'statement_period_start': 'Start of statement period', 'statement_period_end': 'End of statement period', 'ending_balance': 'Closing balance for the period', 'address': 'Address of the account holder', 'bank_account_type': 'Type of bank account - Checking/Saving', 'bank_account_number': 'Bank account number', 'transaction_amount': 'Transaction amount', 'transaction_date': 'Transaction date', 'transaction_type': 'Transaction type', 'transaction_sender': 'Transaction sender'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"account_holder_name\", \"value\": \"MR ROGERS KERRY\"}, {\"field\": \"bank_name\", \"value\": \"Commonwealth Bank\"}, {\"field\": \"statement_period_start\", \"value\": \"2024-04-12\"}, {\"field\": \"statement_period_end\", \"value\": \"2024-05-12\"}, {\"field\": \"ending_balance\", \"value\": \"25608.84\"}, {\"field\": \"address\", \"value\": \"HADLOW STREET WARRAGUL VIC 3820\"}, {\"field\": \"bank_account_type\", \"value\": \"Checking\"}, {\"field\": \"bank_account_number\", \"value\": \"\"}, {\"field\": \"transaction_amount\", \"value\": \"\"}, {\"field\": \"transaction_date\", \"value\": \"\"}, {\"field\": \"transaction_type\", \"value\": \"\"}, {\"field\": \"transaction_sender\", \"value\": \"\"}]}"}]} {"id": "bank_statement_fin_054_p1_classify", "task": "classify", "class": "bank_statement", "doc": "bank_statement_fin_054.pdf", "images": ["images/train/bank_statement/bank_statement_fin_054_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"bank_statement\", \"confidence\": 1.0, \"issuer\": \"Commonwealth Bank\", \"holder\": \"MRS SOPHIE KYRIAKOPOULOS AND MR MICHAEL KYRIAKOPOULOS\", \"page_marker\": \"\"}]}"}]} {"id": "bank_statement_fin_054_extract", "task": "extract", "class": "bank_statement", "doc": "bank_statement_fin_054.pdf", "images": ["images/train/bank_statement/bank_statement_fin_054_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('bank_statement', 'The document class.')\nfields = (['account_holder_name', 'bank_name', 'statement_period_start', 'statement_period_end', 'ending_balance', 'address', 'bank_account_type', 'bank_account_number', 'transaction_amount', 'transaction_date', 'transaction_type', 'transaction_sender'], 'The exact field names to extract.')\nhints = ({'account_holder_name': 'Name on the account', 'bank_name': 'Name of the bank', 'statement_period_start': 'Start of statement period', 'statement_period_end': 'End of statement period', 'ending_balance': 'Closing balance for the period', 'address': 'Address of the account holder', 'bank_account_type': 'Type of bank account - Checking/Saving', 'bank_account_number': 'Bank account number', 'transaction_amount': 'Transaction amount', 'transaction_date': 'Transaction date', 'transaction_type': 'Transaction type', 'transaction_sender': 'Transaction sender'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"account_holder_name\", \"value\": \"MRS SOPHIE KYRIAKOPOULOS AND MR MICHAEL KYRIAKOPOULOS\"}, {\"field\": \"bank_name\", \"value\": \"Commonwealth Bank\"}, {\"field\": \"statement_period_start\", \"value\": \"2014-07-12\"}, {\"field\": \"statement_period_end\", \"value\": \"2014-08-11\"}, {\"field\": \"ending_balance\", \"value\": \"7891.64\"}, {\"field\": \"address\", \"value\": \"12 PENDEREL WAY BULLEEN VIC 3105\"}, {\"field\": \"bank_account_type\", \"value\": \"\"}, {\"field\": \"bank_account_number\", \"value\": \"06 7167 11251029\"}, {\"field\": \"transaction_amount\", \"value\": \"2208.27\"}, {\"field\": \"transaction_date\", \"value\": \"2014-07-23\"}, {\"field\": \"transaction_type\", \"value\": \"Credit\"}, {\"field\": \"transaction_sender\", \"value\": \"Transfer from xx1701 NetBank MandS B PDN Balanc\"}]}"}]} {"id": "bank_statement_fin_055_p1_classify", "task": "classify", "class": "bank_statement", "doc": "bank_statement_fin_055.pdf", "images": ["images/train/bank_statement/bank_statement_fin_055_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"bank_statement\", \"confidence\": 1.0, \"issuer\": \"Commonwealth Bank of Australia\", \"holder\": \"Mr John Doe\", \"page_marker\": \"\"}]}"}]} {"id": "bank_statement_fin_055_extract", "task": "extract", "class": "bank_statement", "doc": "bank_statement_fin_055.pdf", "images": ["images/train/bank_statement/bank_statement_fin_055_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('bank_statement', 'The document class.')\nfields = (['account_holder_name', 'bank_name', 'statement_period_start', 'statement_period_end', 'ending_balance', 'address', 'bank_account_type', 'bank_account_number', 'transaction_amount', 'transaction_date', 'transaction_type', 'transaction_sender'], 'The exact field names to extract.')\nhints = ({'account_holder_name': 'Name on the account', 'bank_name': 'Name of the bank', 'statement_period_start': 'Start of statement period', 'statement_period_end': 'End of statement period', 'ending_balance': 'Closing balance for the period', 'address': 'Address of the account holder', 'bank_account_type': 'Type of bank account - Checking/Saving', 'bank_account_number': 'Bank account number', 'transaction_amount': 'Transaction amount', 'transaction_date': 'Transaction date', 'transaction_type': 'Transaction type', 'transaction_sender': 'Transaction sender'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"account_holder_name\", \"value\": \"Mr John Doe\"}, {\"field\": \"bank_name\", \"value\": \"Commonwealth Bank of Australia\"}, {\"field\": \"statement_period_start\", \"value\": \"2018-06-19\"}, {\"field\": \"statement_period_end\", \"value\": \"2018-09-29\"}, {\"field\": \"ending_balance\", \"value\": \"11709.75\"}, {\"field\": \"address\", \"value\": \"522 Flinders St, Melbourne VIC 3000\"}, {\"field\": \"bank_account_type\", \"value\": \"Checking\"}, {\"field\": \"bank_account_number\", \"value\": \"123 456 12345678\"}, {\"field\": \"transaction_amount\", \"value\": \"50.95\"}, {\"field\": \"transaction_date\", \"value\": \"2018-06-19\"}, {\"field\": \"transaction_type\", \"value\": \"Debit\"}, {\"field\": \"transaction_sender\", \"value\": \"WOOLWORTHS W3174 ST_KILDA AUS Card xx7937\"}]}"}]} {"id": "bank_statement_fin_056_p1_classify", "task": "classify", "class": "bank_statement", "doc": "bank_statement_fin_056.pdf", "images": ["images/train/bank_statement/bank_statement_fin_056_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"bank_statement\", \"confidence\": 1.0, \"issuer\": \"Commonwealth Bank of Australia\", \"holder\": \"MR R D GEE\", \"page_marker\": \"\"}]}"}]} {"id": "bank_statement_fin_056_extract", "task": "extract", "class": "bank_statement", "doc": "bank_statement_fin_056.pdf", "images": ["images/train/bank_statement/bank_statement_fin_056_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('bank_statement', 'The document class.')\nfields = (['account_holder_name', 'bank_name', 'statement_period_start', 'statement_period_end', 'ending_balance', 'address', 'bank_account_type', 'bank_account_number', 'transaction_amount', 'transaction_date', 'transaction_type', 'transaction_sender'], 'The exact field names to extract.')\nhints = ({'account_holder_name': 'Name on the account', 'bank_name': 'Name of the bank', 'statement_period_start': 'Start of statement period', 'statement_period_end': 'End of statement period', 'ending_balance': 'Closing balance for the period', 'address': 'Address of the account holder', 'bank_account_type': 'Type of bank account - Checking/Saving', 'bank_account_number': 'Bank account number', 'transaction_amount': 'Transaction amount', 'transaction_date': 'Transaction date', 'transaction_type': 'Transaction type', 'transaction_sender': 'Transaction sender'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"account_holder_name\", \"value\": \"MR R D GEE\"}, {\"field\": \"bank_name\", \"value\": \"Commonwealth Bank of Australia\"}, {\"field\": \"statement_period_start\", \"value\": \"2018-07-01\"}, {\"field\": \"statement_period_end\", \"value\": \"2018-07-31\"}, {\"field\": \"ending_balance\", \"value\": \"7710.0\"}, {\"field\": \"address\", \"value\": \"14 STODA ST HEATHMONT VIC 3135\"}, {\"field\": \"bank_account_type\", \"value\": \"\"}, {\"field\": \"bank_account_number\", \"value\": \"06 3167 10781391\"}, {\"field\": \"transaction_amount\", \"value\": \"1000.0\"}, {\"field\": \"transaction_date\", \"value\": \"2018-07-02\"}, {\"field\": \"transaction_type\", \"value\": \"Deposit\"}, {\"field\": \"transaction_sender\", \"value\": \"\"}]}"}]} {"id": "bank_statement_fin_057_p1_classify", "task": "classify", "class": "bank_statement", "doc": "bank_statement_fin_057.pdf", "images": ["images/train/bank_statement/bank_statement_fin_057_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"bank_statement\", \"confidence\": 1.0, \"issuer\": \"HSBC UK\", \"holder\": \"Mr Jack Smith\", \"page_marker\": \"\"}]}"}]} {"id": "bank_statement_fin_057_extract", "task": "extract", "class": "bank_statement", "doc": "bank_statement_fin_057.pdf", "images": ["images/train/bank_statement/bank_statement_fin_057_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('bank_statement', 'The document class.')\nfields = (['account_holder_name', 'bank_name', 'statement_period_start', 'statement_period_end', 'ending_balance', 'address', 'bank_account_type', 'bank_account_number', 'transaction_amount', 'transaction_date', 'transaction_type', 'transaction_sender'], 'The exact field names to extract.')\nhints = ({'account_holder_name': 'Name on the account', 'bank_name': 'Name of the bank', 'statement_period_start': 'Start of statement period', 'statement_period_end': 'End of statement period', 'ending_balance': 'Closing balance for the period', 'address': 'Address of the account holder', 'bank_account_type': 'Type of bank account - Checking/Saving', 'bank_account_number': 'Bank account number', 'transaction_amount': 'Transaction amount', 'transaction_date': 'Transaction date', 'transaction_type': 'Transaction type', 'transaction_sender': 'Transaction sender'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"account_holder_name\", \"value\": \"Mr Jack Smith\"}, {\"field\": \"bank_name\", \"value\": \"HSBC UK\"}, {\"field\": \"statement_period_start\", \"value\": \"2020-10-15\"}, {\"field\": \"statement_period_end\", \"value\": \"2020-12-15\"}, {\"field\": \"ending_balance\", \"value\": \"\"}, {\"field\": \"address\", \"value\": \"Deanshanger Memorial, Little London, Milton Keynes, MK19 6HT\"}, {\"field\": \"bank_account_type\", \"value\": \"\"}, {\"field\": \"bank_account_number\", \"value\": \"01692682\"}, {\"field\": \"transaction_amount\", \"value\": \"\"}, {\"field\": \"transaction_date\", \"value\": \"2020-10-18\"}, {\"field\": \"transaction_type\", \"value\": \"CR\"}, {\"field\": \"transaction_sender\", \"value\": \"PRET A MANGER\"}]}"}]} {"id": "bank_statement_fin_058_p1_classify", "task": "classify", "class": "bank_statement", "doc": "bank_statement_fin_058.pdf", "images": ["images/train/bank_statement/bank_statement_fin_058_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"bank_statement\", \"confidence\": 1.0, \"issuer\": \"Commonwealth Bank\", \"holder\": \"MR JOHN CITIZEN\", \"page_marker\": \"\"}]}"}]} {"id": "bank_statement_fin_058_extract", "task": "extract", "class": "bank_statement", "doc": "bank_statement_fin_058.pdf", "images": ["images/train/bank_statement/bank_statement_fin_058_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('bank_statement', 'The document class.')\nfields = (['account_holder_name', 'bank_name', 'statement_period_start', 'statement_period_end', 'ending_balance', 'address', 'bank_account_type', 'bank_account_number', 'transaction_amount', 'transaction_date', 'transaction_type', 'transaction_sender'], 'The exact field names to extract.')\nhints = ({'account_holder_name': 'Name on the account', 'bank_name': 'Name of the bank', 'statement_period_start': 'Start of statement period', 'statement_period_end': 'End of statement period', 'ending_balance': 'Closing balance for the period', 'address': 'Address of the account holder', 'bank_account_type': 'Type of bank account - Checking/Saving', 'bank_account_number': 'Bank account number', 'transaction_amount': 'Transaction amount', 'transaction_date': 'Transaction date', 'transaction_type': 'Transaction type', 'transaction_sender': 'Transaction sender'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"account_holder_name\", \"value\": \"MR JOHN CITIZEN\"}, {\"field\": \"bank_name\", \"value\": \"Commonwealth Bank\"}, {\"field\": \"statement_period_start\", \"value\": \"\"}, {\"field\": \"statement_period_end\", \"value\": \"\"}, {\"field\": \"ending_balance\", \"value\": \"4.81\"}, {\"field\": \"address\", \"value\": \"14 STODA ST HEATHMONT VIC 3135\"}, {\"field\": \"bank_account_type\", \"value\": \"Checking\"}, {\"field\": \"bank_account_number\", \"value\": \"123456789\"}, {\"field\": \"transaction_amount\", \"value\": \"200.0\"}, {\"field\": \"transaction_date\", \"value\": \"2021-03-01\"}, {\"field\": \"transaction_type\", \"value\": \"Credit\"}, {\"field\": \"transaction_sender\", \"value\": \"Cash Dep Branch Karingal\"}]}"}]} {"id": "bank_statement_fin_059_p1_classify", "task": "classify", "class": "bank_statement", "doc": "bank_statement_fin_059.pdf", "images": ["images/train/bank_statement/bank_statement_fin_059_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"bank_statement\", \"confidence\": 1.0, \"issuer\": \"Commonwealth Bank\", \"holder\": \"MS FIRST NAME LAST NAME\", \"page_marker\": \"\"}]}"}]} {"id": "bank_statement_fin_059_extract", "task": "extract", "class": "bank_statement", "doc": "bank_statement_fin_059.pdf", "images": ["images/train/bank_statement/bank_statement_fin_059_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('bank_statement', 'The document class.')\nfields = (['account_holder_name', 'bank_name', 'statement_period_start', 'statement_period_end', 'ending_balance', 'address', 'bank_account_type', 'bank_account_number', 'transaction_amount', 'transaction_date', 'transaction_type', 'transaction_sender'], 'The exact field names to extract.')\nhints = ({'account_holder_name': 'Name on the account', 'bank_name': 'Name of the bank', 'statement_period_start': 'Start of statement period', 'statement_period_end': 'End of statement period', 'ending_balance': 'Closing balance for the period', 'address': 'Address of the account holder', 'bank_account_type': 'Type of bank account - Checking/Saving', 'bank_account_number': 'Bank account number', 'transaction_amount': 'Transaction amount', 'transaction_date': 'Transaction date', 'transaction_type': 'Transaction type', 'transaction_sender': 'Transaction sender'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"account_holder_name\", \"value\": \"MS FIRST NAME LAST NAME\"}, {\"field\": \"bank_name\", \"value\": \"Commonwealth Bank\"}, {\"field\": \"statement_period_start\", \"value\": \"2023-02-15\"}, {\"field\": \"statement_period_end\", \"value\": \"2023-03-15\"}, {\"field\": \"ending_balance\", \"value\": \"3095.03\"}, {\"field\": \"address\", \"value\": \"STREET, CITY STATE ZIP\"}, {\"field\": \"bank_account_type\", \"value\": \"\"}, {\"field\": \"bank_account_number\", \"value\": \"999999\"}, {\"field\": \"transaction_amount\", \"value\": \"\"}, {\"field\": \"transaction_date\", \"value\": \"\"}, {\"field\": \"transaction_type\", \"value\": \"\"}, {\"field\": \"transaction_sender\", \"value\": \"\"}]}"}]} {"id": "bank_statement_fin_061_p1_classify", "task": "classify", "class": "bank_statement", "doc": "bank_statement_fin_061.pdf", "images": ["images/train/bank_statement/bank_statement_fin_061_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"bank_statement\", \"confidence\": 1.0, \"issuer\": \"Commonwealth Bank\", \"holder\": \"WESTERN AUSTRALIAN CONTRACTING PTY LTD\", \"page_marker\": \"\"}]}"}]} {"id": "bank_statement_fin_061_extract", "task": "extract", "class": "bank_statement", "doc": "bank_statement_fin_061.pdf", "images": ["images/train/bank_statement/bank_statement_fin_061_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('bank_statement', 'The document class.')\nfields = (['account_holder_name', 'bank_name', 'statement_period_start', 'statement_period_end', 'ending_balance', 'address', 'bank_account_type', 'bank_account_number', 'transaction_amount', 'transaction_date', 'transaction_type', 'transaction_sender'], 'The exact field names to extract.')\nhints = ({'account_holder_name': 'Name on the account', 'bank_name': 'Name of the bank', 'statement_period_start': 'Start of statement period', 'statement_period_end': 'End of statement period', 'ending_balance': 'Closing balance for the period', 'address': 'Address of the account holder', 'bank_account_type': 'Type of bank account - Checking/Saving', 'bank_account_number': 'Bank account number', 'transaction_amount': 'Transaction amount', 'transaction_date': 'Transaction date', 'transaction_type': 'Transaction type', 'transaction_sender': 'Transaction sender'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"account_holder_name\", \"value\": \"WESTERN AUSTRALIAN CONTRACTING PTY LTD\"}, {\"field\": \"bank_name\", \"value\": \"Commonwealth Bank\"}, {\"field\": \"statement_period_start\", \"value\": \"2018-07-01\"}, {\"field\": \"statement_period_end\", \"value\": \"2018-08-31\"}, {\"field\": \"ending_balance\", \"value\": \"84011.89\"}, {\"field\": \"address\", \"value\": \"\"}, {\"field\": \"bank_account_type\", \"value\": \"Checking\"}, {\"field\": \"bank_account_number\", \"value\": \"\"}, {\"field\": \"transaction_amount\", \"value\": \"\"}, {\"field\": \"transaction_date\", \"value\": \"\"}, {\"field\": \"transaction_type\", \"value\": \"\"}, {\"field\": \"transaction_sender\", \"value\": \"\"}]}"}]} {"id": "bank_statement_fin_062_p1_classify", "task": "classify", "class": "bank_statement", "doc": "bank_statement_fin_062.pdf", "images": ["images/train/bank_statement/bank_statement_fin_062_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"bank_statement\", \"confidence\": 1.0, \"issuer\": \"CTBC HOLDING\", \"holder\": \"Mr John Doe\", \"page_marker\": \"\"}]}"}]} {"id": "bank_statement_fin_062_extract", "task": "extract", "class": "bank_statement", "doc": "bank_statement_fin_062.pdf", "images": ["images/train/bank_statement/bank_statement_fin_062_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('bank_statement', 'The document class.')\nfields = (['account_holder_name', 'bank_name', 'statement_period_start', 'statement_period_end', 'ending_balance', 'address', 'bank_account_type', 'bank_account_number', 'transaction_amount', 'transaction_date', 'transaction_type', 'transaction_sender'], 'The exact field names to extract.')\nhints = ({'account_holder_name': 'Name on the account', 'bank_name': 'Name of the bank', 'statement_period_start': 'Start of statement period', 'statement_period_end': 'End of statement period', 'ending_balance': 'Closing balance for the period', 'address': 'Address of the account holder', 'bank_account_type': 'Type of bank account - Checking/Saving', 'bank_account_number': 'Bank account number', 'transaction_amount': 'Transaction amount', 'transaction_date': 'Transaction date', 'transaction_type': 'Transaction type', 'transaction_sender': 'Transaction sender'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"account_holder_name\", \"value\": \"Mr John Doe\"}, {\"field\": \"bank_name\", \"value\": \"CTBC HOLDING\"}, {\"field\": \"statement_period_start\", \"value\": \"2018-06-19\"}, {\"field\": \"statement_period_end\", \"value\": \"2018-09-29\"}, {\"field\": \"ending_balance\", \"value\": \"11709.75\"}, {\"field\": \"address\", \"value\": \"522 Flinders St, Melbourne VIC 3000\"}, {\"field\": \"bank_account_type\", \"value\": \"\"}, {\"field\": \"bank_account_number\", \"value\": \"123 456 12345678\"}, {\"field\": \"transaction_amount\", \"value\": \"50.95\"}, {\"field\": \"transaction_date\", \"value\": \"2018-06-19\"}, {\"field\": \"transaction_type\", \"value\": \"Debit\"}, {\"field\": \"transaction_sender\", \"value\": \"WOOLWORTHS W3174 ST_KILDA AUS Card xx7937\"}]}"}]} {"id": "bank_statement_fin_063_p1_classify", "task": "classify", "class": "bank_statement", "doc": "bank_statement_fin_063.pdf", "images": ["images/train/bank_statement/bank_statement_fin_063_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"bank_statement\", \"confidence\": 1.0, \"issuer\": \"BARCLAYS\", \"holder\": \"\", \"page_marker\": \"\"}]}"}]} {"id": "bank_statement_fin_063_extract", "task": "extract", "class": "bank_statement", "doc": "bank_statement_fin_063.pdf", "images": ["images/train/bank_statement/bank_statement_fin_063_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('bank_statement', 'The document class.')\nfields = (['account_holder_name', 'bank_name', 'statement_period_start', 'statement_period_end', 'ending_balance', 'address', 'bank_account_type', 'bank_account_number', 'transaction_amount', 'transaction_date', 'transaction_type', 'transaction_sender'], 'The exact field names to extract.')\nhints = ({'account_holder_name': 'Name on the account', 'bank_name': 'Name of the bank', 'statement_period_start': 'Start of statement period', 'statement_period_end': 'End of statement period', 'ending_balance': 'Closing balance for the period', 'address': 'Address of the account holder', 'bank_account_type': 'Type of bank account - Checking/Saving', 'bank_account_number': 'Bank account number', 'transaction_amount': 'Transaction amount', 'transaction_date': 'Transaction date', 'transaction_type': 'Transaction type', 'transaction_sender': 'Transaction sender'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"account_holder_name\", \"value\": \"\"}, {\"field\": \"bank_name\", \"value\": \"BARCLAYS\"}, {\"field\": \"statement_period_start\", \"value\": \"\"}, {\"field\": \"statement_period_end\", \"value\": \"\"}, {\"field\": \"ending_balance\", \"value\": \"3219531.71\"}, {\"field\": \"address\", \"value\": \"\"}, {\"field\": \"bank_account_type\", \"value\": \"Checking\"}, {\"field\": \"bank_account_number\", \"value\": \"\"}, {\"field\": \"transaction_amount\", \"value\": \"100000.0\"}, {\"field\": \"transaction_date\", \"value\": \"\"}, {\"field\": \"transaction_type\", \"value\": \"Direct Debit\"}, {\"field\": \"transaction_sender\", \"value\": \"\"}]}"}]} {"id": "bank_statement_fin_064_p1_classify", "task": "classify", "class": "bank_statement", "doc": "bank_statement_fin_064.pdf", "images": ["images/train/bank_statement/bank_statement_fin_064_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"bank_statement\", \"confidence\": 1.0, \"issuer\": \"BARCLAYS\", \"holder\": \"Mr Ahad Abdul\", \"page_marker\": \"\"}]}"}]} {"id": "bank_statement_fin_064_extract", "task": "extract", "class": "bank_statement", "doc": "bank_statement_fin_064.pdf", "images": ["images/train/bank_statement/bank_statement_fin_064_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('bank_statement', 'The document class.')\nfields = (['account_holder_name', 'bank_name', 'statement_period_start', 'statement_period_end', 'ending_balance', 'address', 'bank_account_type', 'bank_account_number', 'transaction_amount', 'transaction_date', 'transaction_type', 'transaction_sender'], 'The exact field names to extract.')\nhints = ({'account_holder_name': 'Name on the account', 'bank_name': 'Name of the bank', 'statement_period_start': 'Start of statement period', 'statement_period_end': 'End of statement period', 'ending_balance': 'Closing balance for the period', 'address': 'Address of the account holder', 'bank_account_type': 'Type of bank account - Checking/Saving', 'bank_account_number': 'Bank account number', 'transaction_amount': 'Transaction amount', 'transaction_date': 'Transaction date', 'transaction_type': 'Transaction type', 'transaction_sender': 'Transaction sender'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"account_holder_name\", \"value\": \"Mr Ahad Abdul\"}, {\"field\": \"bank_name\", \"value\": \"BARCLAYS\"}, {\"field\": \"statement_period_start\", \"value\": \"2024-07-02\"}, {\"field\": \"statement_period_end\", \"value\": \"2024-07-18\"}, {\"field\": \"ending_balance\", \"value\": \"37.27\"}, {\"field\": \"address\", \"value\": \"38 UNIT 10 ST PAUL ST LONDON N1 7AB\"}, {\"field\": \"bank_account_type\", \"value\": \"Checking\"}, {\"field\": \"bank_account_number\", \"value\": \"06766862\"}, {\"field\": \"transaction_amount\", \"value\": \"73.8\"}, {\"field\": \"transaction_date\", \"value\": \"2024-07-02\"}, {\"field\": \"transaction_type\", \"value\": \"Card Payment\"}, {\"field\": \"transaction_sender\", \"value\": \"B & Q 1255\"}]}"}]} {"id": "bank_statement_fin_065_p1_classify", "task": "classify", "class": "bank_statement", "doc": "bank_statement_fin_065.pdf", "images": ["images/train/bank_statement/bank_statement_fin_065_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"bank_statement\", \"confidence\": 1.0, \"issuer\": \"BARCLAYS\", \"holder\": \"Miss Zainab Qayyum\", \"page_marker\": \"\"}]}"}]} {"id": "bank_statement_fin_065_extract", "task": "extract", "class": "bank_statement", "doc": "bank_statement_fin_065.pdf", "images": ["images/train/bank_statement/bank_statement_fin_065_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('bank_statement', 'The document class.')\nfields = (['account_holder_name', 'bank_name', 'statement_period_start', 'statement_period_end', 'ending_balance', 'address', 'bank_account_type', 'bank_account_number', 'transaction_amount', 'transaction_date', 'transaction_type', 'transaction_sender'], 'The exact field names to extract.')\nhints = ({'account_holder_name': 'Name on the account', 'bank_name': 'Name of the bank', 'statement_period_start': 'Start of statement period', 'statement_period_end': 'End of statement period', 'ending_balance': 'Closing balance for the period', 'address': 'Address of the account holder', 'bank_account_type': 'Type of bank account - Checking/Saving', 'bank_account_number': 'Bank account number', 'transaction_amount': 'Transaction amount', 'transaction_date': 'Transaction date', 'transaction_type': 'Transaction type', 'transaction_sender': 'Transaction sender'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"account_holder_name\", \"value\": \"Miss Zainab Qayyum\"}, {\"field\": \"bank_name\", \"value\": \"BARCLAYS\"}, {\"field\": \"statement_period_start\", \"value\": \"2024-05-19\"}, {\"field\": \"statement_period_end\", \"value\": \"2024-05-28\"}, {\"field\": \"ending_balance\", \"value\": \"12.12\"}, {\"field\": \"address\", \"value\": \"MR AC DA ID LOUIS A LIN 2 LISLE LACE GRAYS RM 5A\"}, {\"field\": \"bank_account_type\", \"value\": \"Checking\"}, {\"field\": \"bank_account_number\", \"value\": \"79176504\"}, {\"field\": \"transaction_amount\", \"value\": \"1598.07\"}, {\"field\": \"transaction_date\", \"value\": \"2024-05-20\"}, {\"field\": \"transaction_type\", \"value\": \"Bank Giro\"}, {\"field\": \"transaction_sender\", \"value\": \"100D39816 Dwp UC\"}]}"}]} {"id": "bank_statement_fin_066_p1_classify", "task": "classify", "class": "bank_statement", "doc": "bank_statement_fin_066.pdf", "images": ["images/train/bank_statement/bank_statement_fin_066_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"bank_statement\", \"confidence\": 1.0, \"issuer\": \"BARCLAYS\", \"holder\": \"MR JOHN CITIZEN\", \"page_marker\": \"\"}]}"}]} {"id": "bank_statement_fin_066_extract", "task": "extract", "class": "bank_statement", "doc": "bank_statement_fin_066.pdf", "images": ["images/train/bank_statement/bank_statement_fin_066_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('bank_statement', 'The document class.')\nfields = (['account_holder_name', 'bank_name', 'statement_period_start', 'statement_period_end', 'ending_balance', 'address', 'bank_account_type', 'bank_account_number', 'transaction_amount', 'transaction_date', 'transaction_type', 'transaction_sender'], 'The exact field names to extract.')\nhints = ({'account_holder_name': 'Name on the account', 'bank_name': 'Name of the bank', 'statement_period_start': 'Start of statement period', 'statement_period_end': 'End of statement period', 'ending_balance': 'Closing balance for the period', 'address': 'Address of the account holder', 'bank_account_type': 'Type of bank account - Checking/Saving', 'bank_account_number': 'Bank account number', 'transaction_amount': 'Transaction amount', 'transaction_date': 'Transaction date', 'transaction_type': 'Transaction type', 'transaction_sender': 'Transaction sender'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"account_holder_name\", \"value\": \"MR JOHN CITIZEN\"}, {\"field\": \"bank_name\", \"value\": \"BARCLAYS\"}, {\"field\": \"statement_period_start\", \"value\": \"2020-11-10\"}, {\"field\": \"statement_period_end\", \"value\": \"2020-12-07\"}, {\"field\": \"ending_balance\", \"value\": \"235.28\"}, {\"field\": \"address\", \"value\": \"33 CONRAD HOUSE, CLIFTON GROVE, DULSTON, EB 1DI\"}, {\"field\": \"bank_account_type\", \"value\": \"Checking\"}, {\"field\": \"bank_account_number\", \"value\": \"10534978\"}, {\"field\": \"transaction_amount\", \"value\": \"50.0\"}, {\"field\": \"transaction_date\", \"value\": \"2020-11-13\"}, {\"field\": \"transaction_type\", \"value\": \"Bill Payment\"}, {\"field\": \"transaction_sender\", \"value\": \"MBNA\"}]}"}]} {"id": "bank_statement_fin_069_p1_classify", "task": "classify", "class": "bank_statement", "doc": "bank_statement_fin_069.pdf", "images": ["images/train/bank_statement/bank_statement_fin_069_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"bank_statement\", \"confidence\": 1.0, \"issuer\": \"BARCLAYS\", \"holder\": \"JOE DANIEL LYNCH\", \"page_marker\": \"\"}]}"}]} {"id": "bank_statement_fin_069_extract", "task": "extract", "class": "bank_statement", "doc": "bank_statement_fin_069.pdf", "images": ["images/train/bank_statement/bank_statement_fin_069_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('bank_statement', 'The document class.')\nfields = (['account_holder_name', 'bank_name', 'statement_period_start', 'statement_period_end', 'ending_balance', 'address', 'bank_account_type', 'bank_account_number', 'transaction_amount', 'transaction_date', 'transaction_type', 'transaction_sender'], 'The exact field names to extract.')\nhints = ({'account_holder_name': 'Name on the account', 'bank_name': 'Name of the bank', 'statement_period_start': 'Start of statement period', 'statement_period_end': 'End of statement period', 'ending_balance': 'Closing balance for the period', 'address': 'Address of the account holder', 'bank_account_type': 'Type of bank account - Checking/Saving', 'bank_account_number': 'Bank account number', 'transaction_amount': 'Transaction amount', 'transaction_date': 'Transaction date', 'transaction_type': 'Transaction type', 'transaction_sender': 'Transaction sender'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"account_holder_name\", \"value\": \"JOE DANIEL LYNCH\"}, {\"field\": \"bank_name\", \"value\": \"BARCLAYS\"}, {\"field\": \"statement_period_start\", \"value\": \"2025-03-08\"}, {\"field\": \"statement_period_end\", \"value\": \"2025-04-05\"}, {\"field\": \"ending_balance\", \"value\": \"713.12\"}, {\"field\": \"address\", \"value\": \"42 CORONATION AVENUE TAMWORTH NOTTINGHAM POLESWORTH B78 1ED\"}, {\"field\": \"bank_account_type\", \"value\": \"Saving\"}, {\"field\": \"bank_account_number\", \"value\": \"13310418\"}, {\"field\": \"transaction_amount\", \"value\": \"1500.0\"}, {\"field\": \"transaction_date\", \"value\": \"2025-03-27\"}, {\"field\": \"transaction_type\", \"value\": \"Giro\"}, {\"field\": \"transaction_sender\", \"value\": \"A Kayran\"}]}"}]} {"id": "bank_statement_fin_072_p1_classify", "task": "classify", "class": "bank_statement", "doc": "bank_statement_fin_072.pdf", "images": ["images/train/bank_statement/bank_statement_fin_072_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"bank_statement\", \"confidence\": 1.0, \"issuer\": \"BARCLAYS\", \"holder\": \"Mr Mozammel Hossain Khan\", \"page_marker\": \"\"}]}"}]} {"id": "bank_statement_fin_072_extract", "task": "extract", "class": "bank_statement", "doc": "bank_statement_fin_072.pdf", "images": ["images/train/bank_statement/bank_statement_fin_072_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('bank_statement', 'The document class.')\nfields = (['account_holder_name', 'bank_name', 'statement_period_start', 'statement_period_end', 'ending_balance', 'address', 'bank_account_type', 'bank_account_number', 'transaction_amount', 'transaction_date', 'transaction_type', 'transaction_sender'], 'The exact field names to extract.')\nhints = ({'account_holder_name': 'Name on the account', 'bank_name': 'Name of the bank', 'statement_period_start': 'Start of statement period', 'statement_period_end': 'End of statement period', 'ending_balance': 'Closing balance for the period', 'address': 'Address of the account holder', 'bank_account_type': 'Type of bank account - Checking/Saving', 'bank_account_number': 'Bank account number', 'transaction_amount': 'Transaction amount', 'transaction_date': 'Transaction date', 'transaction_type': 'Transaction type', 'transaction_sender': 'Transaction sender'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"account_holder_name\", \"value\": \"Mr Mozammel Hossain Khan\"}, {\"field\": \"bank_name\", \"value\": \"BARCLAYS\"}, {\"field\": \"statement_period_start\", \"value\": \"2022-04-30\"}, {\"field\": \"statement_period_end\", \"value\": \"2022-06-01\"}, {\"field\": \"ending_balance\", \"value\": \"4083.49\"}, {\"field\": \"address\", \"value\": \"6 NETHERCOTT HOUSE, BRUCE ROAD, BOW, LONDON, E3 3HU\"}, {\"field\": \"bank_account_type\", \"value\": \"Checking\"}, {\"field\": \"bank_account_number\", \"value\": \"43362329\"}, {\"field\": \"transaction_amount\", \"value\": \"13.37\"}, {\"field\": \"transaction_date\", \"value\": \"2022-05-03\"}, {\"field\": \"transaction_type\", \"value\": \"Direct Debit\"}, {\"field\": \"transaction_sender\", \"value\": \"TV Licence MBP\"}]}"}]} {"id": "bank_statement_fin_073_p1_classify", "task": "classify", "class": "bank_statement", "doc": "bank_statement_fin_073.pdf", "images": ["images/train/bank_statement/bank_statement_fin_073_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"bank_statement\", \"confidence\": 1.0, \"issuer\": \"BARCLAYS\", \"holder\": \"Habib Asif\", \"page_marker\": \"\"}]}"}]} {"id": "bank_statement_fin_073_extract", "task": "extract", "class": "bank_statement", "doc": "bank_statement_fin_073.pdf", "images": ["images/train/bank_statement/bank_statement_fin_073_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('bank_statement', 'The document class.')\nfields = (['account_holder_name', 'bank_name', 'statement_period_start', 'statement_period_end', 'ending_balance', 'address', 'bank_account_type', 'bank_account_number', 'transaction_amount', 'transaction_date', 'transaction_type', 'transaction_sender'], 'The exact field names to extract.')\nhints = ({'account_holder_name': 'Name on the account', 'bank_name': 'Name of the bank', 'statement_period_start': 'Start of statement period', 'statement_period_end': 'End of statement period', 'ending_balance': 'Closing balance for the period', 'address': 'Address of the account holder', 'bank_account_type': 'Type of bank account - Checking/Saving', 'bank_account_number': 'Bank account number', 'transaction_amount': 'Transaction amount', 'transaction_date': 'Transaction date', 'transaction_type': 'Transaction type', 'transaction_sender': 'Transaction sender'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"account_holder_name\", \"value\": \"Habib Asif\"}, {\"field\": \"bank_name\", \"value\": \"BARCLAYS\"}, {\"field\": \"statement_period_start\", \"value\": \"2022-07-01\"}, {\"field\": \"statement_period_end\", \"value\": \"2022-08-01\"}, {\"field\": \"ending_balance\", \"value\": \"4853.72\"}, {\"field\": \"address\", \"value\": \"33 ALMA ROAD, WANDSWORTH, SW18 1AA\"}, {\"field\": \"bank_account_type\", \"value\": \"Checking\"}, {\"field\": \"bank_account_number\", \"value\": \"38293243\"}, {\"field\": \"transaction_amount\", \"value\": \"1150.0\"}, {\"field\": \"transaction_date\", \"value\": \"2022-07-18\"}, {\"field\": \"transaction_type\", \"value\": \"Payment\"}, {\"field\": \"transaction_sender\", \"value\": \"S Balasingham\"}]}"}]} {"id": "bank_statement_fin_074_p1_classify", "task": "classify", "class": "bank_statement", "doc": "bank_statement_fin_074.pdf", "images": ["images/train/bank_statement/bank_statement_fin_074_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"bank_statement\", \"confidence\": 1.0, \"issuer\": \"BARCLAYS\", \"holder\": \"Mr John Citizen and Mrs Jane Citizen\", \"page_marker\": \"\"}]}"}]} {"id": "bank_statement_fin_074_extract", "task": "extract", "class": "bank_statement", "doc": "bank_statement_fin_074.pdf", "images": ["images/train/bank_statement/bank_statement_fin_074_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('bank_statement', 'The document class.')\nfields = (['account_holder_name', 'bank_name', 'statement_period_start', 'statement_period_end', 'ending_balance', 'address', 'bank_account_type', 'bank_account_number', 'transaction_amount', 'transaction_date', 'transaction_type', 'transaction_sender'], 'The exact field names to extract.')\nhints = ({'account_holder_name': 'Name on the account', 'bank_name': 'Name of the bank', 'statement_period_start': 'Start of statement period', 'statement_period_end': 'End of statement period', 'ending_balance': 'Closing balance for the period', 'address': 'Address of the account holder', 'bank_account_type': 'Type of bank account - Checking/Saving', 'bank_account_number': 'Bank account number', 'transaction_amount': 'Transaction amount', 'transaction_date': 'Transaction date', 'transaction_type': 'Transaction type', 'transaction_sender': 'Transaction sender'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"account_holder_name\", \"value\": \"Mr John Citizen and Mrs Jane Citizen\"}, {\"field\": \"bank_name\", \"value\": \"BARCLAYS\"}, {\"field\": \"statement_period_start\", \"value\": \"\"}, {\"field\": \"statement_period_end\", \"value\": \"2020-11-04\"}, {\"field\": \"ending_balance\", \"value\": \"-382.85\"}, {\"field\": \"address\", \"value\": \"34 LONGFELLOW AVENUE, MIDDLE HUTTON, BOLTON, BL3 3TH\"}, {\"field\": \"bank_account_type\", \"value\": \"Checking\"}, {\"field\": \"bank_account_number\", \"value\": \"00999407\"}, {\"field\": \"transaction_amount\", \"value\": \"\"}, {\"field\": \"transaction_date\", \"value\": \"\"}, {\"field\": \"transaction_type\", \"value\": \"\"}, {\"field\": \"transaction_sender\", \"value\": \"\"}]}"}]} {"id": "bank_statement_fin_075_p1_classify", "task": "classify", "class": "bank_statement", "doc": "bank_statement_fin_075.pdf", "images": ["images/train/bank_statement/bank_statement_fin_075_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"bank_statement\", \"confidence\": 1.0, \"issuer\": \"BARCLAYS\", \"holder\": \"MD SAIFUL ISLAM MUNNA\", \"page_marker\": \"\"}]}"}]} {"id": "bank_statement_fin_075_extract", "task": "extract", "class": "bank_statement", "doc": "bank_statement_fin_075.pdf", "images": ["images/train/bank_statement/bank_statement_fin_075_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('bank_statement', 'The document class.')\nfields = (['account_holder_name', 'bank_name', 'statement_period_start', 'statement_period_end', 'ending_balance', 'address', 'bank_account_type', 'bank_account_number', 'transaction_amount', 'transaction_date', 'transaction_type', 'transaction_sender'], 'The exact field names to extract.')\nhints = ({'account_holder_name': 'Name on the account', 'bank_name': 'Name of the bank', 'statement_period_start': 'Start of statement period', 'statement_period_end': 'End of statement period', 'ending_balance': 'Closing balance for the period', 'address': 'Address of the account holder', 'bank_account_type': 'Type of bank account - Checking/Saving', 'bank_account_number': 'Bank account number', 'transaction_amount': 'Transaction amount', 'transaction_date': 'Transaction date', 'transaction_type': 'Transaction type', 'transaction_sender': 'Transaction sender'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"account_holder_name\", \"value\": \"MD SAIFUL ISLAM MUNNA\"}, {\"field\": \"bank_name\", \"value\": \"BARCLAYS\"}, {\"field\": \"statement_period_start\", \"value\": \"2024-07-03\"}, {\"field\": \"statement_period_end\", \"value\": \"2024-08-06\"}, {\"field\": \"ending_balance\", \"value\": \"1474.36\"}, {\"field\": \"address\", \"value\": \"VIKAADANEGRI 55, ROMA (RM) 00137\"}, {\"field\": \"bank_account_type\", \"value\": \"Checking\"}, {\"field\": \"bank_account_number\", \"value\": \"53157294\"}, {\"field\": \"transaction_amount\", \"value\": \"2.77\"}, {\"field\": \"transaction_date\", \"value\": \"2024-07-07\"}, {\"field\": \"transaction_type\", \"value\": \"Cash Machine Withdrawal\"}, {\"field\": \"transaction_sender\", \"value\": \"Barclays Hounslow 2\"}]}"}]} {"id": "bank_statement_fin_077_p1_classify", "task": "classify", "class": "bank_statement", "doc": "bank_statement_fin_077.pdf", "images": ["images/train/bank_statement/bank_statement_fin_077_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"bank_statement\", \"confidence\": 1.0, \"issuer\": \"BARCLAYS\", \"holder\": \"Mister Pavlo Kucher\", \"page_marker\": \"\"}]}"}]} {"id": "bank_statement_fin_077_extract", "task": "extract", "class": "bank_statement", "doc": "bank_statement_fin_077.pdf", "images": ["images/train/bank_statement/bank_statement_fin_077_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('bank_statement', 'The document class.')\nfields = (['account_holder_name', 'bank_name', 'statement_period_start', 'statement_period_end', 'ending_balance', 'address', 'bank_account_type', 'bank_account_number', 'transaction_amount', 'transaction_date', 'transaction_type', 'transaction_sender'], 'The exact field names to extract.')\nhints = ({'account_holder_name': 'Name on the account', 'bank_name': 'Name of the bank', 'statement_period_start': 'Start of statement period', 'statement_period_end': 'End of statement period', 'ending_balance': 'Closing balance for the period', 'address': 'Address of the account holder', 'bank_account_type': 'Type of bank account - Checking/Saving', 'bank_account_number': 'Bank account number', 'transaction_amount': 'Transaction amount', 'transaction_date': 'Transaction date', 'transaction_type': 'Transaction type', 'transaction_sender': 'Transaction sender'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"account_holder_name\", \"value\": \"Mister Pavlo Kucher\"}, {\"field\": \"bank_name\", \"value\": \"BARCLAYS\"}, {\"field\": \"statement_period_start\", \"value\": \"\"}, {\"field\": \"statement_period_end\", \"value\": \"\"}, {\"field\": \"ending_balance\", \"value\": \"919.99\"}, {\"field\": \"address\", \"value\": \"31 LOUVAIN AVENUE MOUNT ROSKILL AUCKLAND 1041 NEW ZEALAND\"}, {\"field\": \"bank_account_type\", \"value\": \"Checking\"}, {\"field\": \"bank_account_number\", \"value\": \"73186504\"}, {\"field\": \"transaction_amount\", \"value\": \"1598.07\"}, {\"field\": \"transaction_date\", \"value\": \"\"}, {\"field\": \"transaction_type\", \"value\": \"Bank Giro\"}, {\"field\": \"transaction_sender\", \"value\": \"100C47D1U Dwp UC\"}]}"}]} {"id": "bank_statement_fin_078_p1_classify", "task": "classify", "class": "bank_statement", "doc": "bank_statement_fin_078.pdf", "images": ["images/train/bank_statement/bank_statement_fin_078_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"bank_statement\", \"confidence\": 1.0, \"issuer\": \"BARCLAYS\", \"holder\": \"ANTHONY DAVID LORAINS\", \"page_marker\": \"\"}]}"}]} {"id": "bank_statement_fin_078_extract", "task": "extract", "class": "bank_statement", "doc": "bank_statement_fin_078.pdf", "images": ["images/train/bank_statement/bank_statement_fin_078_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('bank_statement', 'The document class.')\nfields = (['account_holder_name', 'bank_name', 'statement_period_start', 'statement_period_end', 'ending_balance', 'address', 'bank_account_type', 'bank_account_number', 'transaction_amount', 'transaction_date', 'transaction_type', 'transaction_sender'], 'The exact field names to extract.')\nhints = ({'account_holder_name': 'Name on the account', 'bank_name': 'Name of the bank', 'statement_period_start': 'Start of statement period', 'statement_period_end': 'End of statement period', 'ending_balance': 'Closing balance for the period', 'address': 'Address of the account holder', 'bank_account_type': 'Type of bank account - Checking/Saving', 'bank_account_number': 'Bank account number', 'transaction_amount': 'Transaction amount', 'transaction_date': 'Transaction date', 'transaction_type': 'Transaction type', 'transaction_sender': 'Transaction sender'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"account_holder_name\", \"value\": \"ANTHONY DAVID LORAINS\"}, {\"field\": \"bank_name\", \"value\": \"BARCLAYS\"}, {\"field\": \"statement_period_start\", \"value\": \"2023-03-01\"}, {\"field\": \"statement_period_end\", \"value\": \"2023-03-31\"}, {\"field\": \"ending_balance\", \"value\": \"3491.99\"}, {\"field\": \"address\", \"value\": \"19 MAGS BARROW, WEST, BH22 8PB\"}, {\"field\": \"bank_account_type\", \"value\": \"Checking\"}, {\"field\": \"bank_account_number\", \"value\": \"70612588\"}, {\"field\": \"transaction_amount\", \"value\": \"124.62\"}, {\"field\": \"transaction_date\", \"value\": \"2023-03-01\"}, {\"field\": \"transaction_type\", \"value\": \"Card Payment\"}, {\"field\": \"transaction_sender\", \"value\": \"Hemotion UK\"}]}"}]} {"id": "bank_statement_fin_080_p1_classify", "task": "classify", "class": "bank_statement", "doc": "bank_statement_fin_080.pdf", "images": ["images/train/bank_statement/bank_statement_fin_080_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"bank_statement\", \"confidence\": 1.0, \"issuer\": \"Barclays Bank\", \"holder\": \"Caumartin Christine Lucienne\", \"page_marker\": \"\"}]}"}]} {"id": "bank_statement_fin_080_extract", "task": "extract", "class": "bank_statement", "doc": "bank_statement_fin_080.pdf", "images": ["images/train/bank_statement/bank_statement_fin_080_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('bank_statement', 'The document class.')\nfields = (['account_holder_name', 'bank_name', 'statement_period_start', 'statement_period_end', 'ending_balance', 'address', 'bank_account_type', 'bank_account_number', 'transaction_amount', 'transaction_date', 'transaction_type', 'transaction_sender'], 'The exact field names to extract.')\nhints = ({'account_holder_name': 'Name on the account', 'bank_name': 'Name of the bank', 'statement_period_start': 'Start of statement period', 'statement_period_end': 'End of statement period', 'ending_balance': 'Closing balance for the period', 'address': 'Address of the account holder', 'bank_account_type': 'Type of bank account - Checking/Saving', 'bank_account_number': 'Bank account number', 'transaction_amount': 'Transaction amount', 'transaction_date': 'Transaction date', 'transaction_type': 'Transaction type', 'transaction_sender': 'Transaction sender'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"account_holder_name\", \"value\": \"Caumartin Christine Lucienne\"}, {\"field\": \"bank_name\", \"value\": \"Barclays Bank\"}, {\"field\": \"statement_period_start\", \"value\": \"2019-11-10\"}, {\"field\": \"statement_period_end\", \"value\": \"2019-12-07\"}, {\"field\": \"ending_balance\", \"value\": \"235.28\"}, {\"field\": \"address\", \"value\": \"8 CHEMIN DES EPINAUDNS, 03630 DESERTINES\"}, {\"field\": \"bank_account_type\", \"value\": \"Checking\"}, {\"field\": \"bank_account_number\", \"value\": \"05769354\"}, {\"field\": \"transaction_amount\", \"value\": \"50.0\"}, {\"field\": \"transaction_date\", \"value\": \"2019-11-13\"}, {\"field\": \"transaction_type\", \"value\": \"Bill payment\"}, {\"field\": \"transaction_sender\", \"value\": \"MBNA\"}]}"}]} {"id": "bank_statement_fin_081_p1_classify", "task": "classify", "class": "bank_statement", "doc": "bank_statement_fin_081.pdf", "images": ["images/train/bank_statement/bank_statement_fin_081_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"bank_statement\", \"confidence\": 1.0, \"issuer\": \"Barclays Bank\", \"holder\": \"Miss Margaret Laing\", \"page_marker\": \"\"}]}"}]} {"id": "bank_statement_fin_081_extract", "task": "extract", "class": "bank_statement", "doc": "bank_statement_fin_081.pdf", "images": ["images/train/bank_statement/bank_statement_fin_081_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('bank_statement', 'The document class.')\nfields = (['account_holder_name', 'bank_name', 'statement_period_start', 'statement_period_end', 'ending_balance', 'address', 'bank_account_type', 'bank_account_number', 'transaction_amount', 'transaction_date', 'transaction_type', 'transaction_sender'], 'The exact field names to extract.')\nhints = ({'account_holder_name': 'Name on the account', 'bank_name': 'Name of the bank', 'statement_period_start': 'Start of statement period', 'statement_period_end': 'End of statement period', 'ending_balance': 'Closing balance for the period', 'address': 'Address of the account holder', 'bank_account_type': 'Type of bank account - Checking/Saving', 'bank_account_number': 'Bank account number', 'transaction_amount': 'Transaction amount', 'transaction_date': 'Transaction date', 'transaction_type': 'Transaction type', 'transaction_sender': 'Transaction sender'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"account_holder_name\", \"value\": \"Miss Margaret Laing\"}, {\"field\": \"bank_name\", \"value\": \"Barclays Bank\"}, {\"field\": \"statement_period_start\", \"value\": \"2012-11-10\"}, {\"field\": \"statement_period_end\", \"value\": \"2012-12-07\"}, {\"field\": \"ending_balance\", \"value\": \"235.28\"}, {\"field\": \"address\", \"value\": \"33 CONRAD HOUSE, CLIFTON GROVE, DULSTON, EB 1DL\"}, {\"field\": \"bank_account_type\", \"value\": \"Checking\"}, {\"field\": \"bank_account_number\", \"value\": \"10534978\"}, {\"field\": \"transaction_amount\", \"value\": \"50.0\"}, {\"field\": \"transaction_date\", \"value\": \"2012-11-13\"}, {\"field\": \"transaction_type\", \"value\": \"Bill payment\"}, {\"field\": \"transaction_sender\", \"value\": \"MBNA\"}]}"}]} {"id": "bank_statement_fin_082_p1_classify", "task": "classify", "class": "bank_statement", "doc": "bank_statement_fin_082.pdf", "images": ["images/train/bank_statement/bank_statement_fin_082_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"bank_statement\", \"confidence\": 1.0, \"issuer\": \"BARCLAYS\", \"holder\": \"Mr Preet Jitendra Patel\", \"page_marker\": \"\"}]}"}]} {"id": "bank_statement_fin_082_extract", "task": "extract", "class": "bank_statement", "doc": "bank_statement_fin_082.pdf", "images": ["images/train/bank_statement/bank_statement_fin_082_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('bank_statement', 'The document class.')\nfields = (['account_holder_name', 'bank_name', 'statement_period_start', 'statement_period_end', 'ending_balance', 'address', 'bank_account_type', 'bank_account_number', 'transaction_amount', 'transaction_date', 'transaction_type', 'transaction_sender'], 'The exact field names to extract.')\nhints = ({'account_holder_name': 'Name on the account', 'bank_name': 'Name of the bank', 'statement_period_start': 'Start of statement period', 'statement_period_end': 'End of statement period', 'ending_balance': 'Closing balance for the period', 'address': 'Address of the account holder', 'bank_account_type': 'Type of bank account - Checking/Saving', 'bank_account_number': 'Bank account number', 'transaction_amount': 'Transaction amount', 'transaction_date': 'Transaction date', 'transaction_type': 'Transaction type', 'transaction_sender': 'Transaction sender'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"account_holder_name\", \"value\": \"Mr Preet Jitendra Patel\"}, {\"field\": \"bank_name\", \"value\": \"BARCLAYS\"}, {\"field\": \"statement_period_start\", \"value\": \"\"}, {\"field\": \"statement_period_end\", \"value\": \"2025-04-10\"}, {\"field\": \"ending_balance\", \"value\": \"27165.31\"}, {\"field\": \"address\", \"value\": \"14 OKADALE ROAD FOREST GATE LONDON E7 8JU\"}, {\"field\": \"bank_account_type\", \"value\": \"Checking\"}, {\"field\": \"bank_account_number\", \"value\": \"00538124\"}, {\"field\": \"transaction_amount\", \"value\": \"\"}, {\"field\": \"transaction_date\", \"value\": \"\"}, {\"field\": \"transaction_type\", \"value\": \"\"}, {\"field\": \"transaction_sender\", \"value\": \"\"}]}"}]} {"id": "bank_statement_fin_083_p1_classify", "task": "classify", "class": "bank_statement", "doc": "bank_statement_fin_083.pdf", "images": ["images/train/bank_statement/bank_statement_fin_083_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"bank_statement\", \"confidence\": 1.0, \"issuer\": \"Barclays\", \"holder\": \"Mrs Elismara de Franca Goncalves Mendes\", \"page_marker\": \"\"}]}"}]} {"id": "bank_statement_fin_083_extract", "task": "extract", "class": "bank_statement", "doc": "bank_statement_fin_083.pdf", "images": ["images/train/bank_statement/bank_statement_fin_083_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('bank_statement', 'The document class.')\nfields = (['account_holder_name', 'bank_name', 'statement_period_start', 'statement_period_end', 'ending_balance', 'address', 'bank_account_type', 'bank_account_number', 'transaction_amount', 'transaction_date', 'transaction_type', 'transaction_sender'], 'The exact field names to extract.')\nhints = ({'account_holder_name': 'Name on the account', 'bank_name': 'Name of the bank', 'statement_period_start': 'Start of statement period', 'statement_period_end': 'End of statement period', 'ending_balance': 'Closing balance for the period', 'address': 'Address of the account holder', 'bank_account_type': 'Type of bank account - Checking/Saving', 'bank_account_number': 'Bank account number', 'transaction_amount': 'Transaction amount', 'transaction_date': 'Transaction date', 'transaction_type': 'Transaction type', 'transaction_sender': 'Transaction sender'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"account_holder_name\", \"value\": \"Mrs Elismara de Franca Goncalves Mendes\"}, {\"field\": \"bank_name\", \"value\": \"Barclays\"}, {\"field\": \"statement_period_start\", \"value\": \"2021-01-28\"}, {\"field\": \"statement_period_end\", \"value\": \"2021-02-26\"}, {\"field\": \"ending_balance\", \"value\": \"10.99\"}, {\"field\": \"address\", \"value\": \"159 HIGH STREET SOUTH, EAST HAM, LONDON, E6 3PA\"}, {\"field\": \"bank_account_type\", \"value\": \"Checking\"}, {\"field\": \"bank_account_number\", \"value\": \"60717576\"}, {\"field\": \"transaction_amount\", \"value\": \"1.15\"}, {\"field\": \"transaction_date\", \"value\": \"2021-02-02\"}, {\"field\": \"transaction_type\", \"value\": \"Card Payment\"}, {\"field\": \"transaction_sender\", \"value\": \"Padoca Panificador Brazil\"}]}"}]} {"id": "bank_statement_fin_084_p1_classify", "task": "classify", "class": "bank_statement", "doc": "bank_statement_fin_084.pdf", "images": ["images/train/bank_statement/bank_statement_fin_084_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"bank_statement\", \"confidence\": 1.0, \"issuer\": \"BARCLAYS\", \"holder\": \"Mrs Sarah Ann Hopley\", \"page_marker\": \"\"}]}"}]} {"id": "bank_statement_fin_084_extract", "task": "extract", "class": "bank_statement", "doc": "bank_statement_fin_084.pdf", "images": ["images/train/bank_statement/bank_statement_fin_084_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('bank_statement', 'The document class.')\nfields = (['account_holder_name', 'bank_name', 'statement_period_start', 'statement_period_end', 'ending_balance', 'address', 'bank_account_type', 'bank_account_number', 'transaction_amount', 'transaction_date', 'transaction_type', 'transaction_sender'], 'The exact field names to extract.')\nhints = ({'account_holder_name': 'Name on the account', 'bank_name': 'Name of the bank', 'statement_period_start': 'Start of statement period', 'statement_period_end': 'End of statement period', 'ending_balance': 'Closing balance for the period', 'address': 'Address of the account holder', 'bank_account_type': 'Type of bank account - Checking/Saving', 'bank_account_number': 'Bank account number', 'transaction_amount': 'Transaction amount', 'transaction_date': 'Transaction date', 'transaction_type': 'Transaction type', 'transaction_sender': 'Transaction sender'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"account_holder_name\", \"value\": \"Mrs Sarah Ann Hopley\"}, {\"field\": \"bank_name\", \"value\": \"BARCLAYS\"}, {\"field\": \"statement_period_start\", \"value\": \"2022-02-25\"}, {\"field\": \"statement_period_end\", \"value\": \"2022-03-24\"}, {\"field\": \"ending_balance\", \"value\": \"3.16\"}, {\"field\": \"address\", \"value\": \"6 CAPTAINS PARADE, EAST COWES, PO32 6GT\"}, {\"field\": \"bank_account_type\", \"value\": \"Checking\"}, {\"field\": \"bank_account_number\", \"value\": \"00342270\"}, {\"field\": \"transaction_amount\", \"value\": \"5.75\"}, {\"field\": \"transaction_date\", \"value\": \"2022-02-25\"}, {\"field\": \"transaction_type\", \"value\": \"Card Payment\"}, {\"field\": \"transaction_sender\", \"value\": \"Co-Op Group 070623\"}]}"}]} {"id": "bank_statement_fin_085_p1_classify", "task": "classify", "class": "bank_statement", "doc": "bank_statement_fin_085.pdf", "images": ["images/train/bank_statement/bank_statement_fin_085_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"bank_statement\", \"confidence\": 1.0, \"issuer\": \"Barclays Bank\", \"holder\": \"Mr Michael Granger\", \"page_marker\": \"\"}]}"}]} {"id": "bank_statement_fin_085_extract", "task": "extract", "class": "bank_statement", "doc": "bank_statement_fin_085.pdf", "images": ["images/train/bank_statement/bank_statement_fin_085_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('bank_statement', 'The document class.')\nfields = (['account_holder_name', 'bank_name', 'statement_period_start', 'statement_period_end', 'ending_balance', 'address', 'bank_account_type', 'bank_account_number', 'transaction_amount', 'transaction_date', 'transaction_type', 'transaction_sender'], 'The exact field names to extract.')\nhints = ({'account_holder_name': 'Name on the account', 'bank_name': 'Name of the bank', 'statement_period_start': 'Start of statement period', 'statement_period_end': 'End of statement period', 'ending_balance': 'Closing balance for the period', 'address': 'Address of the account holder', 'bank_account_type': 'Type of bank account - Checking/Saving', 'bank_account_number': 'Bank account number', 'transaction_amount': 'Transaction amount', 'transaction_date': 'Transaction date', 'transaction_type': 'Transaction type', 'transaction_sender': 'Transaction sender'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"account_holder_name\", \"value\": \"Mr Michael Granger\"}, {\"field\": \"bank_name\", \"value\": \"Barclays Bank\"}, {\"field\": \"statement_period_start\", \"value\": \"2024-10-01\"}, {\"field\": \"statement_period_end\", \"value\": \"2024-10-31\"}, {\"field\": \"ending_balance\", \"value\": \"5603.02\"}, {\"field\": \"address\", \"value\": \"48 EVELYN STREET PLYMOUTH PL5 1QB\"}, {\"field\": \"bank_account_type\", \"value\": \"Checking\"}, {\"field\": \"bank_account_number\", \"value\": \"00584444\"}, {\"field\": \"transaction_amount\", \"value\": \"420.0\"}, {\"field\": \"transaction_date\", \"value\": \"2024-10-01\"}, {\"field\": \"transaction_type\", \"value\": \"Standing Order\"}, {\"field\": \"transaction_sender\", \"value\": \"Plymouth Highbury\"}]}"}]} {"id": "bank_statement_fin_086_p1_classify", "task": "classify", "class": "bank_statement", "doc": "bank_statement_fin_086.pdf", "images": ["images/train/bank_statement/bank_statement_fin_086_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"bank_statement\", \"confidence\": 1.0, \"issuer\": \"BARCLAYS\", \"holder\": \"Mr Evgenii Kovalev\", \"page_marker\": \"\"}]}"}]} {"id": "bank_statement_fin_086_extract", "task": "extract", "class": "bank_statement", "doc": "bank_statement_fin_086.pdf", "images": ["images/train/bank_statement/bank_statement_fin_086_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('bank_statement', 'The document class.')\nfields = (['account_holder_name', 'bank_name', 'statement_period_start', 'statement_period_end', 'ending_balance', 'address', 'bank_account_type', 'bank_account_number', 'transaction_amount', 'transaction_date', 'transaction_type', 'transaction_sender'], 'The exact field names to extract.')\nhints = ({'account_holder_name': 'Name on the account', 'bank_name': 'Name of the bank', 'statement_period_start': 'Start of statement period', 'statement_period_end': 'End of statement period', 'ending_balance': 'Closing balance for the period', 'address': 'Address of the account holder', 'bank_account_type': 'Type of bank account - Checking/Saving', 'bank_account_number': 'Bank account number', 'transaction_amount': 'Transaction amount', 'transaction_date': 'Transaction date', 'transaction_type': 'Transaction type', 'transaction_sender': 'Transaction sender'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"account_holder_name\", \"value\": \"Mr Evgenii Kovalev\"}, {\"field\": \"bank_name\", \"value\": \"BARCLAYS\"}, {\"field\": \"statement_period_start\", \"value\": \"2025-03-01\"}, {\"field\": \"statement_period_end\", \"value\": \"2025-03-24\"}, {\"field\": \"ending_balance\", \"value\": \"9659.7\"}, {\"field\": \"address\", \"value\": \"60 PARK AVENUE, LONDON, E6 2SR\"}, {\"field\": \"bank_account_type\", \"value\": \"Checking\"}, {\"field\": \"bank_account_number\", \"value\": \"50882755\"}, {\"field\": \"transaction_amount\", \"value\": \"\"}, {\"field\": \"transaction_date\", \"value\": \"2025-03-02\"}, {\"field\": \"transaction_type\", \"value\": \"Card Payment\"}, {\"field\": \"transaction_sender\", \"value\": \"Favourable Chicken\"}]}"}]} {"id": "bank_statement_fin_088_p1_classify", "task": "classify", "class": "bank_statement", "doc": "bank_statement_fin_088.pdf", "images": ["images/train/bank_statement/bank_statement_fin_088_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"bank_statement\", \"confidence\": 1.0, \"issuer\": \"BARCLAYS\", \"holder\": \"Mrs Sarah Ann Hopley\", \"page_marker\": \"\"}]}"}]} {"id": "bank_statement_fin_088_extract", "task": "extract", "class": "bank_statement", "doc": "bank_statement_fin_088.pdf", "images": ["images/train/bank_statement/bank_statement_fin_088_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('bank_statement', 'The document class.')\nfields = (['account_holder_name', 'bank_name', 'statement_period_start', 'statement_period_end', 'ending_balance', 'address', 'bank_account_type', 'bank_account_number', 'transaction_amount', 'transaction_date', 'transaction_type', 'transaction_sender'], 'The exact field names to extract.')\nhints = ({'account_holder_name': 'Name on the account', 'bank_name': 'Name of the bank', 'statement_period_start': 'Start of statement period', 'statement_period_end': 'End of statement period', 'ending_balance': 'Closing balance for the period', 'address': 'Address of the account holder', 'bank_account_type': 'Type of bank account - Checking/Saving', 'bank_account_number': 'Bank account number', 'transaction_amount': 'Transaction amount', 'transaction_date': 'Transaction date', 'transaction_type': 'Transaction type', 'transaction_sender': 'Transaction sender'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"account_holder_name\", \"value\": \"Mrs Sarah Ann Hopley\"}, {\"field\": \"bank_name\", \"value\": \"BARCLAYS\"}, {\"field\": \"statement_period_start\", \"value\": \"2022-02-25\"}, {\"field\": \"statement_period_end\", \"value\": \"2022-03-24\"}, {\"field\": \"ending_balance\", \"value\": \"3.16\"}, {\"field\": \"address\", \"value\": \"6 CAPTAINS PARADE, EAST COWES, PO32 6GT\"}, {\"field\": \"bank_account_type\", \"value\": \"Checking\"}, {\"field\": \"bank_account_number\", \"value\": \"00342270\"}, {\"field\": \"transaction_amount\", \"value\": \"9.0\"}, {\"field\": \"transaction_date\", \"value\": \"2022-03-01\"}, {\"field\": \"transaction_type\", \"value\": \"Bank Giro\"}, {\"field\": \"transaction_sender\", \"value\": \"Sarah Cambridge\"}]}"}]} {"id": "bank_statement_fin_089_p1_classify", "task": "classify", "class": "bank_statement", "doc": "bank_statement_fin_089.pdf", "images": ["images/train/bank_statement/bank_statement_fin_089_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"bank_statement\", \"confidence\": 1.0, \"issuer\": \"Barclays\", \"holder\": \"Mr Joseph Wagura\", \"page_marker\": \"\"}]}"}]} {"id": "bank_statement_fin_089_extract", "task": "extract", "class": "bank_statement", "doc": "bank_statement_fin_089.pdf", "images": ["images/train/bank_statement/bank_statement_fin_089_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('bank_statement', 'The document class.')\nfields = (['account_holder_name', 'bank_name', 'statement_period_start', 'statement_period_end', 'ending_balance', 'address', 'bank_account_type', 'bank_account_number', 'transaction_amount', 'transaction_date', 'transaction_type', 'transaction_sender'], 'The exact field names to extract.')\nhints = ({'account_holder_name': 'Name on the account', 'bank_name': 'Name of the bank', 'statement_period_start': 'Start of statement period', 'statement_period_end': 'End of statement period', 'ending_balance': 'Closing balance for the period', 'address': 'Address of the account holder', 'bank_account_type': 'Type of bank account - Checking/Saving', 'bank_account_number': 'Bank account number', 'transaction_amount': 'Transaction amount', 'transaction_date': 'Transaction date', 'transaction_type': 'Transaction type', 'transaction_sender': 'Transaction sender'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"account_holder_name\", \"value\": \"Mr Joseph Wagura\"}, {\"field\": \"bank_name\", \"value\": \"Barclays\"}, {\"field\": \"statement_period_start\", \"value\": \"2023-12-06\"}, {\"field\": \"statement_period_end\", \"value\": \"2023-12-22\"}, {\"field\": \"ending_balance\", \"value\": \"2421.32\"}, {\"field\": \"address\", \"value\": \"9 CREEKSIDE, RAINHAM, RM13 8QS\"}, {\"field\": \"bank_account_type\", \"value\": \"Checking\"}, {\"field\": \"bank_account_number\", \"value\": \"90992941\"}, {\"field\": \"transaction_amount\", \"value\": \"2100.0\"}, {\"field\": \"transaction_date\", \"value\": \"2023-12-06\"}, {\"field\": \"transaction_type\", \"value\": \"Account Credit: Deposit at Barclays\"}, {\"field\": \"transaction_sender\", \"value\": \"77 79 Wennington R\"}]}"}]} {"id": "bank_statement_fin_091_p1_classify", "task": "classify", "class": "bank_statement", "doc": "bank_statement_fin_091.pdf", "images": ["images/train/bank_statement/bank_statement_fin_091_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"bank_statement\", \"confidence\": 1.0, \"issuer\": \"BARCLAYS\", \"holder\": \"MISTER TARAS KUZKA\", \"page_marker\": \"\"}]}"}]} {"id": "bank_statement_fin_091_extract", "task": "extract", "class": "bank_statement", "doc": "bank_statement_fin_091.pdf", "images": ["images/train/bank_statement/bank_statement_fin_091_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('bank_statement', 'The document class.')\nfields = (['account_holder_name', 'bank_name', 'statement_period_start', 'statement_period_end', 'ending_balance', 'address', 'bank_account_type', 'bank_account_number', 'transaction_amount', 'transaction_date', 'transaction_type', 'transaction_sender'], 'The exact field names to extract.')\nhints = ({'account_holder_name': 'Name on the account', 'bank_name': 'Name of the bank', 'statement_period_start': 'Start of statement period', 'statement_period_end': 'End of statement period', 'ending_balance': 'Closing balance for the period', 'address': 'Address of the account holder', 'bank_account_type': 'Type of bank account - Checking/Saving', 'bank_account_number': 'Bank account number', 'transaction_amount': 'Transaction amount', 'transaction_date': 'Transaction date', 'transaction_type': 'Transaction type', 'transaction_sender': 'Transaction sender'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"account_holder_name\", \"value\": \"MISTER TARAS KUZKA\"}, {\"field\": \"bank_name\", \"value\": \"BARCLAYS\"}, {\"field\": \"statement_period_start\", \"value\": \"2024-12-05\"}, {\"field\": \"statement_period_end\", \"value\": \"2025-01-06\"}, {\"field\": \"ending_balance\", \"value\": \"919.99\"}, {\"field\": \"address\", \"value\": \"7 BYWORTH CLOSE, BRIGHTON, BN2 5HG\"}, {\"field\": \"bank_account_type\", \"value\": \"Checking\"}, {\"field\": \"bank_account_number\", \"value\": \"73186504\"}, {\"field\": \"transaction_amount\", \"value\": \"1.2\"}, {\"field\": \"transaction_date\", \"value\": \"2024-12-05\"}, {\"field\": \"transaction_type\", \"value\": \"Card Payment\"}, {\"field\": \"transaction_sender\", \"value\": \"Hegan Telecom Ltd\"}]}"}]} {"id": "bank_statement_fin_092_p1_classify", "task": "classify", "class": "bank_statement", "doc": "bank_statement_fin_092.pdf", "images": ["images/train/bank_statement/bank_statement_fin_092_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"bank_statement\", \"confidence\": 1.0, \"issuer\": \"BARCLAYS\", \"holder\": \"Rene L Test\", \"page_marker\": \"\"}]}"}]} {"id": "bank_statement_fin_092_extract", "task": "extract", "class": "bank_statement", "doc": "bank_statement_fin_092.pdf", "images": ["images/train/bank_statement/bank_statement_fin_092_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('bank_statement', 'The document class.')\nfields = (['account_holder_name', 'bank_name', 'statement_period_start', 'statement_period_end', 'ending_balance', 'address', 'bank_account_type', 'bank_account_number', 'transaction_amount', 'transaction_date', 'transaction_type', 'transaction_sender'], 'The exact field names to extract.')\nhints = ({'account_holder_name': 'Name on the account', 'bank_name': 'Name of the bank', 'statement_period_start': 'Start of statement period', 'statement_period_end': 'End of statement period', 'ending_balance': 'Closing balance for the period', 'address': 'Address of the account holder', 'bank_account_type': 'Type of bank account - Checking/Saving', 'bank_account_number': 'Bank account number', 'transaction_amount': 'Transaction amount', 'transaction_date': 'Transaction date', 'transaction_type': 'Transaction type', 'transaction_sender': 'Transaction sender'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"account_holder_name\", \"value\": \"Rene L Test\"}, {\"field\": \"bank_name\", \"value\": \"BARCLAYS\"}, {\"field\": \"statement_period_start\", \"value\": \"2020-06-01\"}, {\"field\": \"statement_period_end\", \"value\": \"2020-06-30\"}, {\"field\": \"ending_balance\", \"value\": \"57940.48\"}, {\"field\": \"address\", \"value\": \"1999 Jim Daniel Road 1937 Athens GA 30606\"}, {\"field\": \"bank_account_type\", \"value\": \"Saving\"}, {\"field\": \"bank_account_number\", \"value\": \"xxxxxxxx9999\"}, {\"field\": \"transaction_amount\", \"value\": \"1200.0\"}, {\"field\": \"transaction_date\", \"value\": \"2020-06-01\"}, {\"field\": \"transaction_type\", \"value\": \"ACH Deposit\"}, {\"field\": \"transaction_sender\", \"value\": \"WELLS FARGO BANK DDA account ******5866 Rene Test\"}]}"}]} {"id": "bank_statement_fin_093_p1_classify", "task": "classify", "class": "bank_statement", "doc": "bank_statement_fin_093.pdf", "images": ["images/train/bank_statement/bank_statement_fin_093_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"bank_statement\", \"confidence\": 1.0, \"issuer\": \"BARCLAYS\", \"holder\": \"MR FUHAD OLATEJU\", \"page_marker\": \"\"}]}"}]} {"id": "bank_statement_fin_093_extract", "task": "extract", "class": "bank_statement", "doc": "bank_statement_fin_093.pdf", "images": ["images/train/bank_statement/bank_statement_fin_093_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('bank_statement', 'The document class.')\nfields = (['account_holder_name', 'bank_name', 'statement_period_start', 'statement_period_end', 'ending_balance', 'address', 'bank_account_type', 'bank_account_number', 'transaction_amount', 'transaction_date', 'transaction_type', 'transaction_sender'], 'The exact field names to extract.')\nhints = ({'account_holder_name': 'Name on the account', 'bank_name': 'Name of the bank', 'statement_period_start': 'Start of statement period', 'statement_period_end': 'End of statement period', 'ending_balance': 'Closing balance for the period', 'address': 'Address of the account holder', 'bank_account_type': 'Type of bank account - Checking/Saving', 'bank_account_number': 'Bank account number', 'transaction_amount': 'Transaction amount', 'transaction_date': 'Transaction date', 'transaction_type': 'Transaction type', 'transaction_sender': 'Transaction sender'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"account_holder_name\", \"value\": \"MR FUHAD OLATEJU\"}, {\"field\": \"bank_name\", \"value\": \"BARCLAYS\"}, {\"field\": \"statement_period_start\", \"value\": \"\"}, {\"field\": \"statement_period_end\", \"value\": \"2024-11-15\"}, {\"field\": \"ending_balance\", \"value\": \"478.21\"}, {\"field\": \"address\", \"value\": \"23 OUTRAM ROAD, LONDON, E6 1JP\"}, {\"field\": \"bank_account_type\", \"value\": \"Checking\"}, {\"field\": \"bank_account_number\", \"value\": \"23138704\"}, {\"field\": \"transaction_amount\", \"value\": \"\"}, {\"field\": \"transaction_date\", \"value\": \"\"}, {\"field\": \"transaction_type\", \"value\": \"\"}, {\"field\": \"transaction_sender\", \"value\": \"\"}]}"}]} {"id": "bank_statement_fin_094_p1_classify", "task": "classify", "class": "bank_statement", "doc": "bank_statement_fin_094.pdf", "images": ["images/train/bank_statement/bank_statement_fin_094_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"bank_statement\", \"confidence\": 1.0, \"issuer\": \"BARCLAYS\", \"holder\": \"AMJAD FAROOQ WAH WAH MIRCHI FROZEN FOODS LIMITED\", \"page_marker\": \"\"}]}"}]} {"id": "bank_statement_fin_094_extract", "task": "extract", "class": "bank_statement", "doc": "bank_statement_fin_094.pdf", "images": ["images/train/bank_statement/bank_statement_fin_094_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('bank_statement', 'The document class.')\nfields = (['account_holder_name', 'bank_name', 'statement_period_start', 'statement_period_end', 'ending_balance', 'address', 'bank_account_type', 'bank_account_number', 'transaction_amount', 'transaction_date', 'transaction_type', 'transaction_sender'], 'The exact field names to extract.')\nhints = ({'account_holder_name': 'Name on the account', 'bank_name': 'Name of the bank', 'statement_period_start': 'Start of statement period', 'statement_period_end': 'End of statement period', 'ending_balance': 'Closing balance for the period', 'address': 'Address of the account holder', 'bank_account_type': 'Type of bank account - Checking/Saving', 'bank_account_number': 'Bank account number', 'transaction_amount': 'Transaction amount', 'transaction_date': 'Transaction date', 'transaction_type': 'Transaction type', 'transaction_sender': 'Transaction sender'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"account_holder_name\", \"value\": \"AMJAD FAROOQ WAH WAH MIRCHI FROZEN FOODS LIMITED\"}, {\"field\": \"bank_name\", \"value\": \"BARCLAYS\"}, {\"field\": \"statement_period_start\", \"value\": \"2023-07-25\"}, {\"field\": \"statement_period_end\", \"value\": \"2023-08-26\"}, {\"field\": \"ending_balance\", \"value\": \"8736.08\"}, {\"field\": \"address\", \"value\": \"147-149 SOHO ROAD, HANDSWORTH, BIRMINGHAM, B21 9ST\"}, {\"field\": \"bank_account_type\", \"value\": \"Checking\"}, {\"field\": \"bank_account_number\", \"value\": \"33111601\"}, {\"field\": \"transaction_amount\", \"value\": \"140.0\"}, {\"field\": \"transaction_date\", \"value\": \"2023-07-27\"}, {\"field\": \"transaction_type\", \"value\": \"On-Line Banking Bill Payment\"}, {\"field\": \"transaction_sender\", \"value\": \"Brain Research UK\"}]}"}]} {"id": "bank_statement_fin_095_p1_classify", "task": "classify", "class": "bank_statement", "doc": "bank_statement_fin_095.pdf", "images": ["images/train/bank_statement/bank_statement_fin_095_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"bank_statement\", \"confidence\": 1.0, \"issuer\": \"OCBC Bank (Malaysia) Berhad\", \"holder\": \"\", \"page_marker\": \"\"}]}"}]} {"id": "bank_statement_fin_095_extract", "task": "extract", "class": "bank_statement", "doc": "bank_statement_fin_095.pdf", "images": ["images/train/bank_statement/bank_statement_fin_095_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('bank_statement', 'The document class.')\nfields = (['account_holder_name', 'bank_name', 'statement_period_start', 'statement_period_end', 'ending_balance', 'address', 'bank_account_type', 'bank_account_number', 'transaction_amount', 'transaction_date', 'transaction_type', 'transaction_sender'], 'The exact field names to extract.')\nhints = ({'account_holder_name': 'Name on the account', 'bank_name': 'Name of the bank', 'statement_period_start': 'Start of statement period', 'statement_period_end': 'End of statement period', 'ending_balance': 'Closing balance for the period', 'address': 'Address of the account holder', 'bank_account_type': 'Type of bank account - Checking/Saving', 'bank_account_number': 'Bank account number', 'transaction_amount': 'Transaction amount', 'transaction_date': 'Transaction date', 'transaction_type': 'Transaction type', 'transaction_sender': 'Transaction sender'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"account_holder_name\", \"value\": \"\"}, {\"field\": \"bank_name\", \"value\": \"OCBC Bank (Malaysia) Berhad\"}, {\"field\": \"statement_period_start\", \"value\": \"2013-01-01\"}, {\"field\": \"statement_period_end\", \"value\": \"2013-06-30\"}, {\"field\": \"ending_balance\", \"value\": \"444109.64\"}, {\"field\": \"address\", \"value\": \"\"}, {\"field\": \"bank_account_type\", \"value\": \"\"}, {\"field\": \"bank_account_number\", \"value\": \"7074xxxxxx-000000\"}, {\"field\": \"transaction_amount\", \"value\": \"1860.88\"}, {\"field\": \"transaction_date\", \"value\": \"2013-01-03\"}, {\"field\": \"transaction_type\", \"value\": \"Int-Auto Split (Chq)\"}, {\"field\": \"transaction_sender\", \"value\": \"\"}]}"}]} {"id": "bank_statement_fin_096_p1_classify", "task": "classify", "class": "bank_statement", "doc": "bank_statement_fin_096.pdf", "images": ["images/train/bank_statement/bank_statement_fin_096_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"bank_statement\", \"confidence\": 1.0, \"issuer\": \"Barclays\", \"holder\": \"Mr Mohammed Zeeshan Iqbal\", \"page_marker\": \"\"}]}"}]} {"id": "bank_statement_fin_096_extract", "task": "extract", "class": "bank_statement", "doc": "bank_statement_fin_096.pdf", "images": ["images/train/bank_statement/bank_statement_fin_096_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('bank_statement', 'The document class.')\nfields = (['account_holder_name', 'bank_name', 'statement_period_start', 'statement_period_end', 'ending_balance', 'address', 'bank_account_type', 'bank_account_number', 'transaction_amount', 'transaction_date', 'transaction_type', 'transaction_sender'], 'The exact field names to extract.')\nhints = ({'account_holder_name': 'Name on the account', 'bank_name': 'Name of the bank', 'statement_period_start': 'Start of statement period', 'statement_period_end': 'End of statement period', 'ending_balance': 'Closing balance for the period', 'address': 'Address of the account holder', 'bank_account_type': 'Type of bank account - Checking/Saving', 'bank_account_number': 'Bank account number', 'transaction_amount': 'Transaction amount', 'transaction_date': 'Transaction date', 'transaction_type': 'Transaction type', 'transaction_sender': 'Transaction sender'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"account_holder_name\", \"value\": \"Mr Mohammed Zeeshan Iqbal\"}, {\"field\": \"bank_name\", \"value\": \"Barclays\"}, {\"field\": \"statement_period_start\", \"value\": \"2023-08-25\"}, {\"field\": \"statement_period_end\", \"value\": \"2023-09-22\"}, {\"field\": \"ending_balance\", \"value\": \"174.21\"}, {\"field\": \"address\", \"value\": \"101 HEIGHAM ROAD, EAST HAM, LONDON, E6 2JJ\"}, {\"field\": \"bank_account_type\", \"value\": \"Checking\"}, {\"field\": \"bank_account_number\", \"value\": \"00849243\"}, {\"field\": \"transaction_amount\", \"value\": \"7.09\"}, {\"field\": \"transaction_date\", \"value\": \"2023-08-29\"}, {\"field\": \"transaction_type\", \"value\": \"Card Payment\"}, {\"field\": \"transaction_sender\", \"value\": \"Boylesports\"}]}"}]} {"id": "bank_statement_fin_097_p1_classify", "task": "classify", "class": "bank_statement", "doc": "bank_statement_fin_097.pdf", "images": ["images/train/bank_statement/bank_statement_fin_097_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"bank_statement\", \"confidence\": 1.0, \"issuer\": \"UOB\", \"holder\": \"ABDUL ARIF BIN MAHAMAD HANAFIAH\", \"page_marker\": \"\"}]}"}]} {"id": "bank_statement_fin_097_extract", "task": "extract", "class": "bank_statement", "doc": "bank_statement_fin_097.pdf", "images": ["images/train/bank_statement/bank_statement_fin_097_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('bank_statement', 'The document class.')\nfields = (['account_holder_name', 'bank_name', 'statement_period_start', 'statement_period_end', 'ending_balance', 'address', 'bank_account_type', 'bank_account_number', 'transaction_amount', 'transaction_date', 'transaction_type', 'transaction_sender'], 'The exact field names to extract.')\nhints = ({'account_holder_name': 'Name on the account', 'bank_name': 'Name of the bank', 'statement_period_start': 'Start of statement period', 'statement_period_end': 'End of statement period', 'ending_balance': 'Closing balance for the period', 'address': 'Address of the account holder', 'bank_account_type': 'Type of bank account - Checking/Saving', 'bank_account_number': 'Bank account number', 'transaction_amount': 'Transaction amount', 'transaction_date': 'Transaction date', 'transaction_type': 'Transaction type', 'transaction_sender': 'Transaction sender'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"account_holder_name\", \"value\": \"ABDUL ARIF BIN MAHAMAD HANAFIAH\"}, {\"field\": \"bank_name\", \"value\": \"UOB\"}, {\"field\": \"statement_period_start\", \"value\": \"2023-05-01\"}, {\"field\": \"statement_period_end\", \"value\": \"2023-05-31\"}, {\"field\": \"ending_balance\", \"value\": \"19124.98\"}, {\"field\": \"address\", \"value\": \"NO 2 KAMPUNG KUAK HULU 33100 PENGKALAN HULU PERAK\"}, {\"field\": \"bank_account_type\", \"value\": \"Current Account\"}, {\"field\": \"bank_account_number\", \"value\": \"3803291258\"}, {\"field\": \"transaction_amount\", \"value\": \"68.96\"}, {\"field\": \"transaction_date\", \"value\": \"2023-05-30\"}, {\"field\": \"transaction_type\", \"value\": \"Misc DR-Debit Card\"}, {\"field\": \"transaction_sender\", \"value\": \"ZAPIER.COM/CHARGE\"}]}"}]} {"id": "bank_statement_fin_098_p1_classify", "task": "classify", "class": "bank_statement", "doc": "bank_statement_fin_098.pdf", "images": ["images/train/bank_statement/bank_statement_fin_098_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"bank_statement\", \"confidence\": 1.0, \"issuer\": \"UOB\", \"holder\": \"MR UOB VIP\", \"page_marker\": \"\"}]}"}]} {"id": "bank_statement_fin_098_extract", "task": "extract", "class": "bank_statement", "doc": "bank_statement_fin_098.pdf", "images": ["images/train/bank_statement/bank_statement_fin_098_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('bank_statement', 'The document class.')\nfields = (['account_holder_name', 'bank_name', 'statement_period_start', 'statement_period_end', 'ending_balance', 'address', 'bank_account_type', 'bank_account_number', 'transaction_amount', 'transaction_date', 'transaction_type', 'transaction_sender'], 'The exact field names to extract.')\nhints = ({'account_holder_name': 'Name on the account', 'bank_name': 'Name of the bank', 'statement_period_start': 'Start of statement period', 'statement_period_end': 'End of statement period', 'ending_balance': 'Closing balance for the period', 'address': 'Address of the account holder', 'bank_account_type': 'Type of bank account - Checking/Saving', 'bank_account_number': 'Bank account number', 'transaction_amount': 'Transaction amount', 'transaction_date': 'Transaction date', 'transaction_type': 'Transaction type', 'transaction_sender': 'Transaction sender'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"account_holder_name\", \"value\": \"MR UOB VIP\"}, {\"field\": \"bank_name\", \"value\": \"UOB\"}, {\"field\": \"statement_period_start\", \"value\": \"\"}, {\"field\": \"statement_period_end\", \"value\": \"2025-04-04\"}, {\"field\": \"ending_balance\", \"value\": \"25281.45\"}, {\"field\": \"address\", \"value\": \"ADDRESS LINE 1 SEQ 001 ADDRESS LINE 2 ADDRESS LINE 3 ADDRESS LINE 4\"}, {\"field\": \"bank_account_type\", \"value\": \"\"}, {\"field\": \"bank_account_number\", \"value\": \"\"}, {\"field\": \"transaction_amount\", \"value\": \"\"}, {\"field\": \"transaction_date\", \"value\": \"\"}, {\"field\": \"transaction_type\", \"value\": \"\"}, {\"field\": \"transaction_sender\", \"value\": \"\"}]}"}]} {"id": "bank_statement_fin_099_p1_classify", "task": "classify", "class": "bank_statement", "doc": "bank_statement_fin_099.pdf", "images": ["images/train/bank_statement/bank_statement_fin_099_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"bank_statement\", \"confidence\": 1.0, \"issuer\": \"UOB\", \"holder\": \"DENI SAPUTRA\", \"page_marker\": \"\"}]}"}]} {"id": "bank_statement_fin_099_extract", "task": "extract", "class": "bank_statement", "doc": "bank_statement_fin_099.pdf", "images": ["images/train/bank_statement/bank_statement_fin_099_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('bank_statement', 'The document class.')\nfields = (['account_holder_name', 'bank_name', 'statement_period_start', 'statement_period_end', 'ending_balance', 'address', 'bank_account_type', 'bank_account_number', 'transaction_amount', 'transaction_date', 'transaction_type', 'transaction_sender'], 'The exact field names to extract.')\nhints = ({'account_holder_name': 'Name on the account', 'bank_name': 'Name of the bank', 'statement_period_start': 'Start of statement period', 'statement_period_end': 'End of statement period', 'ending_balance': 'Closing balance for the period', 'address': 'Address of the account holder', 'bank_account_type': 'Type of bank account - Checking/Saving', 'bank_account_number': 'Bank account number', 'transaction_amount': 'Transaction amount', 'transaction_date': 'Transaction date', 'transaction_type': 'Transaction type', 'transaction_sender': 'Transaction sender'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"account_holder_name\", \"value\": \"DENI SAPUTRA\"}, {\"field\": \"bank_name\", \"value\": \"UOB\"}, {\"field\": \"statement_period_start\", \"value\": \"2024-06-01\"}, {\"field\": \"statement_period_end\", \"value\": \"2024-06-30\"}, {\"field\": \"ending_balance\", \"value\": \"20.0\"}, {\"field\": \"address\", \"value\": \"D\\u0142ugo\\u0142\\u0119ka-\\u015awierkla 11, Podegrodzie, 33-386, poland\"}, {\"field\": \"bank_account_type\", \"value\": \"Saving\"}, {\"field\": \"bank_account_number\", \"value\": \"749-387-941-7\"}, {\"field\": \"transaction_amount\", \"value\": \"\"}, {\"field\": \"transaction_date\", \"value\": \"\"}, {\"field\": \"transaction_type\", \"value\": \"\"}, {\"field\": \"transaction_sender\", \"value\": \"\"}]}"}]} {"id": "bank_statement_fin_100_p1_classify", "task": "classify", "class": "bank_statement", "doc": "bank_statement_fin_100.pdf", "images": ["images/train/bank_statement/bank_statement_fin_100_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"bank_statement\", \"confidence\": 1.0, \"issuer\": \"UOB\", \"holder\": \"John Citizen\", \"page_marker\": \"\"}]}"}]} {"id": "bank_statement_fin_100_extract", "task": "extract", "class": "bank_statement", "doc": "bank_statement_fin_100.pdf", "images": ["images/train/bank_statement/bank_statement_fin_100_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('bank_statement', 'The document class.')\nfields = (['account_holder_name', 'bank_name', 'statement_period_start', 'statement_period_end', 'ending_balance', 'address', 'bank_account_type', 'bank_account_number', 'transaction_amount', 'transaction_date', 'transaction_type', 'transaction_sender'], 'The exact field names to extract.')\nhints = ({'account_holder_name': 'Name on the account', 'bank_name': 'Name of the bank', 'statement_period_start': 'Start of statement period', 'statement_period_end': 'End of statement period', 'ending_balance': 'Closing balance for the period', 'address': 'Address of the account holder', 'bank_account_type': 'Type of bank account - Checking/Saving', 'bank_account_number': 'Bank account number', 'transaction_amount': 'Transaction amount', 'transaction_date': 'Transaction date', 'transaction_type': 'Transaction type', 'transaction_sender': 'Transaction sender'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"account_holder_name\", \"value\": \"John Citizen\"}, {\"field\": \"bank_name\", \"value\": \"UOB\"}, {\"field\": \"statement_period_start\", \"value\": \"2021-12-19\"}, {\"field\": \"statement_period_end\", \"value\": \"2022-01-03\"}, {\"field\": \"ending_balance\", \"value\": \"\"}, {\"field\": \"address\", \"value\": \"No. 80 Raffles Pl. 1 UOB Plaza. Singapore 048624\"}, {\"field\": \"bank_account_type\", \"value\": \"\"}, {\"field\": \"bank_account_number\", \"value\": \"422803******0489\"}, {\"field\": \"transaction_amount\", \"value\": \"0.07\"}, {\"field\": \"transaction_date\", \"value\": \"2021-12-19\"}, {\"field\": \"transaction_type\", \"value\": \"DDA Credit VERIFYBANK PAYPAL PAYPAL\"}, {\"field\": \"transaction_sender\", \"value\": \"PAYPAL\"}]}"}]} {"id": "drivers_license_fin_001_p1_classify", "task": "classify", "class": "drivers_license", "doc": "drivers_license_fin_001.pdf", "images": ["images/train/drivers_license/drivers_license_fin_001_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"drivers_license\", \"confidence\": 1.0, \"issuer\": \"NEW YORK\", \"holder\": \"DOCUMENT SAMPLE, LICENSE\", \"page_marker\": \"\"}]}"}]} {"id": "drivers_license_fin_001_extract", "task": "extract", "class": "drivers_license", "doc": "drivers_license_fin_001.pdf", "images": ["images/train/drivers_license/drivers_license_fin_001_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('drivers_license', 'The document class.')\nfields = (['full_name', 'date_of_birth', 'address', 'issue_date', 'expiration_date', 'state'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name', 'date_of_birth': 'Date of Birth', 'address': 'Address', 'issue_date': 'Issued Date', 'expiration_date': 'Valid Until Date', 'state': 'Issuing state'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"DOCUMENT SAMPLE, LICENSE\"}, {\"field\": \"date_of_birth\", \"value\": \"1985-12-30\"}, {\"field\": \"address\", \"value\": \"2345 ANYPLACE AVE, ANYTOWN NY 12345\"}, {\"field\": \"issue_date\", \"value\": \"2008-09-30\"}, {\"field\": \"expiration_date\", \"value\": \"2016-12-30\"}, {\"field\": \"state\", \"value\": \"NEW YORK\"}]}"}]} {"id": "drivers_license_fin_002_p1_classify", "task": "classify", "class": "drivers_license", "doc": "drivers_license_fin_002.pdf", "images": ["images/train/drivers_license/drivers_license_fin_002_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"drivers_license\", \"confidence\": 1.0, \"issuer\": \"New York\", \"holder\": \"MURPHY, BRENNA ANNA C\", \"page_marker\": \"\"}]}"}]} {"id": "drivers_license_fin_002_extract", "task": "extract", "class": "drivers_license", "doc": "drivers_license_fin_002.pdf", "images": ["images/train/drivers_license/drivers_license_fin_002_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('drivers_license', 'The document class.')\nfields = (['full_name', 'date_of_birth', 'address', 'issue_date', 'expiration_date', 'state'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name', 'date_of_birth': 'Date of Birth', 'address': 'Address', 'issue_date': 'Issued Date', 'expiration_date': 'Valid Until Date', 'state': 'Issuing state'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"MURPHY, BRENNA ANNA C\"}, {\"field\": \"date_of_birth\", \"value\": \"1970-01-01\"}, {\"field\": \"address\", \"value\": \"111 ANYWHERE STREET, CITY, DL 67890-101\"}, {\"field\": \"issue_date\", \"value\": \"2012-08-20\"}, {\"field\": \"expiration_date\", \"value\": \"2020-08-20\"}, {\"field\": \"state\", \"value\": \"New York\"}]}"}]} {"id": "drivers_license_fin_003_p1_classify", "task": "classify", "class": "drivers_license", "doc": "drivers_license_fin_003.pdf", "images": ["images/train/drivers_license/drivers_license_fin_003_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"drivers_license\", \"confidence\": 1.0, \"issuer\": \"Michigan\", \"holder\": \"JUNE MARIE SAMPLE\", \"page_marker\": \"\"}]}"}]} {"id": "drivers_license_fin_003_extract", "task": "extract", "class": "drivers_license", "doc": "drivers_license_fin_003.pdf", "images": ["images/train/drivers_license/drivers_license_fin_003_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('drivers_license', 'The document class.')\nfields = (['full_name', 'date_of_birth', 'address', 'issue_date', 'expiration_date', 'state'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name', 'date_of_birth': 'Date of Birth', 'address': 'Address', 'issue_date': 'Issued Date', 'expiration_date': 'Valid Until Date', 'state': 'Issuing state'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"JUNE MARIE SAMPLE\"}, {\"field\": \"date_of_birth\", \"value\": \"1984-05-01\"}, {\"field\": \"address\", \"value\": \"123 NORTH STATE ST. LANSING, MI 48918-0000\"}, {\"field\": \"issue_date\", \"value\": \"2018-04-01\"}, {\"field\": \"expiration_date\", \"value\": \"2022-05-01\"}, {\"field\": \"state\", \"value\": \"Michigan\"}]}"}]} {"id": "drivers_license_fin_005_p1_classify", "task": "classify", "class": "drivers_license", "doc": "drivers_license_fin_005.pdf", "images": ["images/train/drivers_license/drivers_license_fin_005_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"drivers_license\", \"confidence\": 1.0, \"issuer\": \"Oregon\", \"holder\": \"CARD OREGON SAMPLE, SR\", \"page_marker\": \"\"}]}"}]} {"id": "drivers_license_fin_005_extract", "task": "extract", "class": "drivers_license", "doc": "drivers_license_fin_005.pdf", "images": ["images/train/drivers_license/drivers_license_fin_005_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('drivers_license', 'The document class.')\nfields = (['full_name', 'date_of_birth', 'address', 'issue_date', 'expiration_date', 'state'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name', 'date_of_birth': 'Date of Birth', 'address': 'Address', 'issue_date': 'Issued Date', 'expiration_date': 'Valid Until Date', 'state': 'Issuing state'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"CARD OREGON SAMPLE, SR\"}, {\"field\": \"date_of_birth\", \"value\": \"1979-12-12\"}, {\"field\": \"address\", \"value\": \"1906 LANA AVE NE, SALEM, OR, 97314\"}, {\"field\": \"issue_date\", \"value\": \"2018-03-16\"}, {\"field\": \"expiration_date\", \"value\": \"2026-12-12\"}, {\"field\": \"state\", \"value\": \"Oregon\"}]}"}]} {"id": "drivers_license_fin_006_p1_classify", "task": "classify", "class": "drivers_license", "doc": "drivers_license_fin_006.pdf", "images": ["images/train/drivers_license/drivers_license_fin_006_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"drivers_license\", \"confidence\": 1.0, \"issuer\": \"North Carolina\", \"holder\": \"PUBLIC JANE Q\", \"page_marker\": \"\"}]}"}]} {"id": "drivers_license_fin_006_extract", "task": "extract", "class": "drivers_license", "doc": "drivers_license_fin_006.pdf", "images": ["images/train/drivers_license/drivers_license_fin_006_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('drivers_license', 'The document class.')\nfields = (['full_name', 'date_of_birth', 'address', 'issue_date', 'expiration_date', 'state'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name', 'date_of_birth': 'Date of Birth', 'address': 'Address', 'issue_date': 'Issued Date', 'expiration_date': 'Valid Until Date', 'state': 'Issuing state'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"PUBLIC JANE Q\"}, {\"field\": \"date_of_birth\", \"value\": \"1962-10-31\"}, {\"field\": \"address\", \"value\": \"12345 YOUR STREET, YOUR CITY, NC 99999-1234\"}, {\"field\": \"issue_date\", \"value\": \"2024-05-28\"}, {\"field\": \"expiration_date\", \"value\": \"2025-08-18\"}, {\"field\": \"state\", \"value\": \"North Carolina\"}]}"}]} {"id": "drivers_license_fin_007_p1_classify", "task": "classify", "class": "drivers_license", "doc": "drivers_license_fin_007.pdf", "images": ["images/train/drivers_license/drivers_license_fin_007_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"drivers_license\", \"confidence\": 1.0, \"issuer\": \"California\", \"holder\": \"GUITERREZ CARLOS\", \"page_marker\": \"\"}]}"}]} {"id": "drivers_license_fin_007_extract", "task": "extract", "class": "drivers_license", "doc": "drivers_license_fin_007.pdf", "images": ["images/train/drivers_license/drivers_license_fin_007_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('drivers_license', 'The document class.')\nfields = (['full_name', 'date_of_birth', 'address', 'issue_date', 'expiration_date', 'state'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name', 'date_of_birth': 'Date of Birth', 'address': 'Address', 'issue_date': 'Issued Date', 'expiration_date': 'Valid Until Date', 'state': 'Issuing state'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"GUITERREZ CARLOS\"}, {\"field\": \"date_of_birth\", \"value\": \"1987-04-23\"}, {\"field\": \"address\", \"value\": \"15212 COPPERMAN ST #202, LOS ANGELES, CA 90045\"}, {\"field\": \"issue_date\", \"value\": \"\"}, {\"field\": \"expiration_date\", \"value\": \"2018-04-23\"}, {\"field\": \"state\", \"value\": \"California\"}]}"}]} {"id": "drivers_license_fin_008_p1_classify", "task": "classify", "class": "drivers_license", "doc": "drivers_license_fin_008.pdf", "images": ["images/train/drivers_license/drivers_license_fin_008_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"drivers_license\", \"confidence\": 1.0, \"issuer\": \"United States Department of State\", \"holder\": \"DIPLOMAT, AMANDA\", \"page_marker\": \"\"}]}"}]} {"id": "drivers_license_fin_008_extract", "task": "extract", "class": "drivers_license", "doc": "drivers_license_fin_008.pdf", "images": ["images/train/drivers_license/drivers_license_fin_008_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('drivers_license', 'The document class.')\nfields = (['full_name', 'date_of_birth', 'address', 'issue_date', 'expiration_date', 'state'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name', 'date_of_birth': 'Date of Birth', 'address': 'Address', 'issue_date': 'Issued Date', 'expiration_date': 'Valid Until Date', 'state': 'Issuing state'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"DIPLOMAT, AMANDA\"}, {\"field\": \"date_of_birth\", \"value\": \"2002-01-01\"}, {\"field\": \"address\", \"value\": \"3507 INTERNATIONAL PL. NW, WASHINGTON, DC 20008\"}, {\"field\": \"issue_date\", \"value\": \"2002-05-22\"}, {\"field\": \"expiration_date\", \"value\": \"2007-05-31\"}, {\"field\": \"state\", \"value\": \"United States Department of State\"}]}"}]} {"id": "drivers_license_fin_009_p1_classify", "task": "classify", "class": "drivers_license", "doc": "drivers_license_fin_009.pdf", "images": ["images/train/drivers_license/drivers_license_fin_009_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"drivers_license\", \"confidence\": 1.0, \"issuer\": \"Florida\", \"holder\": \"SAMPLE NICK\", \"page_marker\": \"\"}]}"}]} {"id": "drivers_license_fin_009_extract", "task": "extract", "class": "drivers_license", "doc": "drivers_license_fin_009.pdf", "images": ["images/train/drivers_license/drivers_license_fin_009_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('drivers_license', 'The document class.')\nfields = (['full_name', 'date_of_birth', 'address', 'issue_date', 'expiration_date', 'state'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name', 'date_of_birth': 'Date of Birth', 'address': 'Address', 'issue_date': 'Issued Date', 'expiration_date': 'Valid Until Date', 'state': 'Issuing state'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"SAMPLE NICK\"}, {\"field\": \"date_of_birth\", \"value\": \"1957-01-12\"}, {\"field\": \"address\", \"value\": \"123 MAIN STREET, TALLAHASSEE, FL 00000-1234\"}, {\"field\": \"issue_date\", \"value\": \"2019-07-27\"}, {\"field\": \"expiration_date\", \"value\": \"2027-01-12\"}, {\"field\": \"state\", \"value\": \"Florida\"}]}"}]} {"id": "drivers_license_fin_010_p1_classify", "task": "classify", "class": "drivers_license", "doc": "drivers_license_fin_010.pdf", "images": ["images/train/drivers_license/drivers_license_fin_010_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"drivers_license\", \"confidence\": 1.0, \"issuer\": \"\", \"holder\": \"NAME SURNAME\", \"page_marker\": \"\"}]}"}]} {"id": "drivers_license_fin_010_extract", "task": "extract", "class": "drivers_license", "doc": "drivers_license_fin_010.pdf", "images": ["images/train/drivers_license/drivers_license_fin_010_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('drivers_license', 'The document class.')\nfields = (['full_name', 'date_of_birth', 'address', 'issue_date', 'expiration_date', 'state'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name', 'date_of_birth': 'Date of Birth', 'address': 'Address', 'issue_date': 'Issued Date', 'expiration_date': 'Valid Until Date', 'state': 'Issuing state'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"NAME SURNAME\"}, {\"field\": \"date_of_birth\", \"value\": \"1985-02-02\"}, {\"field\": \"address\", \"value\": \"123 ANYWAY Street, CITY, 34567-890\"}, {\"field\": \"issue_date\", \"value\": \"\"}, {\"field\": \"expiration_date\", \"value\": \"\"}, {\"field\": \"state\", \"value\": \"\"}]}"}]} {"id": "drivers_license_fin_013_p1_classify", "task": "classify", "class": "drivers_license", "doc": "drivers_license_fin_013.pdf", "images": ["images/train/drivers_license/drivers_license_fin_013_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"drivers_license\", \"confidence\": 1.0, \"issuer\": \"DISTRICT OF COLUMBIA\", \"holder\": \"THOMAS ALEXANDER SAMPLECARD\", \"page_marker\": \"\"}]}"}]} {"id": "drivers_license_fin_013_extract", "task": "extract", "class": "drivers_license", "doc": "drivers_license_fin_013.pdf", "images": ["images/train/drivers_license/drivers_license_fin_013_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('drivers_license', 'The document class.')\nfields = (['full_name', 'date_of_birth', 'address', 'issue_date', 'expiration_date', 'state'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name', 'date_of_birth': 'Date of Birth', 'address': 'Address', 'issue_date': 'Issued Date', 'expiration_date': 'Valid Until Date', 'state': 'Issuing state'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"THOMAS ALEXANDER SAMPLECARD\"}, {\"field\": \"date_of_birth\", \"value\": \"1984-02-21\"}, {\"field\": \"address\", \"value\": \"1234 COMMODORE JOSHUA BARNEY DRIVE, NE #1234 WASHINGTON, DC 00000-0000\"}, {\"field\": \"issue_date\", \"value\": \"2013-12-03\"}, {\"field\": \"expiration_date\", \"value\": \"2021-02-21\"}, {\"field\": \"state\", \"value\": \"DISTRICT OF COLUMBIA\"}]}"}]} {"id": "drivers_license_fin_015_p1_classify", "task": "classify", "class": "drivers_license", "doc": "drivers_license_fin_015.pdf", "images": ["images/train/drivers_license/drivers_license_fin_015_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"drivers_license\", \"confidence\": 1.0, \"issuer\": \"CALIFORNIA\", \"holder\": \"ORENTHAL JAMES SIMPSON\", \"page_marker\": \"\"}]}"}]} {"id": "drivers_license_fin_015_extract", "task": "extract", "class": "drivers_license", "doc": "drivers_license_fin_015.pdf", "images": ["images/train/drivers_license/drivers_license_fin_015_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('drivers_license', 'The document class.')\nfields = (['full_name', 'date_of_birth', 'address', 'issue_date', 'expiration_date', 'state'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name', 'date_of_birth': 'Date of Birth', 'address': 'Address', 'issue_date': 'Issued Date', 'expiration_date': 'Valid Until Date', 'state': 'Issuing state'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"ORENTHAL JAMES SIMPSON\"}, {\"field\": \"date_of_birth\", \"value\": \"1947-07-09\"}, {\"field\": \"address\", \"value\": \"11661 SAN VICENTE BL STE 632, LOS ANGELES CA 90049\"}, {\"field\": \"issue_date\", \"value\": \"1995-06-08\"}, {\"field\": \"expiration_date\", \"value\": \"1999-07-09\"}, {\"field\": \"state\", \"value\": \"CALIFORNIA\"}]}"}]} {"id": "drivers_license_fin_016_p1_classify", "task": "classify", "class": "drivers_license", "doc": "drivers_license_fin_016.pdf", "images": ["images/train/drivers_license/drivers_license_fin_016_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"drivers_license\", \"confidence\": 1.0, \"issuer\": \"Oklahoma\", \"holder\": \"SAMPLE DRIVER\", \"page_marker\": \"\"}]}"}]} {"id": "drivers_license_fin_016_extract", "task": "extract", "class": "drivers_license", "doc": "drivers_license_fin_016.pdf", "images": ["images/train/drivers_license/drivers_license_fin_016_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('drivers_license', 'The document class.')\nfields = (['full_name', 'date_of_birth', 'address', 'issue_date', 'expiration_date', 'state'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name', 'date_of_birth': 'Date of Birth', 'address': 'Address', 'issue_date': 'Issued Date', 'expiration_date': 'Valid Until Date', 'state': 'Issuing state'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"SAMPLE DRIVER\"}, {\"field\": \"date_of_birth\", \"value\": \"1951-10-15\"}, {\"field\": \"address\", \"value\": \"123 ANYSTREET, ANYCITY, OK 12345-0000\"}, {\"field\": \"issue_date\", \"value\": \"2026-10-15\"}, {\"field\": \"expiration_date\", \"value\": \"2034-10-31\"}, {\"field\": \"state\", \"value\": \"Oklahoma\"}]}"}]} {"id": "drivers_license_fin_017_p1_classify", "task": "classify", "class": "drivers_license", "doc": "drivers_license_fin_017.pdf", "images": ["images/train/drivers_license/drivers_license_fin_017_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"drivers_license\", \"confidence\": 1.0, \"issuer\": \"South Dakota\", \"holder\": \"SAMPLE SAMMY JACOB JOHN\", \"page_marker\": \"\"}]}"}]} {"id": "drivers_license_fin_017_extract", "task": "extract", "class": "drivers_license", "doc": "drivers_license_fin_017.pdf", "images": ["images/train/drivers_license/drivers_license_fin_017_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('drivers_license', 'The document class.')\nfields = (['full_name', 'date_of_birth', 'address', 'issue_date', 'expiration_date', 'state'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name', 'date_of_birth': 'Date of Birth', 'address': 'Address', 'issue_date': 'Issued Date', 'expiration_date': 'Valid Until Date', 'state': 'Issuing state'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"SAMPLE SAMMY JACOB JOHN\"}, {\"field\": \"date_of_birth\", \"value\": \"1982-08-08\"}, {\"field\": \"address\", \"value\": \"118 W CAPITOL AVE, PIERRE, SD 57501-2000\"}, {\"field\": \"issue_date\", \"value\": \"2021-09-09\"}, {\"field\": \"expiration_date\", \"value\": \"2026-08-08\"}, {\"field\": \"state\", \"value\": \"South Dakota\"}]}"}]} {"id": "drivers_license_fin_018_p1_classify", "task": "classify", "class": "drivers_license", "doc": "drivers_license_fin_018.pdf", "images": ["images/train/drivers_license/drivers_license_fin_018_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"drivers_license\", \"confidence\": 1.0, \"issuer\": \"Maine\", \"holder\": \"SUMMERS JULIE W\", \"page_marker\": \"\"}]}"}]} {"id": "drivers_license_fin_018_extract", "task": "extract", "class": "drivers_license", "doc": "drivers_license_fin_018.pdf", "images": ["images/train/drivers_license/drivers_license_fin_018_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('drivers_license', 'The document class.')\nfields = (['full_name', 'date_of_birth', 'address', 'issue_date', 'expiration_date', 'state'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name', 'date_of_birth': 'Date of Birth', 'address': 'Address', 'issue_date': 'Issued Date', 'expiration_date': 'Valid Until Date', 'state': 'Issuing state'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"SUMMERS JULIE W\"}, {\"field\": \"date_of_birth\", \"value\": \"1981-02-17\"}, {\"field\": \"address\", \"value\": \"6969 SUMMERSIDE BLVD, PORTLAND, ME 04101\"}, {\"field\": \"issue_date\", \"value\": \"2011-08-05\"}, {\"field\": \"expiration_date\", \"value\": \"2019-02-17\"}, {\"field\": \"state\", \"value\": \"Maine\"}]}"}]} {"id": "drivers_license_fin_020_p1_classify", "task": "classify", "class": "drivers_license", "doc": "drivers_license_fin_020.pdf", "images": ["images/train/drivers_license/drivers_license_fin_020_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"drivers_license\", \"confidence\": 1.0, \"issuer\": \"Wisconsin\", \"holder\": \"SAMPLE DEREK THOMAS\", \"page_marker\": \"\"}]}"}]} {"id": "drivers_license_fin_020_extract", "task": "extract", "class": "drivers_license", "doc": "drivers_license_fin_020.pdf", "images": ["images/train/drivers_license/drivers_license_fin_020_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('drivers_license', 'The document class.')\nfields = (['full_name', 'date_of_birth', 'address', 'issue_date', 'expiration_date', 'state'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name', 'date_of_birth': 'Date of Birth', 'address': 'Address', 'issue_date': 'Issued Date', 'expiration_date': 'Valid Until Date', 'state': 'Issuing state'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"SAMPLE DEREK THOMAS\"}, {\"field\": \"date_of_birth\", \"value\": \"1974-02-11\"}, {\"field\": \"address\", \"value\": \"4802 SHEBOYGAN AVE, MADISON, WI 53705\"}, {\"field\": \"issue_date\", \"value\": \"2015-10-31\"}, {\"field\": \"expiration_date\", \"value\": \"2023-02-11\"}, {\"field\": \"state\", \"value\": \"Wisconsin\"}]}"}]} {"id": "drivers_license_fin_021_p1_classify", "task": "classify", "class": "drivers_license", "doc": "drivers_license_fin_021.pdf", "images": ["images/train/drivers_license/drivers_license_fin_021_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"drivers_license\", \"confidence\": 1.0, \"issuer\": \"Alberta\", \"holder\": \"SAMPLE, Sally\", \"page_marker\": \"\"}]}"}]} {"id": "drivers_license_fin_021_extract", "task": "extract", "class": "drivers_license", "doc": "drivers_license_fin_021.pdf", "images": ["images/train/drivers_license/drivers_license_fin_021_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('drivers_license', 'The document class.')\nfields = (['full_name', 'date_of_birth', 'address', 'issue_date', 'expiration_date', 'state'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name', 'date_of_birth': 'Date of Birth', 'address': 'Address', 'issue_date': 'Issued Date', 'expiration_date': 'Valid Until Date', 'state': 'Issuing state'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"SAMPLE, Sally\"}, {\"field\": \"date_of_birth\", \"value\": \"1988-06-05\"}, {\"field\": \"address\", \"value\": \"12345 Somewhere Street, Anywhere AB T5N 2F5\"}, {\"field\": \"issue_date\", \"value\": \"2018-05-17\"}, {\"field\": \"expiration_date\", \"value\": \"2023-06-05\"}, {\"field\": \"state\", \"value\": \"Alberta\"}]}"}]} {"id": "drivers_license_fin_022_p1_classify", "task": "classify", "class": "drivers_license", "doc": "drivers_license_fin_022.pdf", "images": ["images/train/drivers_license/drivers_license_fin_022_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"drivers_license\", \"confidence\": 1.0, \"issuer\": \"New Jersey\", \"holder\": \"SAMPLE HEIDI\", \"page_marker\": \"\"}]}"}]} {"id": "drivers_license_fin_022_extract", "task": "extract", "class": "drivers_license", "doc": "drivers_license_fin_022.pdf", "images": ["images/train/drivers_license/drivers_license_fin_022_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('drivers_license', 'The document class.')\nfields = (['full_name', 'date_of_birth', 'address', 'issue_date', 'expiration_date', 'state'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name', 'date_of_birth': 'Date of Birth', 'address': 'Address', 'issue_date': 'Issued Date', 'expiration_date': 'Valid Until Date', 'state': 'Issuing state'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"SAMPLE HEIDI\"}, {\"field\": \"date_of_birth\", \"value\": \"1967-01-12\"}, {\"field\": \"address\", \"value\": \"123 NORTH STATE STREET, TRENTON, NJ 08666-1234\"}, {\"field\": \"issue_date\", \"value\": \"2022-01-05\"}, {\"field\": \"expiration_date\", \"value\": \"2026-01-12\"}, {\"field\": \"state\", \"value\": \"New Jersey\"}]}"}]} {"id": "drivers_license_fin_024_p1_classify", "task": "classify", "class": "drivers_license", "doc": "drivers_license_fin_024.pdf", "images": ["images/train/drivers_license/drivers_license_fin_024_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"drivers_license\", \"confidence\": 1.0, \"issuer\": \"Connecticut\", \"holder\": \"SAMPLE SUSAN CATHERINE\", \"page_marker\": \"\"}]}"}]} {"id": "drivers_license_fin_024_extract", "task": "extract", "class": "drivers_license", "doc": "drivers_license_fin_024.pdf", "images": ["images/train/drivers_license/drivers_license_fin_024_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('drivers_license', 'The document class.')\nfields = (['full_name', 'date_of_birth', 'address', 'issue_date', 'expiration_date', 'state'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name', 'date_of_birth': 'Date of Birth', 'address': 'Address', 'issue_date': 'Issued Date', 'expiration_date': 'Valid Until Date', 'state': 'Issuing state'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"SAMPLE SUSAN CATHERINE\"}, {\"field\": \"date_of_birth\", \"value\": \"1978-05-30\"}, {\"field\": \"address\", \"value\": \"60 STATE ST, ROOM 145, WETHERSFIELD CT 06109\"}, {\"field\": \"issue_date\", \"value\": \"2009-05-30\"}, {\"field\": \"expiration_date\", \"value\": \"2015-05-30\"}, {\"field\": \"state\", \"value\": \"Connecticut\"}]}"}]} {"id": "drivers_license_fin_025_p1_classify", "task": "classify", "class": "drivers_license", "doc": "drivers_license_fin_025.pdf", "images": ["images/train/drivers_license/drivers_license_fin_025_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"drivers_license\", \"confidence\": 1.0, \"issuer\": \"GEORGIA\", \"holder\": \"IMA GEORGIA SAMPLE\", \"page_marker\": \"\"}]}"}]} {"id": "drivers_license_fin_025_extract", "task": "extract", "class": "drivers_license", "doc": "drivers_license_fin_025.pdf", "images": ["images/train/drivers_license/drivers_license_fin_025_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('drivers_license', 'The document class.')\nfields = (['full_name', 'date_of_birth', 'address', 'issue_date', 'expiration_date', 'state'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name', 'date_of_birth': 'Date of Birth', 'address': 'Address', 'issue_date': 'Issued Date', 'expiration_date': 'Valid Until Date', 'state': 'Issuing state'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"IMA GEORGIA SAMPLE\"}, {\"field\": \"date_of_birth\", \"value\": \"1954-02-22\"}, {\"field\": \"address\", \"value\": \"123 MAIN STREET, ANYTOWN, GA 39999, ANYCOUNTY\"}, {\"field\": \"issue_date\", \"value\": \"2018-02-22\"}, {\"field\": \"expiration_date\", \"value\": \"2028-02-22\"}, {\"field\": \"state\", \"value\": \"GEORGIA\"}]}"}]} {"id": "drivers_license_fin_027_p1_classify", "task": "classify", "class": "drivers_license", "doc": "drivers_license_fin_027.pdf", "images": ["images/train/drivers_license/drivers_license_fin_027_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"drivers_license\", \"confidence\": 1.0, \"issuer\": \"MARYLAND\", \"holder\": \"JANE DIANE SAMPLE\", \"page_marker\": \"\"}]}"}]} {"id": "drivers_license_fin_027_extract", "task": "extract", "class": "drivers_license", "doc": "drivers_license_fin_027.pdf", "images": ["images/train/drivers_license/drivers_license_fin_027_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('drivers_license', 'The document class.')\nfields = (['full_name', 'date_of_birth', 'address', 'issue_date', 'expiration_date', 'state'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name', 'date_of_birth': 'Date of Birth', 'address': 'Address', 'issue_date': 'Issued Date', 'expiration_date': 'Valid Until Date', 'state': 'Issuing state'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"JANE DIANE SAMPLE\"}, {\"field\": \"date_of_birth\", \"value\": \"1985-07-07\"}, {\"field\": \"address\", \"value\": \"123 ANY STREET GLEN BURNIE MD 21062\"}, {\"field\": \"issue_date\", \"value\": \"2016-03-25\"}, {\"field\": \"expiration_date\", \"value\": \"2023-07-07\"}, {\"field\": \"state\", \"value\": \"MARYLAND\"}]}"}]} {"id": "drivers_license_fin_028_p1_classify", "task": "classify", "class": "drivers_license", "doc": "drivers_license_fin_028.pdf", "images": ["images/train/drivers_license/drivers_license_fin_028_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"drivers_license\", \"confidence\": 1.0, \"issuer\": \"WASHINGTON\", \"holder\": \"WHITE, LIST ELEVEN\", \"page_marker\": \"\"}]}"}]} {"id": "drivers_license_fin_028_extract", "task": "extract", "class": "drivers_license", "doc": "drivers_license_fin_028.pdf", "images": ["images/train/drivers_license/drivers_license_fin_028_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('drivers_license', 'The document class.')\nfields = (['full_name', 'date_of_birth', 'address', 'issue_date', 'expiration_date', 'state'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name', 'date_of_birth': 'Date of Birth', 'address': 'Address', 'issue_date': 'Issued Date', 'expiration_date': 'Valid Until Date', 'state': 'Issuing state'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"WHITE, LIST ELEVEN\"}, {\"field\": \"date_of_birth\", \"value\": \"1991-01-01\"}, {\"field\": \"address\", \"value\": \"2600 MARTIN WAY E, OLYMPIA WA 98506-4927\"}, {\"field\": \"issue_date\", \"value\": \"2013-01-23\"}, {\"field\": \"expiration_date\", \"value\": \"2018-01-01\"}, {\"field\": \"state\", \"value\": \"WASHINGTON\"}]}"}]} {"id": "drivers_license_fin_029_p1_classify", "task": "classify", "class": "drivers_license", "doc": "drivers_license_fin_029.pdf", "images": ["images/train/drivers_license/drivers_license_fin_029_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"drivers_license\", \"confidence\": 1.0, \"issuer\": \"MICHIGAN\", \"holder\": \"JUNE SAMPLE\", \"page_marker\": \"\"}]}"}]} {"id": "drivers_license_fin_029_extract", "task": "extract", "class": "drivers_license", "doc": "drivers_license_fin_029.pdf", "images": ["images/train/drivers_license/drivers_license_fin_029_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('drivers_license', 'The document class.')\nfields = (['full_name', 'date_of_birth', 'address', 'issue_date', 'expiration_date', 'state'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name', 'date_of_birth': 'Date of Birth', 'address': 'Address', 'issue_date': 'Issued Date', 'expiration_date': 'Valid Until Date', 'state': 'Issuing state'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"JUNE SAMPLE\"}, {\"field\": \"date_of_birth\", \"value\": \"1977-02-01\"}, {\"field\": \"address\", \"value\": \"123 NORTH STATE ST. LANSING, MI 48918-0000\"}, {\"field\": \"issue_date\", \"value\": \"2007-02-01\"}, {\"field\": \"expiration_date\", \"value\": \"2012-02-01\"}, {\"field\": \"state\", \"value\": \"MICHIGAN\"}]}"}]} {"id": "drivers_license_fin_031_p1_classify", "task": "classify", "class": "drivers_license", "doc": "drivers_license_fin_031.pdf", "images": ["images/train/drivers_license/drivers_license_fin_031_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"drivers_license\", \"confidence\": 1.0, \"issuer\": \"NEW YORK STATE\", \"holder\": \"SAMPLE,SALLY\", \"page_marker\": \"\"}]}"}]} {"id": "drivers_license_fin_031_extract", "task": "extract", "class": "drivers_license", "doc": "drivers_license_fin_031.pdf", "images": ["images/train/drivers_license/drivers_license_fin_031_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('drivers_license', 'The document class.')\nfields = (['full_name', 'date_of_birth', 'address', 'issue_date', 'expiration_date', 'state'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name', 'date_of_birth': 'Date of Birth', 'address': 'Address', 'issue_date': 'Issued Date', 'expiration_date': 'Valid Until Date', 'state': 'Issuing state'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"SAMPLE,SALLY\"}, {\"field\": \"date_of_birth\", \"value\": \"1983-07-18\"}, {\"field\": \"address\", \"value\": \"1010 ANYPLACE ST YOURCITY NY 12121\"}, {\"field\": \"issue_date\", \"value\": \"2005-07-18\"}, {\"field\": \"expiration_date\", \"value\": \"2013-07-18\"}, {\"field\": \"state\", \"value\": \"NEW YORK STATE\"}]}"}]} {"id": "drivers_license_fin_032_p1_classify", "task": "classify", "class": "drivers_license", "doc": "drivers_license_fin_032.pdf", "images": ["images/train/drivers_license/drivers_license_fin_032_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"drivers_license\", \"confidence\": 1.0, \"issuer\": \"NEBRASKA\", \"holder\": \"SAMPLE CONNOR M, JR\", \"page_marker\": \"\"}]}"}]} {"id": "drivers_license_fin_032_extract", "task": "extract", "class": "drivers_license", "doc": "drivers_license_fin_032.pdf", "images": ["images/train/drivers_license/drivers_license_fin_032_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('drivers_license', 'The document class.')\nfields = (['full_name', 'date_of_birth', 'address', 'issue_date', 'expiration_date', 'state'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name', 'date_of_birth': 'Date of Birth', 'address': 'Address', 'issue_date': 'Issued Date', 'expiration_date': 'Valid Until Date', 'state': 'Issuing state'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"SAMPLE CONNOR M, JR\"}, {\"field\": \"date_of_birth\", \"value\": \"1967-01-12\"}, {\"field\": \"address\", \"value\": \"123 MAIN STREET, ANYTOWN, NE 12345\"}, {\"field\": \"issue_date\", \"value\": \"2023-05-01\"}, {\"field\": \"expiration_date\", \"value\": \"2028-01-12\"}, {\"field\": \"state\", \"value\": \"NEBRASKA\"}]}"}]} {"id": "drivers_license_fin_033_p1_classify", "task": "classify", "class": "drivers_license", "doc": "drivers_license_fin_033.pdf", "images": ["images/train/drivers_license/drivers_license_fin_033_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"drivers_license\", \"confidence\": 1.0, \"issuer\": \"South Carolina\", \"holder\": \"SAMPLE JANE MARIA\", \"page_marker\": \"\"}]}"}]} {"id": "drivers_license_fin_033_extract", "task": "extract", "class": "drivers_license", "doc": "drivers_license_fin_033.pdf", "images": ["images/train/drivers_license/drivers_license_fin_033_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('drivers_license', 'The document class.')\nfields = (['full_name', 'date_of_birth', 'address', 'issue_date', 'expiration_date', 'state'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name', 'date_of_birth': 'Date of Birth', 'address': 'Address', 'issue_date': 'Issued Date', 'expiration_date': 'Valid Until Date', 'state': 'Issuing state'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"SAMPLE JANE MARIA\"}, {\"field\": \"date_of_birth\", \"value\": \"1970-07-04\"}, {\"field\": \"address\", \"value\": \"123 MAIN STREET, ANYTOWN, SC 123451111\"}, {\"field\": \"issue_date\", \"value\": \"2017-10-01\"}, {\"field\": \"expiration_date\", \"value\": \"2025-07-04\"}, {\"field\": \"state\", \"value\": \"South Carolina\"}]}"}]} {"id": "drivers_license_fin_034_p1_classify", "task": "classify", "class": "drivers_license", "doc": "drivers_license_fin_034.pdf", "images": ["images/train/drivers_license/drivers_license_fin_034_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"drivers_license\", \"confidence\": 1.0, \"issuer\": \"Missouri\", \"holder\": \"SAMPLE JANE DOE\", \"page_marker\": \"\"}]}"}]} {"id": "drivers_license_fin_034_extract", "task": "extract", "class": "drivers_license", "doc": "drivers_license_fin_034.pdf", "images": ["images/train/drivers_license/drivers_license_fin_034_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('drivers_license', 'The document class.')\nfields = (['full_name', 'date_of_birth', 'address', 'issue_date', 'expiration_date', 'state'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name', 'date_of_birth': 'Date of Birth', 'address': 'Address', 'issue_date': 'Issued Date', 'expiration_date': 'Valid Until Date', 'state': 'Issuing state'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"SAMPLE JANE DOE\"}, {\"field\": \"date_of_birth\", \"value\": \"1979-02-01\"}, {\"field\": \"address\", \"value\": \"123 SAMPLE STREET, APT. 4, JEFFERSON CITY, MO 65109\"}, {\"field\": \"issue_date\", \"value\": \"2013-02-01\"}, {\"field\": \"expiration_date\", \"value\": \"2019-02-01\"}, {\"field\": \"state\", \"value\": \"Missouri\"}]}"}]} {"id": "drivers_license_fin_035_p1_classify", "task": "classify", "class": "drivers_license", "doc": "drivers_license_fin_035.pdf", "images": ["images/train/drivers_license/drivers_license_fin_035_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"drivers_license\", \"confidence\": 1.0, \"issuer\": \"Texas\", \"holder\": \"SAMPLE JANICE\", \"page_marker\": \"\"}]}"}]} {"id": "drivers_license_fin_035_extract", "task": "extract", "class": "drivers_license", "doc": "drivers_license_fin_035.pdf", "images": ["images/train/drivers_license/drivers_license_fin_035_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('drivers_license', 'The document class.')\nfields = (['full_name', 'date_of_birth', 'address', 'issue_date', 'expiration_date', 'state'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name', 'date_of_birth': 'Date of Birth', 'address': 'Address', 'issue_date': 'Issued Date', 'expiration_date': 'Valid Until Date', 'state': 'Issuing state'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"SAMPLE JANICE\"}, {\"field\": \"date_of_birth\", \"value\": \"1976-07-30\"}, {\"field\": \"address\", \"value\": \"2120 OLD MAIN STREET ANYTOWN TX 12345-0000\"}, {\"field\": \"issue_date\", \"value\": \"2015-07-30\"}, {\"field\": \"expiration_date\", \"value\": \"2021-07-30\"}, {\"field\": \"state\", \"value\": \"Texas\"}]}"}]} {"id": "drivers_license_fin_036_p1_classify", "task": "classify", "class": "drivers_license", "doc": "drivers_license_fin_036.pdf", "images": ["images/train/drivers_license/drivers_license_fin_036_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"drivers_license\", \"confidence\": 1.0, \"issuer\": \"Florida\", \"holder\": \"JACKSON ELLIS CARTER\", \"page_marker\": \"\"}]}"}]} {"id": "drivers_license_fin_036_extract", "task": "extract", "class": "drivers_license", "doc": "drivers_license_fin_036.pdf", "images": ["images/train/drivers_license/drivers_license_fin_036_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('drivers_license', 'The document class.')\nfields = (['full_name', 'date_of_birth', 'address', 'issue_date', 'expiration_date', 'state'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name', 'date_of_birth': 'Date of Birth', 'address': 'Address', 'issue_date': 'Issued Date', 'expiration_date': 'Valid Until Date', 'state': 'Issuing state'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"JACKSON ELLIS CARTER\"}, {\"field\": \"date_of_birth\", \"value\": \"1993-06-03\"}, {\"field\": \"address\", \"value\": \"5246 BILLINGS ST, LEHIGH ACRES, FL 33971\"}, {\"field\": \"issue_date\", \"value\": \"2020-11-05\"}, {\"field\": \"expiration_date\", \"value\": \"2029-06-03\"}, {\"field\": \"state\", \"value\": \"Florida\"}]}"}]} {"id": "drivers_license_fin_037_p1_classify", "task": "classify", "class": "drivers_license", "doc": "drivers_license_fin_037.pdf", "images": ["images/train/drivers_license/drivers_license_fin_037_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"drivers_license\", \"confidence\": 1.0, \"issuer\": \"California\", \"holder\": \"SMOYER DAVID\", \"page_marker\": \"\"}]}"}]} {"id": "drivers_license_fin_037_extract", "task": "extract", "class": "drivers_license", "doc": "drivers_license_fin_037.pdf", "images": ["images/train/drivers_license/drivers_license_fin_037_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('drivers_license', 'The document class.')\nfields = (['full_name', 'date_of_birth', 'address', 'issue_date', 'expiration_date', 'state'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name', 'date_of_birth': 'Date of Birth', 'address': 'Address', 'issue_date': 'Issued Date', 'expiration_date': 'Valid Until Date', 'state': 'Issuing state'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"SMOYER DAVID\"}, {\"field\": \"date_of_birth\", \"value\": \"1981-05-23\"}, {\"field\": \"address\", \"value\": \"13313 HIGBY HILL ROAD, TOWNVILLE, CA 16360\"}, {\"field\": \"issue_date\", \"value\": \"2015-02-24\"}, {\"field\": \"expiration_date\", \"value\": \"2020-05-23\"}, {\"field\": \"state\", \"value\": \"California\"}]}"}]} {"id": "drivers_license_fin_038_p1_classify", "task": "classify", "class": "drivers_license", "doc": "drivers_license_fin_038.pdf", "images": ["images/train/drivers_license/drivers_license_fin_038_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"drivers_license\", \"confidence\": 1.0, \"issuer\": \"California\", \"holder\": \"DAVID GUAN\", \"page_marker\": \"\"}]}"}]} {"id": "drivers_license_fin_038_extract", "task": "extract", "class": "drivers_license", "doc": "drivers_license_fin_038.pdf", "images": ["images/train/drivers_license/drivers_license_fin_038_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('drivers_license', 'The document class.')\nfields = (['full_name', 'date_of_birth', 'address', 'issue_date', 'expiration_date', 'state'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name', 'date_of_birth': 'Date of Birth', 'address': 'Address', 'issue_date': 'Issued Date', 'expiration_date': 'Valid Until Date', 'state': 'Issuing state'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"DAVID GUAN\"}, {\"field\": \"date_of_birth\", \"value\": \"1974-03-04\"}, {\"field\": \"address\", \"value\": \"2570 24TH STREET, SACRAMENTO, CA 95818\"}, {\"field\": \"issue_date\", \"value\": \"2013-10-05\"}, {\"field\": \"expiration_date\", \"value\": \"2018-10-06\"}, {\"field\": \"state\", \"value\": \"California\"}]}"}]} {"id": "drivers_license_fin_039_p1_classify", "task": "classify", "class": "drivers_license", "doc": "drivers_license_fin_039.pdf", "images": ["images/train/drivers_license/drivers_license_fin_039_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"drivers_license\", \"confidence\": 1.0, \"issuer\": \"AZ\", \"holder\": \"ROBERT JASON BLUM\", \"page_marker\": \"\"}]}"}]} {"id": "drivers_license_fin_039_extract", "task": "extract", "class": "drivers_license", "doc": "drivers_license_fin_039.pdf", "images": ["images/train/drivers_license/drivers_license_fin_039_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('drivers_license', 'The document class.')\nfields = (['full_name', 'date_of_birth', 'address', 'issue_date', 'expiration_date', 'state'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name', 'date_of_birth': 'Date of Birth', 'address': 'Address', 'issue_date': 'Issued Date', 'expiration_date': 'Valid Until Date', 'state': 'Issuing state'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"ROBERT JASON BLUM\"}, {\"field\": \"date_of_birth\", \"value\": \"1980-05-14\"}, {\"field\": \"address\", \"value\": \"2033 ROCK BEACH DR. CLEARLAKE, AZ 85129\"}, {\"field\": \"issue_date\", \"value\": \"2013-02-17\"}, {\"field\": \"expiration_date\", \"value\": \"2024-05-14\"}, {\"field\": \"state\", \"value\": \"AZ\"}]}"}]} {"id": "drivers_license_fin_040_p1_classify", "task": "classify", "class": "drivers_license", "doc": "drivers_license_fin_040.pdf", "images": ["images/train/drivers_license/drivers_license_fin_040_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"drivers_license\", \"confidence\": 1.0, \"issuer\": \"OREGON\", \"holder\": \"SAMPLE BRENDA LYNN\", \"page_marker\": \"\"}]}"}]} {"id": "drivers_license_fin_040_extract", "task": "extract", "class": "drivers_license", "doc": "drivers_license_fin_040.pdf", "images": ["images/train/drivers_license/drivers_license_fin_040_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('drivers_license', 'The document class.')\nfields = (['full_name', 'date_of_birth', 'address', 'issue_date', 'expiration_date', 'state'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name', 'date_of_birth': 'Date of Birth', 'address': 'Address', 'issue_date': 'Issued Date', 'expiration_date': 'Valid Until Date', 'state': 'Issuing state'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"SAMPLE BRENDA LYNN\"}, {\"field\": \"date_of_birth\", \"value\": \"1980-06-30\"}, {\"field\": \"address\", \"value\": \"1906 LANA AVE NE, SALEM, OR, 97314\"}, {\"field\": \"issue_date\", \"value\": \"2023-06-30\"}, {\"field\": \"expiration_date\", \"value\": \"2031-06-30\"}, {\"field\": \"state\", \"value\": \"OREGON\"}]}"}]} {"id": "drivers_license_fin_042_p1_classify", "task": "classify", "class": "drivers_license", "doc": "drivers_license_fin_042.pdf", "images": ["images/train/drivers_license/drivers_license_fin_042_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"drivers_license\", \"confidence\": 1.0, \"issuer\": \"South Carolina\", \"holder\": \"SAMPLE JANE MARIA\", \"page_marker\": \"\"}]}"}]} {"id": "drivers_license_fin_042_extract", "task": "extract", "class": "drivers_license", "doc": "drivers_license_fin_042.pdf", "images": ["images/train/drivers_license/drivers_license_fin_042_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('drivers_license', 'The document class.')\nfields = (['full_name', 'date_of_birth', 'address', 'issue_date', 'expiration_date', 'state'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name', 'date_of_birth': 'Date of Birth', 'address': 'Address', 'issue_date': 'Issued Date', 'expiration_date': 'Valid Until Date', 'state': 'Issuing state'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"SAMPLE JANE MARIA\"}, {\"field\": \"date_of_birth\", \"value\": \"1970-07-04\"}, {\"field\": \"address\", \"value\": \"123 MAIN STREET ANYTOWN, SC 123451111\"}, {\"field\": \"issue_date\", \"value\": \"2017-10-01\"}, {\"field\": \"expiration_date\", \"value\": \"2025-07-04\"}, {\"field\": \"state\", \"value\": \"South Carolina\"}]}"}]} {"id": "drivers_license_fin_044_p1_classify", "task": "classify", "class": "drivers_license", "doc": "drivers_license_fin_044.pdf", "images": ["images/train/drivers_license/drivers_license_fin_044_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"drivers_license\", \"confidence\": 1.0, \"issuer\": \"Georgia\", \"holder\": \"IMA GEORGIA SAMPLE\", \"page_marker\": \"\"}]}"}]} {"id": "drivers_license_fin_044_extract", "task": "extract", "class": "drivers_license", "doc": "drivers_license_fin_044.pdf", "images": ["images/train/drivers_license/drivers_license_fin_044_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('drivers_license', 'The document class.')\nfields = (['full_name', 'date_of_birth', 'address', 'issue_date', 'expiration_date', 'state'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name', 'date_of_birth': 'Date of Birth', 'address': 'Address', 'issue_date': 'Issued Date', 'expiration_date': 'Valid Until Date', 'state': 'Issuing state'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"IMA GEORGIA SAMPLE\"}, {\"field\": \"date_of_birth\", \"value\": \"1968-07-01\"}, {\"field\": \"address\", \"value\": \"1234 ANY ST, ANY TOWN, GA 33333\"}, {\"field\": \"issue_date\", \"value\": \"2012-03-22\"}, {\"field\": \"expiration_date\", \"value\": \"2017-07-01\"}, {\"field\": \"state\", \"value\": \"Georgia\"}]}"}]} {"id": "drivers_license_fin_045_p1_classify", "task": "classify", "class": "drivers_license", "doc": "drivers_license_fin_045.pdf", "images": ["images/train/drivers_license/drivers_license_fin_045_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"drivers_license\", \"confidence\": 1.0, \"issuer\": \"CALIFORNIA\", \"holder\": \"SAMPLE, SEIJI\", \"page_marker\": \"\"}]}"}]} {"id": "drivers_license_fin_045_extract", "task": "extract", "class": "drivers_license", "doc": "drivers_license_fin_045.pdf", "images": ["images/train/drivers_license/drivers_license_fin_045_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('drivers_license', 'The document class.')\nfields = (['full_name', 'date_of_birth', 'address', 'issue_date', 'expiration_date', 'state'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name', 'date_of_birth': 'Date of Birth', 'address': 'Address', 'issue_date': 'Issued Date', 'expiration_date': 'Valid Until Date', 'state': 'Issuing state'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"SAMPLE, SEIJI\"}, {\"field\": \"date_of_birth\", \"value\": \"1977-08-31\"}, {\"field\": \"address\", \"value\": \"2570 24TH STREET, SACRAMENTO, CA 95818\"}, {\"field\": \"issue_date\", \"value\": \"2010-09-30\"}, {\"field\": \"expiration_date\", \"value\": \"2015-08-31\"}, {\"field\": \"state\", \"value\": \"CALIFORNIA\"}]}"}]} {"id": "drivers_license_fin_046_p1_classify", "task": "classify", "class": "drivers_license", "doc": "drivers_license_fin_046.pdf", "images": ["images/train/drivers_license/drivers_license_fin_046_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"drivers_license\", \"confidence\": 1.0, \"issuer\": \"Ontario\", \"holder\": \"DAVID, TAGGART C\", \"page_marker\": \"\"}]}"}]} {"id": "drivers_license_fin_046_extract", "task": "extract", "class": "drivers_license", "doc": "drivers_license_fin_046.pdf", "images": ["images/train/drivers_license/drivers_license_fin_046_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('drivers_license', 'The document class.')\nfields = (['full_name', 'date_of_birth', 'address', 'issue_date', 'expiration_date', 'state'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name', 'date_of_birth': 'Date of Birth', 'address': 'Address', 'issue_date': 'Issued Date', 'expiration_date': 'Valid Until Date', 'state': 'Issuing state'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"DAVID, TAGGART C\"}, {\"field\": \"date_of_birth\", \"value\": \"1969-05-30\"}, {\"field\": \"address\", \"value\": \"10489 CLARK RD, MOUNTAIN ONTARIO K0E1S0\"}, {\"field\": \"issue_date\", \"value\": \"2018-05-30\"}, {\"field\": \"expiration_date\", \"value\": \"2023-05-30\"}, {\"field\": \"state\", \"value\": \"Ontario\"}]}"}]} {"id": "drivers_license_fin_047_p1_classify", "task": "classify", "class": "drivers_license", "doc": "drivers_license_fin_047.pdf", "images": ["images/train/drivers_license/drivers_license_fin_047_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"drivers_license\", \"confidence\": 1.0, \"issuer\": \"ALABAMA\", \"holder\": \"HEIDI SAMPLE\", \"page_marker\": \"\"}]}"}]} {"id": "drivers_license_fin_047_extract", "task": "extract", "class": "drivers_license", "doc": "drivers_license_fin_047.pdf", "images": ["images/train/drivers_license/drivers_license_fin_047_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('drivers_license', 'The document class.')\nfields = (['full_name', 'date_of_birth', 'address', 'issue_date', 'expiration_date', 'state'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name', 'date_of_birth': 'Date of Birth', 'address': 'Address', 'issue_date': 'Issued Date', 'expiration_date': 'Valid Until Date', 'state': 'Issuing state'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"HEIDI SAMPLE\"}, {\"field\": \"date_of_birth\", \"value\": \"1984-05-05\"}, {\"field\": \"address\", \"value\": \"1234 SAMPLE STREET, MOBILE AL 36608-5658\"}, {\"field\": \"issue_date\", \"value\": \"2018-09-04\"}, {\"field\": \"expiration_date\", \"value\": \"2022-09-04\"}, {\"field\": \"state\", \"value\": \"ALABAMA\"}]}"}]} {"id": "drivers_license_fin_048_p1_classify", "task": "classify", "class": "drivers_license", "doc": "drivers_license_fin_048.pdf", "images": ["images/train/drivers_license/drivers_license_fin_048_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"drivers_license\", \"confidence\": 1.0, \"issuer\": \"Wyoming\", \"holder\": \"DOE JOHN LEE\", \"page_marker\": \"\"}]}"}]} {"id": "drivers_license_fin_048_extract", "task": "extract", "class": "drivers_license", "doc": "drivers_license_fin_048.pdf", "images": ["images/train/drivers_license/drivers_license_fin_048_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('drivers_license', 'The document class.')\nfields = (['full_name', 'date_of_birth', 'address', 'issue_date', 'expiration_date', 'state'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name', 'date_of_birth': 'Date of Birth', 'address': 'Address', 'issue_date': 'Issued Date', 'expiration_date': 'Valid Until Date', 'state': 'Issuing state'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"DOE JOHN LEE\"}, {\"field\": \"date_of_birth\", \"value\": \"1986-10-16\"}, {\"field\": \"address\", \"value\": \"123 OLD MAIN STREET, ANYTONWN, WY 12345\"}, {\"field\": \"issue_date\", \"value\": \"2020-12-15\"}, {\"field\": \"expiration_date\", \"value\": \"2025-10-16\"}, {\"field\": \"state\", \"value\": \"Wyoming\"}]}"}]} {"id": "drivers_license_fin_050_p1_classify", "task": "classify", "class": "drivers_license", "doc": "drivers_license_fin_050.pdf", "images": ["images/train/drivers_license/drivers_license_fin_050_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"drivers_license\", \"confidence\": 1.0, \"issuer\": \"Wisconsin\", \"holder\": \"SAMPLE GEOFFRY JAMES, II\", \"page_marker\": \"\"}]}"}]} {"id": "drivers_license_fin_050_extract", "task": "extract", "class": "drivers_license", "doc": "drivers_license_fin_050.pdf", "images": ["images/train/drivers_license/drivers_license_fin_050_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('drivers_license', 'The document class.')\nfields = (['full_name', 'date_of_birth', 'address', 'issue_date', 'expiration_date', 'state'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name', 'date_of_birth': 'Date of Birth', 'address': 'Address', 'issue_date': 'Issued Date', 'expiration_date': 'Valid Until Date', 'state': 'Issuing state'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"SAMPLE GEOFFRY JAMES, II\"}, {\"field\": \"date_of_birth\", \"value\": \"1985-01-19\"}, {\"field\": \"address\", \"value\": \"4822 MADISON YARDS WAY, SUITE 1205, MADISON, WI 53705\"}, {\"field\": \"issue_date\", \"value\": \"2023-05-31\"}, {\"field\": \"expiration_date\", \"value\": \"2030-01-19\"}, {\"field\": \"state\", \"value\": \"Wisconsin\"}]}"}]} {"id": "drivers_license_fin_051_p1_classify", "task": "classify", "class": "drivers_license", "doc": "drivers_license_fin_051.pdf", "images": ["images/train/drivers_license/drivers_license_fin_051_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"drivers_license\", \"confidence\": 1.0, \"issuer\": \"California\", \"holder\": \"JOHN CITIZEN WILLIAMS\", \"page_marker\": \"\"}]}"}]} {"id": "drivers_license_fin_051_extract", "task": "extract", "class": "drivers_license", "doc": "drivers_license_fin_051.pdf", "images": ["images/train/drivers_license/drivers_license_fin_051_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('drivers_license', 'The document class.')\nfields = (['full_name', 'date_of_birth', 'address', 'issue_date', 'expiration_date', 'state'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name', 'date_of_birth': 'Date of Birth', 'address': 'Address', 'issue_date': 'Issued Date', 'expiration_date': 'Valid Until Date', 'state': 'Issuing state'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"JOHN CITIZEN WILLIAMS\"}, {\"field\": \"date_of_birth\", \"value\": \"1980-12-12\"}, {\"field\": \"address\", \"value\": \"2711 12TH STREET, SACRAMENTO, AA 63727\"}, {\"field\": \"issue_date\", \"value\": \"2010-12-12\"}, {\"field\": \"expiration_date\", \"value\": \"2026-12-12\"}, {\"field\": \"state\", \"value\": \"California\"}]}"}]} {"id": "drivers_license_fin_052_p1_classify", "task": "classify", "class": "drivers_license", "doc": "drivers_license_fin_052.pdf", "images": ["images/train/drivers_license/drivers_license_fin_052_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"drivers_license\", \"confidence\": 1.0, \"issuer\": \"Georgia\", \"holder\": \"IMA GEORGIA SAMPLE\", \"page_marker\": \"\"}]}"}]} {"id": "drivers_license_fin_052_extract", "task": "extract", "class": "drivers_license", "doc": "drivers_license_fin_052.pdf", "images": ["images/train/drivers_license/drivers_license_fin_052_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('drivers_license', 'The document class.')\nfields = (['full_name', 'date_of_birth', 'address', 'issue_date', 'expiration_date', 'state'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name', 'date_of_birth': 'Date of Birth', 'address': 'Address', 'issue_date': 'Issued Date', 'expiration_date': 'Valid Until Date', 'state': 'Issuing state'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"IMA GEORGIA SAMPLE\"}, {\"field\": \"date_of_birth\", \"value\": \"1971-10-23\"}, {\"field\": \"address\", \"value\": \"123 MAIN STREET, ANYTOWN, GA 33333\"}, {\"field\": \"issue_date\", \"value\": \"2018-04-11\"}, {\"field\": \"expiration_date\", \"value\": \"2022-04-11\"}, {\"field\": \"state\", \"value\": \"Georgia\"}]}"}]} {"id": "drivers_license_fin_053_p1_classify", "task": "classify", "class": "drivers_license", "doc": "drivers_license_fin_053.pdf", "images": ["images/train/drivers_license/drivers_license_fin_053_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"drivers_license\", \"confidence\": 1.0, \"issuer\": \"Idaho\", \"holder\": \"ARMSTRONG THOMAS\", \"page_marker\": \"\"}]}"}]} {"id": "drivers_license_fin_053_extract", "task": "extract", "class": "drivers_license", "doc": "drivers_license_fin_053.pdf", "images": ["images/train/drivers_license/drivers_license_fin_053_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('drivers_license', 'The document class.')\nfields = (['full_name', 'date_of_birth', 'address', 'issue_date', 'expiration_date', 'state'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name', 'date_of_birth': 'Date of Birth', 'address': 'Address', 'issue_date': 'Issued Date', 'expiration_date': 'Valid Until Date', 'state': 'Issuing state'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"ARMSTRONG THOMAS\"}, {\"field\": \"date_of_birth\", \"value\": \"1980-10-01\"}, {\"field\": \"address\", \"value\": \"123 W. STATE STREET, BOISE, ID 83707\"}, {\"field\": \"issue_date\", \"value\": \"2016-10-01\"}, {\"field\": \"expiration_date\", \"value\": \"2021-10-01\"}, {\"field\": \"state\", \"value\": \"Idaho\"}]}"}]} {"id": "drivers_license_fin_054_p1_classify", "task": "classify", "class": "drivers_license", "doc": "drivers_license_fin_054.pdf", "images": ["images/train/drivers_license/drivers_license_fin_054_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"drivers_license\", \"confidence\": 1.0, \"issuer\": \"WASHINGTON, DC\", \"holder\": \"ANGELINA GABRIELA JONES\", \"page_marker\": \"\"}]}"}]} {"id": "drivers_license_fin_054_extract", "task": "extract", "class": "drivers_license", "doc": "drivers_license_fin_054.pdf", "images": ["images/train/drivers_license/drivers_license_fin_054_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('drivers_license', 'The document class.')\nfields = (['full_name', 'date_of_birth', 'address', 'issue_date', 'expiration_date', 'state'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name', 'date_of_birth': 'Date of Birth', 'address': 'Address', 'issue_date': 'Issued Date', 'expiration_date': 'Valid Until Date', 'state': 'Issuing state'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"ANGELINA GABRIELA JONES\"}, {\"field\": \"date_of_birth\", \"value\": \"1984-02-21\"}, {\"field\": \"address\", \"value\": \"1234 COMMODORE JOSHUA BARNEY DRIVE, NE #1234, WASHINGTON, DC 00000-0000\"}, {\"field\": \"issue_date\", \"value\": \"2010-02-17\"}, {\"field\": \"expiration_date\", \"value\": \"2018-02-21\"}, {\"field\": \"state\", \"value\": \"WASHINGTON, DC\"}]}"}]} {"id": "drivers_license_fin_055_p1_classify", "task": "classify", "class": "drivers_license", "doc": "drivers_license_fin_055.pdf", "images": ["images/train/drivers_license/drivers_license_fin_055_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"drivers_license\", \"confidence\": 1.0, \"issuer\": \"ALASKA\", \"holder\": \"SAMPLE AM D\", \"page_marker\": \"\"}]}"}]} {"id": "drivers_license_fin_055_extract", "task": "extract", "class": "drivers_license", "doc": "drivers_license_fin_055.pdf", "images": ["images/train/drivers_license/drivers_license_fin_055_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('drivers_license', 'The document class.')\nfields = (['full_name', 'date_of_birth', 'address', 'issue_date', 'expiration_date', 'state'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name', 'date_of_birth': 'Date of Birth', 'address': 'Address', 'issue_date': 'Issued Date', 'expiration_date': 'Valid Until Date', 'state': 'Issuing state'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"SAMPLE AM D\"}, {\"field\": \"date_of_birth\", \"value\": \"1980-10-03\"}, {\"field\": \"address\", \"value\": \"12345 TEST SAMPLE, JUNEAU, AK 99801\"}, {\"field\": \"issue_date\", \"value\": \"2018-08-03\"}, {\"field\": \"expiration_date\", \"value\": \"2022-03-13\"}, {\"field\": \"state\", \"value\": \"ALASKA\"}]}"}]} {"id": "drivers_license_fin_056_p1_classify", "task": "classify", "class": "drivers_license", "doc": "drivers_license_fin_056.pdf", "images": ["images/train/drivers_license/drivers_license_fin_056_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"drivers_license\", \"confidence\": 1.0, \"issuer\": \"Wyoming\", \"holder\": \"SAMPLE TINA A\", \"page_marker\": \"\"}]}"}]} {"id": "drivers_license_fin_056_extract", "task": "extract", "class": "drivers_license", "doc": "drivers_license_fin_056.pdf", "images": ["images/train/drivers_license/drivers_license_fin_056_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('drivers_license', 'The document class.')\nfields = (['full_name', 'date_of_birth', 'address', 'issue_date', 'expiration_date', 'state'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name', 'date_of_birth': 'Date of Birth', 'address': 'Address', 'issue_date': 'Issued Date', 'expiration_date': 'Valid Until Date', 'state': 'Issuing state'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"SAMPLE TINA A\"}, {\"field\": \"date_of_birth\", \"value\": \"1974-02-14\"}, {\"field\": \"address\", \"value\": \"123 ANYSTREET ANYTOWN, WY 82000\"}, {\"field\": \"issue_date\", \"value\": \"2023-11-30\"}, {\"field\": \"expiration_date\", \"value\": \"2028-11-29\"}, {\"field\": \"state\", \"value\": \"Wyoming\"}]}"}]} {"id": "drivers_license_fin_057_p1_classify", "task": "classify", "class": "drivers_license", "doc": "drivers_license_fin_057.pdf", "images": ["images/train/drivers_license/drivers_license_fin_057_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"drivers_license\", \"confidence\": 1.0, \"issuer\": \"Oklahoma\", \"holder\": \"SAMPLE EMILY\", \"page_marker\": \"\"}]}"}]} {"id": "drivers_license_fin_057_extract", "task": "extract", "class": "drivers_license", "doc": "drivers_license_fin_057.pdf", "images": ["images/train/drivers_license/drivers_license_fin_057_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('drivers_license', 'The document class.')\nfields = (['full_name', 'date_of_birth', 'address', 'issue_date', 'expiration_date', 'state'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name', 'date_of_birth': 'Date of Birth', 'address': 'Address', 'issue_date': 'Issued Date', 'expiration_date': 'Valid Until Date', 'state': 'Issuing state'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"SAMPLE EMILY\"}, {\"field\": \"date_of_birth\", \"value\": \"1951-06-12\"}, {\"field\": \"address\", \"value\": \"12 PATTERSON ROAD OKLAHOMA CITY OK 73160-0000\"}, {\"field\": \"issue_date\", \"value\": \"2011-06-18\"}, {\"field\": \"expiration_date\", \"value\": \"2015-06-18\"}, {\"field\": \"state\", \"value\": \"Oklahoma\"}]}"}]} {"id": "drivers_license_fin_060_p1_classify", "task": "classify", "class": "drivers_license", "doc": "drivers_license_fin_060.pdf", "images": ["images/train/drivers_license/drivers_license_fin_060_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"drivers_license\", \"confidence\": 1.0, \"issuer\": \"Virginia\", \"holder\": \"REYES RICHARD BENJAMIN\", \"page_marker\": \"\"}]}"}]} {"id": "drivers_license_fin_060_extract", "task": "extract", "class": "drivers_license", "doc": "drivers_license_fin_060.pdf", "images": ["images/train/drivers_license/drivers_license_fin_060_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('drivers_license', 'The document class.')\nfields = (['full_name', 'date_of_birth', 'address', 'issue_date', 'expiration_date', 'state'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name', 'date_of_birth': 'Date of Birth', 'address': 'Address', 'issue_date': 'Issued Date', 'expiration_date': 'Valid Until Date', 'state': 'Issuing state'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"REYES RICHARD BENJAMIN\"}, {\"field\": \"date_of_birth\", \"value\": \"1987-01-05\"}, {\"field\": \"address\", \"value\": \"5235 JOHN TYLER HWY, WILLIAMSBURG, VA 23185\"}, {\"field\": \"issue_date\", \"value\": \"2023-05-06\"}, {\"field\": \"expiration_date\", \"value\": \"2031-01-05\"}, {\"field\": \"state\", \"value\": \"Virginia\"}]}"}]} {"id": "drivers_license_fin_062_p1_classify", "task": "classify", "class": "drivers_license", "doc": "drivers_license_fin_062.pdf", "images": ["images/train/drivers_license/drivers_license_fin_062_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"drivers_license\", \"confidence\": 1.0, \"issuer\": \"COLORADO\", \"holder\": \"SAMPLE JOHN DRIVERLICENSE\", \"page_marker\": \"\"}]}"}]} {"id": "drivers_license_fin_062_extract", "task": "extract", "class": "drivers_license", "doc": "drivers_license_fin_062.pdf", "images": ["images/train/drivers_license/drivers_license_fin_062_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('drivers_license', 'The document class.')\nfields = (['full_name', 'date_of_birth', 'address', 'issue_date', 'expiration_date', 'state'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name', 'date_of_birth': 'Date of Birth', 'address': 'Address', 'issue_date': 'Issued Date', 'expiration_date': 'Valid Until Date', 'state': 'Issuing state'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"SAMPLE JOHN DRIVERLICENSE\"}, {\"field\": \"date_of_birth\", \"value\": \"1976-06-08\"}, {\"field\": \"address\", \"value\": \"1881 PIERCE ST LAKEWOOD, CO 80214\"}, {\"field\": \"issue_date\", \"value\": \"2020-03-04\"}, {\"field\": \"expiration_date\", \"value\": \"2023-03-04\"}, {\"field\": \"state\", \"value\": \"COLORADO\"}]}"}]} {"id": "drivers_license_fin_063_p1_classify", "task": "classify", "class": "drivers_license", "doc": "drivers_license_fin_063.pdf", "images": ["images/train/drivers_license/drivers_license_fin_063_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"drivers_license\", \"confidence\": 1.0, \"issuer\": \"Michigan\", \"holder\": \"JUNE MARIE SAMPLE\", \"page_marker\": \"\"}]}"}]} {"id": "drivers_license_fin_063_extract", "task": "extract", "class": "drivers_license", "doc": "drivers_license_fin_063.pdf", "images": ["images/train/drivers_license/drivers_license_fin_063_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('drivers_license', 'The document class.')\nfields = (['full_name', 'date_of_birth', 'address', 'issue_date', 'expiration_date', 'state'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name', 'date_of_birth': 'Date of Birth', 'address': 'Address', 'issue_date': 'Issued Date', 'expiration_date': 'Valid Until Date', 'state': 'Issuing state'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"JUNE MARIE SAMPLE\"}, {\"field\": \"date_of_birth\", \"value\": \"1984-05-01\"}, {\"field\": \"address\", \"value\": \"123 NORTH STATE ST. LANSING, MI 48918-0000\"}, {\"field\": \"issue_date\", \"value\": \"2018-04-01\"}, {\"field\": \"expiration_date\", \"value\": \"2019-10-15\"}, {\"field\": \"state\", \"value\": \"Michigan\"}]}"}]} {"id": "drivers_license_fin_064_p1_classify", "task": "classify", "class": "drivers_license", "doc": "drivers_license_fin_064.pdf", "images": ["images/train/drivers_license/drivers_license_fin_064_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"drivers_license\", \"confidence\": 1.0, \"issuer\": \"Ontario\", \"holder\": \"CHURCH BLAKE MORGAN\", \"page_marker\": \"\"}]}"}]} {"id": "drivers_license_fin_064_extract", "task": "extract", "class": "drivers_license", "doc": "drivers_license_fin_064.pdf", "images": ["images/train/drivers_license/drivers_license_fin_064_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('drivers_license', 'The document class.')\nfields = (['full_name', 'date_of_birth', 'address', 'issue_date', 'expiration_date', 'state'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name', 'date_of_birth': 'Date of Birth', 'address': 'Address', 'issue_date': 'Issued Date', 'expiration_date': 'Valid Until Date', 'state': 'Issuing state'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"CHURCH BLAKE MORGAN\"}, {\"field\": \"date_of_birth\", \"value\": \"1975-06-09\"}, {\"field\": \"address\", \"value\": \"30 ROEHAMPTON AVENUE TORONTO, ON M4P 1R2\"}, {\"field\": \"issue_date\", \"value\": \"2017-12-13\"}, {\"field\": \"expiration_date\", \"value\": \"2023-12-13\"}, {\"field\": \"state\", \"value\": \"Ontario\"}]}"}]} {"id": "drivers_license_fin_065_p1_classify", "task": "classify", "class": "drivers_license", "doc": "drivers_license_fin_065.pdf", "images": ["images/train/drivers_license/drivers_license_fin_065_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"drivers_license\", \"confidence\": 1.0, \"issuer\": \"Virginia\", \"holder\": \"MAURY JUSTIN, WILLIAM\", \"page_marker\": \"\"}]}"}]} {"id": "drivers_license_fin_065_extract", "task": "extract", "class": "drivers_license", "doc": "drivers_license_fin_065.pdf", "images": ["images/train/drivers_license/drivers_license_fin_065_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('drivers_license', 'The document class.')\nfields = (['full_name', 'date_of_birth', 'address', 'issue_date', 'expiration_date', 'state'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name', 'date_of_birth': 'Date of Birth', 'address': 'Address', 'issue_date': 'Issued Date', 'expiration_date': 'Valid Until Date', 'state': 'Issuing state'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"MAURY JUSTIN, WILLIAM\"}, {\"field\": \"date_of_birth\", \"value\": \"1958-07-15\"}, {\"field\": \"address\", \"value\": \"17 FIRST STREET STAUNTON, VA 24401\"}, {\"field\": \"issue_date\", \"value\": \"2009-08-14\"}, {\"field\": \"expiration_date\", \"value\": \"2016-07-15\"}, {\"field\": \"state\", \"value\": \"Virginia\"}]}"}]} {"id": "drivers_license_fin_066_p1_classify", "task": "classify", "class": "drivers_license", "doc": "drivers_license_fin_066.pdf", "images": ["images/train/drivers_license/drivers_license_fin_066_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"drivers_license\", \"confidence\": 1.0, \"issuer\": \"ARKANSAS\", \"holder\": \"NICK SAMPLE\", \"page_marker\": \"\"}]}"}]} {"id": "drivers_license_fin_066_extract", "task": "extract", "class": "drivers_license", "doc": "drivers_license_fin_066.pdf", "images": ["images/train/drivers_license/drivers_license_fin_066_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('drivers_license', 'The document class.')\nfields = (['full_name', 'date_of_birth', 'address', 'issue_date', 'expiration_date', 'state'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name', 'date_of_birth': 'Date of Birth', 'address': 'Address', 'issue_date': 'Issued Date', 'expiration_date': 'Valid Until Date', 'state': 'Issuing state'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"NICK SAMPLE\"}, {\"field\": \"date_of_birth\", \"value\": \"1960-03-05\"}, {\"field\": \"address\", \"value\": \"123 NORTH STREET, CITY, AR 12345\"}, {\"field\": \"issue_date\", \"value\": \"2018-03-05\"}, {\"field\": \"expiration_date\", \"value\": \"2026-03-05\"}, {\"field\": \"state\", \"value\": \"ARKANSAS\"}]}"}]} {"id": "drivers_license_fin_067_p1_classify", "task": "classify", "class": "drivers_license", "doc": "drivers_license_fin_067.pdf", "images": ["images/train/drivers_license/drivers_license_fin_067_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"drivers_license\", \"confidence\": 1.0, \"issuer\": \"Virginia\", \"holder\": \"BAILEY WESTON COLE\", \"page_marker\": \"\"}]}"}]} {"id": "drivers_license_fin_067_extract", "task": "extract", "class": "drivers_license", "doc": "drivers_license_fin_067.pdf", "images": ["images/train/drivers_license/drivers_license_fin_067_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('drivers_license', 'The document class.')\nfields = (['full_name', 'date_of_birth', 'address', 'issue_date', 'expiration_date', 'state'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name', 'date_of_birth': 'Date of Birth', 'address': 'Address', 'issue_date': 'Issued Date', 'expiration_date': 'Valid Until Date', 'state': 'Issuing state'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"BAILEY WESTON COLE\"}, {\"field\": \"date_of_birth\", \"value\": \"1965-05-01\"}, {\"field\": \"address\", \"value\": \"4058 VALLEY PIKE WINCHESTER, VA 22602\"}, {\"field\": \"issue_date\", \"value\": \"2023-05-21\"}, {\"field\": \"expiration_date\", \"value\": \"2031-05-01\"}, {\"field\": \"state\", \"value\": \"Virginia\"}]}"}]} {"id": "drivers_license_fin_069_p1_classify", "task": "classify", "class": "drivers_license", "doc": "drivers_license_fin_069.pdf", "images": ["images/train/drivers_license/drivers_license_fin_069_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"drivers_license\", \"confidence\": 1.0, \"issuer\": \"REPUBLIC OF TRINIDAD AND TOBAGO\", \"holder\": \"JANET ANNABETH KYLEIGH EDWARDS\", \"page_marker\": \"\"}]}"}]} {"id": "drivers_license_fin_069_extract", "task": "extract", "class": "drivers_license", "doc": "drivers_license_fin_069.pdf", "images": ["images/train/drivers_license/drivers_license_fin_069_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('drivers_license', 'The document class.')\nfields = (['full_name', 'date_of_birth', 'address', 'issue_date', 'expiration_date', 'state'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name', 'date_of_birth': 'Date of Birth', 'address': 'Address', 'issue_date': 'Issued Date', 'expiration_date': 'Valid Until Date', 'state': 'Issuing state'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"JANET ANNABETH KYLEIGH EDWARDS\"}, {\"field\": \"date_of_birth\", \"value\": \"1990-11-30\"}, {\"field\": \"address\", \"value\": \"100 CHARMING AVENUE, SOUTH EASTERN MAIN ROAD, PLEASANTVILLE SAN FERNANDO\"}, {\"field\": \"issue_date\", \"value\": \"2008-03-12\"}, {\"field\": \"expiration_date\", \"value\": \"2023-06-10\"}, {\"field\": \"state\", \"value\": \"REPUBLIC OF TRINIDAD AND TOBAGO\"}]}"}]} {"id": "drivers_license_fin_072_p1_classify", "task": "classify", "class": "drivers_license", "doc": "drivers_license_fin_072.pdf", "images": ["images/train/drivers_license/drivers_license_fin_072_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"drivers_license\", \"confidence\": 1.0, \"issuer\": \"California\", \"holder\": \"JUDY JANE DOE\", \"page_marker\": \"\"}]}"}]} {"id": "drivers_license_fin_072_extract", "task": "extract", "class": "drivers_license", "doc": "drivers_license_fin_072.pdf", "images": ["images/train/drivers_license/drivers_license_fin_072_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('drivers_license', 'The document class.')\nfields = (['full_name', 'date_of_birth', 'address', 'issue_date', 'expiration_date', 'state'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name', 'date_of_birth': 'Date of Birth', 'address': 'Address', 'issue_date': 'Issued Date', 'expiration_date': 'Valid Until Date', 'state': 'Issuing state'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"JUDY JANE DOE\"}, {\"field\": \"date_of_birth\", \"value\": \"1997-01-01\"}, {\"field\": \"address\", \"value\": \"2244 2ND AVE, SAN DIEGO, CA 92101\"}, {\"field\": \"issue_date\", \"value\": \"2015-10-25\"}, {\"field\": \"expiration_date\", \"value\": \"2020-01-01\"}, {\"field\": \"state\", \"value\": \"California\"}]}"}]} {"id": "drivers_license_fin_073_p1_classify", "task": "classify", "class": "drivers_license", "doc": "drivers_license_fin_073.pdf", "images": ["images/train/drivers_license/drivers_license_fin_073_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"drivers_license\", \"confidence\": 1.0, \"issuer\": \"Washington, DC\", \"holder\": \"SAMPLECARD BEVERLY SPRINGS\", \"page_marker\": \"\"}]}"}]} {"id": "drivers_license_fin_073_extract", "task": "extract", "class": "drivers_license", "doc": "drivers_license_fin_073.pdf", "images": ["images/train/drivers_license/drivers_license_fin_073_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('drivers_license', 'The document class.')\nfields = (['full_name', 'date_of_birth', 'address', 'issue_date', 'expiration_date', 'state'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name', 'date_of_birth': 'Date of Birth', 'address': 'Address', 'issue_date': 'Issued Date', 'expiration_date': 'Valid Until Date', 'state': 'Issuing state'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"SAMPLECARD BEVERLY SPRINGS\"}, {\"field\": \"date_of_birth\", \"value\": \"1984-03-23\"}, {\"field\": \"address\", \"value\": \"789 PATRICIA ROBERTS HARRIS PL NE WASHINGTON, DC 20001-0000\"}, {\"field\": \"issue_date\", \"value\": \"2023-06-19\"}, {\"field\": \"expiration_date\", \"value\": \"2031-03-23\"}, {\"field\": \"state\", \"value\": \"Washington, DC\"}]}"}]} {"id": "drivers_license_fin_074_p1_classify", "task": "classify", "class": "drivers_license", "doc": "drivers_license_fin_074.pdf", "images": ["images/train/drivers_license/drivers_license_fin_074_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"drivers_license\", \"confidence\": 1.0, \"issuer\": \"Ontario\", \"holder\": \"DOE JOHN\", \"page_marker\": \"\"}]}"}]} {"id": "drivers_license_fin_074_extract", "task": "extract", "class": "drivers_license", "doc": "drivers_license_fin_074.pdf", "images": ["images/train/drivers_license/drivers_license_fin_074_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('drivers_license', 'The document class.')\nfields = (['full_name', 'date_of_birth', 'address', 'issue_date', 'expiration_date', 'state'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name', 'date_of_birth': 'Date of Birth', 'address': 'Address', 'issue_date': 'Issued Date', 'expiration_date': 'Valid Until Date', 'state': 'Issuing state'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"DOE JOHN\"}, {\"field\": \"date_of_birth\", \"value\": \"1966-09-05\"}, {\"field\": \"address\", \"value\": \"123 ANY STREET, TORONTO, ON, M0M 0M0\"}, {\"field\": \"issue_date\", \"value\": \"2009-04-23\"}, {\"field\": \"expiration_date\", \"value\": \"2014-04-23\"}, {\"field\": \"state\", \"value\": \"Ontario\"}]}"}]} {"id": "drivers_license_fin_075_p1_classify", "task": "classify", "class": "drivers_license", "doc": "drivers_license_fin_075.pdf", "images": ["images/train/drivers_license/drivers_license_fin_075_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"drivers_license\", \"confidence\": 1.0, \"issuer\": \"Arizona\", \"holder\": \"SAMPLE DRIVER\", \"page_marker\": \"\"}]}"}]} {"id": "drivers_license_fin_075_extract", "task": "extract", "class": "drivers_license", "doc": "drivers_license_fin_075.pdf", "images": ["images/train/drivers_license/drivers_license_fin_075_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('drivers_license', 'The document class.')\nfields = (['full_name', 'date_of_birth', 'address', 'issue_date', 'expiration_date', 'state'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name', 'date_of_birth': 'Date of Birth', 'address': 'Address', 'issue_date': 'Issued Date', 'expiration_date': 'Valid Until Date', 'state': 'Issuing state'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"SAMPLE DRIVER\"}, {\"field\": \"date_of_birth\", \"value\": \"1974-02-14\"}, {\"field\": \"address\", \"value\": \"1234 EXAMPLE ST, PHOENIX, AZ 85007\"}, {\"field\": \"issue_date\", \"value\": \"2023-02-01\"}, {\"field\": \"expiration_date\", \"value\": \"2028-02-01\"}, {\"field\": \"state\", \"value\": \"Arizona\"}]}"}]} {"id": "drivers_license_fin_076_p1_classify", "task": "classify", "class": "drivers_license", "doc": "drivers_license_fin_076.pdf", "images": ["images/train/drivers_license/drivers_license_fin_076_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"drivers_license\", \"confidence\": 1.0, \"issuer\": \"Texas\", \"holder\": \"ANDR\\u00c9 REN\\u00c9E A\", \"page_marker\": \"\"}]}"}]} {"id": "drivers_license_fin_076_extract", "task": "extract", "class": "drivers_license", "doc": "drivers_license_fin_076.pdf", "images": ["images/train/drivers_license/drivers_license_fin_076_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('drivers_license', 'The document class.')\nfields = (['full_name', 'date_of_birth', 'address', 'issue_date', 'expiration_date', 'state'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name', 'date_of_birth': 'Date of Birth', 'address': 'Address', 'issue_date': 'Issued Date', 'expiration_date': 'Valid Until Date', 'state': 'Issuing state'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"ANDR\\u00c9 REN\\u00c9E A\"}, {\"field\": \"date_of_birth\", \"value\": \"1959-12-29\"}, {\"field\": \"address\", \"value\": \"3300 ACORN STREET AUSTIN, TX 12345-0000\"}, {\"field\": \"issue_date\", \"value\": \"2019-12-10\"}, {\"field\": \"expiration_date\", \"value\": \"2024-12-20\"}, {\"field\": \"state\", \"value\": \"Texas\"}]}"}]} {"id": "drivers_license_fin_077_p1_classify", "task": "classify", "class": "drivers_license", "doc": "drivers_license_fin_077.pdf", "images": ["images/train/drivers_license/drivers_license_fin_077_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"drivers_license\", \"confidence\": 1.0, \"issuer\": \"Pennsylvania\", \"holder\": \"SAMPLE JANICE ANN\", \"page_marker\": \"\"}]}"}]} {"id": "drivers_license_fin_077_extract", "task": "extract", "class": "drivers_license", "doc": "drivers_license_fin_077.pdf", "images": ["images/train/drivers_license/drivers_license_fin_077_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('drivers_license', 'The document class.')\nfields = (['full_name', 'date_of_birth', 'address', 'issue_date', 'expiration_date', 'state'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name', 'date_of_birth': 'Date of Birth', 'address': 'Address', 'issue_date': 'Issued Date', 'expiration_date': 'Valid Until Date', 'state': 'Issuing state'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"SAMPLE JANICE ANN\"}, {\"field\": \"date_of_birth\", \"value\": \"1969-08-04\"}, {\"field\": \"address\", \"value\": \"123 MAIN STREET APT. 1 HARRISBURG, PA 17101-0000\"}, {\"field\": \"issue_date\", \"value\": \"2016-10-05\"}, {\"field\": \"expiration_date\", \"value\": \"2020-08-05\"}, {\"field\": \"state\", \"value\": \"Pennsylvania\"}]}"}]} {"id": "drivers_license_fin_078_p1_classify", "task": "classify", "class": "drivers_license", "doc": "drivers_license_fin_078.pdf", "images": ["images/train/drivers_license/drivers_license_fin_078_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"drivers_license\", \"confidence\": 1.0, \"issuer\": \"New York\", \"holder\": \"MOTORIST MICHAEL, MATTHEW\", \"page_marker\": \"\"}]}"}]} {"id": "drivers_license_fin_078_extract", "task": "extract", "class": "drivers_license", "doc": "drivers_license_fin_078.pdf", "images": ["images/train/drivers_license/drivers_license_fin_078_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('drivers_license', 'The document class.')\nfields = (['full_name', 'date_of_birth', 'address', 'issue_date', 'expiration_date', 'state'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name', 'date_of_birth': 'Date of Birth', 'address': 'Address', 'issue_date': 'Issued Date', 'expiration_date': 'Valid Until Date', 'state': 'Issuing state'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"MOTORIST MICHAEL, MATTHEW\"}, {\"field\": \"date_of_birth\", \"value\": \"1982-08-31\"}, {\"field\": \"address\", \"value\": \"2345 ANYWHERE STREET, YOUR CITY, NY 12345\"}, {\"field\": \"issue_date\", \"value\": \"2017-08-31\"}, {\"field\": \"expiration_date\", \"value\": \"2022-08-31\"}, {\"field\": \"state\", \"value\": \"New York\"}]}"}]} {"id": "drivers_license_fin_079_p1_classify", "task": "classify", "class": "drivers_license", "doc": "drivers_license_fin_079.pdf", "images": ["images/train/drivers_license/drivers_license_fin_079_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"drivers_license\", \"confidence\": 1.0, \"issuer\": \"New York\", \"holder\": \"JE'NAI ASHLEE MARCELLA MOULON\", \"page_marker\": \"\"}]}"}]} {"id": "drivers_license_fin_079_extract", "task": "extract", "class": "drivers_license", "doc": "drivers_license_fin_079.pdf", "images": ["images/train/drivers_license/drivers_license_fin_079_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('drivers_license', 'The document class.')\nfields = (['full_name', 'date_of_birth', 'address', 'issue_date', 'expiration_date', 'state'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name', 'date_of_birth': 'Date of Birth', 'address': 'Address', 'issue_date': 'Issued Date', 'expiration_date': 'Valid Until Date', 'state': 'Issuing state'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"JE'NAI ASHLEE MARCELLA MOULON\"}, {\"field\": \"date_of_birth\", \"value\": \"1995-09-21\"}, {\"field\": \"address\", \"value\": \"245 VERRAZANO AVE, COPIAGUE NY 11726\"}, {\"field\": \"issue_date\", \"value\": \"2013-09-03\"}, {\"field\": \"expiration_date\", \"value\": \"2021-09-03\"}, {\"field\": \"state\", \"value\": \"New York\"}]}"}]} {"id": "drivers_license_fin_080_p1_classify", "task": "classify", "class": "drivers_license", "doc": "drivers_license_fin_080.pdf", "images": ["images/train/drivers_license/drivers_license_fin_080_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"drivers_license\", \"confidence\": 1.0, \"issuer\": \"New Hampshire\", \"holder\": \"HOLLY SAMPLE\", \"page_marker\": \"\"}]}"}]} {"id": "drivers_license_fin_080_extract", "task": "extract", "class": "drivers_license", "doc": "drivers_license_fin_080.pdf", "images": ["images/train/drivers_license/drivers_license_fin_080_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('drivers_license', 'The document class.')\nfields = (['full_name', 'date_of_birth', 'address', 'issue_date', 'expiration_date', 'state'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name', 'date_of_birth': 'Date of Birth', 'address': 'Address', 'issue_date': 'Issued Date', 'expiration_date': 'Valid Until Date', 'state': 'Issuing state'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"HOLLY SAMPLE\"}, {\"field\": \"date_of_birth\", \"value\": \"1996-01-02\"}, {\"field\": \"address\", \"value\": \"23 HAZEN DR, CONCORD, NH 03305\"}, {\"field\": \"issue_date\", \"value\": \"2017-01-11\"}, {\"field\": \"expiration_date\", \"value\": \"2022-01-02\"}, {\"field\": \"state\", \"value\": \"New Hampshire\"}]}"}]} {"id": "drivers_license_fin_081_p1_classify", "task": "classify", "class": "drivers_license", "doc": "drivers_license_fin_081.pdf", "images": ["images/train/drivers_license/drivers_license_fin_081_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"drivers_license\", \"confidence\": 1.0, \"issuer\": \"New York\", \"holder\": \"MOTORIST MICHELLE, MARIE\", \"page_marker\": \"\"}]}"}]} {"id": "drivers_license_fin_081_extract", "task": "extract", "class": "drivers_license", "doc": "drivers_license_fin_081.pdf", "images": ["images/train/drivers_license/drivers_license_fin_081_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('drivers_license', 'The document class.')\nfields = (['full_name', 'date_of_birth', 'address', 'issue_date', 'expiration_date', 'state'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name', 'date_of_birth': 'Date of Birth', 'address': 'Address', 'issue_date': 'Issued Date', 'expiration_date': 'Valid Until Date', 'state': 'Issuing state'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"MOTORIST MICHELLE, MARIE\"}, {\"field\": \"date_of_birth\", \"value\": \"1990-10-31\"}, {\"field\": \"address\", \"value\": \"2345 ANYWHERE STREET, ALBANY, NY 12222\"}, {\"field\": \"issue_date\", \"value\": \"2022-03-07\"}, {\"field\": \"expiration_date\", \"value\": \"2029-10-31\"}, {\"field\": \"state\", \"value\": \"New York\"}]}"}]} {"id": "drivers_license_fin_083_p1_classify", "task": "classify", "class": "drivers_license", "doc": "drivers_license_fin_083.pdf", "images": ["images/train/drivers_license/drivers_license_fin_083_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"drivers_license\", \"confidence\": 1.0, \"issuer\": \"California\", \"holder\": \"NOWAK ANNA\", \"page_marker\": \"\"}]}"}]} {"id": "drivers_license_fin_083_extract", "task": "extract", "class": "drivers_license", "doc": "drivers_license_fin_083.pdf", "images": ["images/train/drivers_license/drivers_license_fin_083_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('drivers_license', 'The document class.')\nfields = (['full_name', 'date_of_birth', 'address', 'issue_date', 'expiration_date', 'state'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name', 'date_of_birth': 'Date of Birth', 'address': 'Address', 'issue_date': 'Issued Date', 'expiration_date': 'Valid Until Date', 'state': 'Issuing state'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"NOWAK ANNA\"}, {\"field\": \"date_of_birth\", \"value\": \"1977-08-31\"}, {\"field\": \"address\", \"value\": \"2570 24TH STREET, SACRAMENTO, CA 95818\"}, {\"field\": \"issue_date\", \"value\": \"2010-09-30\"}, {\"field\": \"expiration_date\", \"value\": \"2015-08-31\"}, {\"field\": \"state\", \"value\": \"California\"}]}"}]} {"id": "drivers_license_fin_084_p1_classify", "task": "classify", "class": "drivers_license", "doc": "drivers_license_fin_084.pdf", "images": ["images/train/drivers_license/drivers_license_fin_084_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"drivers_license\", \"confidence\": 1.0, \"issuer\": \"CALIFORNIA\", \"holder\": \"MARTIN MCFLY\", \"page_marker\": \"\"}]}"}]} {"id": "drivers_license_fin_084_extract", "task": "extract", "class": "drivers_license", "doc": "drivers_license_fin_084.pdf", "images": ["images/train/drivers_license/drivers_license_fin_084_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('drivers_license', 'The document class.')\nfields = (['full_name', 'date_of_birth', 'address', 'issue_date', 'expiration_date', 'state'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name', 'date_of_birth': 'Date of Birth', 'address': 'Address', 'issue_date': 'Issued Date', 'expiration_date': 'Valid Until Date', 'state': 'Issuing state'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"MARTIN MCFLY\"}, {\"field\": \"date_of_birth\", \"value\": \"1968-06-09\"}, {\"field\": \"address\", \"value\": \"9303 LYON DRIVE, HILL VALLEY, CA 95420\"}, {\"field\": \"issue_date\", \"value\": \"\"}, {\"field\": \"expiration_date\", \"value\": \"1987-06-09\"}, {\"field\": \"state\", \"value\": \"CALIFORNIA\"}]}"}]} {"id": "drivers_license_fin_085_p1_classify", "task": "classify", "class": "drivers_license", "doc": "drivers_license_fin_085.pdf", "images": ["images/train/drivers_license/drivers_license_fin_085_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"drivers_license\", \"confidence\": 1.0, \"issuer\": \"\", \"holder\": \"SARA ADAMS\", \"page_marker\": \"\"}]}"}]} {"id": "drivers_license_fin_085_extract", "task": "extract", "class": "drivers_license", "doc": "drivers_license_fin_085.pdf", "images": ["images/train/drivers_license/drivers_license_fin_085_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('drivers_license', 'The document class.')\nfields = (['full_name', 'date_of_birth', 'address', 'issue_date', 'expiration_date', 'state'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name', 'date_of_birth': 'Date of Birth', 'address': 'Address', 'issue_date': 'Issued Date', 'expiration_date': 'Valid Until Date', 'state': 'Issuing state'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"SARA ADAMS\"}, {\"field\": \"date_of_birth\", \"value\": \"1984-01-02\"}, {\"field\": \"address\", \"value\": \"\"}, {\"field\": \"issue_date\", \"value\": \"2014-02-21\"}, {\"field\": \"expiration_date\", \"value\": \"2024-02-21\"}, {\"field\": \"state\", \"value\": \"\"}]}"}]} {"id": "drivers_license_fin_086_p1_classify", "task": "classify", "class": "drivers_license", "doc": "drivers_license_fin_086.pdf", "images": ["images/train/drivers_license/drivers_license_fin_086_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"drivers_license\", \"confidence\": 1.0, \"issuer\": \"South Carolina\", \"holder\": \"SHWEDO KEVIN ANTHONY\", \"page_marker\": \"\"}]}"}]} {"id": "drivers_license_fin_086_extract", "task": "extract", "class": "drivers_license", "doc": "drivers_license_fin_086.pdf", "images": ["images/train/drivers_license/drivers_license_fin_086_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('drivers_license', 'The document class.')\nfields = (['full_name', 'date_of_birth', 'address', 'issue_date', 'expiration_date', 'state'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name', 'date_of_birth': 'Date of Birth', 'address': 'Address', 'issue_date': 'Issued Date', 'expiration_date': 'Valid Until Date', 'state': 'Issuing state'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"SHWEDO KEVIN ANTHONY\"}, {\"field\": \"date_of_birth\", \"value\": \"1956-07-04\"}, {\"field\": \"address\", \"value\": \"10311 WILSON BLVD COLUMBIA, SC 290161234\"}, {\"field\": \"issue_date\", \"value\": \"2017-08-28\"}, {\"field\": \"expiration_date\", \"value\": \"2025-07-04\"}, {\"field\": \"state\", \"value\": \"South Carolina\"}]}"}]} {"id": "drivers_license_fin_087_p1_classify", "task": "classify", "class": "drivers_license", "doc": "drivers_license_fin_087.pdf", "images": ["images/train/drivers_license/drivers_license_fin_087_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"drivers_license\", \"confidence\": 1.0, \"issuer\": \"South Dakota\", \"holder\": \"LASTNAME FIRSTNAME MIDDLENAME, JR\", \"page_marker\": \"\"}]}"}]} {"id": "drivers_license_fin_087_extract", "task": "extract", "class": "drivers_license", "doc": "drivers_license_fin_087.pdf", "images": ["images/train/drivers_license/drivers_license_fin_087_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('drivers_license', 'The document class.')\nfields = (['full_name', 'date_of_birth', 'address', 'issue_date', 'expiration_date', 'state'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name', 'date_of_birth': 'Date of Birth', 'address': 'Address', 'issue_date': 'Issued Date', 'expiration_date': 'Valid Until Date', 'state': 'Issuing state'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"LASTNAME FIRSTNAME MIDDLENAME, JR\"}, {\"field\": \"date_of_birth\", \"value\": \"1967-01-12\"}, {\"field\": \"address\", \"value\": \"445 EAST CAPITOL AVENUE PIERRE, SD 57501-0000\"}, {\"field\": \"issue_date\", \"value\": \"2017-04-21\"}, {\"field\": \"expiration_date\", \"value\": \"2022-01-12\"}, {\"field\": \"state\", \"value\": \"South Dakota\"}]}"}]} {"id": "drivers_license_fin_088_p1_classify", "task": "classify", "class": "drivers_license", "doc": "drivers_license_fin_088.pdf", "images": ["images/train/drivers_license/drivers_license_fin_088_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"drivers_license\", \"confidence\": 1.0, \"issuer\": \"Virgin Islands of the United States\", \"holder\": \"SAMPLE RECORD M\", \"page_marker\": \"\"}]}"}]} {"id": "drivers_license_fin_088_extract", "task": "extract", "class": "drivers_license", "doc": "drivers_license_fin_088.pdf", "images": ["images/train/drivers_license/drivers_license_fin_088_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('drivers_license', 'The document class.')\nfields = (['full_name', 'date_of_birth', 'address', 'issue_date', 'expiration_date', 'state'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name', 'date_of_birth': 'Date of Birth', 'address': 'Address', 'issue_date': 'Issued Date', 'expiration_date': 'Valid Until Date', 'state': 'Issuing state'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"SAMPLE RECORD M\"}, {\"field\": \"date_of_birth\", \"value\": \"1980-04-15\"}, {\"field\": \"address\", \"value\": \"10065 INDUSTRIAL PARK ST. CROIX VI 00840\"}, {\"field\": \"issue_date\", \"value\": \"2024-09-04\"}, {\"field\": \"expiration_date\", \"value\": \"2029-04-15\"}, {\"field\": \"state\", \"value\": \"Virgin Islands of the United States\"}]}"}]} {"id": "drivers_license_fin_089_p1_classify", "task": "classify", "class": "drivers_license", "doc": "drivers_license_fin_089.pdf", "images": ["images/train/drivers_license/drivers_license_fin_089_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"drivers_license\", \"confidence\": 1.0, \"issuer\": \"Virginia\", \"holder\": \"JAMES JR TYLER, ANDREW\", \"page_marker\": \"\"}]}"}]} {"id": "drivers_license_fin_089_extract", "task": "extract", "class": "drivers_license", "doc": "drivers_license_fin_089.pdf", "images": ["images/train/drivers_license/drivers_license_fin_089_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('drivers_license', 'The document class.')\nfields = (['full_name', 'date_of_birth', 'address', 'issue_date', 'expiration_date', 'state'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name', 'date_of_birth': 'Date of Birth', 'address': 'Address', 'issue_date': 'Issued Date', 'expiration_date': 'Valid Until Date', 'state': 'Issuing state'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"JAMES JR TYLER, ANDREW\"}, {\"field\": \"date_of_birth\", \"value\": \"1992-08-17\"}, {\"field\": \"address\", \"value\": \"103 COMMONWEALTH BLVD. EMPORIA, VA 23847\"}, {\"field\": \"issue_date\", \"value\": \"2017-10-27\"}, {\"field\": \"expiration_date\", \"value\": \"2025-10-27\"}, {\"field\": \"state\", \"value\": \"Virginia\"}]}"}]} {"id": "drivers_license_fin_090_p1_classify", "task": "classify", "class": "drivers_license", "doc": "drivers_license_fin_090.pdf", "images": ["images/train/drivers_license/drivers_license_fin_090_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"drivers_license\", \"confidence\": 1.0, \"issuer\": \"CALIFORNIA\", \"holder\": \"WOOD RALPH B\", \"page_marker\": \"\"}]}"}]} {"id": "drivers_license_fin_090_extract", "task": "extract", "class": "drivers_license", "doc": "drivers_license_fin_090.pdf", "images": ["images/train/drivers_license/drivers_license_fin_090_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('drivers_license', 'The document class.')\nfields = (['full_name', 'date_of_birth', 'address', 'issue_date', 'expiration_date', 'state'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name', 'date_of_birth': 'Date of Birth', 'address': 'Address', 'issue_date': 'Issued Date', 'expiration_date': 'Valid Until Date', 'state': 'Issuing state'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"WOOD RALPH B\"}, {\"field\": \"date_of_birth\", \"value\": \"1940-07-15\"}, {\"field\": \"address\", \"value\": \"2142 LAS TRAMPAS ROAD, ALAMO, CA 94507\"}, {\"field\": \"issue_date\", \"value\": \"2016-10-03\"}, {\"field\": \"expiration_date\", \"value\": \"2021-03-13\"}, {\"field\": \"state\", \"value\": \"CALIFORNIA\"}]}"}]} {"id": "monthly_bill_fin_001_p1_classify", "task": "classify", "class": "monthly_bill", "doc": "monthly_bill_fin_001.pdf", "images": ["images/train/monthly_bill/monthly_bill_fin_001_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"monthly_bill\", \"confidence\": 1.0, \"issuer\": \"Columbia Gas of Ohio\", \"holder\": \"John Q Public\", \"page_marker\": \"\"}]}"}]} {"id": "monthly_bill_fin_001_extract", "task": "extract", "class": "monthly_bill", "doc": "monthly_bill_fin_001.pdf", "images": ["images/train/monthly_bill/monthly_bill_fin_001_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('monthly_bill', 'The document class.')\nfields = (['full_name', 'address', 'issue_date', 'company_name', 'account_number'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name on the bill', 'address': 'Address', 'issue_date': 'Issued date', 'company_name': 'Name of the issuing company', 'account_number': 'Account number'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"John Q Public\"}, {\"field\": \"address\", \"value\": \"123 Main Street, Avon, OH 44011\"}, {\"field\": \"issue_date\", \"value\": \"2013-06-24\"}, {\"field\": \"company_name\", \"value\": \"Columbia Gas of Ohio\"}, {\"field\": \"account_number\", \"value\": \"0000000 000 000 0\"}]}"}]} {"id": "monthly_bill_fin_002_p1_classify", "task": "classify", "class": "monthly_bill", "doc": "monthly_bill_fin_002.pdf", "images": ["images/train/monthly_bill/monthly_bill_fin_002_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"monthly_bill\", \"confidence\": 1.0, \"issuer\": \"EnergyAustralia Pty Ltd\", \"holder\": \"Miss S Sample\", \"page_marker\": \"\"}]}"}]} {"id": "monthly_bill_fin_002_extract", "task": "extract", "class": "monthly_bill", "doc": "monthly_bill_fin_002.pdf", "images": ["images/train/monthly_bill/monthly_bill_fin_002_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('monthly_bill', 'The document class.')\nfields = (['full_name', 'address', 'issue_date', 'company_name', 'account_number'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name on the bill', 'address': 'Address', 'issue_date': 'Issued date', 'company_name': 'Name of the issuing company', 'account_number': 'Account number'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"Miss S Sample\"}, {\"field\": \"address\", \"value\": \"37 Sample Road, SAMPLE SUBURB VIC 3000\"}, {\"field\": \"issue_date\", \"value\": \"2020-08-04\"}, {\"field\": \"company_name\", \"value\": \"EnergyAustralia Pty Ltd\"}, {\"field\": \"account_number\", \"value\": \"1234 567 890\"}]}"}]} {"id": "monthly_bill_fin_003_p1_classify", "task": "classify", "class": "monthly_bill", "doc": "monthly_bill_fin_003.pdf", "images": ["images/train/monthly_bill/monthly_bill_fin_003_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"monthly_bill\", \"confidence\": 1.0, \"issuer\": \"Dialog Axiata PLC (PQ 38)\", \"holder\": \"\", \"page_marker\": \"\"}]}"}]} {"id": "monthly_bill_fin_003_extract", "task": "extract", "class": "monthly_bill", "doc": "monthly_bill_fin_003.pdf", "images": ["images/train/monthly_bill/monthly_bill_fin_003_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('monthly_bill', 'The document class.')\nfields = (['full_name', 'address', 'issue_date', 'company_name', 'account_number'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name on the bill', 'address': 'Address', 'issue_date': 'Issued date', 'company_name': 'Name of the issuing company', 'account_number': 'Account number'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"\"}, {\"field\": \"address\", \"value\": \"MILINI ENTERPRISES,No 2/1/160,Dambuwa Road Thewatta, Ragama 11010\"}, {\"field\": \"issue_date\", \"value\": \"2023-01-05\"}, {\"field\": \"company_name\", \"value\": \"Dialog Axiata PLC (PQ 38)\"}, {\"field\": \"account_number\", \"value\": \"777594691\"}]}"}]} {"id": "monthly_bill_fin_004_p1_classify", "task": "classify", "class": "monthly_bill", "doc": "monthly_bill_fin_004.pdf", "images": ["images/train/monthly_bill/monthly_bill_fin_004_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"monthly_bill\", \"confidence\": 1.0, \"issuer\": \"AGL Sales Pty Limited\", \"holder\": \"Mrs Samantha Sample\", \"page_marker\": \"\"}]}"}]} {"id": "monthly_bill_fin_004_extract", "task": "extract", "class": "monthly_bill", "doc": "monthly_bill_fin_004.pdf", "images": ["images/train/monthly_bill/monthly_bill_fin_004_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('monthly_bill', 'The document class.')\nfields = (['full_name', 'address', 'issue_date', 'company_name', 'account_number'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name on the bill', 'address': 'Address', 'issue_date': 'Issued date', 'company_name': 'Name of the issuing company', 'account_number': 'Account number'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"Mrs Samantha Sample\"}, {\"field\": \"address\", \"value\": \"1 Sample Street, SAMPLETOWN NSW 2222\"}, {\"field\": \"issue_date\", \"value\": \"2017-05-31\"}, {\"field\": \"company_name\", \"value\": \"AGL Sales Pty Limited\"}, {\"field\": \"account_number\", \"value\": \"0000 0000\"}]}"}]} {"id": "monthly_bill_fin_005_p1_classify", "task": "classify", "class": "monthly_bill", "doc": "monthly_bill_fin_005.pdf", "images": ["images/train/monthly_bill/monthly_bill_fin_005_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"monthly_bill\", \"confidence\": 1.0, \"issuer\": \"MERALCO\", \"holder\": \"BUSKOWITZ, HENRY\", \"page_marker\": \"\"}]}"}]} {"id": "monthly_bill_fin_005_extract", "task": "extract", "class": "monthly_bill", "doc": "monthly_bill_fin_005.pdf", "images": ["images/train/monthly_bill/monthly_bill_fin_005_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('monthly_bill', 'The document class.')\nfields = (['full_name', 'address', 'issue_date', 'company_name', 'account_number'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name on the bill', 'address': 'Address', 'issue_date': 'Issued date', 'company_name': 'Name of the issuing company', 'account_number': 'Account number'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"BUSKOWITZ, HENRY\"}, {\"field\": \"address\", \"value\": \"2 J.P Laurel ESTACION ST AYALA AVE., MAKATI CITY METRO MANILA\"}, {\"field\": \"issue_date\", \"value\": \"2023-01-22\"}, {\"field\": \"company_name\", \"value\": \"MERALCO\"}, {\"field\": \"account_number\", \"value\": \"1812942348\"}]}"}]} {"id": "monthly_bill_fin_006_p1_classify", "task": "classify", "class": "monthly_bill", "doc": "monthly_bill_fin_006.pdf", "images": ["images/train/monthly_bill/monthly_bill_fin_006_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"monthly_bill\", \"confidence\": 1.0, \"issuer\": \"Efficient Energy\", \"holder\": \"Mrs P Smith\", \"page_marker\": \"\"}]}"}]} {"id": "monthly_bill_fin_006_extract", "task": "extract", "class": "monthly_bill", "doc": "monthly_bill_fin_006.pdf", "images": ["images/train/monthly_bill/monthly_bill_fin_006_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('monthly_bill', 'The document class.')\nfields = (['full_name', 'address', 'issue_date', 'company_name', 'account_number'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name on the bill', 'address': 'Address', 'issue_date': 'Issued date', 'company_name': 'Name of the issuing company', 'account_number': 'Account number'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"Mrs P Smith\"}, {\"field\": \"address\", \"value\": \"1 Electric Avenue, Brightown, BR0 0AA\"}, {\"field\": \"issue_date\", \"value\": \"2020-11-20\"}, {\"field\": \"company_name\", \"value\": \"Efficient Energy\"}, {\"field\": \"account_number\", \"value\": \"123 123 1234 56\"}]}"}]} {"id": "monthly_bill_fin_007_p1_classify", "task": "classify", "class": "monthly_bill", "doc": "monthly_bill_fin_007.pdf", "images": ["images/train/monthly_bill/monthly_bill_fin_007_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"monthly_bill\", \"confidence\": 1.0, \"issuer\": \"E.ON\", \"holder\": \"Mrs Alicia J Black\", \"page_marker\": \"\"}]}"}]} {"id": "monthly_bill_fin_007_extract", "task": "extract", "class": "monthly_bill", "doc": "monthly_bill_fin_007.pdf", "images": ["images/train/monthly_bill/monthly_bill_fin_007_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('monthly_bill', 'The document class.')\nfields = (['full_name', 'address', 'issue_date', 'company_name', 'account_number'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name on the bill', 'address': 'Address', 'issue_date': 'Issued date', 'company_name': 'Name of the issuing company', 'account_number': 'Account number'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"Mrs Alicia J Black\"}, {\"field\": \"address\", \"value\": \"Manor Works, Manor Road, London, SE25 4TA\"}, {\"field\": \"issue_date\", \"value\": \"2019-08-22\"}, {\"field\": \"company_name\", \"value\": \"E.ON\"}, {\"field\": \"account_number\", \"value\": \"0109 3556 5300\"}]}"}]} {"id": "monthly_bill_fin_008_p1_classify", "task": "classify", "class": "monthly_bill", "doc": "monthly_bill_fin_008.pdf", "images": ["images/train/monthly_bill/monthly_bill_fin_008_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"monthly_bill\", \"confidence\": 1.0, \"issuer\": \"Kenya Power\", \"holder\": \"JOHN CITIZEN\", \"page_marker\": \"\"}]}"}]} {"id": "monthly_bill_fin_008_extract", "task": "extract", "class": "monthly_bill", "doc": "monthly_bill_fin_008.pdf", "images": ["images/train/monthly_bill/monthly_bill_fin_008_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('monthly_bill', 'The document class.')\nfields = (['full_name', 'address', 'issue_date', 'company_name', 'account_number'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name on the bill', 'address': 'Address', 'issue_date': 'Issued date', 'company_name': 'Name of the issuing company', 'account_number': 'Account number'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"JOHN CITIZEN\"}, {\"field\": \"address\", \"value\": \"WAIYAKI WAY, 266 - NAIROBI, KENYA\"}, {\"field\": \"issue_date\", \"value\": \"2020-11-23\"}, {\"field\": \"company_name\", \"value\": \"Kenya Power\"}, {\"field\": \"account_number\", \"value\": \"3017996-01\"}]}"}]} {"id": "monthly_bill_fin_009_p1_classify", "task": "classify", "class": "monthly_bill", "doc": "monthly_bill_fin_009.pdf", "images": ["images/train/monthly_bill/monthly_bill_fin_009_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"monthly_bill\", \"confidence\": 1.0, \"issuer\": \"Kenya Power\", \"holder\": \"KELVIN ETWASI\", \"page_marker\": \"\"}]}"}]} {"id": "monthly_bill_fin_009_extract", "task": "extract", "class": "monthly_bill", "doc": "monthly_bill_fin_009.pdf", "images": ["images/train/monthly_bill/monthly_bill_fin_009_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('monthly_bill', 'The document class.')\nfields = (['full_name', 'address', 'issue_date', 'company_name', 'account_number'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name on the bill', 'address': 'Address', 'issue_date': 'Issued date', 'company_name': 'Name of the issuing company', 'account_number': 'Account number'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"KELVIN ETWASI\"}, {\"field\": \"address\", \"value\": \"Mwale Medical and Teaching Technology, 14, 50100 - BUTERE, KENYA\"}, {\"field\": \"issue_date\", \"value\": \"2023-06-09\"}, {\"field\": \"company_name\", \"value\": \"Kenya Power\"}, {\"field\": \"account_number\", \"value\": \"33281591-04\"}]}"}]} {"id": "monthly_bill_fin_010_p1_classify", "task": "classify", "class": "monthly_bill", "doc": "monthly_bill_fin_010.pdf", "images": ["images/train/monthly_bill/monthly_bill_fin_010_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"monthly_bill\", \"confidence\": 1.0, \"issuer\": \"EnergyAustralia Pty Ltd\", \"holder\": \"SAM SAMPLE\", \"page_marker\": \"\"}]}"}]} {"id": "monthly_bill_fin_010_extract", "task": "extract", "class": "monthly_bill", "doc": "monthly_bill_fin_010.pdf", "images": ["images/train/monthly_bill/monthly_bill_fin_010_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('monthly_bill', 'The document class.')\nfields = (['full_name', 'address', 'issue_date', 'company_name', 'account_number'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name on the bill', 'address': 'Address', 'issue_date': 'Issued date', 'company_name': 'Name of the issuing company', 'account_number': 'Account number'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"SAM SAMPLE\"}, {\"field\": \"address\", \"value\": \"SAMPLE ROAD, SAMPLEVILLE QLD 9999\"}, {\"field\": \"issue_date\", \"value\": \"2019-01-20\"}, {\"field\": \"company_name\", \"value\": \"EnergyAustralia Pty Ltd\"}, {\"field\": \"account_number\", \"value\": \"3333 333 333\"}]}"}]} {"id": "monthly_bill_fin_011_p1_classify", "task": "classify", "class": "monthly_bill", "doc": "monthly_bill_fin_011.pdf", "images": ["images/train/monthly_bill/monthly_bill_fin_011_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"monthly_bill\", \"confidence\": 1.0, \"issuer\": \"Shell Energy Solutions\", \"holder\": \"JOHN DOE\", \"page_marker\": \"\"}]}"}]} {"id": "monthly_bill_fin_011_extract", "task": "extract", "class": "monthly_bill", "doc": "monthly_bill_fin_011.pdf", "images": ["images/train/monthly_bill/monthly_bill_fin_011_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('monthly_bill', 'The document class.')\nfields = (['full_name', 'address', 'issue_date', 'company_name', 'account_number'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name on the bill', 'address': 'Address', 'issue_date': 'Issued date', 'company_name': 'Name of the issuing company', 'account_number': 'Account number'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"JOHN DOE\"}, {\"field\": \"address\", \"value\": \"11000 TEST LANE ANYTOWN, TX 77375\"}, {\"field\": \"issue_date\", \"value\": \"2021-11-30\"}, {\"field\": \"company_name\", \"value\": \"Shell Energy Solutions\"}, {\"field\": \"account_number\", \"value\": \"00001\"}]}"}]} {"id": "monthly_bill_fin_012_p1_classify", "task": "classify", "class": "monthly_bill", "doc": "monthly_bill_fin_012.pdf", "images": ["images/train/monthly_bill/monthly_bill_fin_012_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"monthly_bill\", \"confidence\": 1.0, \"issuer\": \"British Gas\", \"holder\": \"Mr Joseph Williams\", \"page_marker\": \"\"}]}"}]} {"id": "monthly_bill_fin_012_extract", "task": "extract", "class": "monthly_bill", "doc": "monthly_bill_fin_012.pdf", "images": ["images/train/monthly_bill/monthly_bill_fin_012_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('monthly_bill', 'The document class.')\nfields = (['full_name', 'address', 'issue_date', 'company_name', 'account_number'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name on the bill', 'address': 'Address', 'issue_date': 'Issued date', 'company_name': 'Name of the issuing company', 'account_number': 'Account number'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"Mr Joseph Williams\"}, {\"field\": \"address\", \"value\": \"DEANSHANGER MEMORIAL COMMUNITY CENT, LITTLE LONDON, MILTON KEYNES, MK19 6HT\"}, {\"field\": \"issue_date\", \"value\": \"2019-12-20\"}, {\"field\": \"company_name\", \"value\": \"British Gas\"}, {\"field\": \"account_number\", \"value\": \"12 34 56 78 90 25\"}]}"}]} {"id": "monthly_bill_fin_015_p1_classify", "task": "classify", "class": "monthly_bill", "doc": "monthly_bill_fin_015.pdf", "images": ["images/train/monthly_bill/monthly_bill_fin_015_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"monthly_bill\", \"confidence\": 1.0, \"issuer\": \"e.on\", \"holder\": \"Mr A B Customer\", \"page_marker\": \"\"}]}"}]} {"id": "monthly_bill_fin_015_extract", "task": "extract", "class": "monthly_bill", "doc": "monthly_bill_fin_015.pdf", "images": ["images/train/monthly_bill/monthly_bill_fin_015_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('monthly_bill', 'The document class.')\nfields = (['full_name', 'address', 'issue_date', 'company_name', 'account_number'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name on the bill', 'address': 'Address', 'issue_date': 'Issued date', 'company_name': 'Name of the issuing company', 'account_number': 'Account number'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"Mr A B Customer\"}, {\"field\": \"address\", \"value\": \"10 Example Street, Anytown, Countyshire, AB12 3CD\"}, {\"field\": \"issue_date\", \"value\": \"2014-06-16\"}, {\"field\": \"company_name\", \"value\": \"e.on\"}, {\"field\": \"account_number\", \"value\": \"0108 6420 2750\"}]}"}]} {"id": "monthly_bill_fin_016_p1_classify", "task": "classify", "class": "monthly_bill", "doc": "monthly_bill_fin_016.pdf", "images": ["images/train/monthly_bill/monthly_bill_fin_016_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"monthly_bill\", \"confidence\": 1.0, \"issuer\": \"e.on\", \"holder\": \"Mr John Moses\", \"page_marker\": \"\"}]}"}]} {"id": "monthly_bill_fin_016_extract", "task": "extract", "class": "monthly_bill", "doc": "monthly_bill_fin_016.pdf", "images": ["images/train/monthly_bill/monthly_bill_fin_016_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('monthly_bill', 'The document class.')\nfields = (['full_name', 'address', 'issue_date', 'company_name', 'account_number'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name on the bill', 'address': 'Address', 'issue_date': 'Issued date', 'company_name': 'Name of the issuing company', 'account_number': 'Account number'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"Mr John Moses\"}, {\"field\": \"address\", \"value\": \"Grove House, Chevening Cross Road, Chevening, Sevenoaks, Kent, TN14 6HF\"}, {\"field\": \"issue_date\", \"value\": \"2015-03-18\"}, {\"field\": \"company_name\", \"value\": \"e.on\"}, {\"field\": \"account_number\", \"value\": \"0109 3556 5300\"}]}"}]} {"id": "monthly_bill_fin_017_p1_classify", "task": "classify", "class": "monthly_bill", "doc": "monthly_bill_fin_017.pdf", "images": ["images/train/monthly_bill/monthly_bill_fin_017_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"monthly_bill\", \"confidence\": 1.0, \"issuer\": \"Duke Energy\", \"holder\": \"KAMALNAYAN RUPARELIA\", \"page_marker\": \"\"}]}"}]} {"id": "monthly_bill_fin_017_extract", "task": "extract", "class": "monthly_bill", "doc": "monthly_bill_fin_017.pdf", "images": ["images/train/monthly_bill/monthly_bill_fin_017_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('monthly_bill', 'The document class.')\nfields = (['full_name', 'address', 'issue_date', 'company_name', 'account_number'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name on the bill', 'address': 'Address', 'issue_date': 'Issued date', 'company_name': 'Name of the issuing company', 'account_number': 'Account number'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"KAMALNAYAN RUPARELIA\"}, {\"field\": \"address\", \"value\": \"13015 LAKESHORE GROVE DR #D, WINTER GARDEN, FL 34787\"}, {\"field\": \"issue_date\", \"value\": \"\"}, {\"field\": \"company_name\", \"value\": \"Duke Energy\"}, {\"field\": \"account_number\", \"value\": \"6120-2687-07-4\"}]}"}]} {"id": "monthly_bill_fin_018_p1_classify", "task": "classify", "class": "monthly_bill", "doc": "monthly_bill_fin_018.pdf", "images": ["images/train/monthly_bill/monthly_bill_fin_018_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"monthly_bill\", \"confidence\": 1.0, \"issuer\": \"Clark Regional Wastewater District\", \"holder\": \"CUSTOMER NAME\", \"page_marker\": \"\"}]}"}]} {"id": "monthly_bill_fin_018_extract", "task": "extract", "class": "monthly_bill", "doc": "monthly_bill_fin_018.pdf", "images": ["images/train/monthly_bill/monthly_bill_fin_018_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('monthly_bill', 'The document class.')\nfields = (['full_name', 'address', 'issue_date', 'company_name', 'account_number'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name on the bill', 'address': 'Address', 'issue_date': 'Issued date', 'company_name': 'Name of the issuing company', 'account_number': 'Account number'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"CUSTOMER NAME\"}, {\"field\": \"address\", \"value\": \"123 MAIN STREET, VANCOUVER, WA 98661\"}, {\"field\": \"issue_date\", \"value\": \"2017-12-29\"}, {\"field\": \"company_name\", \"value\": \"Clark Regional Wastewater District\"}, {\"field\": \"account_number\", \"value\": \"000000-000\"}]}"}]} {"id": "monthly_bill_fin_019_p1_classify", "task": "classify", "class": "monthly_bill", "doc": "monthly_bill_fin_019.pdf", "images": ["images/train/monthly_bill/monthly_bill_fin_019_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"monthly_bill\", \"confidence\": 1.0, \"issuer\": \"British Gas\", \"holder\": \"Mr Abdur Rakib\", \"page_marker\": \"\"}]}"}]} {"id": "monthly_bill_fin_019_extract", "task": "extract", "class": "monthly_bill", "doc": "monthly_bill_fin_019.pdf", "images": ["images/train/monthly_bill/monthly_bill_fin_019_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('monthly_bill', 'The document class.')\nfields = (['full_name', 'address', 'issue_date', 'company_name', 'account_number'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name on the bill', 'address': 'Address', 'issue_date': 'Issued date', 'company_name': 'Name of the issuing company', 'account_number': 'Account number'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"Mr Abdur Rakib\"}, {\"field\": \"address\", \"value\": \"203C Southgate Road, London N1 3LD\"}, {\"field\": \"issue_date\", \"value\": \"2018-09-29\"}, {\"field\": \"company_name\", \"value\": \"British Gas\"}, {\"field\": \"account_number\", \"value\": \"85 18 22 08 31 17\"}]}"}]} {"id": "monthly_bill_fin_020_p1_classify", "task": "classify", "class": "monthly_bill", "doc": "monthly_bill_fin_020.pdf", "images": ["images/train/monthly_bill/monthly_bill_fin_020_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"monthly_bill\", \"confidence\": 1.0, \"issuer\": \"Liberty Utilities\", \"holder\": \"JOHN SMITH\", \"page_marker\": \"\"}]}"}]} {"id": "monthly_bill_fin_020_extract", "task": "extract", "class": "monthly_bill", "doc": "monthly_bill_fin_020.pdf", "images": ["images/train/monthly_bill/monthly_bill_fin_020_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('monthly_bill', 'The document class.')\nfields = (['full_name', 'address', 'issue_date', 'company_name', 'account_number'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name on the bill', 'address': 'Address', 'issue_date': 'Issued date', 'company_name': 'Name of the issuing company', 'account_number': 'Account number'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"JOHN SMITH\"}, {\"field\": \"address\", \"value\": \"12345 MAIN ST. SUNLAKES, AZ 85248\"}, {\"field\": \"issue_date\", \"value\": \"2020-10-23\"}, {\"field\": \"company_name\", \"value\": \"Liberty Utilities\"}, {\"field\": \"account_number\", \"value\": \"12345678-87654321\"}]}"}]} {"id": "monthly_bill_fin_021_p1_classify", "task": "classify", "class": "monthly_bill", "doc": "monthly_bill_fin_021.pdf", "images": ["images/train/monthly_bill/monthly_bill_fin_021_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"monthly_bill\", \"confidence\": 1.0, \"issuer\": \"Southern Electric\", \"holder\": \"John Citizen\", \"page_marker\": \"\"}]}"}]} {"id": "monthly_bill_fin_021_extract", "task": "extract", "class": "monthly_bill", "doc": "monthly_bill_fin_021.pdf", "images": ["images/train/monthly_bill/monthly_bill_fin_021_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('monthly_bill', 'The document class.')\nfields = (['full_name', 'address', 'issue_date', 'company_name', 'account_number'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name on the bill', 'address': 'Address', 'issue_date': 'Issued date', 'company_name': 'Name of the issuing company', 'account_number': 'Account number'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"John Citizen\"}, {\"field\": \"address\", \"value\": \"3 Tortworth Road, Swindon, SN25 2AJ\"}, {\"field\": \"issue_date\", \"value\": \"2021-11-13\"}, {\"field\": \"company_name\", \"value\": \"Southern Electric\"}, {\"field\": \"account_number\", \"value\": \"607561694\"}]}"}]} {"id": "monthly_bill_fin_022_p1_classify", "task": "classify", "class": "monthly_bill", "doc": "monthly_bill_fin_022.pdf", "images": ["images/train/monthly_bill/monthly_bill_fin_022_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"monthly_bill\", \"confidence\": 1.0, \"issuer\": \"Columbia Gas of Ohio\", \"holder\": \"Columbia Gas of Ohio Customer\", \"page_marker\": \"\"}]}"}]} {"id": "monthly_bill_fin_022_extract", "task": "extract", "class": "monthly_bill", "doc": "monthly_bill_fin_022.pdf", "images": ["images/train/monthly_bill/monthly_bill_fin_022_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('monthly_bill', 'The document class.')\nfields = (['full_name', 'address', 'issue_date', 'company_name', 'account_number'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name on the bill', 'address': 'Address', 'issue_date': 'Issued date', 'company_name': 'Name of the issuing company', 'account_number': 'Account number'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"Columbia Gas of Ohio Customer\"}, {\"field\": \"address\", \"value\": \"123 Main St, Anytown OH 12345-6780\"}, {\"field\": \"issue_date\", \"value\": \"2011-04-29\"}, {\"field\": \"company_name\", \"value\": \"Columbia Gas of Ohio\"}, {\"field\": \"account_number\", \"value\": \"XXXXXXXX XXX XXX X\"}]}"}]} {"id": "monthly_bill_fin_023_p1_classify", "task": "classify", "class": "monthly_bill", "doc": "monthly_bill_fin_023.pdf", "images": ["images/train/monthly_bill/monthly_bill_fin_023_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"monthly_bill\", \"confidence\": 1.0, \"issuer\": \"Egyptian Electric Cooperative Association\", \"holder\": \"JOHN DOE JANE DOE\", \"page_marker\": \"\"}]}"}]} {"id": "monthly_bill_fin_023_extract", "task": "extract", "class": "monthly_bill", "doc": "monthly_bill_fin_023.pdf", "images": ["images/train/monthly_bill/monthly_bill_fin_023_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('monthly_bill', 'The document class.')\nfields = (['full_name', 'address', 'issue_date', 'company_name', 'account_number'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name on the bill', 'address': 'Address', 'issue_date': 'Issued date', 'company_name': 'Name of the issuing company', 'account_number': 'Account number'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"JOHN DOE JANE DOE\"}, {\"field\": \"address\", \"value\": \"72255 ORCHARD TREE RD MURPHYSBORO IL 62966\"}, {\"field\": \"issue_date\", \"value\": \"2016-08-03\"}, {\"field\": \"company_name\", \"value\": \"Egyptian Electric Cooperative Association\"}, {\"field\": \"account_number\", \"value\": \"83XXX001\"}]}"}]} {"id": "monthly_bill_fin_024_p1_classify", "task": "classify", "class": "monthly_bill", "doc": "monthly_bill_fin_024.pdf", "images": ["images/train/monthly_bill/monthly_bill_fin_024_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"monthly_bill\", \"confidence\": 1.0, \"issuer\": \"MARIN WATER\", \"holder\": \"SAMPLE BILL\", \"page_marker\": \"\"}]}"}]} {"id": "monthly_bill_fin_024_extract", "task": "extract", "class": "monthly_bill", "doc": "monthly_bill_fin_024.pdf", "images": ["images/train/monthly_bill/monthly_bill_fin_024_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('monthly_bill', 'The document class.')\nfields = (['full_name', 'address', 'issue_date', 'company_name', 'account_number'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name on the bill', 'address': 'Address', 'issue_date': 'Issued date', 'company_name': 'Name of the issuing company', 'account_number': 'Account number'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"SAMPLE BILL\"}, {\"field\": \"address\", \"value\": \"220 NELLEN AVE, CORTE MADERA, CA 94925\"}, {\"field\": \"issue_date\", \"value\": \"2023-10-01\"}, {\"field\": \"company_name\", \"value\": \"MARIN WATER\"}, {\"field\": \"account_number\", \"value\": \"987654\"}]}"}]} {"id": "monthly_bill_fin_025_p1_classify", "task": "classify", "class": "monthly_bill", "doc": "monthly_bill_fin_025.pdf", "images": ["images/train/monthly_bill/monthly_bill_fin_025_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"monthly_bill\", \"confidence\": 1.0, \"issuer\": \"Sydney Water\", \"holder\": \"MR SAMPLE\", \"page_marker\": \"\"}]}"}]} {"id": "monthly_bill_fin_025_extract", "task": "extract", "class": "monthly_bill", "doc": "monthly_bill_fin_025.pdf", "images": ["images/train/monthly_bill/monthly_bill_fin_025_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('monthly_bill', 'The document class.')\nfields = (['full_name', 'address', 'issue_date', 'company_name', 'account_number'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name on the bill', 'address': 'Address', 'issue_date': 'Issued date', 'company_name': 'Name of the issuing company', 'account_number': 'Account number'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"MR SAMPLE\"}, {\"field\": \"address\", \"value\": \"1 SAMPLE ST SAMPLETOWN NSW 2999\"}, {\"field\": \"issue_date\", \"value\": \"2025-10-07\"}, {\"field\": \"company_name\", \"value\": \"Sydney Water\"}, {\"field\": \"account_number\", \"value\": \"1234 567\"}]}"}]} {"id": "monthly_bill_fin_026_p1_classify", "task": "classify", "class": "monthly_bill", "doc": "monthly_bill_fin_026.pdf", "images": ["images/train/monthly_bill/monthly_bill_fin_026_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"monthly_bill\", \"confidence\": 1.0, \"issuer\": \"BUENAVISTA WATER DISTRICT\", \"holder\": \"Balingit Joe\", \"page_marker\": \"\"}]}"}]} {"id": "monthly_bill_fin_026_extract", "task": "extract", "class": "monthly_bill", "doc": "monthly_bill_fin_026.pdf", "images": ["images/train/monthly_bill/monthly_bill_fin_026_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('monthly_bill', 'The document class.')\nfields = (['full_name', 'address', 'issue_date', 'company_name', 'account_number'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name on the bill', 'address': 'Address', 'issue_date': 'Issued date', 'company_name': 'Name of the issuing company', 'account_number': 'Account number'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"Balingit Joe\"}, {\"field\": \"address\", \"value\": \"Rizal\"}, {\"field\": \"issue_date\", \"value\": \"2021-12-13\"}, {\"field\": \"company_name\", \"value\": \"BUENAVISTA WATER DISTRICT\"}, {\"field\": \"account_number\", \"value\": \"0441-12-022\"}]}"}]} {"id": "monthly_bill_fin_027_p1_classify", "task": "classify", "class": "monthly_bill", "doc": "monthly_bill_fin_027.pdf", "images": ["images/train/monthly_bill/monthly_bill_fin_027_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"monthly_bill\", \"confidence\": 1.0, \"issuer\": \"KENYA WATER COMPANY\", \"holder\": \"CELESTINE MASIKA\", \"page_marker\": \"\"}]}"}]} {"id": "monthly_bill_fin_027_extract", "task": "extract", "class": "monthly_bill", "doc": "monthly_bill_fin_027.pdf", "images": ["images/train/monthly_bill/monthly_bill_fin_027_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('monthly_bill', 'The document class.')\nfields = (['full_name', 'address', 'issue_date', 'company_name', 'account_number'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name on the bill', 'address': 'Address', 'issue_date': 'Issued date', 'company_name': 'Name of the issuing company', 'account_number': 'Account number'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"CELESTINE MASIKA\"}, {\"field\": \"address\", \"value\": \"45300 KIONGO STREET, NAIROBI, 00200\"}, {\"field\": \"issue_date\", \"value\": \"2023-04-01\"}, {\"field\": \"company_name\", \"value\": \"KENYA WATER COMPANY\"}, {\"field\": \"account_number\", \"value\": \"70233890\"}]}"}]} {"id": "monthly_bill_fin_028_p1_classify", "task": "classify", "class": "monthly_bill", "doc": "monthly_bill_fin_028.pdf", "images": ["images/train/monthly_bill/monthly_bill_fin_028_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"monthly_bill\", \"confidence\": 1.0, \"issuer\": \"KISUMU WATER COMPANY\", \"holder\": \"JENEPHER KATUYI\", \"page_marker\": \"\"}]}"}]} {"id": "monthly_bill_fin_028_extract", "task": "extract", "class": "monthly_bill", "doc": "monthly_bill_fin_028.pdf", "images": ["images/train/monthly_bill/monthly_bill_fin_028_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('monthly_bill', 'The document class.')\nfields = (['full_name', 'address', 'issue_date', 'company_name', 'account_number'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name on the bill', 'address': 'Address', 'issue_date': 'Issued date', 'company_name': 'Name of the issuing company', 'account_number': 'Account number'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"JENEPHER KATUYI\"}, {\"field\": \"address\", \"value\": \"MACHAKOS WOTE RD MUU APARTMENT, 5TH FLOOR, ROOM 7 MUUMANDU AREA\"}, {\"field\": \"issue_date\", \"value\": \"2022-10-01\"}, {\"field\": \"company_name\", \"value\": \"KISUMU WATER COMPANY\"}, {\"field\": \"account_number\", \"value\": \"8593681\"}]}"}]} {"id": "monthly_bill_fin_029_p1_classify", "task": "classify", "class": "monthly_bill", "doc": "monthly_bill_fin_029.pdf", "images": ["images/train/monthly_bill/monthly_bill_fin_029_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"monthly_bill\", \"confidence\": 1.0, \"issuer\": \"Clear River Electric & Water District\", \"holder\": \"John Doe Jane Doe\", \"page_marker\": \"\"}]}"}]} {"id": "monthly_bill_fin_029_extract", "task": "extract", "class": "monthly_bill", "doc": "monthly_bill_fin_029.pdf", "images": ["images/train/monthly_bill/monthly_bill_fin_029_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('monthly_bill', 'The document class.')\nfields = (['full_name', 'address', 'issue_date', 'company_name', 'account_number'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name on the bill', 'address': 'Address', 'issue_date': 'Issued date', 'company_name': 'Name of the issuing company', 'account_number': 'Account number'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"John Doe Jane Doe\"}, {\"field\": \"address\", \"value\": \"123 Sample Road PASCOAG RI 02859\"}, {\"field\": \"issue_date\", \"value\": \"2024-11-27\"}, {\"field\": \"company_name\", \"value\": \"Clear River Electric & Water District\"}, {\"field\": \"account_number\", \"value\": \"12345678\"}]}"}]} {"id": "monthly_bill_fin_031_p1_classify", "task": "classify", "class": "monthly_bill", "doc": "monthly_bill_fin_031.pdf", "images": ["images/train/monthly_bill/monthly_bill_fin_031_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"monthly_bill\", \"confidence\": 1.0, \"issuer\": \"R.V.GAS AGENCY (41004895)\", \"holder\": \"SRINIVAS RAO . J\", \"page_marker\": \"\"}]}"}]} {"id": "monthly_bill_fin_031_extract", "task": "extract", "class": "monthly_bill", "doc": "monthly_bill_fin_031.pdf", "images": ["images/train/monthly_bill/monthly_bill_fin_031_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('monthly_bill', 'The document class.')\nfields = (['full_name', 'address', 'issue_date', 'company_name', 'account_number'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name on the bill', 'address': 'Address', 'issue_date': 'Issued date', 'company_name': 'Name of the issuing company', 'account_number': 'Account number'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"SRINIVAS RAO . J\"}, {\"field\": \"address\", \"value\": \"H.NO:35-294/3/1,KALAVATHI NAGAR. SHAPOOR NAGAR,. 8897138272. 500055\"}, {\"field\": \"issue_date\", \"value\": \"2025-06-16\"}, {\"field\": \"company_name\", \"value\": \"R.V.GAS AGENCY (41004895)\"}, {\"field\": \"account_number\", \"value\": \"425038\"}]}"}]} {"id": "monthly_bill_fin_032_p1_classify", "task": "classify", "class": "monthly_bill", "doc": "monthly_bill_fin_032.pdf", "images": ["images/train/monthly_bill/monthly_bill_fin_032_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"monthly_bill\", \"confidence\": 1.0, \"issuer\": \"iD mobile\", \"holder\": \"Mr JOHN SMITH\", \"page_marker\": \"\"}]}"}]} {"id": "monthly_bill_fin_032_extract", "task": "extract", "class": "monthly_bill", "doc": "monthly_bill_fin_032.pdf", "images": ["images/train/monthly_bill/monthly_bill_fin_032_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('monthly_bill', 'The document class.')\nfields = (['full_name', 'address', 'issue_date', 'company_name', 'account_number'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name on the bill', 'address': 'Address', 'issue_date': 'Issued date', 'company_name': 'Name of the issuing company', 'account_number': 'Account number'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"Mr JOHN SMITH\"}, {\"field\": \"address\", \"value\": \"1 LOREM IPSUM, LOREM IPSUM DOLOR, LONDON, UK, NW4 14J\"}, {\"field\": \"issue_date\", \"value\": \"2017-04-07\"}, {\"field\": \"company_name\", \"value\": \"iD mobile\"}, {\"field\": \"account_number\", \"value\": \"12345678\"}]}"}]} {"id": "monthly_bill_fin_033_p1_classify", "task": "classify", "class": "monthly_bill", "doc": "monthly_bill_fin_033.pdf", "images": ["images/train/monthly_bill/monthly_bill_fin_033_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"monthly_bill\", \"confidence\": 1.0, \"issuer\": \"AT&T\", \"holder\": \"AT&T CUSTOMER\", \"page_marker\": \"\"}]}"}]} {"id": "monthly_bill_fin_033_extract", "task": "extract", "class": "monthly_bill", "doc": "monthly_bill_fin_033.pdf", "images": ["images/train/monthly_bill/monthly_bill_fin_033_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('monthly_bill', 'The document class.')\nfields = (['full_name', 'address', 'issue_date', 'company_name', 'account_number'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name on the bill', 'address': 'Address', 'issue_date': 'Issued date', 'company_name': 'Name of the issuing company', 'account_number': 'Account number'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"AT&T CUSTOMER\"}, {\"field\": \"address\", \"value\": \"1234 TELEPHONE LN, AUSTIN, TX 78748-1234\"}, {\"field\": \"issue_date\", \"value\": \"\"}, {\"field\": \"company_name\", \"value\": \"AT&T\"}, {\"field\": \"account_number\", \"value\": \"123456789105\"}]}"}]} {"id": "monthly_bill_fin_035_p1_classify", "task": "classify", "class": "monthly_bill", "doc": "monthly_bill_fin_035.pdf", "images": ["images/train/monthly_bill/monthly_bill_fin_035_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"monthly_bill\", \"confidence\": 1.0, \"issuer\": \"iD Mobile\", \"holder\": \"MR JOHN DOE\", \"page_marker\": \"\"}]}"}]} {"id": "monthly_bill_fin_035_extract", "task": "extract", "class": "monthly_bill", "doc": "monthly_bill_fin_035.pdf", "images": ["images/train/monthly_bill/monthly_bill_fin_035_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('monthly_bill', 'The document class.')\nfields = (['full_name', 'address', 'issue_date', 'company_name', 'account_number'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name on the bill', 'address': 'Address', 'issue_date': 'Issued date', 'company_name': 'Name of the issuing company', 'account_number': 'Account number'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"MR JOHN DOE\"}, {\"field\": \"address\", \"value\": \"123 CASHBACK ROAD LONDON SE32 1ZZ\"}, {\"field\": \"issue_date\", \"value\": \"2023-08-30\"}, {\"field\": \"company_name\", \"value\": \"iD Mobile\"}, {\"field\": \"account_number\", \"value\": \"\"}]}"}]} {"id": "monthly_bill_fin_038_p1_classify", "task": "classify", "class": "monthly_bill", "doc": "monthly_bill_fin_038.pdf", "images": ["images/train/monthly_bill/monthly_bill_fin_038_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"monthly_bill\", \"confidence\": 1.0, \"issuer\": \"Ab Company\", \"holder\": \"Maron Smith\", \"page_marker\": \"\"}]}"}]} {"id": "monthly_bill_fin_038_extract", "task": "extract", "class": "monthly_bill", "doc": "monthly_bill_fin_038.pdf", "images": ["images/train/monthly_bill/monthly_bill_fin_038_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('monthly_bill', 'The document class.')\nfields = (['full_name', 'address', 'issue_date', 'company_name', 'account_number'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name on the bill', 'address': 'Address', 'issue_date': 'Issued date', 'company_name': 'Name of the issuing company', 'account_number': 'Account number'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"Maron Smith\"}, {\"field\": \"address\", \"value\": \"Hope Street no. 26, Building B Mumbai, Maharashtra (MH - 27), India\"}, {\"field\": \"issue_date\", \"value\": \"2018-02-21\"}, {\"field\": \"company_name\", \"value\": \"Ab Company\"}, {\"field\": \"account_number\", \"value\": \"\"}]}"}]} {"id": "monthly_bill_fin_039_p1_classify", "task": "classify", "class": "monthly_bill", "doc": "monthly_bill_fin_039.pdf", "images": ["images/train/monthly_bill/monthly_bill_fin_039_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"monthly_bill\", \"confidence\": 1.0, \"issuer\": \"sky\", \"holder\": \"MISS LILY BROOKER\", \"page_marker\": \"\"}]}"}]} {"id": "monthly_bill_fin_039_extract", "task": "extract", "class": "monthly_bill", "doc": "monthly_bill_fin_039.pdf", "images": ["images/train/monthly_bill/monthly_bill_fin_039_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('monthly_bill', 'The document class.')\nfields = (['full_name', 'address', 'issue_date', 'company_name', 'account_number'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name on the bill', 'address': 'Address', 'issue_date': 'Issued date', 'company_name': 'Name of the issuing company', 'account_number': 'Account number'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"MISS LILY BROOKER\"}, {\"field\": \"address\", \"value\": \"3 LOUP COURT, AXMINSTER, EX13 5FD\"}, {\"field\": \"issue_date\", \"value\": \"\"}, {\"field\": \"company_name\", \"value\": \"sky\"}, {\"field\": \"account_number\", \"value\": \"624743228535\"}]}"}]} {"id": "monthly_bill_fin_040_p1_classify", "task": "classify", "class": "monthly_bill", "doc": "monthly_bill_fin_040.pdf", "images": ["images/train/monthly_bill/monthly_bill_fin_040_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"monthly_bill\", \"confidence\": 1.0, \"issuer\": \"EE\", \"holder\": \"Ms. A. Sample\", \"page_marker\": \"\"}]}"}]} {"id": "monthly_bill_fin_040_extract", "task": "extract", "class": "monthly_bill", "doc": "monthly_bill_fin_040.pdf", "images": ["images/train/monthly_bill/monthly_bill_fin_040_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('monthly_bill', 'The document class.')\nfields = (['full_name', 'address', 'issue_date', 'company_name', 'account_number'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name on the bill', 'address': 'Address', 'issue_date': 'Issued date', 'company_name': 'Name of the issuing company', 'account_number': 'Account number'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"Ms. A. Sample\"}, {\"field\": \"address\", \"value\": \"1 Sample Street, Sample Town, AA11 1AA\"}, {\"field\": \"issue_date\", \"value\": \"2018-09-01\"}, {\"field\": \"company_name\", \"value\": \"EE\"}, {\"field\": \"account_number\", \"value\": \"888888888\"}]}"}]} {"id": "monthly_bill_fin_041_p1_classify", "task": "classify", "class": "monthly_bill", "doc": "monthly_bill_fin_041.pdf", "images": ["images/train/monthly_bill/monthly_bill_fin_041_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"monthly_bill\", \"confidence\": 1.0, \"issuer\": \"Vodafone Idea Limited\", \"holder\": \"MNR SOLUTIONS PVT LTD\", \"page_marker\": \"\"}]}"}]} {"id": "monthly_bill_fin_041_extract", "task": "extract", "class": "monthly_bill", "doc": "monthly_bill_fin_041.pdf", "images": ["images/train/monthly_bill/monthly_bill_fin_041_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('monthly_bill', 'The document class.')\nfields = (['full_name', 'address', 'issue_date', 'company_name', 'account_number'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name on the bill', 'address': 'Address', 'issue_date': 'Issued date', 'company_name': 'Name of the issuing company', 'account_number': 'Account number'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"MNR SOLUTIONS PVT LTD\"}, {\"field\": \"address\", \"value\": \"H-102 3RD FLOOR SECTOR 63 NOIDA NOIDA NCR NOIDA NCR 201301 UTTAR PRADESH\"}, {\"field\": \"issue_date\", \"value\": \"2024-03-06\"}, {\"field\": \"company_name\", \"value\": \"Vodafone Idea Limited\"}, {\"field\": \"account_number\", \"value\": \"174502535\"}]}"}]} {"id": "monthly_bill_fin_043_p1_classify", "task": "classify", "class": "monthly_bill", "doc": "monthly_bill_fin_043.pdf", "images": ["images/train/monthly_bill/monthly_bill_fin_043_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"monthly_bill\", \"confidence\": 1.0, \"issuer\": \"Vodafone\", \"holder\": \"Nalinh Douangphichit\", \"page_marker\": \"\"}]}"}]} {"id": "monthly_bill_fin_043_extract", "task": "extract", "class": "monthly_bill", "doc": "monthly_bill_fin_043.pdf", "images": ["images/train/monthly_bill/monthly_bill_fin_043_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('monthly_bill', 'The document class.')\nfields = (['full_name', 'address', 'issue_date', 'company_name', 'account_number'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name on the bill', 'address': 'Address', 'issue_date': 'Issued date', 'company_name': 'Name of the issuing company', 'account_number': 'Account number'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"Nalinh Douangphichit\"}, {\"field\": \"address\", \"value\": \"FLAT 1, 85 CARRINGTON ROAD, POINT CHEVALIER, AUCKLAND 1025\"}, {\"field\": \"issue_date\", \"value\": \"2014-09-01\"}, {\"field\": \"company_name\", \"value\": \"Vodafone\"}, {\"field\": \"account_number\", \"value\": \"9921881\"}]}"}]} {"id": "monthly_bill_fin_044_p1_classify", "task": "classify", "class": "monthly_bill", "doc": "monthly_bill_fin_044.pdf", "images": ["images/train/monthly_bill/monthly_bill_fin_044_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"monthly_bill\", \"confidence\": 1.0, \"issuer\": \"vodafone\", \"holder\": \"JAMES EDDIE\", \"page_marker\": \"\"}]}"}]} {"id": "monthly_bill_fin_044_extract", "task": "extract", "class": "monthly_bill", "doc": "monthly_bill_fin_044.pdf", "images": ["images/train/monthly_bill/monthly_bill_fin_044_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('monthly_bill', 'The document class.')\nfields = (['full_name', 'address', 'issue_date', 'company_name', 'account_number'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name on the bill', 'address': 'Address', 'issue_date': 'Issued date', 'company_name': 'Name of the issuing company', 'account_number': 'Account number'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"JAMES EDDIE\"}, {\"field\": \"address\", \"value\": \"7 HEATH DRIVE PETERHEAD ABERDEENSHIRE AB42 2FZ\"}, {\"field\": \"issue_date\", \"value\": \"2019-10-29\"}, {\"field\": \"company_name\", \"value\": \"vodafone\"}, {\"field\": \"account_number\", \"value\": \"7040989192\"}]}"}]} {"id": "monthly_bill_fin_045_p1_classify", "task": "classify", "class": "monthly_bill", "doc": "monthly_bill_fin_045.pdf", "images": ["images/train/monthly_bill/monthly_bill_fin_045_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"monthly_bill\", \"confidence\": 1.0, \"issuer\": \"Vodafone Cook Islands\", \"holder\": \"NBCX Go Live Ltd\", \"page_marker\": \"\"}]}"}]} {"id": "monthly_bill_fin_045_extract", "task": "extract", "class": "monthly_bill", "doc": "monthly_bill_fin_045.pdf", "images": ["images/train/monthly_bill/monthly_bill_fin_045_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('monthly_bill', 'The document class.')\nfields = (['full_name', 'address', 'issue_date', 'company_name', 'account_number'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name on the bill', 'address': 'Address', 'issue_date': 'Issued date', 'company_name': 'Name of the issuing company', 'account_number': 'Account number'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"NBCX Go Live Ltd\"}, {\"field\": \"address\", \"value\": \"Avarua -, 2000 Avarua\"}, {\"field\": \"issue_date\", \"value\": \"2021-12-31\"}, {\"field\": \"company_name\", \"value\": \"Vodafone Cook Islands\"}, {\"field\": \"account_number\", \"value\": \"CN00001\"}]}"}]} {"id": "monthly_bill_fin_046_p1_classify", "task": "classify", "class": "monthly_bill", "doc": "monthly_bill_fin_046.pdf", "images": ["images/train/monthly_bill/monthly_bill_fin_046_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"monthly_bill\", \"confidence\": 1.0, \"issuer\": \"Vodafone New Zealand Limited\", \"holder\": \"Mr Rajdeep Dadiala\", \"page_marker\": \"\"}]}"}]} {"id": "monthly_bill_fin_046_extract", "task": "extract", "class": "monthly_bill", "doc": "monthly_bill_fin_046.pdf", "images": ["images/train/monthly_bill/monthly_bill_fin_046_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('monthly_bill', 'The document class.')\nfields = (['full_name', 'address', 'issue_date', 'company_name', 'account_number'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name on the bill', 'address': 'Address', 'issue_date': 'Issued date', 'company_name': 'Name of the issuing company', 'account_number': 'Account number'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"Mr Rajdeep Dadiala\"}, {\"field\": \"address\", \"value\": \"36 William Souter Street, Forrest Hill, AUCKLAND 0620\"}, {\"field\": \"issue_date\", \"value\": \"2023-02-12\"}, {\"field\": \"company_name\", \"value\": \"Vodafone New Zealand Limited\"}, {\"field\": \"account_number\", \"value\": \"484850296\"}]}"}]} {"id": "monthly_bill_fin_047_p1_classify", "task": "classify", "class": "monthly_bill", "doc": "monthly_bill_fin_047.pdf", "images": ["images/train/monthly_bill/monthly_bill_fin_047_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"monthly_bill\", \"confidence\": 1.0, \"issuer\": \"vodafone\", \"holder\": \"John Citizen\", \"page_marker\": \"\"}]}"}]} {"id": "monthly_bill_fin_047_extract", "task": "extract", "class": "monthly_bill", "doc": "monthly_bill_fin_047.pdf", "images": ["images/train/monthly_bill/monthly_bill_fin_047_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('monthly_bill', 'The document class.')\nfields = (['full_name', 'address', 'issue_date', 'company_name', 'account_number'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name on the bill', 'address': 'Address', 'issue_date': 'Issued date', 'company_name': 'Name of the issuing company', 'account_number': 'Account number'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"John Citizen\"}, {\"field\": \"address\", \"value\": \"Ort\\u00edn Norias 121 30009 Murcia Murcia\"}, {\"field\": \"issue_date\", \"value\": \"2019-12-15\"}, {\"field\": \"company_name\", \"value\": \"vodafone\"}, {\"field\": \"account_number\", \"value\": \"926024926\"}]}"}]} {"id": "monthly_bill_fin_048_p1_classify", "task": "classify", "class": "monthly_bill", "doc": "monthly_bill_fin_048.pdf", "images": ["images/train/monthly_bill/monthly_bill_fin_048_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"monthly_bill\", \"confidence\": 1.0, \"issuer\": \"vodafone\", \"holder\": \"U ROSI\", \"page_marker\": \"\"}]}"}]} {"id": "monthly_bill_fin_048_extract", "task": "extract", "class": "monthly_bill", "doc": "monthly_bill_fin_048.pdf", "images": ["images/train/monthly_bill/monthly_bill_fin_048_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('monthly_bill', 'The document class.')\nfields = (['full_name', 'address', 'issue_date', 'company_name', 'account_number'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name on the bill', 'address': 'Address', 'issue_date': 'Issued date', 'company_name': 'Name of the issuing company', 'account_number': 'Account number'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"U ROSI\"}, {\"field\": \"address\", \"value\": \"136 COALPORT CLOSE, HARLOW, ESSEX, CM17 9RA\"}, {\"field\": \"issue_date\", \"value\": \"2023-09-07\"}, {\"field\": \"company_name\", \"value\": \"vodafone\"}, {\"field\": \"account_number\", \"value\": \"7044828003\"}]}"}]} {"id": "monthly_bill_fin_049_p1_classify", "task": "classify", "class": "monthly_bill", "doc": "monthly_bill_fin_049.pdf", "images": ["images/train/monthly_bill/monthly_bill_fin_049_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"monthly_bill\", \"confidence\": 1.0, \"issuer\": \"Vodafone\", \"holder\": \"\", \"page_marker\": \"\"}]}"}]} {"id": "monthly_bill_fin_049_extract", "task": "extract", "class": "monthly_bill", "doc": "monthly_bill_fin_049.pdf", "images": ["images/train/monthly_bill/monthly_bill_fin_049_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('monthly_bill', 'The document class.')\nfields = (['full_name', 'address', 'issue_date', 'company_name', 'account_number'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name on the bill', 'address': 'Address', 'issue_date': 'Issued date', 'company_name': 'Name of the issuing company', 'account_number': 'Account number'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"\"}, {\"field\": \"address\", \"value\": \"\"}, {\"field\": \"issue_date\", \"value\": \"2018-11-23\"}, {\"field\": \"company_name\", \"value\": \"Vodafone\"}, {\"field\": \"account_number\", \"value\": \"12955053\"}]}"}]} {"id": "monthly_bill_fin_050_p1_classify", "task": "classify", "class": "monthly_bill", "doc": "monthly_bill_fin_050.pdf", "images": ["images/train/monthly_bill/monthly_bill_fin_050_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"monthly_bill\", \"confidence\": 1.0, \"issuer\": \"airtel\", \"holder\": \"Pradnya Bagave\", \"page_marker\": \"\"}]}"}]} {"id": "monthly_bill_fin_050_extract", "task": "extract", "class": "monthly_bill", "doc": "monthly_bill_fin_050.pdf", "images": ["images/train/monthly_bill/monthly_bill_fin_050_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('monthly_bill', 'The document class.')\nfields = (['full_name', 'address', 'issue_date', 'company_name', 'account_number'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name on the bill', 'address': 'Address', 'issue_date': 'Issued date', 'company_name': 'Name of the issuing company', 'account_number': 'Account number'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"Pradnya Bagave\"}, {\"field\": \"address\", \"value\": \"Building No 8 Flat No G2,Rakshak Nagar Phase 2,Opp Durgamata Mandir,Kharad i,Pune,Pune,Maharashtra,411014 Pune, Maharashtra, 411014,\"}, {\"field\": \"issue_date\", \"value\": \"2024-02-06\"}, {\"field\": \"company_name\", \"value\": \"airtel\"}, {\"field\": \"account_number\", \"value\": \"02010805695_dsl\"}]}"}]} {"id": "monthly_bill_fin_051_p1_classify", "task": "classify", "class": "monthly_bill", "doc": "monthly_bill_fin_051.pdf", "images": ["images/train/monthly_bill/monthly_bill_fin_051_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"monthly_bill\", \"confidence\": 1.0, \"issuer\": \"airtel\", \"holder\": \"Mr VIVEKANAND AGRAWAL AGRAWAL\", \"page_marker\": \"\"}]}"}]} {"id": "monthly_bill_fin_051_extract", "task": "extract", "class": "monthly_bill", "doc": "monthly_bill_fin_051.pdf", "images": ["images/train/monthly_bill/monthly_bill_fin_051_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('monthly_bill', 'The document class.')\nfields = (['full_name', 'address', 'issue_date', 'company_name', 'account_number'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name on the bill', 'address': 'Address', 'issue_date': 'Issued date', 'company_name': 'Name of the issuing company', 'account_number': 'Account number'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"Mr VIVEKANAND AGRAWAL AGRAWAL\"}, {\"field\": \"address\", \"value\": \"Sec 01 Hig 10 Dd Nagar Thana Ke Piche I F Vidhaya Pati Garden, Raipur 492001, Chattisgarh, Landmark : Garden\"}, {\"field\": \"issue_date\", \"value\": \"2019-01-03\"}, {\"field\": \"company_name\", \"value\": \"airtel\"}, {\"field\": \"account_number\", \"value\": \"7032250482\"}]}"}]} {"id": "monthly_bill_fin_054_p1_classify", "task": "classify", "class": "monthly_bill", "doc": "monthly_bill_fin_054.pdf", "images": ["images/train/monthly_bill/monthly_bill_fin_054_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"monthly_bill\", \"confidence\": 1.0, \"issuer\": \"airtel\", \"holder\": \"Mr Roshankumar Singlesh Singh\", \"page_marker\": \"\"}]}"}]} {"id": "monthly_bill_fin_054_extract", "task": "extract", "class": "monthly_bill", "doc": "monthly_bill_fin_054.pdf", "images": ["images/train/monthly_bill/monthly_bill_fin_054_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('monthly_bill', 'The document class.')\nfields = (['full_name', 'address', 'issue_date', 'company_name', 'account_number'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name on the bill', 'address': 'Address', 'issue_date': 'Issued date', 'company_name': 'Name of the issuing company', 'account_number': 'Account number'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"Mr Roshankumar Singlesh Singh\"}, {\"field\": \"address\", \"value\": \"astRoom No 2,Premchandra Chwl Pal Nagar Pal Nagar,Pathan Wadi,Malad East S O Mumbai 400097 Maharashtra Landmark :Near Shankar Temple\"}, {\"field\": \"issue_date\", \"value\": \"2019-04-27\"}, {\"field\": \"company_name\", \"value\": \"airtel\"}, {\"field\": \"account_number\", \"value\": \"9619512246\"}]}"}]} {"id": "monthly_bill_fin_055_p1_classify", "task": "classify", "class": "monthly_bill", "doc": "monthly_bill_fin_055.pdf", "images": ["images/train/monthly_bill/monthly_bill_fin_055_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"monthly_bill\", \"confidence\": 1.0, \"issuer\": \"INNOVATIVE RETAIL CONCEPTS PRIVATE LIMITED\", \"holder\": \"\", \"page_marker\": \"\"}]}"}]} {"id": "monthly_bill_fin_055_extract", "task": "extract", "class": "monthly_bill", "doc": "monthly_bill_fin_055.pdf", "images": ["images/train/monthly_bill/monthly_bill_fin_055_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('monthly_bill', 'The document class.')\nfields = (['full_name', 'address', 'issue_date', 'company_name', 'account_number'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name on the bill', 'address': 'Address', 'issue_date': 'Issued date', 'company_name': 'Name of the issuing company', 'account_number': 'Account number'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"\"}, {\"field\": \"address\", \"value\": \"No 1 & 2 3rd Main Teachers Colony Jakkasandra Koramangala Bangalore KAR 560034\"}, {\"field\": \"issue_date\", \"value\": \"2024-12-18\"}, {\"field\": \"company_name\", \"value\": \"INNOVATIVE RETAIL CONCEPTS PRIVATE LIMITED\"}, {\"field\": \"account_number\", \"value\": \"20007931144\"}]}"}]} {"id": "monthly_bill_fin_056_p1_classify", "task": "classify", "class": "monthly_bill", "doc": "monthly_bill_fin_056.pdf", "images": ["images/train/monthly_bill/monthly_bill_fin_056_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"monthly_bill\", \"confidence\": 1.0, \"issuer\": \"airtel\", \"holder\": \"Mr. Ajay Amrate\", \"page_marker\": \"\"}]}"}]} {"id": "monthly_bill_fin_056_extract", "task": "extract", "class": "monthly_bill", "doc": "monthly_bill_fin_056.pdf", "images": ["images/train/monthly_bill/monthly_bill_fin_056_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('monthly_bill', 'The document class.')\nfields = (['full_name', 'address', 'issue_date', 'company_name', 'account_number'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name on the bill', 'address': 'Address', 'issue_date': 'Issued date', 'company_name': 'Name of the issuing company', 'account_number': 'Account number'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"Mr. Ajay Amrate\"}, {\"field\": \"address\", \"value\": \"Flat No-201 1st Floor Vimal Nagar Pune Sholapur Road Nr-Jila Parishad School Pune 412307 Maharashtra\"}, {\"field\": \"issue_date\", \"value\": \"2019-04-21\"}, {\"field\": \"company_name\", \"value\": \"airtel\"}, {\"field\": \"account_number\", \"value\": \"8103588866\"}]}"}]} {"id": "monthly_bill_fin_057_p1_classify", "task": "classify", "class": "monthly_bill", "doc": "monthly_bill_fin_057.pdf", "images": ["images/train/monthly_bill/monthly_bill_fin_057_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"monthly_bill\", \"confidence\": 1.0, \"issuer\": \"airtel\", \"holder\": \"Soumyajit Sikdar\", \"page_marker\": \"\"}]}"}]} {"id": "monthly_bill_fin_057_extract", "task": "extract", "class": "monthly_bill", "doc": "monthly_bill_fin_057.pdf", "images": ["images/train/monthly_bill/monthly_bill_fin_057_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('monthly_bill', 'The document class.')\nfields = (['full_name', 'address', 'issue_date', 'company_name', 'account_number'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name on the bill', 'address': 'Address', 'issue_date': 'Issued date', 'company_name': 'Name of the issuing company', 'account_number': 'Account number'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"Soumyajit Sikdar\"}, {\"field\": \"address\", \"value\": \"\"}, {\"field\": \"issue_date\", \"value\": \"2022-04-27\"}, {\"field\": \"company_name\", \"value\": \"airtel\"}, {\"field\": \"account_number\", \"value\": \"10101013953128\"}]}"}]} {"id": "monthly_bill_fin_058_p1_classify", "task": "classify", "class": "monthly_bill", "doc": "monthly_bill_fin_058.pdf", "images": ["images/train/monthly_bill/monthly_bill_fin_058_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"monthly_bill\", \"confidence\": 1.0, \"issuer\": \"airtel\", \"holder\": \"Mr. Rajesh .\", \"page_marker\": \"\"}]}"}]} {"id": "monthly_bill_fin_058_extract", "task": "extract", "class": "monthly_bill", "doc": "monthly_bill_fin_058.pdf", "images": ["images/train/monthly_bill/monthly_bill_fin_058_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('monthly_bill', 'The document class.')\nfields = (['full_name', 'address', 'issue_date', 'company_name', 'account_number'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name on the bill', 'address': 'Address', 'issue_date': 'Issued date', 'company_name': 'Name of the issuing company', 'account_number': 'Account number'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"Mr. Rajesh .\"}, {\"field\": \"address\", \"value\": \"14/156 Dr Ambedkar nagar, Main Road Thiruvottiyur, Tiruvallur, Chennai 600019, Tamil Nadu\"}, {\"field\": \"issue_date\", \"value\": \"2020-03-03\"}, {\"field\": \"company_name\", \"value\": \"airtel\"}, {\"field\": \"account_number\", \"value\": \"1-2831808252803\"}]}"}]} {"id": "monthly_bill_fin_059_p1_classify", "task": "classify", "class": "monthly_bill", "doc": "monthly_bill_fin_059.pdf", "images": ["images/train/monthly_bill/monthly_bill_fin_059_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"monthly_bill\", \"confidence\": 1.0, \"issuer\": \"airtel\", \"holder\": \"Mr ASHWANTH PRATHAPANI\", \"page_marker\": \"\"}]}"}]} {"id": "monthly_bill_fin_059_extract", "task": "extract", "class": "monthly_bill", "doc": "monthly_bill_fin_059.pdf", "images": ["images/train/monthly_bill/monthly_bill_fin_059_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('monthly_bill', 'The document class.')\nfields = (['full_name', 'address', 'issue_date', 'company_name', 'account_number'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name on the bill', 'address': 'Address', 'issue_date': 'Issued date', 'company_name': 'Name of the issuing company', 'account_number': 'Account number'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"Mr ASHWANTH PRATHAPANI\"}, {\"field\": \"address\", \"value\": \"606 Media House Society Sector 47 Subhash Ma, Gurgaon, Haryana, 122001\"}, {\"field\": \"issue_date\", \"value\": \"2024-06-18\"}, {\"field\": \"company_name\", \"value\": \"airtel\"}, {\"field\": \"account_number\", \"value\": \"012445020195_dsl\"}]}"}]} {"id": "monthly_bill_fin_061_p1_classify", "task": "classify", "class": "monthly_bill", "doc": "monthly_bill_fin_061.pdf", "images": ["images/train/monthly_bill/monthly_bill_fin_061_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"monthly_bill\", \"confidence\": 1.0, \"issuer\": \"Bharti Airtel Limited\", \"holder\": \"Mr ANKIT AGGARWAL\", \"page_marker\": \"\"}]}"}]} {"id": "monthly_bill_fin_061_extract", "task": "extract", "class": "monthly_bill", "doc": "monthly_bill_fin_061.pdf", "images": ["images/train/monthly_bill/monthly_bill_fin_061_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('monthly_bill', 'The document class.')\nfields = (['full_name', 'address', 'issue_date', 'company_name', 'account_number'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name on the bill', 'address': 'Address', 'issue_date': 'Issued date', 'company_name': 'Name of the issuing company', 'account_number': 'Account number'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"Mr ANKIT AGGARWAL\"}, {\"field\": \"address\", \"value\": \"H No F 503 C-Wing Happy Home Residency, Mira Road E, Thane 401107, Maharashtra\"}, {\"field\": \"issue_date\", \"value\": \"2021-04-24\"}, {\"field\": \"company_name\", \"value\": \"Bharti Airtel Limited\"}, {\"field\": \"account_number\", \"value\": \"7038383160\"}]}"}]} {"id": "monthly_bill_fin_062_p1_classify", "task": "classify", "class": "monthly_bill", "doc": "monthly_bill_fin_062.pdf", "images": ["images/train/monthly_bill/monthly_bill_fin_062_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"monthly_bill\", \"confidence\": 1.0, \"issuer\": \"Setra\", \"holder\": \"Setra Systems\", \"page_marker\": \"\"}]}"}]} {"id": "monthly_bill_fin_062_extract", "task": "extract", "class": "monthly_bill", "doc": "monthly_bill_fin_062.pdf", "images": ["images/train/monthly_bill/monthly_bill_fin_062_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('monthly_bill', 'The document class.')\nfields = (['full_name', 'address', 'issue_date', 'company_name', 'account_number'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name on the bill', 'address': 'Address', 'issue_date': 'Issued date', 'company_name': 'Name of the issuing company', 'account_number': 'Account number'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"Setra Systems\"}, {\"field\": \"address\", \"value\": \"159 Swason Road Boxbourough MA 01719\"}, {\"field\": \"issue_date\", \"value\": \"2016-11-21\"}, {\"field\": \"company_name\", \"value\": \"Setra\"}, {\"field\": \"account_number\", \"value\": \"007 000 9999 54321\"}]}"}]} {"id": "monthly_bill_fin_063_p1_classify", "task": "classify", "class": "monthly_bill", "doc": "monthly_bill_fin_063.pdf", "images": ["images/train/monthly_bill/monthly_bill_fin_063_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"monthly_bill\", \"confidence\": 1.0, \"issuer\": \"BHARTI AIRTEL SERVICES LTD.\", \"holder\": \"Mr Pankaj Azad\", \"page_marker\": \"\"}]}"}]} {"id": "monthly_bill_fin_063_extract", "task": "extract", "class": "monthly_bill", "doc": "monthly_bill_fin_063.pdf", "images": ["images/train/monthly_bill/monthly_bill_fin_063_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('monthly_bill', 'The document class.')\nfields = (['full_name', 'address', 'issue_date', 'company_name', 'account_number'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name on the bill', 'address': 'Address', 'issue_date': 'Issued date', 'company_name': 'Name of the issuing company', 'account_number': 'Account number'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"Mr Pankaj Azad\"}, {\"field\": \"address\", \"value\": \"BPTP Crest Plot No-15, 5th Floor Udyog Vihar, Phase-4 Near Airtel Building, Gurgaon 122015, Haryana\"}, {\"field\": \"issue_date\", \"value\": \"2019-02-27\"}, {\"field\": \"company_name\", \"value\": \"BHARTI AIRTEL SERVICES LTD.\"}, {\"field\": \"account_number\", \"value\": \"8800166833\"}]}"}]} {"id": "monthly_bill_fin_064_p1_classify", "task": "classify", "class": "monthly_bill", "doc": "monthly_bill_fin_064.pdf", "images": ["images/train/monthly_bill/monthly_bill_fin_064_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"monthly_bill\", \"confidence\": 1.0, \"issuer\": \"AT&T\", \"holder\": \"VICTOR BROOKS\", \"page_marker\": \"\"}]}"}]} {"id": "monthly_bill_fin_064_extract", "task": "extract", "class": "monthly_bill", "doc": "monthly_bill_fin_064.pdf", "images": ["images/train/monthly_bill/monthly_bill_fin_064_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('monthly_bill', 'The document class.')\nfields = (['full_name', 'address', 'issue_date', 'company_name', 'account_number'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name on the bill', 'address': 'Address', 'issue_date': 'Issued date', 'company_name': 'Name of the issuing company', 'account_number': 'Account number'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"VICTOR BROOKS\"}, {\"field\": \"address\", \"value\": \"513 CLARK ST, EATONVILLE FL, 32751\"}, {\"field\": \"issue_date\", \"value\": \"\"}, {\"field\": \"company_name\", \"value\": \"AT&T\"}, {\"field\": \"account_number\", \"value\": \"158237304723\"}]}"}]} {"id": "monthly_bill_fin_065_p1_classify", "task": "classify", "class": "monthly_bill", "doc": "monthly_bill_fin_065.pdf", "images": ["images/train/monthly_bill/monthly_bill_fin_065_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"monthly_bill\", \"confidence\": 1.0, \"issuer\": \"AT&T\", \"holder\": \"JEANETTA MCCARTER\", \"page_marker\": \"\"}]}"}]} {"id": "monthly_bill_fin_065_extract", "task": "extract", "class": "monthly_bill", "doc": "monthly_bill_fin_065.pdf", "images": ["images/train/monthly_bill/monthly_bill_fin_065_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('monthly_bill', 'The document class.')\nfields = (['full_name', 'address', 'issue_date', 'company_name', 'account_number'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name on the bill', 'address': 'Address', 'issue_date': 'Issued date', 'company_name': 'Name of the issuing company', 'account_number': 'Account number'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"JEANETTA MCCARTER\"}, {\"field\": \"address\", \"value\": \"325 EDGEWOOD LN TRACY CA 95376\"}, {\"field\": \"issue_date\", \"value\": \"2022-11-26\"}, {\"field\": \"company_name\", \"value\": \"AT&T\"}, {\"field\": \"account_number\", \"value\": \"141626444\"}]}"}]} {"id": "monthly_bill_fin_067_p1_classify", "task": "classify", "class": "monthly_bill", "doc": "monthly_bill_fin_067.pdf", "images": ["images/train/monthly_bill/monthly_bill_fin_067_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"monthly_bill\", \"confidence\": 1.0, \"issuer\": \"at&t\", \"holder\": \"GWENDOLYN JOAN JOHNSON\", \"page_marker\": \"\"}]}"}]} {"id": "monthly_bill_fin_067_extract", "task": "extract", "class": "monthly_bill", "doc": "monthly_bill_fin_067.pdf", "images": ["images/train/monthly_bill/monthly_bill_fin_067_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('monthly_bill', 'The document class.')\nfields = (['full_name', 'address', 'issue_date', 'company_name', 'account_number'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name on the bill', 'address': 'Address', 'issue_date': 'Issued date', 'company_name': 'Name of the issuing company', 'account_number': 'Account number'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"GWENDOLYN JOAN JOHNSON\"}, {\"field\": \"address\", \"value\": \"2307 DELRIO DR, RICHMOND VA 23223\"}, {\"field\": \"issue_date\", \"value\": \"2018-04-01\"}, {\"field\": \"company_name\", \"value\": \"at&t\"}, {\"field\": \"account_number\", \"value\": \"9010454-4846670399\"}]}"}]} {"id": "monthly_bill_fin_068_p1_classify", "task": "classify", "class": "monthly_bill", "doc": "monthly_bill_fin_068.pdf", "images": ["images/train/monthly_bill/monthly_bill_fin_068_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"monthly_bill\", \"confidence\": 1.0, \"issuer\": \"AT&T\", \"holder\": \"CUSTOMER NAME\", \"page_marker\": \"\"}]}"}]} {"id": "monthly_bill_fin_068_extract", "task": "extract", "class": "monthly_bill", "doc": "monthly_bill_fin_068.pdf", "images": ["images/train/monthly_bill/monthly_bill_fin_068_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('monthly_bill', 'The document class.')\nfields = (['full_name', 'address', 'issue_date', 'company_name', 'account_number'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name on the bill', 'address': 'Address', 'issue_date': 'Issued date', 'company_name': 'Name of the issuing company', 'account_number': 'Account number'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"CUSTOMER NAME\"}, {\"field\": \"address\", \"value\": \"1234 TELEPHONE LN ANY CITY ST 99999-1234\"}, {\"field\": \"issue_date\", \"value\": \"\"}, {\"field\": \"company_name\", \"value\": \"AT&T\"}, {\"field\": \"account_number\", \"value\": \"123456789101\"}]}"}]} {"id": "monthly_bill_fin_069_p1_classify", "task": "classify", "class": "monthly_bill", "doc": "monthly_bill_fin_069.pdf", "images": ["images/train/monthly_bill/monthly_bill_fin_069_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"monthly_bill\", \"confidence\": 1.0, \"issuer\": \"AT&T Authorized Retailer Connect Wireless\", \"holder\": \"CAROLE ONEILL\", \"page_marker\": \"\"}]}"}]} {"id": "monthly_bill_fin_069_extract", "task": "extract", "class": "monthly_bill", "doc": "monthly_bill_fin_069.pdf", "images": ["images/train/monthly_bill/monthly_bill_fin_069_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('monthly_bill', 'The document class.')\nfields = (['full_name', 'address', 'issue_date', 'company_name', 'account_number'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name on the bill', 'address': 'Address', 'issue_date': 'Issued date', 'company_name': 'Name of the issuing company', 'account_number': 'Account number'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"CAROLE ONEILL\"}, {\"field\": \"address\", \"value\": \"29853 E 165TH AVE, BRIGHTON CO United States 80603\"}, {\"field\": \"issue_date\", \"value\": \"2020-01-20\"}, {\"field\": \"company_name\", \"value\": \"AT&T Authorized Retailer Connect Wireless\"}, {\"field\": \"account_number\", \"value\": \"COBRIIN20171\"}]}"}]} {"id": "monthly_bill_fin_071_p1_classify", "task": "classify", "class": "monthly_bill", "doc": "monthly_bill_fin_071.pdf", "images": ["images/train/monthly_bill/monthly_bill_fin_071_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"monthly_bill\", \"confidence\": 1.0, \"issuer\": \"AT&T\", \"holder\": \"JUAN J MORILLO\", \"page_marker\": \"\"}]}"}]} {"id": "monthly_bill_fin_071_extract", "task": "extract", "class": "monthly_bill", "doc": "monthly_bill_fin_071.pdf", "images": ["images/train/monthly_bill/monthly_bill_fin_071_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('monthly_bill', 'The document class.')\nfields = (['full_name', 'address', 'issue_date', 'company_name', 'account_number'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name on the bill', 'address': 'Address', 'issue_date': 'Issued date', 'company_name': 'Name of the issuing company', 'account_number': 'Account number'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"JUAN J MORILLO\"}, {\"field\": \"address\", \"value\": \"9841 EDMONTON DR LAND O LAKES FLORIDA 34638\"}, {\"field\": \"issue_date\", \"value\": \"\"}, {\"field\": \"company_name\", \"value\": \"AT&T\"}, {\"field\": \"account_number\", \"value\": \"727492716\"}]}"}]} {"id": "monthly_bill_fin_072_p1_classify", "task": "classify", "class": "monthly_bill", "doc": "monthly_bill_fin_072.pdf", "images": ["images/train/monthly_bill/monthly_bill_fin_072_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"monthly_bill\", \"confidence\": 1.0, \"issuer\": \"British Gas\", \"holder\": \"Mr A Sample\", \"page_marker\": \"\"}]}"}]} {"id": "monthly_bill_fin_072_extract", "task": "extract", "class": "monthly_bill", "doc": "monthly_bill_fin_072.pdf", "images": ["images/train/monthly_bill/monthly_bill_fin_072_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('monthly_bill', 'The document class.')\nfields = (['full_name', 'address', 'issue_date', 'company_name', 'account_number'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name on the bill', 'address': 'Address', 'issue_date': 'Issued date', 'company_name': 'Name of the issuing company', 'account_number': 'Account number'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"Mr A Sample\"}, {\"field\": \"address\", \"value\": \"2 Sample Street, Sampletown, Sampleshire, AB12 3CD\"}, {\"field\": \"issue_date\", \"value\": \"2010-03-04\"}, {\"field\": \"company_name\", \"value\": \"British Gas\"}, {\"field\": \"account_number\", \"value\": \"1234 1234 1234\"}]}"}]} {"id": "monthly_bill_fin_073_p1_classify", "task": "classify", "class": "monthly_bill", "doc": "monthly_bill_fin_073.pdf", "images": ["images/train/monthly_bill/monthly_bill_fin_073_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"monthly_bill\", \"confidence\": 1.0, \"issuer\": \"Verizon\", \"holder\": \"SANDRA KELLY\", \"page_marker\": \"\"}]}"}]} {"id": "monthly_bill_fin_073_extract", "task": "extract", "class": "monthly_bill", "doc": "monthly_bill_fin_073.pdf", "images": ["images/train/monthly_bill/monthly_bill_fin_073_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('monthly_bill', 'The document class.')\nfields = (['full_name', 'address', 'issue_date', 'company_name', 'account_number'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name on the bill', 'address': 'Address', 'issue_date': 'Issued date', 'company_name': 'Name of the issuing company', 'account_number': 'Account number'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"SANDRA KELLY\"}, {\"field\": \"address\", \"value\": \"13424 TANGIER PL, ROCKVILLE MD 20853\"}, {\"field\": \"issue_date\", \"value\": \"2022-11-26\"}, {\"field\": \"company_name\", \"value\": \"Verizon\"}, {\"field\": \"account_number\", \"value\": \"165269405-0311\"}]}"}]} {"id": "monthly_bill_fin_074_p1_classify", "task": "classify", "class": "monthly_bill", "doc": "monthly_bill_fin_074.pdf", "images": ["images/train/monthly_bill/monthly_bill_fin_074_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"monthly_bill\", \"confidence\": 1.0, \"issuer\": \"verizon\", \"holder\": \"CARLOS A. NASCIMENTO\", \"page_marker\": \"\"}]}"}]} {"id": "monthly_bill_fin_074_extract", "task": "extract", "class": "monthly_bill", "doc": "monthly_bill_fin_074.pdf", "images": ["images/train/monthly_bill/monthly_bill_fin_074_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('monthly_bill', 'The document class.')\nfields = (['full_name', 'address', 'issue_date', 'company_name', 'account_number'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name on the bill', 'address': 'Address', 'issue_date': 'Issued date', 'company_name': 'Name of the issuing company', 'account_number': 'Account number'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"CARLOS A. NASCIMENTO\"}, {\"field\": \"address\", \"value\": \"43 MANCHESTER RD, FLEMINGTON, NJ 08822-6823\"}, {\"field\": \"issue_date\", \"value\": \"2025-03-09\"}, {\"field\": \"company_name\", \"value\": \"verizon\"}, {\"field\": \"account_number\", \"value\": \"790163839-00001\"}]}"}]} {"id": "monthly_bill_fin_077_p1_classify", "task": "classify", "class": "monthly_bill", "doc": "monthly_bill_fin_077.pdf", "images": ["images/train/monthly_bill/monthly_bill_fin_077_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"monthly_bill\", \"confidence\": 1.0, \"issuer\": \"verizon\", \"holder\": \"AMANDA L LERMA\", \"page_marker\": \"\"}]}"}]} {"id": "monthly_bill_fin_077_extract", "task": "extract", "class": "monthly_bill", "doc": "monthly_bill_fin_077.pdf", "images": ["images/train/monthly_bill/monthly_bill_fin_077_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('monthly_bill', 'The document class.')\nfields = (['full_name', 'address', 'issue_date', 'company_name', 'account_number'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name on the bill', 'address': 'Address', 'issue_date': 'Issued date', 'company_name': 'Name of the issuing company', 'account_number': 'Account number'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"AMANDA L LERMA\"}, {\"field\": \"address\", \"value\": \"1123 E POLK AV, PHARR, TX 78577-3313\"}, {\"field\": \"issue_date\", \"value\": \"\"}, {\"field\": \"company_name\", \"value\": \"verizon\"}, {\"field\": \"account_number\", \"value\": \"326341865-00001\"}]}"}]} {"id": "monthly_bill_fin_079_p1_classify", "task": "classify", "class": "monthly_bill", "doc": "monthly_bill_fin_079.pdf", "images": ["images/train/monthly_bill/monthly_bill_fin_079_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"monthly_bill\", \"confidence\": 1.0, \"issuer\": \"verizon\", \"holder\": \"MEDHAT BESHAI\", \"page_marker\": \"\"}]}"}]} {"id": "monthly_bill_fin_079_extract", "task": "extract", "class": "monthly_bill", "doc": "monthly_bill_fin_079.pdf", "images": ["images/train/monthly_bill/monthly_bill_fin_079_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('monthly_bill', 'The document class.')\nfields = (['full_name', 'address', 'issue_date', 'company_name', 'account_number'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name on the bill', 'address': 'Address', 'issue_date': 'Issued date', 'company_name': 'Name of the issuing company', 'account_number': 'Account number'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"MEDHAT BESHAI\"}, {\"field\": \"address\", \"value\": \"382 COUNTRY CLUB DR UNIT D, SIMI VALLEY, CA 93065-7653\"}, {\"field\": \"issue_date\", \"value\": \"\"}, {\"field\": \"company_name\", \"value\": \"verizon\"}, {\"field\": \"account_number\", \"value\": \"970386930-00002\"}]}"}]} {"id": "monthly_bill_fin_080_p1_classify", "task": "classify", "class": "monthly_bill", "doc": "monthly_bill_fin_080.pdf", "images": ["images/train/monthly_bill/monthly_bill_fin_080_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"monthly_bill\", \"confidence\": 1.0, \"issuer\": \"verizon\", \"holder\": \"FIRST NAME LAST NAME\", \"page_marker\": \"\"}]}"}]} {"id": "monthly_bill_fin_080_extract", "task": "extract", "class": "monthly_bill", "doc": "monthly_bill_fin_080.pdf", "images": ["images/train/monthly_bill/monthly_bill_fin_080_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('monthly_bill', 'The document class.')\nfields = (['full_name', 'address', 'issue_date', 'company_name', 'account_number'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name on the bill', 'address': 'Address', 'issue_date': 'Issued date', 'company_name': 'Name of the issuing company', 'account_number': 'Account number'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"FIRST NAME LAST NAME\"}, {\"field\": \"address\", \"value\": \"ADRESS, CITY, CA 99999\"}, {\"field\": \"issue_date\", \"value\": \"\"}, {\"field\": \"company_name\", \"value\": \"verizon\"}, {\"field\": \"account_number\", \"value\": \"999999999 - 00001\"}]}"}]} {"id": "monthly_bill_fin_082_p1_classify", "task": "classify", "class": "monthly_bill", "doc": "monthly_bill_fin_082.pdf", "images": ["images/train/monthly_bill/monthly_bill_fin_082_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"monthly_bill\", \"confidence\": 1.0, \"issuer\": \"Spectrum\", \"holder\": \"JACOB WALLACE\", \"page_marker\": \"\"}]}"}]} {"id": "monthly_bill_fin_082_extract", "task": "extract", "class": "monthly_bill", "doc": "monthly_bill_fin_082.pdf", "images": ["images/train/monthly_bill/monthly_bill_fin_082_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('monthly_bill', 'The document class.')\nfields = (['full_name', 'address', 'issue_date', 'company_name', 'account_number'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name on the bill', 'address': 'Address', 'issue_date': 'Issued date', 'company_name': 'Name of the issuing company', 'account_number': 'Account number'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"JACOB WALLACE\"}, {\"field\": \"address\", \"value\": \"948 JEAN AVE, AKRON, OH 44310\"}, {\"field\": \"issue_date\", \"value\": \"2020-12-25\"}, {\"field\": \"company_name\", \"value\": \"Spectrum\"}, {\"field\": \"account_number\", \"value\": \"8260 19 044 4852144\"}]}"}]} {"id": "monthly_bill_fin_083_p1_classify", "task": "classify", "class": "monthly_bill", "doc": "monthly_bill_fin_083.pdf", "images": ["images/train/monthly_bill/monthly_bill_fin_083_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"monthly_bill\", \"confidence\": 1.0, \"issuer\": \"\", \"holder\": \"J Smith\", \"page_marker\": \"\"}]}"}]} {"id": "monthly_bill_fin_083_extract", "task": "extract", "class": "monthly_bill", "doc": "monthly_bill_fin_083.pdf", "images": ["images/train/monthly_bill/monthly_bill_fin_083_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('monthly_bill', 'The document class.')\nfields = (['full_name', 'address', 'issue_date', 'company_name', 'account_number'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name on the bill', 'address': 'Address', 'issue_date': 'Issued date', 'company_name': 'Name of the issuing company', 'account_number': 'Account number'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"J Smith\"}, {\"field\": \"address\", \"value\": \"123 Sample Rd, Sampletown NSW 2000\"}, {\"field\": \"issue_date\", \"value\": \"2019-06-27\"}, {\"field\": \"company_name\", \"value\": \"\"}, {\"field\": \"account_number\", \"value\": \"2222 2222 2222\"}]}"}]} {"id": "monthly_bill_fin_084_p1_classify", "task": "classify", "class": "monthly_bill", "doc": "monthly_bill_fin_084.pdf", "images": ["images/train/monthly_bill/monthly_bill_fin_084_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"monthly_bill\", \"confidence\": 1.0, \"issuer\": \"Spectrum\", \"holder\": \"Maria Rodriguez\", \"page_marker\": \"\"}]}"}]} {"id": "monthly_bill_fin_084_extract", "task": "extract", "class": "monthly_bill", "doc": "monthly_bill_fin_084.pdf", "images": ["images/train/monthly_bill/monthly_bill_fin_084_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('monthly_bill', 'The document class.')\nfields = (['full_name', 'address', 'issue_date', 'company_name', 'account_number'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name on the bill', 'address': 'Address', 'issue_date': 'Issued date', 'company_name': 'Name of the issuing company', 'account_number': 'Account number'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"Maria Rodriguez\"}, {\"field\": \"address\", \"value\": \"219 45TH ST # 1 BROOKLYN NY 11220-1005\"}, {\"field\": \"issue_date\", \"value\": \"2018-01-04\"}, {\"field\": \"company_name\", \"value\": \"Spectrum\"}, {\"field\": \"account_number\", \"value\": \"8150 13 004 1221132\"}]}"}]} {"id": "monthly_bill_fin_085_p1_classify", "task": "classify", "class": "monthly_bill", "doc": "monthly_bill_fin_085.pdf", "images": ["images/train/monthly_bill/monthly_bill_fin_085_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"monthly_bill\", \"confidence\": 1.0, \"issuer\": \"Spectrum\", \"holder\": \"\", \"page_marker\": \"\"}]}"}]} {"id": "monthly_bill_fin_085_extract", "task": "extract", "class": "monthly_bill", "doc": "monthly_bill_fin_085.pdf", "images": ["images/train/monthly_bill/monthly_bill_fin_085_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('monthly_bill', 'The document class.')\nfields = (['full_name', 'address', 'issue_date', 'company_name', 'account_number'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name on the bill', 'address': 'Address', 'issue_date': 'Issued date', 'company_name': 'Name of the issuing company', 'account_number': 'Account number'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"\"}, {\"field\": \"address\", \"value\": \"\"}, {\"field\": \"issue_date\", \"value\": \"2020-10-11\"}, {\"field\": \"company_name\", \"value\": \"Spectrum\"}, {\"field\": \"account_number\", \"value\": \"\"}]}"}]} {"id": "monthly_bill_fin_086_p1_classify", "task": "classify", "class": "monthly_bill", "doc": "monthly_bill_fin_086.pdf", "images": ["images/train/monthly_bill/monthly_bill_fin_086_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"monthly_bill\", \"confidence\": 1.0, \"issuer\": \"airtel\", \"holder\": \"Mr ROHIT ANAND SINGH\", \"page_marker\": \"\"}]}"}]} {"id": "monthly_bill_fin_086_extract", "task": "extract", "class": "monthly_bill", "doc": "monthly_bill_fin_086.pdf", "images": ["images/train/monthly_bill/monthly_bill_fin_086_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('monthly_bill', 'The document class.')\nfields = (['full_name', 'address', 'issue_date', 'company_name', 'account_number'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name on the bill', 'address': 'Address', 'issue_date': 'Issued date', 'company_name': 'Name of the issuing company', 'account_number': 'Account number'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"Mr ROHIT ANAND SINGH\"}, {\"field\": \"address\", \"value\": \"HN0 919 SEC10A, Gurgaon, Haryana 122001\"}, {\"field\": \"issue_date\", \"value\": \"2017-11-24\"}, {\"field\": \"company_name\", \"value\": \"airtel\"}, {\"field\": \"account_number\", \"value\": \"7031321775\"}]}"}]} {"id": "monthly_bill_fin_087_p1_classify", "task": "classify", "class": "monthly_bill", "doc": "monthly_bill_fin_087.pdf", "images": ["images/train/monthly_bill/monthly_bill_fin_087_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"monthly_bill\", \"confidence\": 1.0, \"issuer\": \"Comcast\", \"holder\": \"Timothy Bradshaw\", \"page_marker\": \"\"}]}"}]} {"id": "monthly_bill_fin_087_extract", "task": "extract", "class": "monthly_bill", "doc": "monthly_bill_fin_087.pdf", "images": ["images/train/monthly_bill/monthly_bill_fin_087_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('monthly_bill', 'The document class.')\nfields = (['full_name', 'address', 'issue_date', 'company_name', 'account_number'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name on the bill', 'address': 'Address', 'issue_date': 'Issued date', 'company_name': 'Name of the issuing company', 'account_number': 'Account number'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"Timothy Bradshaw\"}, {\"field\": \"address\", \"value\": \"392 Weeks Landing Rd, Cameron, SC 29030\"}, {\"field\": \"issue_date\", \"value\": \"2018-07-02\"}, {\"field\": \"company_name\", \"value\": \"Comcast\"}, {\"field\": \"account_number\", \"value\": \"3022 18 400 02084702\"}]}"}]} {"id": "monthly_bill_fin_088_p1_classify", "task": "classify", "class": "monthly_bill", "doc": "monthly_bill_fin_088.pdf", "images": ["images/train/monthly_bill/monthly_bill_fin_088_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"monthly_bill\", \"confidence\": 1.0, \"issuer\": \"ATRIA CONVERGENCE TECHNOLOGIES LIMITED.\", \"holder\": \"AMARNATH RAMARAJ\", \"page_marker\": \"\"}]}"}]} {"id": "monthly_bill_fin_088_extract", "task": "extract", "class": "monthly_bill", "doc": "monthly_bill_fin_088.pdf", "images": ["images/train/monthly_bill/monthly_bill_fin_088_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('monthly_bill', 'The document class.')\nfields = (['full_name', 'address', 'issue_date', 'company_name', 'account_number'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name on the bill', 'address': 'Address', 'issue_date': 'Issued date', 'company_name': 'Name of the issuing company', 'account_number': 'Account number'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"AMARNATH RAMARAJ\"}, {\"field\": \"address\", \"value\": \"L504 ITIINA SOUPERNIKA ,SARJAPUR MAIN ROAD KASAVANAHALLI, Bangalore, Karnataka, India, Zip Code: 560035\"}, {\"field\": \"issue_date\", \"value\": \"2019-04-01\"}, {\"field\": \"company_name\", \"value\": \"ATRIA CONVERGENCE TECHNOLOGIES LIMITED.\"}, {\"field\": \"account_number\", \"value\": \"102009889530\"}]}"}]} {"id": "monthly_bill_fin_090_p1_classify", "task": "classify", "class": "monthly_bill", "doc": "monthly_bill_fin_090.pdf", "images": ["images/train/monthly_bill/monthly_bill_fin_090_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"monthly_bill\", \"confidence\": 1.0, \"issuer\": \"Comcast\", \"holder\": \"Edmond Seeling\", \"page_marker\": \"\"}]}"}]} {"id": "monthly_bill_fin_090_extract", "task": "extract", "class": "monthly_bill", "doc": "monthly_bill_fin_090.pdf", "images": ["images/train/monthly_bill/monthly_bill_fin_090_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('monthly_bill', 'The document class.')\nfields = (['full_name', 'address', 'issue_date', 'company_name', 'account_number'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name on the bill', 'address': 'Address', 'issue_date': 'Issued date', 'company_name': 'Name of the issuing company', 'account_number': 'Account number'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"Edmond Seeling\"}, {\"field\": \"address\", \"value\": \"10307 72ND AVE NW, GIG HARBOR WA 98332-8578\"}, {\"field\": \"issue_date\", \"value\": \"2011-04-08\"}, {\"field\": \"company_name\", \"value\": \"Comcast\"}, {\"field\": \"account_number\", \"value\": \"8498 36 010 0053354\"}]}"}]} {"id": "monthly_bill_fin_091_p1_classify", "task": "classify", "class": "monthly_bill", "doc": "monthly_bill_fin_091.pdf", "images": ["images/train/monthly_bill/monthly_bill_fin_091_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"monthly_bill\", \"confidence\": 1.0, \"issuer\": \"Comcast\", \"holder\": \"Liliana Gilo\", \"page_marker\": \"\"}]}"}]} {"id": "monthly_bill_fin_091_extract", "task": "extract", "class": "monthly_bill", "doc": "monthly_bill_fin_091.pdf", "images": ["images/train/monthly_bill/monthly_bill_fin_091_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('monthly_bill', 'The document class.')\nfields = (['full_name', 'address', 'issue_date', 'company_name', 'account_number'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name on the bill', 'address': 'Address', 'issue_date': 'Issued date', 'company_name': 'Name of the issuing company', 'account_number': 'Account number'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"Liliana Gilo\"}, {\"field\": \"address\", \"value\": \"53 CASSIA ST PITTSBURG CA 94565-7244\"}, {\"field\": \"issue_date\", \"value\": \"2014-11-03\"}, {\"field\": \"company_name\", \"value\": \"Comcast\"}, {\"field\": \"account_number\", \"value\": \"8155 40 032 0793613\"}]}"}]} {"id": "monthly_bill_fin_093_p1_classify", "task": "classify", "class": "monthly_bill", "doc": "monthly_bill_fin_093.pdf", "images": ["images/train/monthly_bill/monthly_bill_fin_093_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"monthly_bill\", \"confidence\": 1.0, \"issuer\": \"Belluomini Farms, Inc.\", \"holder\": \"\", \"page_marker\": \"\"}]}"}]} {"id": "monthly_bill_fin_093_extract", "task": "extract", "class": "monthly_bill", "doc": "monthly_bill_fin_093.pdf", "images": ["images/train/monthly_bill/monthly_bill_fin_093_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('monthly_bill', 'The document class.')\nfields = (['full_name', 'address', 'issue_date', 'company_name', 'account_number'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name on the bill', 'address': 'Address', 'issue_date': 'Issued date', 'company_name': 'Name of the issuing company', 'account_number': 'Account number'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"\"}, {\"field\": \"address\", \"value\": \"1542 N FAY AVE, FRESNO CA 93728-1312\"}, {\"field\": \"issue_date\", \"value\": \"2025-09-17\"}, {\"field\": \"company_name\", \"value\": \"Belluomini Farms, Inc.\"}, {\"field\": \"account_number\", \"value\": \"8155 50 053 0585681\"}]}"}]} {"id": "paystub_fin_001_p1_classify", "task": "classify", "class": "paystub", "doc": "paystub_fin_001.pdf", "images": ["images/train/paystub/paystub_fin_001_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"paystub\", \"confidence\": 1.0, \"issuer\": \"\", \"holder\": \"Last Name, First\", \"page_marker\": \"\"}]}"}]} {"id": "paystub_fin_001_extract", "task": "extract", "class": "paystub", "doc": "paystub_fin_001.pdf", "images": ["images/train/paystub/paystub_fin_001_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('paystub', 'The document class.')\nfields = (['pay_date', 'employee_name', 'employer_name', 'gross_pay', 'ytd_gross', 'pay_frequency', 'pay_period_start', 'pay_period_end', 'federal_tax', 'state_tax', 'social_security_tax', 'medicare_tax', 'hire_date', 'direct_deposit_bank_type'], 'The exact field names to extract.')\nhints = ({'pay_date': 'Date the paycheck was issued', 'employee_name': 'Name of Employee', 'employer_name': 'Name of Employer', 'gross_pay': 'Gross pay per period (current period, not YTD)', 'ytd_gross': 'Gross pay on paystub year to date (cumulative, not current period)', 'pay_frequency': 'The frequency if provided: weekly, biweekly, monthly, semi-monthly', 'pay_period_start': 'Pay period start date', 'pay_period_end': 'Pay period end date', 'hire_date': 'Job hire date', 'direct_deposit_bank_type': 'Direct deposit bank type'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"pay_date\", \"value\": \"2023-05-04\"}, {\"field\": \"employee_name\", \"value\": \"Last Name, First\"}, {\"field\": \"employer_name\", \"value\": \"\"}, {\"field\": \"gross_pay\", \"value\": \"5000.0\"}, {\"field\": \"ytd_gross\", \"value\": \"20000.0\"}, {\"field\": \"pay_frequency\", \"value\": \"\"}, {\"field\": \"pay_period_start\", \"value\": \"2023-04-01\"}, {\"field\": \"pay_period_end\", \"value\": \"2023-05-01\"}, {\"field\": \"federal_tax\", \"value\": \"600.0\"}, {\"field\": \"state_tax\", \"value\": \"225.0\"}, {\"field\": \"social_security_tax\", \"value\": \"310.0\"}, {\"field\": \"medicare_tax\", \"value\": \"65.0\"}, {\"field\": \"hire_date\", \"value\": \"\"}, {\"field\": \"direct_deposit_bank_type\", \"value\": \"\"}]}"}]} {"id": "paystub_fin_002_p1_classify", "task": "classify", "class": "paystub", "doc": "paystub_fin_002.pdf", "images": ["images/train/paystub/paystub_fin_002_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"paystub\", \"confidence\": 1.0, \"issuer\": \"Acme LLC\", \"holder\": \"John Smith\", \"page_marker\": \"\"}]}"}]} {"id": "paystub_fin_002_extract", "task": "extract", "class": "paystub", "doc": "paystub_fin_002.pdf", "images": ["images/train/paystub/paystub_fin_002_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('paystub', 'The document class.')\nfields = (['pay_date', 'employee_name', 'employer_name', 'gross_pay', 'ytd_gross', 'pay_frequency', 'pay_period_start', 'pay_period_end', 'federal_tax', 'state_tax', 'social_security_tax', 'medicare_tax', 'hire_date', 'direct_deposit_bank_type'], 'The exact field names to extract.')\nhints = ({'pay_date': 'Date the paycheck was issued', 'employee_name': 'Name of Employee', 'employer_name': 'Name of Employer', 'gross_pay': 'Gross pay per period (current period, not YTD)', 'ytd_gross': 'Gross pay on paystub year to date (cumulative, not current period)', 'pay_frequency': 'The frequency if provided: weekly, biweekly, monthly, semi-monthly', 'pay_period_start': 'Pay period start date', 'pay_period_end': 'Pay period end date', 'hire_date': 'Job hire date', 'direct_deposit_bank_type': 'Direct deposit bank type'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"pay_date\", \"value\": \"2017-10-15\"}, {\"field\": \"employee_name\", \"value\": \"John Smith\"}, {\"field\": \"employer_name\", \"value\": \"Acme LLC\"}, {\"field\": \"gross_pay\", \"value\": \"1.0\"}, {\"field\": \"ytd_gross\", \"value\": \"3000.0\"}, {\"field\": \"pay_frequency\", \"value\": \"\"}, {\"field\": \"pay_period_start\", \"value\": \"\"}, {\"field\": \"pay_period_end\", \"value\": \"2017-09-28\"}, {\"field\": \"federal_tax\", \"value\": \"289.3\"}, {\"field\": \"state_tax\", \"value\": \"15.0\"}, {\"field\": \"social_security_tax\", \"value\": \"50.0\"}, {\"field\": \"medicare_tax\", \"value\": \"75.1\"}, {\"field\": \"hire_date\", \"value\": \"\"}, {\"field\": \"direct_deposit_bank_type\", \"value\": \"\"}]}"}]} {"id": "paystub_fin_003_p1_classify", "task": "classify", "class": "paystub", "doc": "paystub_fin_003.pdf", "images": ["images/train/paystub/paystub_fin_003_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"paystub\", \"confidence\": 1.0, \"issuer\": \"Company Name\", \"holder\": \"Employee Name\", \"page_marker\": \"\"}]}"}]} {"id": "paystub_fin_003_extract", "task": "extract", "class": "paystub", "doc": "paystub_fin_003.pdf", "images": ["images/train/paystub/paystub_fin_003_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('paystub', 'The document class.')\nfields = (['pay_date', 'employee_name', 'employer_name', 'gross_pay', 'ytd_gross', 'pay_frequency', 'pay_period_start', 'pay_period_end', 'federal_tax', 'state_tax', 'social_security_tax', 'medicare_tax', 'hire_date', 'direct_deposit_bank_type'], 'The exact field names to extract.')\nhints = ({'pay_date': 'Date the paycheck was issued', 'employee_name': 'Name of Employee', 'employer_name': 'Name of Employer', 'gross_pay': 'Gross pay per period (current period, not YTD)', 'ytd_gross': 'Gross pay on paystub year to date (cumulative, not current period)', 'pay_frequency': 'The frequency if provided: weekly, biweekly, monthly, semi-monthly', 'pay_period_start': 'Pay period start date', 'pay_period_end': 'Pay period end date', 'hire_date': 'Job hire date', 'direct_deposit_bank_type': 'Direct deposit bank type'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"pay_date\", \"value\": \"2023-04-04\"}, {\"field\": \"employee_name\", \"value\": \"Employee Name\"}, {\"field\": \"employer_name\", \"value\": \"Company Name\"}, {\"field\": \"gross_pay\", \"value\": \"16.0\"}, {\"field\": \"ytd_gross\", \"value\": \"1488.0\"}, {\"field\": \"pay_frequency\", \"value\": \"\"}, {\"field\": \"pay_period_start\", \"value\": \"2023-04-03\"}, {\"field\": \"pay_period_end\", \"value\": \"2023-04-03\"}, {\"field\": \"federal_tax\", \"value\": \"\"}, {\"field\": \"state_tax\", \"value\": \"\"}, {\"field\": \"social_security_tax\", \"value\": \"0.99\"}, {\"field\": \"medicare_tax\", \"value\": \"0.23\"}, {\"field\": \"hire_date\", \"value\": \"\"}, {\"field\": \"direct_deposit_bank_type\", \"value\": \"\"}]}"}]} {"id": "paystub_fin_004_p1_classify", "task": "classify", "class": "paystub", "doc": "paystub_fin_004.pdf", "images": ["images/train/paystub/paystub_fin_004_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"paystub\", \"confidence\": 1.0, \"issuer\": \"Company Name\", \"holder\": \"Employee Name\", \"page_marker\": \"\"}]}"}]} {"id": "paystub_fin_004_extract", "task": "extract", "class": "paystub", "doc": "paystub_fin_004.pdf", "images": ["images/train/paystub/paystub_fin_004_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('paystub', 'The document class.')\nfields = (['pay_date', 'employee_name', 'employer_name', 'gross_pay', 'ytd_gross', 'pay_frequency', 'pay_period_start', 'pay_period_end', 'federal_tax', 'state_tax', 'social_security_tax', 'medicare_tax', 'hire_date', 'direct_deposit_bank_type'], 'The exact field names to extract.')\nhints = ({'pay_date': 'Date the paycheck was issued', 'employee_name': 'Name of Employee', 'employer_name': 'Name of Employer', 'gross_pay': 'Gross pay per period (current period, not YTD)', 'ytd_gross': 'Gross pay on paystub year to date (cumulative, not current period)', 'pay_frequency': 'The frequency if provided: weekly, biweekly, monthly, semi-monthly', 'pay_period_start': 'Pay period start date', 'pay_period_end': 'Pay period end date', 'hire_date': 'Job hire date', 'direct_deposit_bank_type': 'Direct deposit bank type'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"pay_date\", \"value\": \"2023-04-04\"}, {\"field\": \"employee_name\", \"value\": \"Employee Name\"}, {\"field\": \"employer_name\", \"value\": \"Company Name\"}, {\"field\": \"gross_pay\", \"value\": \"16.0\"}, {\"field\": \"ytd_gross\", \"value\": \"1488.0\"}, {\"field\": \"pay_frequency\", \"value\": \"\"}, {\"field\": \"pay_period_start\", \"value\": \"2023-04-03\"}, {\"field\": \"pay_period_end\", \"value\": \"2023-04-03\"}, {\"field\": \"federal_tax\", \"value\": \"\"}, {\"field\": \"state_tax\", \"value\": \"\"}, {\"field\": \"social_security_tax\", \"value\": \"0.99\"}, {\"field\": \"medicare_tax\", \"value\": \"0.23\"}, {\"field\": \"hire_date\", \"value\": \"\"}, {\"field\": \"direct_deposit_bank_type\", \"value\": \"\"}]}"}]} {"id": "paystub_fin_005_p1_classify", "task": "classify", "class": "paystub", "doc": "paystub_fin_005.pdf", "images": ["images/train/paystub/paystub_fin_005_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"paystub\", \"confidence\": 1.0, \"issuer\": \"Lab Samples Inc\", \"holder\": \"Penelope Reed\", \"page_marker\": \"\"}]}"}]} {"id": "paystub_fin_005_extract", "task": "extract", "class": "paystub", "doc": "paystub_fin_005.pdf", "images": ["images/train/paystub/paystub_fin_005_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('paystub', 'The document class.')\nfields = (['pay_date', 'employee_name', 'employer_name', 'gross_pay', 'ytd_gross', 'pay_frequency', 'pay_period_start', 'pay_period_end', 'federal_tax', 'state_tax', 'social_security_tax', 'medicare_tax', 'hire_date', 'direct_deposit_bank_type'], 'The exact field names to extract.')\nhints = ({'pay_date': 'Date the paycheck was issued', 'employee_name': 'Name of Employee', 'employer_name': 'Name of Employer', 'gross_pay': 'Gross pay per period (current period, not YTD)', 'ytd_gross': 'Gross pay on paystub year to date (cumulative, not current period)', 'pay_frequency': 'The frequency if provided: weekly, biweekly, monthly, semi-monthly', 'pay_period_start': 'Pay period start date', 'pay_period_end': 'Pay period end date', 'hire_date': 'Job hire date', 'direct_deposit_bank_type': 'Direct deposit bank type'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"pay_date\", \"value\": \"\"}, {\"field\": \"employee_name\", \"value\": \"Penelope Reed\"}, {\"field\": \"employer_name\", \"value\": \"Lab Samples Inc\"}, {\"field\": \"gross_pay\", \"value\": \"7380.0\"}, {\"field\": \"ytd_gross\", \"value\": \"7380.0\"}, {\"field\": \"pay_frequency\", \"value\": \"\"}, {\"field\": \"pay_period_start\", \"value\": \"2024-02-01\"}, {\"field\": \"pay_period_end\", \"value\": \"2024-02-30\"}, {\"field\": \"federal_tax\", \"value\": \"12.0\"}, {\"field\": \"state_tax\", \"value\": \"10.0\"}, {\"field\": \"social_security_tax\", \"value\": \"15.0\"}, {\"field\": \"medicare_tax\", \"value\": \"\"}, {\"field\": \"hire_date\", \"value\": \"\"}, {\"field\": \"direct_deposit_bank_type\", \"value\": \"\"}]}"}]} {"id": "paystub_fin_006_p1_classify", "task": "classify", "class": "paystub", "doc": "paystub_fin_006.pdf", "images": ["images/train/paystub/paystub_fin_006_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"paystub\", \"confidence\": 1.0, \"issuer\": \"Good Behavior Stragies LLC\", \"holder\": \"Brandon Tyson\", \"page_marker\": \"\"}]}"}]} {"id": "paystub_fin_006_extract", "task": "extract", "class": "paystub", "doc": "paystub_fin_006.pdf", "images": ["images/train/paystub/paystub_fin_006_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('paystub', 'The document class.')\nfields = (['pay_date', 'employee_name', 'employer_name', 'gross_pay', 'ytd_gross', 'pay_frequency', 'pay_period_start', 'pay_period_end', 'federal_tax', 'state_tax', 'social_security_tax', 'medicare_tax', 'hire_date', 'direct_deposit_bank_type'], 'The exact field names to extract.')\nhints = ({'pay_date': 'Date the paycheck was issued', 'employee_name': 'Name of Employee', 'employer_name': 'Name of Employer', 'gross_pay': 'Gross pay per period (current period, not YTD)', 'ytd_gross': 'Gross pay on paystub year to date (cumulative, not current period)', 'pay_frequency': 'The frequency if provided: weekly, biweekly, monthly, semi-monthly', 'pay_period_start': 'Pay period start date', 'pay_period_end': 'Pay period end date', 'hire_date': 'Job hire date', 'direct_deposit_bank_type': 'Direct deposit bank type'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"pay_date\", \"value\": \"2024-08-31\"}, {\"field\": \"employee_name\", \"value\": \"Brandon Tyson\"}, {\"field\": \"employer_name\", \"value\": \"Good Behavior Stragies LLC\"}, {\"field\": \"gross_pay\", \"value\": \"2080.0\"}, {\"field\": \"ytd_gross\", \"value\": \"33280.0\"}, {\"field\": \"pay_frequency\", \"value\": \"\"}, {\"field\": \"pay_period_start\", \"value\": \"2024-07-23\"}, {\"field\": \"pay_period_end\", \"value\": \"2024-08-05\"}, {\"field\": \"federal_tax\", \"value\": \"173.3\"}, {\"field\": \"state_tax\", \"value\": \"59.13\"}, {\"field\": \"social_security_tax\", \"value\": \"128.96\"}, {\"field\": \"medicare_tax\", \"value\": \"30.16\"}, {\"field\": \"hire_date\", \"value\": \"\"}, {\"field\": \"direct_deposit_bank_type\", \"value\": \"\"}]}"}]} {"id": "paystub_fin_007_p1_classify", "task": "classify", "class": "paystub", "doc": "paystub_fin_007.pdf", "images": ["images/train/paystub/paystub_fin_007_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"paystub\", \"confidence\": 1.0, \"issuer\": \"Good Behavior Stragies LLC\", \"holder\": \"Brandon Tyson\", \"page_marker\": \"\"}]}"}]} {"id": "paystub_fin_007_extract", "task": "extract", "class": "paystub", "doc": "paystub_fin_007.pdf", "images": ["images/train/paystub/paystub_fin_007_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('paystub', 'The document class.')\nfields = (['pay_date', 'employee_name', 'employer_name', 'gross_pay', 'ytd_gross', 'pay_frequency', 'pay_period_start', 'pay_period_end', 'federal_tax', 'state_tax', 'social_security_tax', 'medicare_tax', 'hire_date', 'direct_deposit_bank_type'], 'The exact field names to extract.')\nhints = ({'pay_date': 'Date the paycheck was issued', 'employee_name': 'Name of Employee', 'employer_name': 'Name of Employer', 'gross_pay': 'Gross pay per period (current period, not YTD)', 'ytd_gross': 'Gross pay on paystub year to date (cumulative, not current period)', 'pay_frequency': 'The frequency if provided: weekly, biweekly, monthly, semi-monthly', 'pay_period_start': 'Pay period start date', 'pay_period_end': 'Pay period end date', 'hire_date': 'Job hire date', 'direct_deposit_bank_type': 'Direct deposit bank type'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"pay_date\", \"value\": \"2024-08-31\"}, {\"field\": \"employee_name\", \"value\": \"Brandon Tyson\"}, {\"field\": \"employer_name\", \"value\": \"Good Behavior Stragies LLC\"}, {\"field\": \"gross_pay\", \"value\": \"2080.0\"}, {\"field\": \"ytd_gross\", \"value\": \"33280.0\"}, {\"field\": \"pay_frequency\", \"value\": \"\"}, {\"field\": \"pay_period_start\", \"value\": \"2024-07-23\"}, {\"field\": \"pay_period_end\", \"value\": \"2024-08-05\"}, {\"field\": \"federal_tax\", \"value\": \"173.3\"}, {\"field\": \"state_tax\", \"value\": \"59.13\"}, {\"field\": \"social_security_tax\", \"value\": \"128.96\"}, {\"field\": \"medicare_tax\", \"value\": \"30.16\"}, {\"field\": \"hire_date\", \"value\": \"\"}, {\"field\": \"direct_deposit_bank_type\", \"value\": \"\"}]}"}]} {"id": "paystub_fin_009_p1_classify", "task": "classify", "class": "paystub", "doc": "paystub_fin_009.pdf", "images": ["images/train/paystub/paystub_fin_009_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"paystub\", \"confidence\": 1.0, \"issuer\": \"\", \"holder\": \"Jacklyn Lockman\", \"page_marker\": \"\"}]}"}]} {"id": "paystub_fin_009_extract", "task": "extract", "class": "paystub", "doc": "paystub_fin_009.pdf", "images": ["images/train/paystub/paystub_fin_009_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('paystub', 'The document class.')\nfields = (['pay_date', 'employee_name', 'employer_name', 'gross_pay', 'ytd_gross', 'pay_frequency', 'pay_period_start', 'pay_period_end', 'federal_tax', 'state_tax', 'social_security_tax', 'medicare_tax', 'hire_date', 'direct_deposit_bank_type'], 'The exact field names to extract.')\nhints = ({'pay_date': 'Date the paycheck was issued', 'employee_name': 'Name of Employee', 'employer_name': 'Name of Employer', 'gross_pay': 'Gross pay per period (current period, not YTD)', 'ytd_gross': 'Gross pay on paystub year to date (cumulative, not current period)', 'pay_frequency': 'The frequency if provided: weekly, biweekly, monthly, semi-monthly', 'pay_period_start': 'Pay period start date', 'pay_period_end': 'Pay period end date', 'hire_date': 'Job hire date', 'direct_deposit_bank_type': 'Direct deposit bank type'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"pay_date\", \"value\": \"\"}, {\"field\": \"employee_name\", \"value\": \"Jacklyn Lockman\"}, {\"field\": \"employer_name\", \"value\": \"\"}, {\"field\": \"gross_pay\", \"value\": \"1800.0\"}, {\"field\": \"ytd_gross\", \"value\": \"\"}, {\"field\": \"pay_frequency\", \"value\": \"\"}, {\"field\": \"pay_period_start\", \"value\": \"2050-01-01\"}, {\"field\": \"pay_period_end\", \"value\": \"2050-01-15\"}, {\"field\": \"federal_tax\", \"value\": \"200.0\"}, {\"field\": \"state_tax\", \"value\": \"100.0\"}, {\"field\": \"social_security_tax\", \"value\": \"\"}, {\"field\": \"medicare_tax\", \"value\": \"\"}, {\"field\": \"hire_date\", \"value\": \"\"}, {\"field\": \"direct_deposit_bank_type\", \"value\": \"\"}]}"}]} {"id": "paystub_fin_010_p1_classify", "task": "classify", "class": "paystub", "doc": "paystub_fin_010.pdf", "images": ["images/train/paystub/paystub_fin_010_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"paystub\", \"confidence\": 1.0, \"issuer\": \"Acme LLC\", \"holder\": \"Peter Sims\", \"page_marker\": \"\"}]}"}]} {"id": "paystub_fin_010_extract", "task": "extract", "class": "paystub", "doc": "paystub_fin_010.pdf", "images": ["images/train/paystub/paystub_fin_010_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('paystub', 'The document class.')\nfields = (['pay_date', 'employee_name', 'employer_name', 'gross_pay', 'ytd_gross', 'pay_frequency', 'pay_period_start', 'pay_period_end', 'federal_tax', 'state_tax', 'social_security_tax', 'medicare_tax', 'hire_date', 'direct_deposit_bank_type'], 'The exact field names to extract.')\nhints = ({'pay_date': 'Date the paycheck was issued', 'employee_name': 'Name of Employee', 'employer_name': 'Name of Employer', 'gross_pay': 'Gross pay per period (current period, not YTD)', 'ytd_gross': 'Gross pay on paystub year to date (cumulative, not current period)', 'pay_frequency': 'The frequency if provided: weekly, biweekly, monthly, semi-monthly', 'pay_period_start': 'Pay period start date', 'pay_period_end': 'Pay period end date', 'hire_date': 'Job hire date', 'direct_deposit_bank_type': 'Direct deposit bank type'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"pay_date\", \"value\": \"2017-06-07\"}, {\"field\": \"employee_name\", \"value\": \"Peter Sims\"}, {\"field\": \"employer_name\", \"value\": \"Acme LLC\"}, {\"field\": \"gross_pay\", \"value\": \"2000.0\"}, {\"field\": \"ytd_gross\", \"value\": \"24000.0\"}, {\"field\": \"pay_frequency\", \"value\": \"\"}, {\"field\": \"pay_period_start\", \"value\": \"2017-05-22\"}, {\"field\": \"pay_period_end\", \"value\": \"2017-06-04\"}, {\"field\": \"federal_tax\", \"value\": \"316.11\"}, {\"field\": \"state_tax\", \"value\": \"71.56\"}, {\"field\": \"social_security_tax\", \"value\": \"124.0\"}, {\"field\": \"medicare_tax\", \"value\": \"\"}, {\"field\": \"hire_date\", \"value\": \"\"}, {\"field\": \"direct_deposit_bank_type\", \"value\": \"\"}]}"}]} {"id": "paystub_fin_011_p1_classify", "task": "classify", "class": "paystub", "doc": "paystub_fin_011.pdf", "images": ["images/train/paystub/paystub_fin_011_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"paystub\", \"confidence\": 1.0, \"issuer\": \"Acme LLC\", \"holder\": \"Peter Sims\", \"page_marker\": \"\"}]}"}]} {"id": "paystub_fin_011_extract", "task": "extract", "class": "paystub", "doc": "paystub_fin_011.pdf", "images": ["images/train/paystub/paystub_fin_011_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('paystub', 'The document class.')\nfields = (['pay_date', 'employee_name', 'employer_name', 'gross_pay', 'ytd_gross', 'pay_frequency', 'pay_period_start', 'pay_period_end', 'federal_tax', 'state_tax', 'social_security_tax', 'medicare_tax', 'hire_date', 'direct_deposit_bank_type'], 'The exact field names to extract.')\nhints = ({'pay_date': 'Date the paycheck was issued', 'employee_name': 'Name of Employee', 'employer_name': 'Name of Employer', 'gross_pay': 'Gross pay per period (current period, not YTD)', 'ytd_gross': 'Gross pay on paystub year to date (cumulative, not current period)', 'pay_frequency': 'The frequency if provided: weekly, biweekly, monthly, semi-monthly', 'pay_period_start': 'Pay period start date', 'pay_period_end': 'Pay period end date', 'hire_date': 'Job hire date', 'direct_deposit_bank_type': 'Direct deposit bank type'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"pay_date\", \"value\": \"2017-06-07\"}, {\"field\": \"employee_name\", \"value\": \"Peter Sims\"}, {\"field\": \"employer_name\", \"value\": \"Acme LLC\"}, {\"field\": \"gross_pay\", \"value\": \"2000.0\"}, {\"field\": \"ytd_gross\", \"value\": \"24000.0\"}, {\"field\": \"pay_frequency\", \"value\": \"\"}, {\"field\": \"pay_period_start\", \"value\": \"2017-05-22\"}, {\"field\": \"pay_period_end\", \"value\": \"2017-06-04\"}, {\"field\": \"federal_tax\", \"value\": \"316.11\"}, {\"field\": \"state_tax\", \"value\": \"71.56\"}, {\"field\": \"social_security_tax\", \"value\": \"124.0\"}, {\"field\": \"medicare_tax\", \"value\": \"\"}, {\"field\": \"hire_date\", \"value\": \"\"}, {\"field\": \"direct_deposit_bank_type\", \"value\": \"\"}]}"}]} {"id": "paystub_fin_012_p1_classify", "task": "classify", "class": "paystub", "doc": "paystub_fin_012.pdf", "images": ["images/train/paystub/paystub_fin_012_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"paystub\", \"confidence\": 1.0, \"issuer\": \"Acme LLC\", \"holder\": \"Peter Sims\", \"page_marker\": \"\"}]}"}]} {"id": "paystub_fin_012_extract", "task": "extract", "class": "paystub", "doc": "paystub_fin_012.pdf", "images": ["images/train/paystub/paystub_fin_012_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('paystub', 'The document class.')\nfields = (['pay_date', 'employee_name', 'employer_name', 'gross_pay', 'ytd_gross', 'pay_frequency', 'pay_period_start', 'pay_period_end', 'federal_tax', 'state_tax', 'social_security_tax', 'medicare_tax', 'hire_date', 'direct_deposit_bank_type'], 'The exact field names to extract.')\nhints = ({'pay_date': 'Date the paycheck was issued', 'employee_name': 'Name of Employee', 'employer_name': 'Name of Employer', 'gross_pay': 'Gross pay per period (current period, not YTD)', 'ytd_gross': 'Gross pay on paystub year to date (cumulative, not current period)', 'pay_frequency': 'The frequency if provided: weekly, biweekly, monthly, semi-monthly', 'pay_period_start': 'Pay period start date', 'pay_period_end': 'Pay period end date', 'hire_date': 'Job hire date', 'direct_deposit_bank_type': 'Direct deposit bank type'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"pay_date\", \"value\": \"2017-06-07\"}, {\"field\": \"employee_name\", \"value\": \"Peter Sims\"}, {\"field\": \"employer_name\", \"value\": \"Acme LLC\"}, {\"field\": \"gross_pay\", \"value\": \"2000.0\"}, {\"field\": \"ytd_gross\", \"value\": \"24000.0\"}, {\"field\": \"pay_frequency\", \"value\": \"\"}, {\"field\": \"pay_period_start\", \"value\": \"2017-05-22\"}, {\"field\": \"pay_period_end\", \"value\": \"2017-06-04\"}, {\"field\": \"federal_tax\", \"value\": \"316.11\"}, {\"field\": \"state_tax\", \"value\": \"71.56\"}, {\"field\": \"social_security_tax\", \"value\": \"124.0\"}, {\"field\": \"medicare_tax\", \"value\": \"\"}, {\"field\": \"hire_date\", \"value\": \"\"}, {\"field\": \"direct_deposit_bank_type\", \"value\": \"\"}]}"}]} {"id": "paystub_fin_013_p1_classify", "task": "classify", "class": "paystub", "doc": "paystub_fin_013.pdf", "images": ["images/train/paystub/paystub_fin_013_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"paystub\", \"confidence\": 1.0, \"issuer\": \"Acme LLC\", \"holder\": \"Peter Sims\", \"page_marker\": \"\"}]}"}]} {"id": "paystub_fin_013_extract", "task": "extract", "class": "paystub", "doc": "paystub_fin_013.pdf", "images": ["images/train/paystub/paystub_fin_013_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('paystub', 'The document class.')\nfields = (['pay_date', 'employee_name', 'employer_name', 'gross_pay', 'ytd_gross', 'pay_frequency', 'pay_period_start', 'pay_period_end', 'federal_tax', 'state_tax', 'social_security_tax', 'medicare_tax', 'hire_date', 'direct_deposit_bank_type'], 'The exact field names to extract.')\nhints = ({'pay_date': 'Date the paycheck was issued', 'employee_name': 'Name of Employee', 'employer_name': 'Name of Employer', 'gross_pay': 'Gross pay per period (current period, not YTD)', 'ytd_gross': 'Gross pay on paystub year to date (cumulative, not current period)', 'pay_frequency': 'The frequency if provided: weekly, biweekly, monthly, semi-monthly', 'pay_period_start': 'Pay period start date', 'pay_period_end': 'Pay period end date', 'hire_date': 'Job hire date', 'direct_deposit_bank_type': 'Direct deposit bank type'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"pay_date\", \"value\": \"2017-06-07\"}, {\"field\": \"employee_name\", \"value\": \"Peter Sims\"}, {\"field\": \"employer_name\", \"value\": \"Acme LLC\"}, {\"field\": \"gross_pay\", \"value\": \"2000.0\"}, {\"field\": \"ytd_gross\", \"value\": \"24000.0\"}, {\"field\": \"pay_frequency\", \"value\": \"\"}, {\"field\": \"pay_period_start\", \"value\": \"2017-05-22\"}, {\"field\": \"pay_period_end\", \"value\": \"2017-06-04\"}, {\"field\": \"federal_tax\", \"value\": \"316.11\"}, {\"field\": \"state_tax\", \"value\": \"71.56\"}, {\"field\": \"social_security_tax\", \"value\": \"124.0\"}, {\"field\": \"medicare_tax\", \"value\": \"\"}, {\"field\": \"hire_date\", \"value\": \"\"}, {\"field\": \"direct_deposit_bank_type\", \"value\": \"\"}]}"}]} {"id": "paystub_fin_014_p1_classify", "task": "classify", "class": "paystub", "doc": "paystub_fin_014.pdf", "images": ["images/train/paystub/paystub_fin_014_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"paystub\", \"confidence\": 1.0, \"issuer\": \"Acme LLC\", \"holder\": \"Peter Sims\", \"page_marker\": \"\"}]}"}]} {"id": "paystub_fin_014_extract", "task": "extract", "class": "paystub", "doc": "paystub_fin_014.pdf", "images": ["images/train/paystub/paystub_fin_014_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('paystub', 'The document class.')\nfields = (['pay_date', 'employee_name', 'employer_name', 'gross_pay', 'ytd_gross', 'pay_frequency', 'pay_period_start', 'pay_period_end', 'federal_tax', 'state_tax', 'social_security_tax', 'medicare_tax', 'hire_date', 'direct_deposit_bank_type'], 'The exact field names to extract.')\nhints = ({'pay_date': 'Date the paycheck was issued', 'employee_name': 'Name of Employee', 'employer_name': 'Name of Employer', 'gross_pay': 'Gross pay per period (current period, not YTD)', 'ytd_gross': 'Gross pay on paystub year to date (cumulative, not current period)', 'pay_frequency': 'The frequency if provided: weekly, biweekly, monthly, semi-monthly', 'pay_period_start': 'Pay period start date', 'pay_period_end': 'Pay period end date', 'hire_date': 'Job hire date', 'direct_deposit_bank_type': 'Direct deposit bank type'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"pay_date\", \"value\": \"2017-06-07\"}, {\"field\": \"employee_name\", \"value\": \"Peter Sims\"}, {\"field\": \"employer_name\", \"value\": \"Acme LLC\"}, {\"field\": \"gross_pay\", \"value\": \"2000.0\"}, {\"field\": \"ytd_gross\", \"value\": \"24000.0\"}, {\"field\": \"pay_frequency\", \"value\": \"\"}, {\"field\": \"pay_period_start\", \"value\": \"2017-05-22\"}, {\"field\": \"pay_period_end\", \"value\": \"2017-06-04\"}, {\"field\": \"federal_tax\", \"value\": \"316.11\"}, {\"field\": \"state_tax\", \"value\": \"71.56\"}, {\"field\": \"social_security_tax\", \"value\": \"124.0\"}, {\"field\": \"medicare_tax\", \"value\": \"\"}, {\"field\": \"hire_date\", \"value\": \"\"}, {\"field\": \"direct_deposit_bank_type\", \"value\": \"\"}]}"}]} {"id": "paystub_fin_015_p1_classify", "task": "classify", "class": "paystub", "doc": "paystub_fin_015.pdf", "images": ["images/train/paystub/paystub_fin_015_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"paystub\", \"confidence\": 1.0, \"issuer\": \"Acme Corporation\", \"holder\": \"John M. Smith\", \"page_marker\": \"\"}]}"}]} {"id": "paystub_fin_015_extract", "task": "extract", "class": "paystub", "doc": "paystub_fin_015.pdf", "images": ["images/train/paystub/paystub_fin_015_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('paystub', 'The document class.')\nfields = (['pay_date', 'employee_name', 'employer_name', 'gross_pay', 'ytd_gross', 'pay_frequency', 'pay_period_start', 'pay_period_end', 'federal_tax', 'state_tax', 'social_security_tax', 'medicare_tax', 'hire_date', 'direct_deposit_bank_type'], 'The exact field names to extract.')\nhints = ({'pay_date': 'Date the paycheck was issued', 'employee_name': 'Name of Employee', 'employer_name': 'Name of Employer', 'gross_pay': 'Gross pay per period (current period, not YTD)', 'ytd_gross': 'Gross pay on paystub year to date (cumulative, not current period)', 'pay_frequency': 'The frequency if provided: weekly, biweekly, monthly, semi-monthly', 'pay_period_start': 'Pay period start date', 'pay_period_end': 'Pay period end date', 'hire_date': 'Job hire date', 'direct_deposit_bank_type': 'Direct deposit bank type'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"pay_date\", \"value\": \"2024-12-20\"}, {\"field\": \"employee_name\", \"value\": \"John M. Smith\"}, {\"field\": \"employer_name\", \"value\": \"Acme Corporation\"}, {\"field\": \"gross_pay\", \"value\": \"2000.0\"}, {\"field\": \"ytd_gross\", \"value\": \"48000.0\"}, {\"field\": \"pay_frequency\", \"value\": \"biweekly\"}, {\"field\": \"pay_period_start\", \"value\": \"2024-12-01\"}, {\"field\": \"pay_period_end\", \"value\": \"2024-12-15\"}, {\"field\": \"federal_tax\", \"value\": \"240.0\"}, {\"field\": \"state_tax\", \"value\": \"100.0\"}, {\"field\": \"social_security_tax\", \"value\": \"124.0\"}, {\"field\": \"medicare_tax\", \"value\": \"29.0\"}, {\"field\": \"hire_date\", \"value\": \"\"}, {\"field\": \"direct_deposit_bank_type\", \"value\": \"\"}]}"}]} {"id": "paystub_fin_016_p1_classify", "task": "classify", "class": "paystub", "doc": "paystub_fin_016.pdf", "images": ["images/train/paystub/paystub_fin_016_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"paystub\", \"confidence\": 1.0, \"issuer\": \"Acme Corporation\", \"holder\": \"John M. Smith\", \"page_marker\": \"\"}]}"}]} {"id": "paystub_fin_016_extract", "task": "extract", "class": "paystub", "doc": "paystub_fin_016.pdf", "images": ["images/train/paystub/paystub_fin_016_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('paystub', 'The document class.')\nfields = (['pay_date', 'employee_name', 'employer_name', 'gross_pay', 'ytd_gross', 'pay_frequency', 'pay_period_start', 'pay_period_end', 'federal_tax', 'state_tax', 'social_security_tax', 'medicare_tax', 'hire_date', 'direct_deposit_bank_type'], 'The exact field names to extract.')\nhints = ({'pay_date': 'Date the paycheck was issued', 'employee_name': 'Name of Employee', 'employer_name': 'Name of Employer', 'gross_pay': 'Gross pay per period (current period, not YTD)', 'ytd_gross': 'Gross pay on paystub year to date (cumulative, not current period)', 'pay_frequency': 'The frequency if provided: weekly, biweekly, monthly, semi-monthly', 'pay_period_start': 'Pay period start date', 'pay_period_end': 'Pay period end date', 'hire_date': 'Job hire date', 'direct_deposit_bank_type': 'Direct deposit bank type'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"pay_date\", \"value\": \"2024-12-20\"}, {\"field\": \"employee_name\", \"value\": \"John M. Smith\"}, {\"field\": \"employer_name\", \"value\": \"Acme Corporation\"}, {\"field\": \"gross_pay\", \"value\": \"2000.0\"}, {\"field\": \"ytd_gross\", \"value\": \"48000.0\"}, {\"field\": \"pay_frequency\", \"value\": \"biweekly\"}, {\"field\": \"pay_period_start\", \"value\": \"2024-12-01\"}, {\"field\": \"pay_period_end\", \"value\": \"2024-12-15\"}, {\"field\": \"federal_tax\", \"value\": \"240.0\"}, {\"field\": \"state_tax\", \"value\": \"100.0\"}, {\"field\": \"social_security_tax\", \"value\": \"124.0\"}, {\"field\": \"medicare_tax\", \"value\": \"29.0\"}, {\"field\": \"hire_date\", \"value\": \"\"}, {\"field\": \"direct_deposit_bank_type\", \"value\": \"Direct Deposit\"}]}"}]} {"id": "paystub_fin_017_p1_classify", "task": "classify", "class": "paystub", "doc": "paystub_fin_017.pdf", "images": ["images/train/paystub/paystub_fin_017_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"paystub\", \"confidence\": 1.0, \"issuer\": \"Acme Corporation\", \"holder\": \"John M. Smith\", \"page_marker\": \"\"}]}"}]} {"id": "paystub_fin_017_extract", "task": "extract", "class": "paystub", "doc": "paystub_fin_017.pdf", "images": ["images/train/paystub/paystub_fin_017_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('paystub', 'The document class.')\nfields = (['pay_date', 'employee_name', 'employer_name', 'gross_pay', 'ytd_gross', 'pay_frequency', 'pay_period_start', 'pay_period_end', 'federal_tax', 'state_tax', 'social_security_tax', 'medicare_tax', 'hire_date', 'direct_deposit_bank_type'], 'The exact field names to extract.')\nhints = ({'pay_date': 'Date the paycheck was issued', 'employee_name': 'Name of Employee', 'employer_name': 'Name of Employer', 'gross_pay': 'Gross pay per period (current period, not YTD)', 'ytd_gross': 'Gross pay on paystub year to date (cumulative, not current period)', 'pay_frequency': 'The frequency if provided: weekly, biweekly, monthly, semi-monthly', 'pay_period_start': 'Pay period start date', 'pay_period_end': 'Pay period end date', 'hire_date': 'Job hire date', 'direct_deposit_bank_type': 'Direct deposit bank type'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"pay_date\", \"value\": \"2024-12-20\"}, {\"field\": \"employee_name\", \"value\": \"John M. Smith\"}, {\"field\": \"employer_name\", \"value\": \"Acme Corporation\"}, {\"field\": \"gross_pay\", \"value\": \"2000.0\"}, {\"field\": \"ytd_gross\", \"value\": \"48000.0\"}, {\"field\": \"pay_frequency\", \"value\": \"biweekly\"}, {\"field\": \"pay_period_start\", \"value\": \"2024-12-01\"}, {\"field\": \"pay_period_end\", \"value\": \"2024-12-15\"}, {\"field\": \"federal_tax\", \"value\": \"240.0\"}, {\"field\": \"state_tax\", \"value\": \"100.0\"}, {\"field\": \"social_security_tax\", \"value\": \"124.0\"}, {\"field\": \"medicare_tax\", \"value\": \"29.0\"}, {\"field\": \"hire_date\", \"value\": \"\"}, {\"field\": \"direct_deposit_bank_type\", \"value\": \"Direct Deposit\"}]}"}]} {"id": "paystub_fin_018_p1_classify", "task": "classify", "class": "paystub", "doc": "paystub_fin_018.pdf", "images": ["images/train/paystub/paystub_fin_018_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"paystub\", \"confidence\": 1.0, \"issuer\": \"Acme Corporation\", \"holder\": \"John M. Smith\", \"page_marker\": \"\"}]}"}]} {"id": "paystub_fin_018_extract", "task": "extract", "class": "paystub", "doc": "paystub_fin_018.pdf", "images": ["images/train/paystub/paystub_fin_018_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('paystub', 'The document class.')\nfields = (['pay_date', 'employee_name', 'employer_name', 'gross_pay', 'ytd_gross', 'pay_frequency', 'pay_period_start', 'pay_period_end', 'federal_tax', 'state_tax', 'social_security_tax', 'medicare_tax', 'hire_date', 'direct_deposit_bank_type'], 'The exact field names to extract.')\nhints = ({'pay_date': 'Date the paycheck was issued', 'employee_name': 'Name of Employee', 'employer_name': 'Name of Employer', 'gross_pay': 'Gross pay per period (current period, not YTD)', 'ytd_gross': 'Gross pay on paystub year to date (cumulative, not current period)', 'pay_frequency': 'The frequency if provided: weekly, biweekly, monthly, semi-monthly', 'pay_period_start': 'Pay period start date', 'pay_period_end': 'Pay period end date', 'hire_date': 'Job hire date', 'direct_deposit_bank_type': 'Direct deposit bank type'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"pay_date\", \"value\": \"2024-12-20\"}, {\"field\": \"employee_name\", \"value\": \"John M. Smith\"}, {\"field\": \"employer_name\", \"value\": \"Acme Corporation\"}, {\"field\": \"gross_pay\", \"value\": \"2500.0\"}, {\"field\": \"ytd_gross\", \"value\": \"60000.0\"}, {\"field\": \"pay_frequency\", \"value\": \"biweekly\"}, {\"field\": \"pay_period_start\", \"value\": \"2024-12-01\"}, {\"field\": \"pay_period_end\", \"value\": \"2024-12-15\"}, {\"field\": \"federal_tax\", \"value\": \"300.0\"}, {\"field\": \"state_tax\", \"value\": \"125.0\"}, {\"field\": \"social_security_tax\", \"value\": \"155.0\"}, {\"field\": \"medicare_tax\", \"value\": \"36.25\"}, {\"field\": \"hire_date\", \"value\": \"\"}, {\"field\": \"direct_deposit_bank_type\", \"value\": \"Direct Deposit\"}]}"}]} {"id": "paystub_fin_019_p1_classify", "task": "classify", "class": "paystub", "doc": "paystub_fin_019.pdf", "images": ["images/train/paystub/paystub_fin_019_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"paystub\", \"confidence\": 1.0, \"issuer\": \"\", \"holder\": \"Ora W. D'Amato\", \"page_marker\": \"\"}]}"}]} {"id": "paystub_fin_019_extract", "task": "extract", "class": "paystub", "doc": "paystub_fin_019.pdf", "images": ["images/train/paystub/paystub_fin_019_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('paystub', 'The document class.')\nfields = (['pay_date', 'employee_name', 'employer_name', 'gross_pay', 'ytd_gross', 'pay_frequency', 'pay_period_start', 'pay_period_end', 'federal_tax', 'state_tax', 'social_security_tax', 'medicare_tax', 'hire_date', 'direct_deposit_bank_type'], 'The exact field names to extract.')\nhints = ({'pay_date': 'Date the paycheck was issued', 'employee_name': 'Name of Employee', 'employer_name': 'Name of Employer', 'gross_pay': 'Gross pay per period (current period, not YTD)', 'ytd_gross': 'Gross pay on paystub year to date (cumulative, not current period)', 'pay_frequency': 'The frequency if provided: weekly, biweekly, monthly, semi-monthly', 'pay_period_start': 'Pay period start date', 'pay_period_end': 'Pay period end date', 'hire_date': 'Job hire date', 'direct_deposit_bank_type': 'Direct deposit bank type'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"pay_date\", \"value\": \"2023-09-03\"}, {\"field\": \"employee_name\", \"value\": \"Ora W. D'Amato\"}, {\"field\": \"employer_name\", \"value\": \"\"}, {\"field\": \"gross_pay\", \"value\": \"4954.0\"}, {\"field\": \"ytd_gross\", \"value\": \"42054.0\"}, {\"field\": \"pay_frequency\", \"value\": \"biweekly\"}, {\"field\": \"pay_period_start\", \"value\": \"2023-08-01\"}, {\"field\": \"pay_period_end\", \"value\": \"2023-08-31\"}, {\"field\": \"federal_tax\", \"value\": \"256.0\"}, {\"field\": \"state_tax\", \"value\": \"50.0\"}, {\"field\": \"social_security_tax\", \"value\": \"150.0\"}, {\"field\": \"medicare_tax\", \"value\": \"71.83\"}, {\"field\": \"hire_date\", \"value\": \"\"}, {\"field\": \"direct_deposit_bank_type\", \"value\": \"\"}]}"}]} {"id": "paystub_fin_020_p1_classify", "task": "classify", "class": "paystub", "doc": "paystub_fin_020.pdf", "images": ["images/train/paystub/paystub_fin_020_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"paystub\", \"confidence\": 1.0, \"issuer\": \"Acme Corporation\", \"holder\": \"John M. Smith\", \"page_marker\": \"\"}]}"}]} {"id": "paystub_fin_020_extract", "task": "extract", "class": "paystub", "doc": "paystub_fin_020.pdf", "images": ["images/train/paystub/paystub_fin_020_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('paystub', 'The document class.')\nfields = (['pay_date', 'employee_name', 'employer_name', 'gross_pay', 'ytd_gross', 'pay_frequency', 'pay_period_start', 'pay_period_end', 'federal_tax', 'state_tax', 'social_security_tax', 'medicare_tax', 'hire_date', 'direct_deposit_bank_type'], 'The exact field names to extract.')\nhints = ({'pay_date': 'Date the paycheck was issued', 'employee_name': 'Name of Employee', 'employer_name': 'Name of Employer', 'gross_pay': 'Gross pay per period (current period, not YTD)', 'ytd_gross': 'Gross pay on paystub year to date (cumulative, not current period)', 'pay_frequency': 'The frequency if provided: weekly, biweekly, monthly, semi-monthly', 'pay_period_start': 'Pay period start date', 'pay_period_end': 'Pay period end date', 'hire_date': 'Job hire date', 'direct_deposit_bank_type': 'Direct deposit bank type'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"pay_date\", \"value\": \"2024-12-20\"}, {\"field\": \"employee_name\", \"value\": \"John M. Smith\"}, {\"field\": \"employer_name\", \"value\": \"Acme Corporation\"}, {\"field\": \"gross_pay\", \"value\": \"2000.0\"}, {\"field\": \"ytd_gross\", \"value\": \"48000.0\"}, {\"field\": \"pay_frequency\", \"value\": \"biweekly\"}, {\"field\": \"pay_period_start\", \"value\": \"2024-12-01\"}, {\"field\": \"pay_period_end\", \"value\": \"2024-12-15\"}, {\"field\": \"federal_tax\", \"value\": \"240.0\"}, {\"field\": \"state_tax\", \"value\": \"100.0\"}, {\"field\": \"social_security_tax\", \"value\": \"124.0\"}, {\"field\": \"medicare_tax\", \"value\": \"29.0\"}, {\"field\": \"hire_date\", \"value\": \"\"}, {\"field\": \"direct_deposit_bank_type\", \"value\": \"Direct Deposit\"}]}"}]} {"id": "paystub_fin_021_p1_classify", "task": "classify", "class": "paystub", "doc": "paystub_fin_021.pdf", "images": ["images/train/paystub/paystub_fin_021_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"paystub\", \"confidence\": 1.0, \"issuer\": \"Acme Corporation\", \"holder\": \"John M. Smith\", \"page_marker\": \"\"}]}"}]} {"id": "paystub_fin_021_extract", "task": "extract", "class": "paystub", "doc": "paystub_fin_021.pdf", "images": ["images/train/paystub/paystub_fin_021_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('paystub', 'The document class.')\nfields = (['pay_date', 'employee_name', 'employer_name', 'gross_pay', 'ytd_gross', 'pay_frequency', 'pay_period_start', 'pay_period_end', 'federal_tax', 'state_tax', 'social_security_tax', 'medicare_tax', 'hire_date', 'direct_deposit_bank_type'], 'The exact field names to extract.')\nhints = ({'pay_date': 'Date the paycheck was issued', 'employee_name': 'Name of Employee', 'employer_name': 'Name of Employer', 'gross_pay': 'Gross pay per period (current period, not YTD)', 'ytd_gross': 'Gross pay on paystub year to date (cumulative, not current period)', 'pay_frequency': 'The frequency if provided: weekly, biweekly, monthly, semi-monthly', 'pay_period_start': 'Pay period start date', 'pay_period_end': 'Pay period end date', 'hire_date': 'Job hire date', 'direct_deposit_bank_type': 'Direct deposit bank type'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"pay_date\", \"value\": \"2024-12-20\"}, {\"field\": \"employee_name\", \"value\": \"John M. Smith\"}, {\"field\": \"employer_name\", \"value\": \"Acme Corporation\"}, {\"field\": \"gross_pay\", \"value\": \"2000.0\"}, {\"field\": \"ytd_gross\", \"value\": \"48000.0\"}, {\"field\": \"pay_frequency\", \"value\": \"biweekly\"}, {\"field\": \"pay_period_start\", \"value\": \"2024-12-01\"}, {\"field\": \"pay_period_end\", \"value\": \"2024-12-15\"}, {\"field\": \"federal_tax\", \"value\": \"240.0\"}, {\"field\": \"state_tax\", \"value\": \"100.0\"}, {\"field\": \"social_security_tax\", \"value\": \"124.0\"}, {\"field\": \"medicare_tax\", \"value\": \"29.0\"}, {\"field\": \"hire_date\", \"value\": \"\"}, {\"field\": \"direct_deposit_bank_type\", \"value\": \"\"}]}"}]} {"id": "paystub_fin_022_p1_classify", "task": "classify", "class": "paystub", "doc": "paystub_fin_022.pdf", "images": ["images/train/paystub/paystub_fin_022_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"paystub\", \"confidence\": 1.0, \"issuer\": \"Acme Corporation\", \"holder\": \"John M. Smith\", \"page_marker\": \"\"}]}"}]} {"id": "paystub_fin_022_extract", "task": "extract", "class": "paystub", "doc": "paystub_fin_022.pdf", "images": ["images/train/paystub/paystub_fin_022_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('paystub', 'The document class.')\nfields = (['pay_date', 'employee_name', 'employer_name', 'gross_pay', 'ytd_gross', 'pay_frequency', 'pay_period_start', 'pay_period_end', 'federal_tax', 'state_tax', 'social_security_tax', 'medicare_tax', 'hire_date', 'direct_deposit_bank_type'], 'The exact field names to extract.')\nhints = ({'pay_date': 'Date the paycheck was issued', 'employee_name': 'Name of Employee', 'employer_name': 'Name of Employer', 'gross_pay': 'Gross pay per period (current period, not YTD)', 'ytd_gross': 'Gross pay on paystub year to date (cumulative, not current period)', 'pay_frequency': 'The frequency if provided: weekly, biweekly, monthly, semi-monthly', 'pay_period_start': 'Pay period start date', 'pay_period_end': 'Pay period end date', 'hire_date': 'Job hire date', 'direct_deposit_bank_type': 'Direct deposit bank type'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"pay_date\", \"value\": \"2024-12-20\"}, {\"field\": \"employee_name\", \"value\": \"John M. Smith\"}, {\"field\": \"employer_name\", \"value\": \"Acme Corporation\"}, {\"field\": \"gross_pay\", \"value\": \"2450.0\"}, {\"field\": \"ytd_gross\", \"value\": \"50450.0\"}, {\"field\": \"pay_frequency\", \"value\": \"biweekly\"}, {\"field\": \"pay_period_start\", \"value\": \"2024-12-01\"}, {\"field\": \"pay_period_end\", \"value\": \"2024-12-15\"}, {\"field\": \"federal_tax\", \"value\": \"294.0\"}, {\"field\": \"state_tax\", \"value\": \"122.5\"}, {\"field\": \"social_security_tax\", \"value\": \"151.9\"}, {\"field\": \"medicare_tax\", \"value\": \"35.53\"}, {\"field\": \"hire_date\", \"value\": \"\"}, {\"field\": \"direct_deposit_bank_type\", \"value\": \"Direct Deposit\"}]}"}]} {"id": "paystub_fin_024_p1_classify", "task": "classify", "class": "paystub", "doc": "paystub_fin_024.pdf", "images": ["images/train/paystub/paystub_fin_024_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"paystub\", \"confidence\": 1.0, \"issuer\": \"Acme Corporation\", \"holder\": \"John M. Smith\", \"page_marker\": \"\"}]}"}]} {"id": "paystub_fin_024_extract", "task": "extract", "class": "paystub", "doc": "paystub_fin_024.pdf", "images": ["images/train/paystub/paystub_fin_024_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('paystub', 'The document class.')\nfields = (['pay_date', 'employee_name', 'employer_name', 'gross_pay', 'ytd_gross', 'pay_frequency', 'pay_period_start', 'pay_period_end', 'federal_tax', 'state_tax', 'social_security_tax', 'medicare_tax', 'hire_date', 'direct_deposit_bank_type'], 'The exact field names to extract.')\nhints = ({'pay_date': 'Date the paycheck was issued', 'employee_name': 'Name of Employee', 'employer_name': 'Name of Employer', 'gross_pay': 'Gross pay per period (current period, not YTD)', 'ytd_gross': 'Gross pay on paystub year to date (cumulative, not current period)', 'pay_frequency': 'The frequency if provided: weekly, biweekly, monthly, semi-monthly', 'pay_period_start': 'Pay period start date', 'pay_period_end': 'Pay period end date', 'hire_date': 'Job hire date', 'direct_deposit_bank_type': 'Direct deposit bank type'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"pay_date\", \"value\": \"2024-12-20\"}, {\"field\": \"employee_name\", \"value\": \"John M. Smith\"}, {\"field\": \"employer_name\", \"value\": \"Acme Corporation\"}, {\"field\": \"gross_pay\", \"value\": \"2450.0\"}, {\"field\": \"ytd_gross\", \"value\": \"50450.0\"}, {\"field\": \"pay_frequency\", \"value\": \"biweekly\"}, {\"field\": \"pay_period_start\", \"value\": \"2024-12-01\"}, {\"field\": \"pay_period_end\", \"value\": \"2024-12-15\"}, {\"field\": \"federal_tax\", \"value\": \"294.0\"}, {\"field\": \"state_tax\", \"value\": \"122.5\"}, {\"field\": \"social_security_tax\", \"value\": \"151.9\"}, {\"field\": \"medicare_tax\", \"value\": \"35.53\"}, {\"field\": \"hire_date\", \"value\": \"\"}, {\"field\": \"direct_deposit_bank_type\", \"value\": \"Direct Deposit\"}]}"}]} {"id": "paystub_fin_025_p1_classify", "task": "classify", "class": "paystub", "doc": "paystub_fin_025.pdf", "images": ["images/train/paystub/paystub_fin_025_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"paystub\", \"confidence\": 1.0, \"issuer\": \"Acme Corporation\", \"holder\": \"John M. Smith\", \"page_marker\": \"\"}]}"}]} {"id": "paystub_fin_025_extract", "task": "extract", "class": "paystub", "doc": "paystub_fin_025.pdf", "images": ["images/train/paystub/paystub_fin_025_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('paystub', 'The document class.')\nfields = (['pay_date', 'employee_name', 'employer_name', 'gross_pay', 'ytd_gross', 'pay_frequency', 'pay_period_start', 'pay_period_end', 'federal_tax', 'state_tax', 'social_security_tax', 'medicare_tax', 'hire_date', 'direct_deposit_bank_type'], 'The exact field names to extract.')\nhints = ({'pay_date': 'Date the paycheck was issued', 'employee_name': 'Name of Employee', 'employer_name': 'Name of Employer', 'gross_pay': 'Gross pay per period (current period, not YTD)', 'ytd_gross': 'Gross pay on paystub year to date (cumulative, not current period)', 'pay_frequency': 'The frequency if provided: weekly, biweekly, monthly, semi-monthly', 'pay_period_start': 'Pay period start date', 'pay_period_end': 'Pay period end date', 'hire_date': 'Job hire date', 'direct_deposit_bank_type': 'Direct deposit bank type'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"pay_date\", \"value\": \"2024-12-20\"}, {\"field\": \"employee_name\", \"value\": \"John M. Smith\"}, {\"field\": \"employer_name\", \"value\": \"Acme Corporation\"}, {\"field\": \"gross_pay\", \"value\": \"2450.0\"}, {\"field\": \"ytd_gross\", \"value\": \"50450.0\"}, {\"field\": \"pay_frequency\", \"value\": \"biweekly\"}, {\"field\": \"pay_period_start\", \"value\": \"2024-12-01\"}, {\"field\": \"pay_period_end\", \"value\": \"2024-12-15\"}, {\"field\": \"federal_tax\", \"value\": \"294.0\"}, {\"field\": \"state_tax\", \"value\": \"122.5\"}, {\"field\": \"social_security_tax\", \"value\": \"151.9\"}, {\"field\": \"medicare_tax\", \"value\": \"35.53\"}, {\"field\": \"hire_date\", \"value\": \"\"}, {\"field\": \"direct_deposit_bank_type\", \"value\": \"\"}]}"}]} {"id": "paystub_fin_027_p1_classify", "task": "classify", "class": "paystub", "doc": "paystub_fin_027.pdf", "images": ["images/train/paystub/paystub_fin_027_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"paystub\", \"confidence\": 1.0, \"issuer\": \"Acme Corporation\", \"holder\": \"John M. Smith\", \"page_marker\": \"\"}]}"}]} {"id": "paystub_fin_027_extract", "task": "extract", "class": "paystub", "doc": "paystub_fin_027.pdf", "images": ["images/train/paystub/paystub_fin_027_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('paystub', 'The document class.')\nfields = (['pay_date', 'employee_name', 'employer_name', 'gross_pay', 'ytd_gross', 'pay_frequency', 'pay_period_start', 'pay_period_end', 'federal_tax', 'state_tax', 'social_security_tax', 'medicare_tax', 'hire_date', 'direct_deposit_bank_type'], 'The exact field names to extract.')\nhints = ({'pay_date': 'Date the paycheck was issued', 'employee_name': 'Name of Employee', 'employer_name': 'Name of Employer', 'gross_pay': 'Gross pay per period (current period, not YTD)', 'ytd_gross': 'Gross pay on paystub year to date (cumulative, not current period)', 'pay_frequency': 'The frequency if provided: weekly, biweekly, monthly, semi-monthly', 'pay_period_start': 'Pay period start date', 'pay_period_end': 'Pay period end date', 'hire_date': 'Job hire date', 'direct_deposit_bank_type': 'Direct deposit bank type'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"pay_date\", \"value\": \"2024-12-20\"}, {\"field\": \"employee_name\", \"value\": \"John M. Smith\"}, {\"field\": \"employer_name\", \"value\": \"Acme Corporation\"}, {\"field\": \"gross_pay\", \"value\": \"3000.0\"}, {\"field\": \"ytd_gross\", \"value\": \"72000.0\"}, {\"field\": \"pay_frequency\", \"value\": \"\"}, {\"field\": \"pay_period_start\", \"value\": \"2024-12-01\"}, {\"field\": \"pay_period_end\", \"value\": \"2024-12-15\"}, {\"field\": \"federal_tax\", \"value\": \"\"}, {\"field\": \"state_tax\", \"value\": \"\"}, {\"field\": \"social_security_tax\", \"value\": \"\"}, {\"field\": \"medicare_tax\", \"value\": \"\"}, {\"field\": \"hire_date\", \"value\": \"\"}, {\"field\": \"direct_deposit_bank_type\", \"value\": \"ACH Transfer\"}]}"}]} {"id": "paystub_fin_030_p1_classify", "task": "classify", "class": "paystub", "doc": "paystub_fin_030.pdf", "images": ["images/train/paystub/paystub_fin_030_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"paystub\", \"confidence\": 1.0, \"issuer\": \"Acme Corporation\", \"holder\": \"Lisa Anderson\", \"page_marker\": \"\"}]}"}]} {"id": "paystub_fin_030_extract", "task": "extract", "class": "paystub", "doc": "paystub_fin_030.pdf", "images": ["images/train/paystub/paystub_fin_030_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('paystub', 'The document class.')\nfields = (['pay_date', 'employee_name', 'employer_name', 'gross_pay', 'ytd_gross', 'pay_frequency', 'pay_period_start', 'pay_period_end', 'federal_tax', 'state_tax', 'social_security_tax', 'medicare_tax', 'hire_date', 'direct_deposit_bank_type'], 'The exact field names to extract.')\nhints = ({'pay_date': 'Date the paycheck was issued', 'employee_name': 'Name of Employee', 'employer_name': 'Name of Employer', 'gross_pay': 'Gross pay per period (current period, not YTD)', 'ytd_gross': 'Gross pay on paystub year to date (cumulative, not current period)', 'pay_frequency': 'The frequency if provided: weekly, biweekly, monthly, semi-monthly', 'pay_period_start': 'Pay period start date', 'pay_period_end': 'Pay period end date', 'hire_date': 'Job hire date', 'direct_deposit_bank_type': 'Direct deposit bank type'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"pay_date\", \"value\": \"2024-12-20\"}, {\"field\": \"employee_name\", \"value\": \"Lisa Anderson\"}, {\"field\": \"employer_name\", \"value\": \"Acme Corporation\"}, {\"field\": \"gross_pay\", \"value\": \"2200.0\"}, {\"field\": \"ytd_gross\", \"value\": \"52800.0\"}, {\"field\": \"pay_frequency\", \"value\": \"biweekly\"}, {\"field\": \"pay_period_start\", \"value\": \"2024-12-01\"}, {\"field\": \"pay_period_end\", \"value\": \"2024-12-15\"}, {\"field\": \"federal_tax\", \"value\": \"264.0\"}, {\"field\": \"state_tax\", \"value\": \"110.0\"}, {\"field\": \"social_security_tax\", \"value\": \"136.4\"}, {\"field\": \"medicare_tax\", \"value\": \"31.9\"}, {\"field\": \"hire_date\", \"value\": \"\"}, {\"field\": \"direct_deposit_bank_type\", \"value\": \"First National Bank\"}]}"}]} {"id": "paystub_fin_031_p1_classify", "task": "classify", "class": "paystub", "doc": "paystub_fin_031.pdf", "images": ["images/train/paystub/paystub_fin_031_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"paystub\", \"confidence\": 1.0, \"issuer\": \"Acme Sales Corp\", \"holder\": \"Sarah Johnson\", \"page_marker\": \"\"}]}"}]} {"id": "paystub_fin_031_extract", "task": "extract", "class": "paystub", "doc": "paystub_fin_031.pdf", "images": ["images/train/paystub/paystub_fin_031_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('paystub', 'The document class.')\nfields = (['pay_date', 'employee_name', 'employer_name', 'gross_pay', 'ytd_gross', 'pay_frequency', 'pay_period_start', 'pay_period_end', 'federal_tax', 'state_tax', 'social_security_tax', 'medicare_tax', 'hire_date', 'direct_deposit_bank_type'], 'The exact field names to extract.')\nhints = ({'pay_date': 'Date the paycheck was issued', 'employee_name': 'Name of Employee', 'employer_name': 'Name of Employer', 'gross_pay': 'Gross pay per period (current period, not YTD)', 'ytd_gross': 'Gross pay on paystub year to date (cumulative, not current period)', 'pay_frequency': 'The frequency if provided: weekly, biweekly, monthly, semi-monthly', 'pay_period_start': 'Pay period start date', 'pay_period_end': 'Pay period end date', 'hire_date': 'Job hire date', 'direct_deposit_bank_type': 'Direct deposit bank type'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"pay_date\", \"value\": \"2024-12-20\"}, {\"field\": \"employee_name\", \"value\": \"Sarah Johnson\"}, {\"field\": \"employer_name\", \"value\": \"Acme Sales Corp\"}, {\"field\": \"gross_pay\", \"value\": \"3500.0\"}, {\"field\": \"ytd_gross\", \"value\": \"84000.0\"}, {\"field\": \"pay_frequency\", \"value\": \"biweekly\"}, {\"field\": \"pay_period_start\", \"value\": \"2024-12-01\"}, {\"field\": \"pay_period_end\", \"value\": \"2024-12-15\"}, {\"field\": \"federal_tax\", \"value\": \"420.0\"}, {\"field\": \"state_tax\", \"value\": \"175.0\"}, {\"field\": \"social_security_tax\", \"value\": \"217.0\"}, {\"field\": \"medicare_tax\", \"value\": \"50.75\"}, {\"field\": \"hire_date\", \"value\": \"\"}, {\"field\": \"direct_deposit_bank_type\", \"value\": \"Direct Deposit\"}]}"}]} {"id": "paystub_fin_032_p1_classify", "task": "classify", "class": "paystub", "doc": "paystub_fin_032.pdf", "images": ["images/train/paystub/paystub_fin_032_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"paystub\", \"confidence\": 1.0, \"issuer\": \"Acme Corporation\", \"holder\": \"Michael Chen\", \"page_marker\": \"\"}]}"}]} {"id": "paystub_fin_032_extract", "task": "extract", "class": "paystub", "doc": "paystub_fin_032.pdf", "images": ["images/train/paystub/paystub_fin_032_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('paystub', 'The document class.')\nfields = (['pay_date', 'employee_name', 'employer_name', 'gross_pay', 'ytd_gross', 'pay_frequency', 'pay_period_start', 'pay_period_end', 'federal_tax', 'state_tax', 'social_security_tax', 'medicare_tax', 'hire_date', 'direct_deposit_bank_type'], 'The exact field names to extract.')\nhints = ({'pay_date': 'Date the paycheck was issued', 'employee_name': 'Name of Employee', 'employer_name': 'Name of Employer', 'gross_pay': 'Gross pay per period (current period, not YTD)', 'ytd_gross': 'Gross pay on paystub year to date (cumulative, not current period)', 'pay_frequency': 'The frequency if provided: weekly, biweekly, monthly, semi-monthly', 'pay_period_start': 'Pay period start date', 'pay_period_end': 'Pay period end date', 'hire_date': 'Job hire date', 'direct_deposit_bank_type': 'Direct deposit bank type'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"pay_date\", \"value\": \"2024-12-20\"}, {\"field\": \"employee_name\", \"value\": \"Michael Chen\"}, {\"field\": \"employer_name\", \"value\": \"Acme Corporation\"}, {\"field\": \"gross_pay\", \"value\": \"5900.0\"}, {\"field\": \"ytd_gross\", \"value\": \"72000.0\"}, {\"field\": \"pay_frequency\", \"value\": \"biweekly\"}, {\"field\": \"pay_period_start\", \"value\": \"2024-12-01\"}, {\"field\": \"pay_period_end\", \"value\": \"2024-12-15\"}, {\"field\": \"federal_tax\", \"value\": \"708.0\"}, {\"field\": \"state_tax\", \"value\": \"295.0\"}, {\"field\": \"social_security_tax\", \"value\": \"365.8\"}, {\"field\": \"medicare_tax\", \"value\": \"85.55\"}, {\"field\": \"hire_date\", \"value\": \"\"}, {\"field\": \"direct_deposit_bank_type\", \"value\": \"Direct Deposit\"}]}"}]} {"id": "paystub_fin_033_p1_classify", "task": "classify", "class": "paystub", "doc": "paystub_fin_033.pdf", "images": ["images/train/paystub/paystub_fin_033_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"paystub\", \"confidence\": 1.0, \"issuer\": \"Downtown Bistro\", \"holder\": \"Emily Rodriguez\", \"page_marker\": \"\"}]}"}]} {"id": "paystub_fin_033_extract", "task": "extract", "class": "paystub", "doc": "paystub_fin_033.pdf", "images": ["images/train/paystub/paystub_fin_033_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('paystub', 'The document class.')\nfields = (['pay_date', 'employee_name', 'employer_name', 'gross_pay', 'ytd_gross', 'pay_frequency', 'pay_period_start', 'pay_period_end', 'federal_tax', 'state_tax', 'social_security_tax', 'medicare_tax', 'hire_date', 'direct_deposit_bank_type'], 'The exact field names to extract.')\nhints = ({'pay_date': 'Date the paycheck was issued', 'employee_name': 'Name of Employee', 'employer_name': 'Name of Employer', 'gross_pay': 'Gross pay per period (current period, not YTD)', 'ytd_gross': 'Gross pay on paystub year to date (cumulative, not current period)', 'pay_frequency': 'The frequency if provided: weekly, biweekly, monthly, semi-monthly', 'pay_period_start': 'Pay period start date', 'pay_period_end': 'Pay period end date', 'hire_date': 'Job hire date', 'direct_deposit_bank_type': 'Direct deposit bank type'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"pay_date\", \"value\": \"2024-12-20\"}, {\"field\": \"employee_name\", \"value\": \"Emily Rodriguez\"}, {\"field\": \"employer_name\", \"value\": \"Downtown Bistro\"}, {\"field\": \"gross_pay\", \"value\": \"1635.0\"}, {\"field\": \"ytd_gross\", \"value\": \"39240.0\"}, {\"field\": \"pay_frequency\", \"value\": \"biweekly\"}, {\"field\": \"pay_period_start\", \"value\": \"2024-12-01\"}, {\"field\": \"pay_period_end\", \"value\": \"2024-12-15\"}, {\"field\": \"federal_tax\", \"value\": \"163.5\"}, {\"field\": \"state_tax\", \"value\": \"81.75\"}, {\"field\": \"social_security_tax\", \"value\": \"101.37\"}, {\"field\": \"medicare_tax\", \"value\": \"23.71\"}, {\"field\": \"hire_date\", \"value\": \"\"}, {\"field\": \"direct_deposit_bank_type\", \"value\": \"Direct Deposit\"}]}"}]} {"id": "paystub_fin_034_p1_classify", "task": "classify", "class": "paystub", "doc": "paystub_fin_034.pdf", "images": ["images/train/paystub/paystub_fin_034_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"paystub\", \"confidence\": 1.0, \"issuer\": \"Acme Corporation\", \"holder\": \"John Smith\", \"page_marker\": \"\"}]}"}]} {"id": "paystub_fin_034_extract", "task": "extract", "class": "paystub", "doc": "paystub_fin_034.pdf", "images": ["images/train/paystub/paystub_fin_034_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('paystub', 'The document class.')\nfields = (['pay_date', 'employee_name', 'employer_name', 'gross_pay', 'ytd_gross', 'pay_frequency', 'pay_period_start', 'pay_period_end', 'federal_tax', 'state_tax', 'social_security_tax', 'medicare_tax', 'hire_date', 'direct_deposit_bank_type'], 'The exact field names to extract.')\nhints = ({'pay_date': 'Date the paycheck was issued', 'employee_name': 'Name of Employee', 'employer_name': 'Name of Employer', 'gross_pay': 'Gross pay per period (current period, not YTD)', 'ytd_gross': 'Gross pay on paystub year to date (cumulative, not current period)', 'pay_frequency': 'The frequency if provided: weekly, biweekly, monthly, semi-monthly', 'pay_period_start': 'Pay period start date', 'pay_period_end': 'Pay period end date', 'hire_date': 'Job hire date', 'direct_deposit_bank_type': 'Direct deposit bank type'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"pay_date\", \"value\": \"2024-12-20\"}, {\"field\": \"employee_name\", \"value\": \"John Smith\"}, {\"field\": \"employer_name\", \"value\": \"Acme Corporation\"}, {\"field\": \"gross_pay\", \"value\": \"2308.0\"}, {\"field\": \"ytd_gross\", \"value\": \"55392.0\"}, {\"field\": \"pay_frequency\", \"value\": \"biweekly\"}, {\"field\": \"pay_period_start\", \"value\": \"2024-12-01\"}, {\"field\": \"pay_period_end\", \"value\": \"2024-12-14\"}, {\"field\": \"federal_tax\", \"value\": \"276.96\"}, {\"field\": \"state_tax\", \"value\": \"115.4\"}, {\"field\": \"social_security_tax\", \"value\": \"143.1\"}, {\"field\": \"medicare_tax\", \"value\": \"33.47\"}, {\"field\": \"hire_date\", \"value\": \"\"}, {\"field\": \"direct_deposit_bank_type\", \"value\": \"Direct Deposit\"}]}"}]} {"id": "paystub_fin_035_p1_classify", "task": "classify", "class": "paystub", "doc": "paystub_fin_035.pdf", "images": ["images/train/paystub/paystub_fin_035_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"paystub\", \"confidence\": 1.0, \"issuer\": \"Quick Service Inc\", \"holder\": \"David Martinez\", \"page_marker\": \"\"}]}"}]} {"id": "paystub_fin_035_extract", "task": "extract", "class": "paystub", "doc": "paystub_fin_035.pdf", "images": ["images/train/paystub/paystub_fin_035_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('paystub', 'The document class.')\nfields = (['pay_date', 'employee_name', 'employer_name', 'gross_pay', 'ytd_gross', 'pay_frequency', 'pay_period_start', 'pay_period_end', 'federal_tax', 'state_tax', 'social_security_tax', 'medicare_tax', 'hire_date', 'direct_deposit_bank_type'], 'The exact field names to extract.')\nhints = ({'pay_date': 'Date the paycheck was issued', 'employee_name': 'Name of Employee', 'employer_name': 'Name of Employer', 'gross_pay': 'Gross pay per period (current period, not YTD)', 'ytd_gross': 'Gross pay on paystub year to date (cumulative, not current period)', 'pay_frequency': 'The frequency if provided: weekly, biweekly, monthly, semi-monthly', 'pay_period_start': 'Pay period start date', 'pay_period_end': 'Pay period end date', 'hire_date': 'Job hire date', 'direct_deposit_bank_type': 'Direct deposit bank type'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"pay_date\", \"value\": \"2024-12-20\"}, {\"field\": \"employee_name\", \"value\": \"David Martinez\"}, {\"field\": \"employer_name\", \"value\": \"Quick Service Inc\"}, {\"field\": \"gross_pay\", \"value\": \"1045.0\"}, {\"field\": \"ytd_gross\", \"value\": \"50160.0\"}, {\"field\": \"pay_frequency\", \"value\": \"weekly\"}, {\"field\": \"pay_period_start\", \"value\": \"2024-12-09\"}, {\"field\": \"pay_period_end\", \"value\": \"2024-12-15\"}, {\"field\": \"federal_tax\", \"value\": \"104.5\"}, {\"field\": \"state_tax\", \"value\": \"52.25\"}, {\"field\": \"social_security_tax\", \"value\": \"64.79\"}, {\"field\": \"medicare_tax\", \"value\": \"15.15\"}, {\"field\": \"hire_date\", \"value\": \"\"}, {\"field\": \"direct_deposit_bank_type\", \"value\": \"Direct Deposit\"}]}"}]} {"id": "paystub_fin_036_p1_classify", "task": "classify", "class": "paystub", "doc": "paystub_fin_036.pdf", "images": ["images/train/paystub/paystub_fin_036_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"paystub\", \"confidence\": 1.0, \"issuer\": \"Professional Services LLC\", \"holder\": \"Rachel Thompson\", \"page_marker\": \"\"}]}"}]} {"id": "paystub_fin_036_extract", "task": "extract", "class": "paystub", "doc": "paystub_fin_036.pdf", "images": ["images/train/paystub/paystub_fin_036_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('paystub', 'The document class.')\nfields = (['pay_date', 'employee_name', 'employer_name', 'gross_pay', 'ytd_gross', 'pay_frequency', 'pay_period_start', 'pay_period_end', 'federal_tax', 'state_tax', 'social_security_tax', 'medicare_tax', 'hire_date', 'direct_deposit_bank_type'], 'The exact field names to extract.')\nhints = ({'pay_date': 'Date the paycheck was issued', 'employee_name': 'Name of Employee', 'employer_name': 'Name of Employer', 'gross_pay': 'Gross pay per period (current period, not YTD)', 'ytd_gross': 'Gross pay on paystub year to date (cumulative, not current period)', 'pay_frequency': 'The frequency if provided: weekly, biweekly, monthly, semi-monthly', 'pay_period_start': 'Pay period start date', 'pay_period_end': 'Pay period end date', 'hire_date': 'Job hire date', 'direct_deposit_bank_type': 'Direct deposit bank type'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"pay_date\", \"value\": \"2024-12-15\"}, {\"field\": \"employee_name\", \"value\": \"Rachel Thompson\"}, {\"field\": \"employer_name\", \"value\": \"Professional Services LLC\"}, {\"field\": \"gross_pay\", \"value\": \"3000.0\"}, {\"field\": \"ytd_gross\", \"value\": \"69000.0\"}, {\"field\": \"pay_frequency\", \"value\": \"semimonthly\"}, {\"field\": \"pay_period_start\", \"value\": \"2024-12-01\"}, {\"field\": \"pay_period_end\", \"value\": \"2024-12-15\"}, {\"field\": \"federal_tax\", \"value\": \"360.0\"}, {\"field\": \"state_tax\", \"value\": \"150.0\"}, {\"field\": \"social_security_tax\", \"value\": \"186.0\"}, {\"field\": \"medicare_tax\", \"value\": \"43.5\"}, {\"field\": \"hire_date\", \"value\": \"\"}, {\"field\": \"direct_deposit_bank_type\", \"value\": \"Direct Deposit\"}]}"}]} {"id": "paystub_fin_037_p1_classify", "task": "classify", "class": "paystub", "doc": "paystub_fin_037.pdf", "images": ["images/train/paystub/paystub_fin_037_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"paystub\", \"confidence\": 1.0, \"issuer\": \"Johnson Family\", \"holder\": \"Maria Garcia\", \"page_marker\": \"\"}]}"}]} {"id": "paystub_fin_037_extract", "task": "extract", "class": "paystub", "doc": "paystub_fin_037.pdf", "images": ["images/train/paystub/paystub_fin_037_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('paystub', 'The document class.')\nfields = (['pay_date', 'employee_name', 'employer_name', 'gross_pay', 'ytd_gross', 'pay_frequency', 'pay_period_start', 'pay_period_end', 'federal_tax', 'state_tax', 'social_security_tax', 'medicare_tax', 'hire_date', 'direct_deposit_bank_type'], 'The exact field names to extract.')\nhints = ({'pay_date': 'Date the paycheck was issued', 'employee_name': 'Name of Employee', 'employer_name': 'Name of Employer', 'gross_pay': 'Gross pay per period (current period, not YTD)', 'ytd_gross': 'Gross pay on paystub year to date (cumulative, not current period)', 'pay_frequency': 'The frequency if provided: weekly, biweekly, monthly, semi-monthly', 'pay_period_start': 'Pay period start date', 'pay_period_end': 'Pay period end date', 'hire_date': 'Job hire date', 'direct_deposit_bank_type': 'Direct deposit bank type'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"pay_date\", \"value\": \"2024-12-20\"}, {\"field\": \"employee_name\", \"value\": \"Maria Garcia\"}, {\"field\": \"employer_name\", \"value\": \"Johnson Family\"}, {\"field\": \"gross_pay\", \"value\": \"2090.0\"}, {\"field\": \"ytd_gross\", \"value\": \"50160.0\"}, {\"field\": \"pay_frequency\", \"value\": \"biweekly\"}, {\"field\": \"pay_period_start\", \"value\": \"2024-12-01\"}, {\"field\": \"pay_period_end\", \"value\": \"2024-12-15\"}, {\"field\": \"federal_tax\", \"value\": \"209.0\"}, {\"field\": \"state_tax\", \"value\": \"125.4\"}, {\"field\": \"social_security_tax\", \"value\": \"129.58\"}, {\"field\": \"medicare_tax\", \"value\": \"30.31\"}, {\"field\": \"hire_date\", \"value\": \"\"}, {\"field\": \"direct_deposit_bank_type\", \"value\": \"Direct Deposit\"}]}"}]} {"id": "paystub_fin_038_p1_classify", "task": "classify", "class": "paystub", "doc": "paystub_fin_038.pdf", "images": ["images/train/paystub/paystub_fin_038_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"paystub\", \"confidence\": 1.0, \"issuer\": \"Urban Grill Restaurant\", \"holder\": \"Alex Rivera\", \"page_marker\": \"\"}]}"}]} {"id": "paystub_fin_038_extract", "task": "extract", "class": "paystub", "doc": "paystub_fin_038.pdf", "images": ["images/train/paystub/paystub_fin_038_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('paystub', 'The document class.')\nfields = (['pay_date', 'employee_name', 'employer_name', 'gross_pay', 'ytd_gross', 'pay_frequency', 'pay_period_start', 'pay_period_end', 'federal_tax', 'state_tax', 'social_security_tax', 'medicare_tax', 'hire_date', 'direct_deposit_bank_type'], 'The exact field names to extract.')\nhints = ({'pay_date': 'Date the paycheck was issued', 'employee_name': 'Name of Employee', 'employer_name': 'Name of Employer', 'gross_pay': 'Gross pay per period (current period, not YTD)', 'ytd_gross': 'Gross pay on paystub year to date (cumulative, not current period)', 'pay_frequency': 'The frequency if provided: weekly, biweekly, monthly, semi-monthly', 'pay_period_start': 'Pay period start date', 'pay_period_end': 'Pay period end date', 'hire_date': 'Job hire date', 'direct_deposit_bank_type': 'Direct deposit bank type'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"pay_date\", \"value\": \"2024-12-20\"}, {\"field\": \"employee_name\", \"value\": \"Alex Rivera\"}, {\"field\": \"employer_name\", \"value\": \"Urban Grill Restaurant\"}, {\"field\": \"gross_pay\", \"value\": \"1473.6\"}, {\"field\": \"ytd_gross\", \"value\": \"35366.4\"}, {\"field\": \"pay_frequency\", \"value\": \"biweekly\"}, {\"field\": \"pay_period_start\", \"value\": \"2024-12-01\"}, {\"field\": \"pay_period_end\", \"value\": \"2024-12-15\"}, {\"field\": \"federal_tax\", \"value\": \"147.36\"}, {\"field\": \"state_tax\", \"value\": \"\"}, {\"field\": \"social_security_tax\", \"value\": \"91.36\"}, {\"field\": \"medicare_tax\", \"value\": \"21.37\"}, {\"field\": \"hire_date\", \"value\": \"\"}, {\"field\": \"direct_deposit_bank_type\", \"value\": \"\"}]}"}]} {"id": "paystub_fin_039_p1_classify", "task": "classify", "class": "paystub", "doc": "paystub_fin_039.pdf", "images": ["images/train/paystub/paystub_fin_039_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"paystub\", \"confidence\": 1.0, \"issuer\": \"ABC Company\", \"holder\": \"John Doe\", \"page_marker\": \"\"}]}"}]} {"id": "paystub_fin_039_extract", "task": "extract", "class": "paystub", "doc": "paystub_fin_039.pdf", "images": ["images/train/paystub/paystub_fin_039_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('paystub', 'The document class.')\nfields = (['pay_date', 'employee_name', 'employer_name', 'gross_pay', 'ytd_gross', 'pay_frequency', 'pay_period_start', 'pay_period_end', 'federal_tax', 'state_tax', 'social_security_tax', 'medicare_tax', 'hire_date', 'direct_deposit_bank_type'], 'The exact field names to extract.')\nhints = ({'pay_date': 'Date the paycheck was issued', 'employee_name': 'Name of Employee', 'employer_name': 'Name of Employer', 'gross_pay': 'Gross pay per period (current period, not YTD)', 'ytd_gross': 'Gross pay on paystub year to date (cumulative, not current period)', 'pay_frequency': 'The frequency if provided: weekly, biweekly, monthly, semi-monthly', 'pay_period_start': 'Pay period start date', 'pay_period_end': 'Pay period end date', 'hire_date': 'Job hire date', 'direct_deposit_bank_type': 'Direct deposit bank type'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"pay_date\", \"value\": \"2021-07-01\"}, {\"field\": \"employee_name\", \"value\": \"John Doe\"}, {\"field\": \"employer_name\", \"value\": \"ABC Company\"}, {\"field\": \"gross_pay\", \"value\": \"5000.0\"}, {\"field\": \"ytd_gross\", \"value\": \"65000.0\"}, {\"field\": \"pay_frequency\", \"value\": \"\"}, {\"field\": \"pay_period_start\", \"value\": \"2021-06-16\"}, {\"field\": \"pay_period_end\", \"value\": \"2021-06-30\"}, {\"field\": \"federal_tax\", \"value\": \"825.38\"}, {\"field\": \"state_tax\", \"value\": \"360.28\"}, {\"field\": \"social_security_tax\", \"value\": \"310.0\"}, {\"field\": \"medicare_tax\", \"value\": \"72.5\"}, {\"field\": \"hire_date\", \"value\": \"\"}, {\"field\": \"direct_deposit_bank_type\", \"value\": \"\"}]}"}]} {"id": "paystub_fin_040_p1_classify", "task": "classify", "class": "paystub", "doc": "paystub_fin_040.pdf", "images": ["images/train/paystub/paystub_fin_040_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"paystub\", \"confidence\": 1.0, \"issuer\": \"ABC Company\", \"holder\": \"John Smith\", \"page_marker\": \"\"}]}"}]} {"id": "paystub_fin_040_extract", "task": "extract", "class": "paystub", "doc": "paystub_fin_040.pdf", "images": ["images/train/paystub/paystub_fin_040_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('paystub', 'The document class.')\nfields = (['pay_date', 'employee_name', 'employer_name', 'gross_pay', 'ytd_gross', 'pay_frequency', 'pay_period_start', 'pay_period_end', 'federal_tax', 'state_tax', 'social_security_tax', 'medicare_tax', 'hire_date', 'direct_deposit_bank_type'], 'The exact field names to extract.')\nhints = ({'pay_date': 'Date the paycheck was issued', 'employee_name': 'Name of Employee', 'employer_name': 'Name of Employer', 'gross_pay': 'Gross pay per period (current period, not YTD)', 'ytd_gross': 'Gross pay on paystub year to date (cumulative, not current period)', 'pay_frequency': 'The frequency if provided: weekly, biweekly, monthly, semi-monthly', 'pay_period_start': 'Pay period start date', 'pay_period_end': 'Pay period end date', 'hire_date': 'Job hire date', 'direct_deposit_bank_type': 'Direct deposit bank type'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"pay_date\", \"value\": \"2020-09-01\"}, {\"field\": \"employee_name\", \"value\": \"John Smith\"}, {\"field\": \"employer_name\", \"value\": \"ABC Company\"}, {\"field\": \"gross_pay\", \"value\": \"5820.0\"}, {\"field\": \"ytd_gross\", \"value\": \"46560.0\"}, {\"field\": \"pay_frequency\", \"value\": \"monthly\"}, {\"field\": \"pay_period_start\", \"value\": \"2020-08-01\"}, {\"field\": \"pay_period_end\", \"value\": \"2020-08-31\"}, {\"field\": \"federal_tax\", \"value\": \"12.0\"}, {\"field\": \"state_tax\", \"value\": \"10.0\"}, {\"field\": \"social_security_tax\", \"value\": \"15.0\"}, {\"field\": \"medicare_tax\", \"value\": \"40.0\"}, {\"field\": \"hire_date\", \"value\": \"\"}, {\"field\": \"direct_deposit_bank_type\", \"value\": \"Bank Transfer\"}]}"}]} {"id": "paystub_fin_041_p1_classify", "task": "classify", "class": "paystub", "doc": "paystub_fin_041.pdf", "images": ["images/train/paystub/paystub_fin_041_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"paystub\", \"confidence\": 1.0, \"issuer\": \"ABC Company\", \"holder\": \"John Doe\", \"page_marker\": \"\"}]}"}]} {"id": "paystub_fin_041_extract", "task": "extract", "class": "paystub", "doc": "paystub_fin_041.pdf", "images": ["images/train/paystub/paystub_fin_041_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('paystub', 'The document class.')\nfields = (['pay_date', 'employee_name', 'employer_name', 'gross_pay', 'ytd_gross', 'pay_frequency', 'pay_period_start', 'pay_period_end', 'federal_tax', 'state_tax', 'social_security_tax', 'medicare_tax', 'hire_date', 'direct_deposit_bank_type'], 'The exact field names to extract.')\nhints = ({'pay_date': 'Date the paycheck was issued', 'employee_name': 'Name of Employee', 'employer_name': 'Name of Employer', 'gross_pay': 'Gross pay per period (current period, not YTD)', 'ytd_gross': 'Gross pay on paystub year to date (cumulative, not current period)', 'pay_frequency': 'The frequency if provided: weekly, biweekly, monthly, semi-monthly', 'pay_period_start': 'Pay period start date', 'pay_period_end': 'Pay period end date', 'hire_date': 'Job hire date', 'direct_deposit_bank_type': 'Direct deposit bank type'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"pay_date\", \"value\": \"2021-07-01\"}, {\"field\": \"employee_name\", \"value\": \"John Doe\"}, {\"field\": \"employer_name\", \"value\": \"ABC Company\"}, {\"field\": \"gross_pay\", \"value\": \"5000.0\"}, {\"field\": \"ytd_gross\", \"value\": \"65000.0\"}, {\"field\": \"pay_frequency\", \"value\": \"semimonthly\"}, {\"field\": \"pay_period_start\", \"value\": \"2021-06-16\"}, {\"field\": \"pay_period_end\", \"value\": \"2021-06-30\"}, {\"field\": \"federal_tax\", \"value\": \"825.38\"}, {\"field\": \"state_tax\", \"value\": \"360.28\"}, {\"field\": \"social_security_tax\", \"value\": \"310.0\"}, {\"field\": \"medicare_tax\", \"value\": \"72.5\"}, {\"field\": \"hire_date\", \"value\": \"\"}, {\"field\": \"direct_deposit_bank_type\", \"value\": \"\"}]}"}]} {"id": "paystub_fin_043_p1_classify", "task": "classify", "class": "paystub", "doc": "paystub_fin_043.pdf", "images": ["images/train/paystub/paystub_fin_043_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"paystub\", \"confidence\": 1.0, \"issuer\": \"[COMPANY NAME]\", \"holder\": \"Andrew Stewart Williams\", \"page_marker\": \"\"}]}"}]} {"id": "paystub_fin_043_extract", "task": "extract", "class": "paystub", "doc": "paystub_fin_043.pdf", "images": ["images/train/paystub/paystub_fin_043_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('paystub', 'The document class.')\nfields = (['pay_date', 'employee_name', 'employer_name', 'gross_pay', 'ytd_gross', 'pay_frequency', 'pay_period_start', 'pay_period_end', 'federal_tax', 'state_tax', 'social_security_tax', 'medicare_tax', 'hire_date', 'direct_deposit_bank_type'], 'The exact field names to extract.')\nhints = ({'pay_date': 'Date the paycheck was issued', 'employee_name': 'Name of Employee', 'employer_name': 'Name of Employer', 'gross_pay': 'Gross pay per period (current period, not YTD)', 'ytd_gross': 'Gross pay on paystub year to date (cumulative, not current period)', 'pay_frequency': 'The frequency if provided: weekly, biweekly, monthly, semi-monthly', 'pay_period_start': 'Pay period start date', 'pay_period_end': 'Pay period end date', 'hire_date': 'Job hire date', 'direct_deposit_bank_type': 'Direct deposit bank type'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"pay_date\", \"value\": \"2019-01-31\"}, {\"field\": \"employee_name\", \"value\": \"Andrew Stewart Williams\"}, {\"field\": \"employer_name\", \"value\": \"[COMPANY NAME]\"}, {\"field\": \"gross_pay\", \"value\": \"4400.0\"}, {\"field\": \"ytd_gross\", \"value\": \"\"}, {\"field\": \"pay_frequency\", \"value\": \"\"}, {\"field\": \"pay_period_start\", \"value\": \"2020-01-04\"}, {\"field\": \"pay_period_end\", \"value\": \"2019-01-25\"}, {\"field\": \"federal_tax\", \"value\": \"28.5\"}, {\"field\": \"state_tax\", \"value\": \"\"}, {\"field\": \"social_security_tax\", \"value\": \"50.0\"}, {\"field\": \"medicare_tax\", \"value\": \"\"}, {\"field\": \"hire_date\", \"value\": \"\"}, {\"field\": \"direct_deposit_bank_type\", \"value\": \"PQRS International Bank\"}]}"}]} {"id": "paystub_fin_044_p1_classify", "task": "classify", "class": "paystub", "doc": "paystub_fin_044.pdf", "images": ["images/train/paystub/paystub_fin_044_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"paystub\", \"confidence\": 1.0, \"issuer\": \"EXAMPLE COMPANY, INC.\", \"holder\": \"JOHN DOE\", \"page_marker\": \"\"}]}"}]} {"id": "paystub_fin_044_extract", "task": "extract", "class": "paystub", "doc": "paystub_fin_044.pdf", "images": ["images/train/paystub/paystub_fin_044_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('paystub', 'The document class.')\nfields = (['pay_date', 'employee_name', 'employer_name', 'gross_pay', 'ytd_gross', 'pay_frequency', 'pay_period_start', 'pay_period_end', 'federal_tax', 'state_tax', 'social_security_tax', 'medicare_tax', 'hire_date', 'direct_deposit_bank_type'], 'The exact field names to extract.')\nhints = ({'pay_date': 'Date the paycheck was issued', 'employee_name': 'Name of Employee', 'employer_name': 'Name of Employer', 'gross_pay': 'Gross pay per period (current period, not YTD)', 'ytd_gross': 'Gross pay on paystub year to date (cumulative, not current period)', 'pay_frequency': 'The frequency if provided: weekly, biweekly, monthly, semi-monthly', 'pay_period_start': 'Pay period start date', 'pay_period_end': 'Pay period end date', 'hire_date': 'Job hire date', 'direct_deposit_bank_type': 'Direct deposit bank type'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"pay_date\", \"value\": \"2013-07-23\"}, {\"field\": \"employee_name\", \"value\": \"JOHN DOE\"}, {\"field\": \"employer_name\", \"value\": \"EXAMPLE COMPANY, INC.\"}, {\"field\": \"gross_pay\", \"value\": \"1200.0\"}, {\"field\": \"ytd_gross\", \"value\": \"17622.5\"}, {\"field\": \"pay_frequency\", \"value\": \"\"}, {\"field\": \"pay_period_start\", \"value\": \"2013-06-19\"}, {\"field\": \"pay_period_end\", \"value\": \"2013-07-02\"}, {\"field\": \"federal_tax\", \"value\": \"157.39\"}, {\"field\": \"state_tax\", \"value\": \"40.06\"}, {\"field\": \"social_security_tax\", \"value\": \"71.69\"}, {\"field\": \"medicare_tax\", \"value\": \"16.77\"}, {\"field\": \"hire_date\", \"value\": \"\"}, {\"field\": \"direct_deposit_bank_type\", \"value\": \"SAMPLE BANK\"}]}"}]} {"id": "paystub_fin_046_p1_classify", "task": "classify", "class": "paystub", "doc": "paystub_fin_046.pdf", "images": ["images/train/paystub/paystub_fin_046_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"paystub\", \"confidence\": 1.0, \"issuer\": \"Guardian National Bank\", \"holder\": \"Julie Jones\", \"page_marker\": \"\"}]}"}]} {"id": "paystub_fin_046_extract", "task": "extract", "class": "paystub", "doc": "paystub_fin_046.pdf", "images": ["images/train/paystub/paystub_fin_046_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('paystub', 'The document class.')\nfields = (['pay_date', 'employee_name', 'employer_name', 'gross_pay', 'ytd_gross', 'pay_frequency', 'pay_period_start', 'pay_period_end', 'federal_tax', 'state_tax', 'social_security_tax', 'medicare_tax', 'hire_date', 'direct_deposit_bank_type'], 'The exact field names to extract.')\nhints = ({'pay_date': 'Date the paycheck was issued', 'employee_name': 'Name of Employee', 'employer_name': 'Name of Employer', 'gross_pay': 'Gross pay per period (current period, not YTD)', 'ytd_gross': 'Gross pay on paystub year to date (cumulative, not current period)', 'pay_frequency': 'The frequency if provided: weekly, biweekly, monthly, semi-monthly', 'pay_period_start': 'Pay period start date', 'pay_period_end': 'Pay period end date', 'hire_date': 'Job hire date', 'direct_deposit_bank_type': 'Direct deposit bank type'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"pay_date\", \"value\": \"2020-05-01\"}, {\"field\": \"employee_name\", \"value\": \"Julie Jones\"}, {\"field\": \"employer_name\", \"value\": \"Guardian National Bank\"}, {\"field\": \"gross_pay\", \"value\": \"\"}, {\"field\": \"ytd_gross\", \"value\": \"\"}, {\"field\": \"pay_frequency\", \"value\": \"biweekly\"}, {\"field\": \"pay_period_start\", \"value\": \"2020-04-01\"}, {\"field\": \"pay_period_end\", \"value\": \"2020-04-14\"}, {\"field\": \"federal_tax\", \"value\": \"122.05\"}, {\"field\": \"state_tax\", \"value\": \"42.27\"}, {\"field\": \"social_security_tax\", \"value\": \"\"}, {\"field\": \"medicare_tax\", \"value\": \"\"}, {\"field\": \"hire_date\", \"value\": \"\"}, {\"field\": \"direct_deposit_bank_type\", \"value\": \"\"}]}"}]} {"id": "paystub_fin_047_p1_classify", "task": "classify", "class": "paystub", "doc": "paystub_fin_047.pdf", "images": ["images/train/paystub/paystub_fin_047_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"paystub\", \"confidence\": 1.0, \"issuer\": \"MYERS COMPANY\", \"holder\": \"Carl Y. Anderson\", \"page_marker\": \"\"}]}"}]} {"id": "paystub_fin_047_extract", "task": "extract", "class": "paystub", "doc": "paystub_fin_047.pdf", "images": ["images/train/paystub/paystub_fin_047_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('paystub', 'The document class.')\nfields = (['pay_date', 'employee_name', 'employer_name', 'gross_pay', 'ytd_gross', 'pay_frequency', 'pay_period_start', 'pay_period_end', 'federal_tax', 'state_tax', 'social_security_tax', 'medicare_tax', 'hire_date', 'direct_deposit_bank_type'], 'The exact field names to extract.')\nhints = ({'pay_date': 'Date the paycheck was issued', 'employee_name': 'Name of Employee', 'employer_name': 'Name of Employer', 'gross_pay': 'Gross pay per period (current period, not YTD)', 'ytd_gross': 'Gross pay on paystub year to date (cumulative, not current period)', 'pay_frequency': 'The frequency if provided: weekly, biweekly, monthly, semi-monthly', 'pay_period_start': 'Pay period start date', 'pay_period_end': 'Pay period end date', 'hire_date': 'Job hire date', 'direct_deposit_bank_type': 'Direct deposit bank type'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"pay_date\", \"value\": \"2023-09-01\"}, {\"field\": \"employee_name\", \"value\": \"Carl Y. Anderson\"}, {\"field\": \"employer_name\", \"value\": \"MYERS COMPANY\"}, {\"field\": \"gross_pay\", \"value\": \"5450.0\"}, {\"field\": \"ytd_gross\", \"value\": \"41660.0\"}, {\"field\": \"pay_frequency\", \"value\": \"\"}, {\"field\": \"pay_period_start\", \"value\": \"2023-08-01\"}, {\"field\": \"pay_period_end\", \"value\": \"2023-08-31\"}, {\"field\": \"federal_tax\", \"value\": \"56.0\"}, {\"field\": \"state_tax\", \"value\": \"12.5\"}, {\"field\": \"social_security_tax\", \"value\": \"256.0\"}, {\"field\": \"medicare_tax\", \"value\": \"145.0\"}, {\"field\": \"hire_date\", \"value\": \"\"}, {\"field\": \"direct_deposit_bank_type\", \"value\": \"\"}]}"}]} {"id": "paystub_fin_048_p1_classify", "task": "classify", "class": "paystub", "doc": "paystub_fin_048.pdf", "images": ["images/train/paystub/paystub_fin_048_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"paystub\", \"confidence\": 1.0, \"issuer\": \"University of New Mexico\", \"holder\": \"John Q. Doe\", \"page_marker\": \"\"}]}"}]} {"id": "paystub_fin_048_extract", "task": "extract", "class": "paystub", "doc": "paystub_fin_048.pdf", "images": ["images/train/paystub/paystub_fin_048_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('paystub', 'The document class.')\nfields = (['pay_date', 'employee_name', 'employer_name', 'gross_pay', 'ytd_gross', 'pay_frequency', 'pay_period_start', 'pay_period_end', 'federal_tax', 'state_tax', 'social_security_tax', 'medicare_tax', 'hire_date', 'direct_deposit_bank_type'], 'The exact field names to extract.')\nhints = ({'pay_date': 'Date the paycheck was issued', 'employee_name': 'Name of Employee', 'employer_name': 'Name of Employer', 'gross_pay': 'Gross pay per period (current period, not YTD)', 'ytd_gross': 'Gross pay on paystub year to date (cumulative, not current period)', 'pay_frequency': 'The frequency if provided: weekly, biweekly, monthly, semi-monthly', 'pay_period_start': 'Pay period start date', 'pay_period_end': 'Pay period end date', 'hire_date': 'Job hire date', 'direct_deposit_bank_type': 'Direct deposit bank type'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"pay_date\", \"value\": \"2017-01-31\"}, {\"field\": \"employee_name\", \"value\": \"John Q. Doe\"}, {\"field\": \"employer_name\", \"value\": \"University of New Mexico\"}, {\"field\": \"gross_pay\", \"value\": \"4313.54\"}, {\"field\": \"ytd_gross\", \"value\": \"4313.54\"}, {\"field\": \"pay_frequency\", \"value\": \"monthly\"}, {\"field\": \"pay_period_start\", \"value\": \"2017-01-01\"}, {\"field\": \"pay_period_end\", \"value\": \"2017-01-31\"}, {\"field\": \"federal_tax\", \"value\": \"428.75\"}, {\"field\": \"state_tax\", \"value\": \"129.46\"}, {\"field\": \"social_security_tax\", \"value\": \"254.7\"}, {\"field\": \"medicare_tax\", \"value\": \"59.57\"}, {\"field\": \"hire_date\", \"value\": \"\"}, {\"field\": \"direct_deposit_bank_type\", \"value\": \"My Bank\"}]}"}]} {"id": "paystub_fin_049_p1_classify", "task": "classify", "class": "paystub", "doc": "paystub_fin_049.pdf", "images": ["images/train/paystub/paystub_fin_049_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"paystub\", \"confidence\": 1.0, \"issuer\": \"SMITH AND COMPANY, INC.\", \"holder\": \"Johnson, Bob\", \"page_marker\": \"\"}]}"}]} {"id": "paystub_fin_049_extract", "task": "extract", "class": "paystub", "doc": "paystub_fin_049.pdf", "images": ["images/train/paystub/paystub_fin_049_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('paystub', 'The document class.')\nfields = (['pay_date', 'employee_name', 'employer_name', 'gross_pay', 'ytd_gross', 'pay_frequency', 'pay_period_start', 'pay_period_end', 'federal_tax', 'state_tax', 'social_security_tax', 'medicare_tax', 'hire_date', 'direct_deposit_bank_type'], 'The exact field names to extract.')\nhints = ({'pay_date': 'Date the paycheck was issued', 'employee_name': 'Name of Employee', 'employer_name': 'Name of Employer', 'gross_pay': 'Gross pay per period (current period, not YTD)', 'ytd_gross': 'Gross pay on paystub year to date (cumulative, not current period)', 'pay_frequency': 'The frequency if provided: weekly, biweekly, monthly, semi-monthly', 'pay_period_start': 'Pay period start date', 'pay_period_end': 'Pay period end date', 'hire_date': 'Job hire date', 'direct_deposit_bank_type': 'Direct deposit bank type'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"pay_date\", \"value\": \"\"}, {\"field\": \"employee_name\", \"value\": \"Johnson, Bob\"}, {\"field\": \"employer_name\", \"value\": \"SMITH AND COMPANY, INC.\"}, {\"field\": \"gross_pay\", \"value\": \"774.0\"}, {\"field\": \"ytd_gross\", \"value\": \"\"}, {\"field\": \"pay_frequency\", \"value\": \"weekly\"}, {\"field\": \"pay_period_start\", \"value\": \"\"}, {\"field\": \"pay_period_end\", \"value\": \"\"}, {\"field\": \"federal_tax\", \"value\": \"60.45\"}, {\"field\": \"state_tax\", \"value\": \"10.04\"}, {\"field\": \"social_security_tax\", \"value\": \"47.99\"}, {\"field\": \"medicare_tax\", \"value\": \"11.22\"}, {\"field\": \"hire_date\", \"value\": \"\"}, {\"field\": \"direct_deposit_bank_type\", \"value\": \"\"}]}"}]} {"id": "paystub_fin_050_p1_classify", "task": "classify", "class": "paystub", "doc": "paystub_fin_050.pdf", "images": ["images/train/paystub/paystub_fin_050_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"paystub\", \"confidence\": 1.0, \"issuer\": \"Sample Company Name\", \"holder\": \"\", \"page_marker\": \"\"}]}"}]} {"id": "paystub_fin_050_extract", "task": "extract", "class": "paystub", "doc": "paystub_fin_050.pdf", "images": ["images/train/paystub/paystub_fin_050_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('paystub', 'The document class.')\nfields = (['pay_date', 'employee_name', 'employer_name', 'gross_pay', 'ytd_gross', 'pay_frequency', 'pay_period_start', 'pay_period_end', 'federal_tax', 'state_tax', 'social_security_tax', 'medicare_tax', 'hire_date', 'direct_deposit_bank_type'], 'The exact field names to extract.')\nhints = ({'pay_date': 'Date the paycheck was issued', 'employee_name': 'Name of Employee', 'employer_name': 'Name of Employer', 'gross_pay': 'Gross pay per period (current period, not YTD)', 'ytd_gross': 'Gross pay on paystub year to date (cumulative, not current period)', 'pay_frequency': 'The frequency if provided: weekly, biweekly, monthly, semi-monthly', 'pay_period_start': 'Pay period start date', 'pay_period_end': 'Pay period end date', 'hire_date': 'Job hire date', 'direct_deposit_bank_type': 'Direct deposit bank type'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"pay_date\", \"value\": \"2018-04-19\"}, {\"field\": \"employee_name\", \"value\": \"\"}, {\"field\": \"employer_name\", \"value\": \"Sample Company Name\"}, {\"field\": \"gross_pay\", \"value\": \"400.0\"}, {\"field\": \"ytd_gross\", \"value\": \"6400.0\"}, {\"field\": \"pay_frequency\", \"value\": \"weekly\"}, {\"field\": \"pay_period_start\", \"value\": \"2018-04-12\"}, {\"field\": \"pay_period_end\", \"value\": \"2018-04-18\"}, {\"field\": \"federal_tax\", \"value\": \"35.82\"}, {\"field\": \"state_tax\", \"value\": \"20.0\"}, {\"field\": \"social_security_tax\", \"value\": \"24.8\"}, {\"field\": \"medicare_tax\", \"value\": \"5.8\"}, {\"field\": \"hire_date\", \"value\": \"\"}, {\"field\": \"direct_deposit_bank_type\", \"value\": \"\"}]}"}]} {"id": "paystub_fin_051_p1_classify", "task": "classify", "class": "paystub", "doc": "paystub_fin_051.pdf", "images": ["images/train/paystub/paystub_fin_051_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"paystub\", \"confidence\": 1.0, \"issuer\": \"ABC Company\", \"holder\": \"Jane Dough\", \"page_marker\": \"\"}]}"}]} {"id": "paystub_fin_051_extract", "task": "extract", "class": "paystub", "doc": "paystub_fin_051.pdf", "images": ["images/train/paystub/paystub_fin_051_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('paystub', 'The document class.')\nfields = (['pay_date', 'employee_name', 'employer_name', 'gross_pay', 'ytd_gross', 'pay_frequency', 'pay_period_start', 'pay_period_end', 'federal_tax', 'state_tax', 'social_security_tax', 'medicare_tax', 'hire_date', 'direct_deposit_bank_type'], 'The exact field names to extract.')\nhints = ({'pay_date': 'Date the paycheck was issued', 'employee_name': 'Name of Employee', 'employer_name': 'Name of Employer', 'gross_pay': 'Gross pay per period (current period, not YTD)', 'ytd_gross': 'Gross pay on paystub year to date (cumulative, not current period)', 'pay_frequency': 'The frequency if provided: weekly, biweekly, monthly, semi-monthly', 'pay_period_start': 'Pay period start date', 'pay_period_end': 'Pay period end date', 'hire_date': 'Job hire date', 'direct_deposit_bank_type': 'Direct deposit bank type'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"pay_date\", \"value\": \"\"}, {\"field\": \"employee_name\", \"value\": \"Jane Dough\"}, {\"field\": \"employer_name\", \"value\": \"ABC Company\"}, {\"field\": \"gross_pay\", \"value\": \"1346.15\"}, {\"field\": \"ytd_gross\", \"value\": \"1346.15\"}, {\"field\": \"pay_frequency\", \"value\": \"\"}, {\"field\": \"pay_period_start\", \"value\": \"2014-07-17\"}, {\"field\": \"pay_period_end\", \"value\": \"2014-07-31\"}, {\"field\": \"federal_tax\", \"value\": \"177.04\"}, {\"field\": \"state_tax\", \"value\": \"\"}, {\"field\": \"social_security_tax\", \"value\": \"\"}, {\"field\": \"medicare_tax\", \"value\": \"\"}, {\"field\": \"hire_date\", \"value\": \"\"}, {\"field\": \"direct_deposit_bank_type\", \"value\": \"\"}]}"}]} {"id": "paystub_fin_052_p1_classify", "task": "classify", "class": "paystub", "doc": "paystub_fin_052.pdf", "images": ["images/train/paystub/paystub_fin_052_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"paystub\", \"confidence\": 1.0, \"issuer\": \"Payroll Company\", \"holder\": \"John B Doe\", \"page_marker\": \"\"}]}"}]} {"id": "paystub_fin_052_extract", "task": "extract", "class": "paystub", "doc": "paystub_fin_052.pdf", "images": ["images/train/paystub/paystub_fin_052_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('paystub', 'The document class.')\nfields = (['pay_date', 'employee_name', 'employer_name', 'gross_pay', 'ytd_gross', 'pay_frequency', 'pay_period_start', 'pay_period_end', 'federal_tax', 'state_tax', 'social_security_tax', 'medicare_tax', 'hire_date', 'direct_deposit_bank_type'], 'The exact field names to extract.')\nhints = ({'pay_date': 'Date the paycheck was issued', 'employee_name': 'Name of Employee', 'employer_name': 'Name of Employer', 'gross_pay': 'Gross pay per period (current period, not YTD)', 'ytd_gross': 'Gross pay on paystub year to date (cumulative, not current period)', 'pay_frequency': 'The frequency if provided: weekly, biweekly, monthly, semi-monthly', 'pay_period_start': 'Pay period start date', 'pay_period_end': 'Pay period end date', 'hire_date': 'Job hire date', 'direct_deposit_bank_type': 'Direct deposit bank type'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"pay_date\", \"value\": \"2012-01-16\"}, {\"field\": \"employee_name\", \"value\": \"John B Doe\"}, {\"field\": \"employer_name\", \"value\": \"Payroll Company\"}, {\"field\": \"gross_pay\", \"value\": \"660.0\"}, {\"field\": \"ytd_gross\", \"value\": \"660.0\"}, {\"field\": \"pay_frequency\", \"value\": \"\"}, {\"field\": \"pay_period_start\", \"value\": \"2012-01-01\"}, {\"field\": \"pay_period_end\", \"value\": \"2012-01-15\"}, {\"field\": \"federal_tax\", \"value\": \"\"}, {\"field\": \"state_tax\", \"value\": \"2.53\"}, {\"field\": \"social_security_tax\", \"value\": \"27.72\"}, {\"field\": \"medicare_tax\", \"value\": \"9.57\"}, {\"field\": \"hire_date\", \"value\": \"\"}, {\"field\": \"direct_deposit_bank_type\", \"value\": \"\"}]}"}]} {"id": "paystub_fin_053_p1_classify", "task": "classify", "class": "paystub", "doc": "paystub_fin_053.pdf", "images": ["images/train/paystub/paystub_fin_053_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"paystub\", \"confidence\": 1.0, \"issuer\": \"XYZ Company\", \"holder\": \"\", \"page_marker\": \"\"}]}"}]} {"id": "paystub_fin_053_extract", "task": "extract", "class": "paystub", "doc": "paystub_fin_053.pdf", "images": ["images/train/paystub/paystub_fin_053_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('paystub', 'The document class.')\nfields = (['pay_date', 'employee_name', 'employer_name', 'gross_pay', 'ytd_gross', 'pay_frequency', 'pay_period_start', 'pay_period_end', 'federal_tax', 'state_tax', 'social_security_tax', 'medicare_tax', 'hire_date', 'direct_deposit_bank_type'], 'The exact field names to extract.')\nhints = ({'pay_date': 'Date the paycheck was issued', 'employee_name': 'Name of Employee', 'employer_name': 'Name of Employer', 'gross_pay': 'Gross pay per period (current period, not YTD)', 'ytd_gross': 'Gross pay on paystub year to date (cumulative, not current period)', 'pay_frequency': 'The frequency if provided: weekly, biweekly, monthly, semi-monthly', 'pay_period_start': 'Pay period start date', 'pay_period_end': 'Pay period end date', 'hire_date': 'Job hire date', 'direct_deposit_bank_type': 'Direct deposit bank type'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"pay_date\", \"value\": \"\"}, {\"field\": \"employee_name\", \"value\": \"\"}, {\"field\": \"employer_name\", \"value\": \"XYZ Company\"}, {\"field\": \"gross_pay\", \"value\": \"\"}, {\"field\": \"ytd_gross\", \"value\": \"\"}, {\"field\": \"pay_frequency\", \"value\": \"\"}, {\"field\": \"pay_period_start\", \"value\": \"\"}, {\"field\": \"pay_period_end\", \"value\": \"\"}, {\"field\": \"federal_tax\", \"value\": \"\"}, {\"field\": \"state_tax\", \"value\": \"\"}, {\"field\": \"social_security_tax\", \"value\": \"\"}, {\"field\": \"medicare_tax\", \"value\": \"\"}, {\"field\": \"hire_date\", \"value\": \"\"}, {\"field\": \"direct_deposit_bank_type\", \"value\": \"\"}]}"}]} {"id": "paystub_fin_055_p1_classify", "task": "classify", "class": "paystub", "doc": "paystub_fin_055.pdf", "images": ["images/train/paystub/paystub_fin_055_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"paystub\", \"confidence\": 1.0, \"issuer\": \"ABC Company\", \"holder\": \"Peter Sims\", \"page_marker\": \"\"}]}"}]} {"id": "paystub_fin_055_extract", "task": "extract", "class": "paystub", "doc": "paystub_fin_055.pdf", "images": ["images/train/paystub/paystub_fin_055_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('paystub', 'The document class.')\nfields = (['pay_date', 'employee_name', 'employer_name', 'gross_pay', 'ytd_gross', 'pay_frequency', 'pay_period_start', 'pay_period_end', 'federal_tax', 'state_tax', 'social_security_tax', 'medicare_tax', 'hire_date', 'direct_deposit_bank_type'], 'The exact field names to extract.')\nhints = ({'pay_date': 'Date the paycheck was issued', 'employee_name': 'Name of Employee', 'employer_name': 'Name of Employer', 'gross_pay': 'Gross pay per period (current period, not YTD)', 'ytd_gross': 'Gross pay on paystub year to date (cumulative, not current period)', 'pay_frequency': 'The frequency if provided: weekly, biweekly, monthly, semi-monthly', 'pay_period_start': 'Pay period start date', 'pay_period_end': 'Pay period end date', 'hire_date': 'Job hire date', 'direct_deposit_bank_type': 'Direct deposit bank type'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"pay_date\", \"value\": \"2020-05-16\"}, {\"field\": \"employee_name\", \"value\": \"Peter Sims\"}, {\"field\": \"employer_name\", \"value\": \"ABC Company\"}, {\"field\": \"gross_pay\", \"value\": \"4000.0\"}, {\"field\": \"ytd_gross\", \"value\": \"20000.0\"}, {\"field\": \"pay_frequency\", \"value\": \"\"}, {\"field\": \"pay_period_start\", \"value\": \"2020-05-15\"}, {\"field\": \"pay_period_end\", \"value\": \"2020-05-16\"}, {\"field\": \"federal_tax\", \"value\": \"339.54\"}, {\"field\": \"state_tax\", \"value\": \"264.0\"}, {\"field\": \"social_security_tax\", \"value\": \"248.0\"}, {\"field\": \"medicare_tax\", \"value\": \"58.0\"}, {\"field\": \"hire_date\", \"value\": \"\"}, {\"field\": \"direct_deposit_bank_type\", \"value\": \"\"}]}"}]} {"id": "paystub_fin_057_p1_classify", "task": "classify", "class": "paystub", "doc": "paystub_fin_057.pdf", "images": ["images/train/paystub/paystub_fin_057_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"paystub\", \"confidence\": 1.0, \"issuer\": \"ABC Company\", \"holder\": \"Peter Sims\", \"page_marker\": \"\"}]}"}]} {"id": "paystub_fin_057_extract", "task": "extract", "class": "paystub", "doc": "paystub_fin_057.pdf", "images": ["images/train/paystub/paystub_fin_057_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('paystub', 'The document class.')\nfields = (['pay_date', 'employee_name', 'employer_name', 'gross_pay', 'ytd_gross', 'pay_frequency', 'pay_period_start', 'pay_period_end', 'federal_tax', 'state_tax', 'social_security_tax', 'medicare_tax', 'hire_date', 'direct_deposit_bank_type'], 'The exact field names to extract.')\nhints = ({'pay_date': 'Date the paycheck was issued', 'employee_name': 'Name of Employee', 'employer_name': 'Name of Employer', 'gross_pay': 'Gross pay per period (current period, not YTD)', 'ytd_gross': 'Gross pay on paystub year to date (cumulative, not current period)', 'pay_frequency': 'The frequency if provided: weekly, biweekly, monthly, semi-monthly', 'pay_period_start': 'Pay period start date', 'pay_period_end': 'Pay period end date', 'hire_date': 'Job hire date', 'direct_deposit_bank_type': 'Direct deposit bank type'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"pay_date\", \"value\": \"2020-05-16\"}, {\"field\": \"employee_name\", \"value\": \"Peter Sims\"}, {\"field\": \"employer_name\", \"value\": \"ABC Company\"}, {\"field\": \"gross_pay\", \"value\": \"460.0\"}, {\"field\": \"ytd_gross\", \"value\": \"1300.0\"}, {\"field\": \"pay_frequency\", \"value\": \"weekly\"}, {\"field\": \"pay_period_start\", \"value\": \"2020-05-15\"}, {\"field\": \"pay_period_end\", \"value\": \"2020-05-16\"}, {\"field\": \"federal_tax\", \"value\": \"45.0\"}, {\"field\": \"state_tax\", \"value\": \"25.0\"}, {\"field\": \"social_security_tax\", \"value\": \"20.0\"}, {\"field\": \"medicare_tax\", \"value\": \"10.0\"}, {\"field\": \"hire_date\", \"value\": \"\"}, {\"field\": \"direct_deposit_bank_type\", \"value\": \"\"}]}"}]} {"id": "paystub_fin_058_p1_classify", "task": "classify", "class": "paystub", "doc": "paystub_fin_058.pdf", "images": ["images/train/paystub/paystub_fin_058_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"paystub\", \"confidence\": 1.0, \"issuer\": \"ABC Company\", \"holder\": \"Peter Sims\", \"page_marker\": \"\"}]}"}]} {"id": "paystub_fin_058_extract", "task": "extract", "class": "paystub", "doc": "paystub_fin_058.pdf", "images": ["images/train/paystub/paystub_fin_058_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('paystub', 'The document class.')\nfields = (['pay_date', 'employee_name', 'employer_name', 'gross_pay', 'ytd_gross', 'pay_frequency', 'pay_period_start', 'pay_period_end', 'federal_tax', 'state_tax', 'social_security_tax', 'medicare_tax', 'hire_date', 'direct_deposit_bank_type'], 'The exact field names to extract.')\nhints = ({'pay_date': 'Date the paycheck was issued', 'employee_name': 'Name of Employee', 'employer_name': 'Name of Employer', 'gross_pay': 'Gross pay per period (current period, not YTD)', 'ytd_gross': 'Gross pay on paystub year to date (cumulative, not current period)', 'pay_frequency': 'The frequency if provided: weekly, biweekly, monthly, semi-monthly', 'pay_period_start': 'Pay period start date', 'pay_period_end': 'Pay period end date', 'hire_date': 'Job hire date', 'direct_deposit_bank_type': 'Direct deposit bank type'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"pay_date\", \"value\": \"2020-05-16\"}, {\"field\": \"employee_name\", \"value\": \"Peter Sims\"}, {\"field\": \"employer_name\", \"value\": \"ABC Company\"}, {\"field\": \"gross_pay\", \"value\": \"4000.0\"}, {\"field\": \"ytd_gross\", \"value\": \"20000.0\"}, {\"field\": \"pay_frequency\", \"value\": \"\"}, {\"field\": \"pay_period_start\", \"value\": \"2020-05-15\"}, {\"field\": \"pay_period_end\", \"value\": \"2020-05-16\"}, {\"field\": \"federal_tax\", \"value\": \"339.54\"}, {\"field\": \"state_tax\", \"value\": \"264.0\"}, {\"field\": \"social_security_tax\", \"value\": \"248.0\"}, {\"field\": \"medicare_tax\", \"value\": \"58.0\"}, {\"field\": \"hire_date\", \"value\": \"\"}, {\"field\": \"direct_deposit_bank_type\", \"value\": \"\"}]}"}]} {"id": "paystub_fin_059_p1_classify", "task": "classify", "class": "paystub", "doc": "paystub_fin_059.pdf", "images": ["images/train/paystub/paystub_fin_059_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"paystub\", \"confidence\": 1.0, \"issuer\": \"ABC Company\", \"holder\": \"John Smith\", \"page_marker\": \"\"}]}"}]} {"id": "paystub_fin_059_extract", "task": "extract", "class": "paystub", "doc": "paystub_fin_059.pdf", "images": ["images/train/paystub/paystub_fin_059_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('paystub', 'The document class.')\nfields = (['pay_date', 'employee_name', 'employer_name', 'gross_pay', 'ytd_gross', 'pay_frequency', 'pay_period_start', 'pay_period_end', 'federal_tax', 'state_tax', 'social_security_tax', 'medicare_tax', 'hire_date', 'direct_deposit_bank_type'], 'The exact field names to extract.')\nhints = ({'pay_date': 'Date the paycheck was issued', 'employee_name': 'Name of Employee', 'employer_name': 'Name of Employer', 'gross_pay': 'Gross pay per period (current period, not YTD)', 'ytd_gross': 'Gross pay on paystub year to date (cumulative, not current period)', 'pay_frequency': 'The frequency if provided: weekly, biweekly, monthly, semi-monthly', 'pay_period_start': 'Pay period start date', 'pay_period_end': 'Pay period end date', 'hire_date': 'Job hire date', 'direct_deposit_bank_type': 'Direct deposit bank type'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"pay_date\", \"value\": \"2020-02-26\"}, {\"field\": \"employee_name\", \"value\": \"John Smith\"}, {\"field\": \"employer_name\", \"value\": \"ABC Company\"}, {\"field\": \"gross_pay\", \"value\": \"400.0\"}, {\"field\": \"ytd_gross\", \"value\": \"3600.0\"}, {\"field\": \"pay_frequency\", \"value\": \"weekly\"}, {\"field\": \"pay_period_start\", \"value\": \"2020-02-19\"}, {\"field\": \"pay_period_end\", \"value\": \"2020-02-25\"}, {\"field\": \"federal_tax\", \"value\": \"44.5\"}, {\"field\": \"state_tax\", \"value\": \"20.0\"}, {\"field\": \"social_security_tax\", \"value\": \"24.8\"}, {\"field\": \"medicare_tax\", \"value\": \"5.8\"}, {\"field\": \"hire_date\", \"value\": \"\"}, {\"field\": \"direct_deposit_bank_type\", \"value\": \"\"}]}"}]} {"id": "paystub_fin_060_p1_classify", "task": "classify", "class": "paystub", "doc": "paystub_fin_060.pdf", "images": ["images/train/paystub/paystub_fin_060_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"paystub\", \"confidence\": 1.0, \"issuer\": \"Sample Company Name\", \"holder\": \"Employee Name\", \"page_marker\": \"\"}]}"}]} {"id": "paystub_fin_060_extract", "task": "extract", "class": "paystub", "doc": "paystub_fin_060.pdf", "images": ["images/train/paystub/paystub_fin_060_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('paystub', 'The document class.')\nfields = (['pay_date', 'employee_name', 'employer_name', 'gross_pay', 'ytd_gross', 'pay_frequency', 'pay_period_start', 'pay_period_end', 'federal_tax', 'state_tax', 'social_security_tax', 'medicare_tax', 'hire_date', 'direct_deposit_bank_type'], 'The exact field names to extract.')\nhints = ({'pay_date': 'Date the paycheck was issued', 'employee_name': 'Name of Employee', 'employer_name': 'Name of Employer', 'gross_pay': 'Gross pay per period (current period, not YTD)', 'ytd_gross': 'Gross pay on paystub year to date (cumulative, not current period)', 'pay_frequency': 'The frequency if provided: weekly, biweekly, monthly, semi-monthly', 'pay_period_start': 'Pay period start date', 'pay_period_end': 'Pay period end date', 'hire_date': 'Job hire date', 'direct_deposit_bank_type': 'Direct deposit bank type'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"pay_date\", \"value\": \"2018-04-19\"}, {\"field\": \"employee_name\", \"value\": \"Employee Name\"}, {\"field\": \"employer_name\", \"value\": \"Sample Company Name\"}, {\"field\": \"gross_pay\", \"value\": \"400.0\"}, {\"field\": \"ytd_gross\", \"value\": \"6400.0\"}, {\"field\": \"pay_frequency\", \"value\": \"\"}, {\"field\": \"pay_period_start\", \"value\": \"1970-01-01\"}, {\"field\": \"pay_period_end\", \"value\": \"1970-01-01\"}, {\"field\": \"federal_tax\", \"value\": \"35.82\"}, {\"field\": \"state_tax\", \"value\": \"20.0\"}, {\"field\": \"social_security_tax\", \"value\": \"24.8\"}, {\"field\": \"medicare_tax\", \"value\": \"5.8\"}, {\"field\": \"hire_date\", \"value\": \"\"}, {\"field\": \"direct_deposit_bank_type\", \"value\": \"\"}]}"}]} {"id": "paystub_fin_061_p1_classify", "task": "classify", "class": "paystub", "doc": "paystub_fin_061.pdf", "images": ["images/train/paystub/paystub_fin_061_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"paystub\", \"confidence\": 1.0, \"issuer\": \"Sample Company Name\", \"holder\": \"Employee Name\", \"page_marker\": \"\"}]}"}]} {"id": "paystub_fin_061_extract", "task": "extract", "class": "paystub", "doc": "paystub_fin_061.pdf", "images": ["images/train/paystub/paystub_fin_061_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('paystub', 'The document class.')\nfields = (['pay_date', 'employee_name', 'employer_name', 'gross_pay', 'ytd_gross', 'pay_frequency', 'pay_period_start', 'pay_period_end', 'federal_tax', 'state_tax', 'social_security_tax', 'medicare_tax', 'hire_date', 'direct_deposit_bank_type'], 'The exact field names to extract.')\nhints = ({'pay_date': 'Date the paycheck was issued', 'employee_name': 'Name of Employee', 'employer_name': 'Name of Employer', 'gross_pay': 'Gross pay per period (current period, not YTD)', 'ytd_gross': 'Gross pay on paystub year to date (cumulative, not current period)', 'pay_frequency': 'The frequency if provided: weekly, biweekly, monthly, semi-monthly', 'pay_period_start': 'Pay period start date', 'pay_period_end': 'Pay period end date', 'hire_date': 'Job hire date', 'direct_deposit_bank_type': 'Direct deposit bank type'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"pay_date\", \"value\": \"2018-02-14\"}, {\"field\": \"employee_name\", \"value\": \"Employee Name\"}, {\"field\": \"employer_name\", \"value\": \"Sample Company Name\"}, {\"field\": \"gross_pay\", \"value\": \"800.0\"}, {\"field\": \"ytd_gross\", \"value\": \"5600.0\"}, {\"field\": \"pay_frequency\", \"value\": \"\"}, {\"field\": \"pay_period_start\", \"value\": \"2018-02-07\"}, {\"field\": \"pay_period_end\", \"value\": \"2018-02-13\"}, {\"field\": \"federal_tax\", \"value\": \"108.1\"}, {\"field\": \"state_tax\", \"value\": \"40.0\"}, {\"field\": \"social_security_tax\", \"value\": \"49.6\"}, {\"field\": \"medicare_tax\", \"value\": \"11.6\"}, {\"field\": \"hire_date\", \"value\": \"\"}, {\"field\": \"direct_deposit_bank_type\", \"value\": \"\"}]}"}]} {"id": "paystub_fin_066_p1_classify", "task": "classify", "class": "paystub", "doc": "paystub_fin_066.pdf", "images": ["images/train/paystub/paystub_fin_066_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"paystub\", \"confidence\": 1.0, \"issuer\": \"Sample Company Name\", \"holder\": \"Employee Name\", \"page_marker\": \"\"}]}"}]} {"id": "paystub_fin_066_extract", "task": "extract", "class": "paystub", "doc": "paystub_fin_066.pdf", "images": ["images/train/paystub/paystub_fin_066_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('paystub', 'The document class.')\nfields = (['pay_date', 'employee_name', 'employer_name', 'gross_pay', 'ytd_gross', 'pay_frequency', 'pay_period_start', 'pay_period_end', 'federal_tax', 'state_tax', 'social_security_tax', 'medicare_tax', 'hire_date', 'direct_deposit_bank_type'], 'The exact field names to extract.')\nhints = ({'pay_date': 'Date the paycheck was issued', 'employee_name': 'Name of Employee', 'employer_name': 'Name of Employer', 'gross_pay': 'Gross pay per period (current period, not YTD)', 'ytd_gross': 'Gross pay on paystub year to date (cumulative, not current period)', 'pay_frequency': 'The frequency if provided: weekly, biweekly, monthly, semi-monthly', 'pay_period_start': 'Pay period start date', 'pay_period_end': 'Pay period end date', 'hire_date': 'Job hire date', 'direct_deposit_bank_type': 'Direct deposit bank type'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"pay_date\", \"value\": \"2018-04-19\"}, {\"field\": \"employee_name\", \"value\": \"Employee Name\"}, {\"field\": \"employer_name\", \"value\": \"Sample Company Name\"}, {\"field\": \"gross_pay\", \"value\": \"400.0\"}, {\"field\": \"ytd_gross\", \"value\": \"6400.0\"}, {\"field\": \"pay_frequency\", \"value\": \"\"}, {\"field\": \"pay_period_start\", \"value\": \"2018-04-12\"}, {\"field\": \"pay_period_end\", \"value\": \"2018-04-18\"}, {\"field\": \"federal_tax\", \"value\": \"35.82\"}, {\"field\": \"state_tax\", \"value\": \"20.0\"}, {\"field\": \"social_security_tax\", \"value\": \"24.8\"}, {\"field\": \"medicare_tax\", \"value\": \"5.8\"}, {\"field\": \"hire_date\", \"value\": \"\"}, {\"field\": \"direct_deposit_bank_type\", \"value\": \"\"}]}"}]} {"id": "paystub_fin_067_p1_classify", "task": "classify", "class": "paystub", "doc": "paystub_fin_067.pdf", "images": ["images/train/paystub/paystub_fin_067_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"paystub\", \"confidence\": 1.0, \"issuer\": \"MY COMPANY INC.\", \"holder\": \"JULIA WARNER\", \"page_marker\": \"\"}]}"}]} {"id": "paystub_fin_067_extract", "task": "extract", "class": "paystub", "doc": "paystub_fin_067.pdf", "images": ["images/train/paystub/paystub_fin_067_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('paystub', 'The document class.')\nfields = (['pay_date', 'employee_name', 'employer_name', 'gross_pay', 'ytd_gross', 'pay_frequency', 'pay_period_start', 'pay_period_end', 'federal_tax', 'state_tax', 'social_security_tax', 'medicare_tax', 'hire_date', 'direct_deposit_bank_type'], 'The exact field names to extract.')\nhints = ({'pay_date': 'Date the paycheck was issued', 'employee_name': 'Name of Employee', 'employer_name': 'Name of Employer', 'gross_pay': 'Gross pay per period (current period, not YTD)', 'ytd_gross': 'Gross pay on paystub year to date (cumulative, not current period)', 'pay_frequency': 'The frequency if provided: weekly, biweekly, monthly, semi-monthly', 'pay_period_start': 'Pay period start date', 'pay_period_end': 'Pay period end date', 'hire_date': 'Job hire date', 'direct_deposit_bank_type': 'Direct deposit bank type'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"pay_date\", \"value\": \"2023-06-05\"}, {\"field\": \"employee_name\", \"value\": \"JULIA WARNER\"}, {\"field\": \"employer_name\", \"value\": \"MY COMPANY INC.\"}, {\"field\": \"gross_pay\", \"value\": \"1085.0\"}, {\"field\": \"ytd_gross\", \"value\": \"\"}, {\"field\": \"pay_frequency\", \"value\": \"weekly\"}, {\"field\": \"pay_period_start\", \"value\": \"2023-05-29\"}, {\"field\": \"pay_period_end\", \"value\": \"2023-06-04\"}, {\"field\": \"federal_tax\", \"value\": \"12.6\"}, {\"field\": \"state_tax\", \"value\": \"20.0\"}, {\"field\": \"social_security_tax\", \"value\": \"41.2\"}, {\"field\": \"medicare_tax\", \"value\": \"\"}, {\"field\": \"hire_date\", \"value\": \"\"}, {\"field\": \"direct_deposit_bank_type\", \"value\": \"\"}]}"}]} {"id": "paystub_fin_068_p1_classify", "task": "classify", "class": "paystub", "doc": "paystub_fin_068.pdf", "images": ["images/train/paystub/paystub_fin_068_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"paystub\", \"confidence\": 1.0, \"issuer\": \"CLIENT COMPANY NAME\", \"holder\": \"Donald Chang\", \"page_marker\": \"\"}]}"}]} {"id": "paystub_fin_068_extract", "task": "extract", "class": "paystub", "doc": "paystub_fin_068.pdf", "images": ["images/train/paystub/paystub_fin_068_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('paystub', 'The document class.')\nfields = (['pay_date', 'employee_name', 'employer_name', 'gross_pay', 'ytd_gross', 'pay_frequency', 'pay_period_start', 'pay_period_end', 'federal_tax', 'state_tax', 'social_security_tax', 'medicare_tax', 'hire_date', 'direct_deposit_bank_type'], 'The exact field names to extract.')\nhints = ({'pay_date': 'Date the paycheck was issued', 'employee_name': 'Name of Employee', 'employer_name': 'Name of Employer', 'gross_pay': 'Gross pay per period (current period, not YTD)', 'ytd_gross': 'Gross pay on paystub year to date (cumulative, not current period)', 'pay_frequency': 'The frequency if provided: weekly, biweekly, monthly, semi-monthly', 'pay_period_start': 'Pay period start date', 'pay_period_end': 'Pay period end date', 'hire_date': 'Job hire date', 'direct_deposit_bank_type': 'Direct deposit bank type'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"pay_date\", \"value\": \"2023-06-01\"}, {\"field\": \"employee_name\", \"value\": \"Donald Chang\"}, {\"field\": \"employer_name\", \"value\": \"CLIENT COMPANY NAME\"}, {\"field\": \"gross_pay\", \"value\": \"4480.0\"}, {\"field\": \"ytd_gross\", \"value\": \"\"}, {\"field\": \"pay_frequency\", \"value\": \"\"}, {\"field\": \"pay_period_start\", \"value\": \"2023-05-01\"}, {\"field\": \"pay_period_end\", \"value\": \"2023-05-31\"}, {\"field\": \"federal_tax\", \"value\": \"180.0\"}, {\"field\": \"state_tax\", \"value\": \"202.0\"}, {\"field\": \"social_security_tax\", \"value\": \"155.0\"}, {\"field\": \"medicare_tax\", \"value\": \"85.0\"}, {\"field\": \"hire_date\", \"value\": \"\"}, {\"field\": \"direct_deposit_bank_type\", \"value\": \"West Rich Bank\"}]}"}]} {"id": "paystub_fin_070_p1_classify", "task": "classify", "class": "paystub", "doc": "paystub_fin_070.pdf", "images": ["images/train/paystub/paystub_fin_070_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"paystub\", \"confidence\": 1.0, \"issuer\": \"BUSINESS NAME\", \"holder\": \"Solomon Navarro\", \"page_marker\": \"\"}]}"}]} {"id": "paystub_fin_070_extract", "task": "extract", "class": "paystub", "doc": "paystub_fin_070.pdf", "images": ["images/train/paystub/paystub_fin_070_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('paystub', 'The document class.')\nfields = (['pay_date', 'employee_name', 'employer_name', 'gross_pay', 'ytd_gross', 'pay_frequency', 'pay_period_start', 'pay_period_end', 'federal_tax', 'state_tax', 'social_security_tax', 'medicare_tax', 'hire_date', 'direct_deposit_bank_type'], 'The exact field names to extract.')\nhints = ({'pay_date': 'Date the paycheck was issued', 'employee_name': 'Name of Employee', 'employer_name': 'Name of Employer', 'gross_pay': 'Gross pay per period (current period, not YTD)', 'ytd_gross': 'Gross pay on paystub year to date (cumulative, not current period)', 'pay_frequency': 'The frequency if provided: weekly, biweekly, monthly, semi-monthly', 'pay_period_start': 'Pay period start date', 'pay_period_end': 'Pay period end date', 'hire_date': 'Job hire date', 'direct_deposit_bank_type': 'Direct deposit bank type'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"pay_date\", \"value\": \"2023-06-01\"}, {\"field\": \"employee_name\", \"value\": \"Solomon Navarro\"}, {\"field\": \"employer_name\", \"value\": \"BUSINESS NAME\"}, {\"field\": \"gross_pay\", \"value\": \"5190.0\"}, {\"field\": \"ytd_gross\", \"value\": \"\"}, {\"field\": \"pay_frequency\", \"value\": \"\"}, {\"field\": \"pay_period_start\", \"value\": \"\"}, {\"field\": \"pay_period_end\", \"value\": \"\"}, {\"field\": \"federal_tax\", \"value\": \"185.0\"}, {\"field\": \"state_tax\", \"value\": \"206.0\"}, {\"field\": \"social_security_tax\", \"value\": \"133.0\"}, {\"field\": \"medicare_tax\", \"value\": \"68.0\"}, {\"field\": \"hire_date\", \"value\": \"\"}, {\"field\": \"direct_deposit_bank_type\", \"value\": \"\"}]}"}]} {"id": "paystub_fin_071_p1_classify", "task": "classify", "class": "paystub", "doc": "paystub_fin_071.pdf", "images": ["images/train/paystub/paystub_fin_071_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"paystub\", \"confidence\": 1.0, \"issuer\": \"DEFENSE FINANCE AND ACCOUNTING SERVICE\", \"holder\": \"WILLIAM T. BONES\", \"page_marker\": \"\"}]}"}]} {"id": "paystub_fin_071_extract", "task": "extract", "class": "paystub", "doc": "paystub_fin_071.pdf", "images": ["images/train/paystub/paystub_fin_071_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('paystub', 'The document class.')\nfields = (['pay_date', 'employee_name', 'employer_name', 'gross_pay', 'ytd_gross', 'pay_frequency', 'pay_period_start', 'pay_period_end', 'federal_tax', 'state_tax', 'social_security_tax', 'medicare_tax', 'hire_date', 'direct_deposit_bank_type'], 'The exact field names to extract.')\nhints = ({'pay_date': 'Date the paycheck was issued', 'employee_name': 'Name of Employee', 'employer_name': 'Name of Employer', 'gross_pay': 'Gross pay per period (current period, not YTD)', 'ytd_gross': 'Gross pay on paystub year to date (cumulative, not current period)', 'pay_frequency': 'The frequency if provided: weekly, biweekly, monthly, semi-monthly', 'pay_period_start': 'Pay period start date', 'pay_period_end': 'Pay period end date', 'hire_date': 'Job hire date', 'direct_deposit_bank_type': 'Direct deposit bank type'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"pay_date\", \"value\": \"2023-01-06\"}, {\"field\": \"employee_name\", \"value\": \"WILLIAM T. BONES\"}, {\"field\": \"employer_name\", \"value\": \"DEFENSE FINANCE AND ACCOUNTING SERVICE\"}, {\"field\": \"gross_pay\", \"value\": \"3146.2\"}, {\"field\": \"ytd_gross\", \"value\": \"\"}, {\"field\": \"pay_frequency\", \"value\": \"\"}, {\"field\": \"pay_period_start\", \"value\": \"\"}, {\"field\": \"pay_period_end\", \"value\": \"\"}, {\"field\": \"federal_tax\", \"value\": \"175.25\"}, {\"field\": \"state_tax\", \"value\": \"202.14\"}, {\"field\": \"social_security_tax\", \"value\": \"102.45\"}, {\"field\": \"medicare_tax\", \"value\": \"89.55\"}, {\"field\": \"hire_date\", \"value\": \"2016-02-03\"}, {\"field\": \"direct_deposit_bank_type\", \"value\": \"Military Bank\"}]}"}]} {"id": "paystub_fin_074_p1_classify", "task": "classify", "class": "paystub", "doc": "paystub_fin_074.pdf", "images": ["images/train/paystub/paystub_fin_074_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"paystub\", \"confidence\": 1.0, \"issuer\": \"ABC Company\", \"holder\": \"Jane Dough\", \"page_marker\": \"\"}]}"}]} {"id": "paystub_fin_074_extract", "task": "extract", "class": "paystub", "doc": "paystub_fin_074.pdf", "images": ["images/train/paystub/paystub_fin_074_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('paystub', 'The document class.')\nfields = (['pay_date', 'employee_name', 'employer_name', 'gross_pay', 'ytd_gross', 'pay_frequency', 'pay_period_start', 'pay_period_end', 'federal_tax', 'state_tax', 'social_security_tax', 'medicare_tax', 'hire_date', 'direct_deposit_bank_type'], 'The exact field names to extract.')\nhints = ({'pay_date': 'Date the paycheck was issued', 'employee_name': 'Name of Employee', 'employer_name': 'Name of Employer', 'gross_pay': 'Gross pay per period (current period, not YTD)', 'ytd_gross': 'Gross pay on paystub year to date (cumulative, not current period)', 'pay_frequency': 'The frequency if provided: weekly, biweekly, monthly, semi-monthly', 'pay_period_start': 'Pay period start date', 'pay_period_end': 'Pay period end date', 'hire_date': 'Job hire date', 'direct_deposit_bank_type': 'Direct deposit bank type'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"pay_date\", \"value\": \"\"}, {\"field\": \"employee_name\", \"value\": \"Jane Dough\"}, {\"field\": \"employer_name\", \"value\": \"ABC Company\"}, {\"field\": \"gross_pay\", \"value\": \"1346.15\"}, {\"field\": \"ytd_gross\", \"value\": \"1346.15\"}, {\"field\": \"pay_frequency\", \"value\": \"\"}, {\"field\": \"pay_period_start\", \"value\": \"2014-07-17\"}, {\"field\": \"pay_period_end\", \"value\": \"2014-07-31\"}, {\"field\": \"federal_tax\", \"value\": \"-177.04\"}, {\"field\": \"state_tax\", \"value\": \"\"}, {\"field\": \"social_security_tax\", \"value\": \"\"}, {\"field\": \"medicare_tax\", \"value\": \"\"}, {\"field\": \"hire_date\", \"value\": \"\"}, {\"field\": \"direct_deposit_bank_type\", \"value\": \"\"}]}"}]} {"id": "paystub_fin_075_p1_classify", "task": "classify", "class": "paystub", "doc": "paystub_fin_075.pdf", "images": ["images/train/paystub/paystub_fin_075_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"paystub\", \"confidence\": 1.0, \"issuer\": \"COMPANY NAME\", \"holder\": \"Nelly Green\", \"page_marker\": \"\"}]}"}]} {"id": "paystub_fin_075_extract", "task": "extract", "class": "paystub", "doc": "paystub_fin_075.pdf", "images": ["images/train/paystub/paystub_fin_075_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('paystub', 'The document class.')\nfields = (['pay_date', 'employee_name', 'employer_name', 'gross_pay', 'ytd_gross', 'pay_frequency', 'pay_period_start', 'pay_period_end', 'federal_tax', 'state_tax', 'social_security_tax', 'medicare_tax', 'hire_date', 'direct_deposit_bank_type'], 'The exact field names to extract.')\nhints = ({'pay_date': 'Date the paycheck was issued', 'employee_name': 'Name of Employee', 'employer_name': 'Name of Employer', 'gross_pay': 'Gross pay per period (current period, not YTD)', 'ytd_gross': 'Gross pay on paystub year to date (cumulative, not current period)', 'pay_frequency': 'The frequency if provided: weekly, biweekly, monthly, semi-monthly', 'pay_period_start': 'Pay period start date', 'pay_period_end': 'Pay period end date', 'hire_date': 'Job hire date', 'direct_deposit_bank_type': 'Direct deposit bank type'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"pay_date\", \"value\": \"2023-12-01\"}, {\"field\": \"employee_name\", \"value\": \"Nelly Green\"}, {\"field\": \"employer_name\", \"value\": \"COMPANY NAME\"}, {\"field\": \"gross_pay\", \"value\": \"3600.0\"}, {\"field\": \"ytd_gross\", \"value\": \"\"}, {\"field\": \"pay_frequency\", \"value\": \"\"}, {\"field\": \"pay_period_start\", \"value\": \"2023-11-01\"}, {\"field\": \"pay_period_end\", \"value\": \"2023-11-30\"}, {\"field\": \"federal_tax\", \"value\": \"\"}, {\"field\": \"state_tax\", \"value\": \"\"}, {\"field\": \"social_security_tax\", \"value\": \"\"}, {\"field\": \"medicare_tax\", \"value\": \"\"}, {\"field\": \"hire_date\", \"value\": \"\"}, {\"field\": \"direct_deposit_bank_type\", \"value\": \"\"}]}"}]} {"id": "paystub_fin_077_p1_classify", "task": "classify", "class": "paystub", "doc": "paystub_fin_077.pdf", "images": ["images/train/paystub/paystub_fin_077_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"paystub\", \"confidence\": 1.0, \"issuer\": \"Stub Samples INC.\", \"holder\": \"JOHN Michel\", \"page_marker\": \"\"}]}"}]} {"id": "paystub_fin_077_extract", "task": "extract", "class": "paystub", "doc": "paystub_fin_077.pdf", "images": ["images/train/paystub/paystub_fin_077_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('paystub', 'The document class.')\nfields = (['pay_date', 'employee_name', 'employer_name', 'gross_pay', 'ytd_gross', 'pay_frequency', 'pay_period_start', 'pay_period_end', 'federal_tax', 'state_tax', 'social_security_tax', 'medicare_tax', 'hire_date', 'direct_deposit_bank_type'], 'The exact field names to extract.')\nhints = ({'pay_date': 'Date the paycheck was issued', 'employee_name': 'Name of Employee', 'employer_name': 'Name of Employer', 'gross_pay': 'Gross pay per period (current period, not YTD)', 'ytd_gross': 'Gross pay on paystub year to date (cumulative, not current period)', 'pay_frequency': 'The frequency if provided: weekly, biweekly, monthly, semi-monthly', 'pay_period_start': 'Pay period start date', 'pay_period_end': 'Pay period end date', 'hire_date': 'Job hire date', 'direct_deposit_bank_type': 'Direct deposit bank type'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"pay_date\", \"value\": \"\"}, {\"field\": \"employee_name\", \"value\": \"JOHN Michel\"}, {\"field\": \"employer_name\", \"value\": \"Stub Samples INC.\"}, {\"field\": \"gross_pay\", \"value\": \"2680.74\"}, {\"field\": \"ytd_gross\", \"value\": \"9230.77\"}, {\"field\": \"pay_frequency\", \"value\": \"\"}, {\"field\": \"pay_period_start\", \"value\": \"2015-05-28\"}, {\"field\": \"pay_period_end\", \"value\": \"\"}, {\"field\": \"federal_tax\", \"value\": \"\"}, {\"field\": \"state_tax\", \"value\": \"\"}, {\"field\": \"social_security_tax\", \"value\": \"\"}, {\"field\": \"medicare_tax\", \"value\": \"\"}, {\"field\": \"hire_date\", \"value\": \"\"}, {\"field\": \"direct_deposit_bank_type\", \"value\": \"\"}]}"}]} {"id": "paystub_fin_078_p1_classify", "task": "classify", "class": "paystub", "doc": "paystub_fin_078.pdf", "images": ["images/train/paystub/paystub_fin_078_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"paystub\", \"confidence\": 1.0, \"issuer\": \"Grayline Electric\", \"holder\": \"Angela Smith\", \"page_marker\": \"\"}]}"}]} {"id": "paystub_fin_078_extract", "task": "extract", "class": "paystub", "doc": "paystub_fin_078.pdf", "images": ["images/train/paystub/paystub_fin_078_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('paystub', 'The document class.')\nfields = (['pay_date', 'employee_name', 'employer_name', 'gross_pay', 'ytd_gross', 'pay_frequency', 'pay_period_start', 'pay_period_end', 'federal_tax', 'state_tax', 'social_security_tax', 'medicare_tax', 'hire_date', 'direct_deposit_bank_type'], 'The exact field names to extract.')\nhints = ({'pay_date': 'Date the paycheck was issued', 'employee_name': 'Name of Employee', 'employer_name': 'Name of Employer', 'gross_pay': 'Gross pay per period (current period, not YTD)', 'ytd_gross': 'Gross pay on paystub year to date (cumulative, not current period)', 'pay_frequency': 'The frequency if provided: weekly, biweekly, monthly, semi-monthly', 'pay_period_start': 'Pay period start date', 'pay_period_end': 'Pay period end date', 'hire_date': 'Job hire date', 'direct_deposit_bank_type': 'Direct deposit bank type'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"pay_date\", \"value\": \"2015-09-03\"}, {\"field\": \"employee_name\", \"value\": \"Angela Smith\"}, {\"field\": \"employer_name\", \"value\": \"Grayline Electric\"}, {\"field\": \"gross_pay\", \"value\": \"4125.0\"}, {\"field\": \"ytd_gross\", \"value\": \"8250.0\"}, {\"field\": \"pay_frequency\", \"value\": \"\"}, {\"field\": \"pay_period_start\", \"value\": \"2015-08-01\"}, {\"field\": \"pay_period_end\", \"value\": \"2015-08-28\"}, {\"field\": \"federal_tax\", \"value\": \"598.87\"}, {\"field\": \"state_tax\", \"value\": \"\"}, {\"field\": \"social_security_tax\", \"value\": \"255.75\"}, {\"field\": \"medicare_tax\", \"value\": \"59.81\"}, {\"field\": \"hire_date\", \"value\": \"\"}, {\"field\": \"direct_deposit_bank_type\", \"value\": \"\"}]}"}]} {"id": "paystub_fin_080_p1_classify", "task": "classify", "class": "paystub", "doc": "paystub_fin_080.pdf", "images": ["images/train/paystub/paystub_fin_080_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"paystub\", \"confidence\": 1.0, \"issuer\": \"[YOUR COMPANY NAME]\", \"holder\": \"Macey Erdman\", \"page_marker\": \"\"}]}"}]} {"id": "paystub_fin_080_extract", "task": "extract", "class": "paystub", "doc": "paystub_fin_080.pdf", "images": ["images/train/paystub/paystub_fin_080_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('paystub', 'The document class.')\nfields = (['pay_date', 'employee_name', 'employer_name', 'gross_pay', 'ytd_gross', 'pay_frequency', 'pay_period_start', 'pay_period_end', 'federal_tax', 'state_tax', 'social_security_tax', 'medicare_tax', 'hire_date', 'direct_deposit_bank_type'], 'The exact field names to extract.')\nhints = ({'pay_date': 'Date the paycheck was issued', 'employee_name': 'Name of Employee', 'employer_name': 'Name of Employer', 'gross_pay': 'Gross pay per period (current period, not YTD)', 'ytd_gross': 'Gross pay on paystub year to date (cumulative, not current period)', 'pay_frequency': 'The frequency if provided: weekly, biweekly, monthly, semi-monthly', 'pay_period_start': 'Pay period start date', 'pay_period_end': 'Pay period end date', 'hire_date': 'Job hire date', 'direct_deposit_bank_type': 'Direct deposit bank type'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"pay_date\", \"value\": \"2050-01-15\"}, {\"field\": \"employee_name\", \"value\": \"Macey Erdman\"}, {\"field\": \"employer_name\", \"value\": \"[YOUR COMPANY NAME]\"}, {\"field\": \"gross_pay\", \"value\": \"2850.0\"}, {\"field\": \"ytd_gross\", \"value\": \"\"}, {\"field\": \"pay_frequency\", \"value\": \"\"}, {\"field\": \"pay_period_start\", \"value\": \"2050-01-01\"}, {\"field\": \"pay_period_end\", \"value\": \"2050-01-15\"}, {\"field\": \"federal_tax\", \"value\": \"350.0\"}, {\"field\": \"state_tax\", \"value\": \"150.0\"}, {\"field\": \"social_security_tax\", \"value\": \"176.1\"}, {\"field\": \"medicare_tax\", \"value\": \"\"}, {\"field\": \"hire_date\", \"value\": \"\"}, {\"field\": \"direct_deposit_bank_type\", \"value\": \"\"}]}"}]} {"id": "paystub_fin_081_p1_classify", "task": "classify", "class": "paystub", "doc": "paystub_fin_081.pdf", "images": ["images/train/paystub/paystub_fin_081_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"paystub\", \"confidence\": 1.0, \"issuer\": \"Your Company Name\", \"holder\": \"Jay Cyrus\", \"page_marker\": \"\"}]}"}]} {"id": "paystub_fin_081_extract", "task": "extract", "class": "paystub", "doc": "paystub_fin_081.pdf", "images": ["images/train/paystub/paystub_fin_081_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('paystub', 'The document class.')\nfields = (['pay_date', 'employee_name', 'employer_name', 'gross_pay', 'ytd_gross', 'pay_frequency', 'pay_period_start', 'pay_period_end', 'federal_tax', 'state_tax', 'social_security_tax', 'medicare_tax', 'hire_date', 'direct_deposit_bank_type'], 'The exact field names to extract.')\nhints = ({'pay_date': 'Date the paycheck was issued', 'employee_name': 'Name of Employee', 'employer_name': 'Name of Employer', 'gross_pay': 'Gross pay per period (current period, not YTD)', 'ytd_gross': 'Gross pay on paystub year to date (cumulative, not current period)', 'pay_frequency': 'The frequency if provided: weekly, biweekly, monthly, semi-monthly', 'pay_period_start': 'Pay period start date', 'pay_period_end': 'Pay period end date', 'hire_date': 'Job hire date', 'direct_deposit_bank_type': 'Direct deposit bank type'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"pay_date\", \"value\": \"2024-02-07\"}, {\"field\": \"employee_name\", \"value\": \"Jay Cyrus\"}, {\"field\": \"employer_name\", \"value\": \"Your Company Name\"}, {\"field\": \"gross_pay\", \"value\": \"2820.0\"}, {\"field\": \"ytd_gross\", \"value\": \"9800.0\"}, {\"field\": \"pay_frequency\", \"value\": \"monthly\"}, {\"field\": \"pay_period_start\", \"value\": \"2024-01-01\"}, {\"field\": \"pay_period_end\", \"value\": \"2024-01-31\"}, {\"field\": \"federal_tax\", \"value\": \"100.0\"}, {\"field\": \"state_tax\", \"value\": \"60.0\"}, {\"field\": \"social_security_tax\", \"value\": \"174.84\"}, {\"field\": \"medicare_tax\", \"value\": \"40.89\"}, {\"field\": \"hire_date\", \"value\": \"\"}, {\"field\": \"direct_deposit_bank_type\", \"value\": \"Direct Deposit\"}]}"}]} {"id": "paystub_fin_082_p1_classify", "task": "classify", "class": "paystub", "doc": "paystub_fin_082.pdf", "images": ["images/train/paystub/paystub_fin_082_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"paystub\", \"confidence\": 1.0, \"issuer\": \"Sample Company Name\", \"holder\": \"Sample Employee Name\", \"page_marker\": \"\"}]}"}]} {"id": "paystub_fin_082_extract", "task": "extract", "class": "paystub", "doc": "paystub_fin_082.pdf", "images": ["images/train/paystub/paystub_fin_082_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('paystub', 'The document class.')\nfields = (['pay_date', 'employee_name', 'employer_name', 'gross_pay', 'ytd_gross', 'pay_frequency', 'pay_period_start', 'pay_period_end', 'federal_tax', 'state_tax', 'social_security_tax', 'medicare_tax', 'hire_date', 'direct_deposit_bank_type'], 'The exact field names to extract.')\nhints = ({'pay_date': 'Date the paycheck was issued', 'employee_name': 'Name of Employee', 'employer_name': 'Name of Employer', 'gross_pay': 'Gross pay per period (current period, not YTD)', 'ytd_gross': 'Gross pay on paystub year to date (cumulative, not current period)', 'pay_frequency': 'The frequency if provided: weekly, biweekly, monthly, semi-monthly', 'pay_period_start': 'Pay period start date', 'pay_period_end': 'Pay period end date', 'hire_date': 'Job hire date', 'direct_deposit_bank_type': 'Direct deposit bank type'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"pay_date\", \"value\": \"2023-05-08\"}, {\"field\": \"employee_name\", \"value\": \"Sample Employee Name\"}, {\"field\": \"employer_name\", \"value\": \"Sample Company Name\"}, {\"field\": \"gross_pay\", \"value\": \"1650.0\"}, {\"field\": \"ytd_gross\", \"value\": \"14850.0\"}, {\"field\": \"pay_frequency\", \"value\": \"\"}, {\"field\": \"pay_period_start\", \"value\": \"2023-04-22\"}, {\"field\": \"pay_period_end\", \"value\": \"2023-05-05\"}, {\"field\": \"federal_tax\", \"value\": \"125.62\"}, {\"field\": \"state_tax\", \"value\": \"48.23\"}, {\"field\": \"social_security_tax\", \"value\": \"102.3\"}, {\"field\": \"medicare_tax\", \"value\": \"23.93\"}, {\"field\": \"hire_date\", \"value\": \"\"}, {\"field\": \"direct_deposit_bank_type\", \"value\": \"\"}]}"}]} {"id": "paystub_fin_083_p1_classify", "task": "classify", "class": "paystub", "doc": "paystub_fin_083.pdf", "images": ["images/train/paystub/paystub_fin_083_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"paystub\", \"confidence\": 1.0, \"issuer\": \"Business Name, LLC\", \"holder\": \"Joe Employee\", \"page_marker\": \"\"}]}"}]} {"id": "paystub_fin_083_extract", "task": "extract", "class": "paystub", "doc": "paystub_fin_083.pdf", "images": ["images/train/paystub/paystub_fin_083_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('paystub', 'The document class.')\nfields = (['pay_date', 'employee_name', 'employer_name', 'gross_pay', 'ytd_gross', 'pay_frequency', 'pay_period_start', 'pay_period_end', 'federal_tax', 'state_tax', 'social_security_tax', 'medicare_tax', 'hire_date', 'direct_deposit_bank_type'], 'The exact field names to extract.')\nhints = ({'pay_date': 'Date the paycheck was issued', 'employee_name': 'Name of Employee', 'employer_name': 'Name of Employer', 'gross_pay': 'Gross pay per period (current period, not YTD)', 'ytd_gross': 'Gross pay on paystub year to date (cumulative, not current period)', 'pay_frequency': 'The frequency if provided: weekly, biweekly, monthly, semi-monthly', 'pay_period_start': 'Pay period start date', 'pay_period_end': 'Pay period end date', 'hire_date': 'Job hire date', 'direct_deposit_bank_type': 'Direct deposit bank type'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"pay_date\", \"value\": \"2009-01-23\"}, {\"field\": \"employee_name\", \"value\": \"Joe Employee\"}, {\"field\": \"employer_name\", \"value\": \"Business Name, LLC\"}, {\"field\": \"gross_pay\", \"value\": \"320.0\"}, {\"field\": \"ytd_gross\", \"value\": \"\"}, {\"field\": \"pay_frequency\", \"value\": \"\"}, {\"field\": \"pay_period_start\", \"value\": \"2009-01-17\"}, {\"field\": \"pay_period_end\", \"value\": \"2009-01-23\"}, {\"field\": \"federal_tax\", \"value\": \"12.0\"}, {\"field\": \"state_tax\", \"value\": \"16.0\"}, {\"field\": \"social_security_tax\", \"value\": \"19.2\"}, {\"field\": \"medicare_tax\", \"value\": \"3.2\"}, {\"field\": \"hire_date\", \"value\": \"\"}, {\"field\": \"direct_deposit_bank_type\", \"value\": \"\"}]}"}]} {"id": "paystub_fin_084_p1_classify", "task": "classify", "class": "paystub", "doc": "paystub_fin_084.pdf", "images": ["images/train/paystub/paystub_fin_084_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"paystub\", \"confidence\": 1.0, \"issuer\": \"CANADIAN ONE PLC\", \"holder\": \"BENJAMIN WALKER\", \"page_marker\": \"\"}]}"}]} {"id": "paystub_fin_084_extract", "task": "extract", "class": "paystub", "doc": "paystub_fin_084.pdf", "images": ["images/train/paystub/paystub_fin_084_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('paystub', 'The document class.')\nfields = (['pay_date', 'employee_name', 'employer_name', 'gross_pay', 'ytd_gross', 'pay_frequency', 'pay_period_start', 'pay_period_end', 'federal_tax', 'state_tax', 'social_security_tax', 'medicare_tax', 'hire_date', 'direct_deposit_bank_type'], 'The exact field names to extract.')\nhints = ({'pay_date': 'Date the paycheck was issued', 'employee_name': 'Name of Employee', 'employer_name': 'Name of Employer', 'gross_pay': 'Gross pay per period (current period, not YTD)', 'ytd_gross': 'Gross pay on paystub year to date (cumulative, not current period)', 'pay_frequency': 'The frequency if provided: weekly, biweekly, monthly, semi-monthly', 'pay_period_start': 'Pay period start date', 'pay_period_end': 'Pay period end date', 'hire_date': 'Job hire date', 'direct_deposit_bank_type': 'Direct deposit bank type'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"pay_date\", \"value\": \"2025-02-01\"}, {\"field\": \"employee_name\", \"value\": \"BENJAMIN WALKER\"}, {\"field\": \"employer_name\", \"value\": \"CANADIAN ONE PLC\"}, {\"field\": \"gross_pay\", \"value\": \"4278.0\"}, {\"field\": \"ytd_gross\", \"value\": \"4278.0\"}, {\"field\": \"pay_frequency\", \"value\": \"monthly\"}, {\"field\": \"pay_period_start\", \"value\": \"2025-01-01\"}, {\"field\": \"pay_period_end\", \"value\": \"2025-01-31\"}, {\"field\": \"federal_tax\", \"value\": \"385.77\"}, {\"field\": \"state_tax\", \"value\": \"214.55\"}, {\"field\": \"social_security_tax\", \"value\": \"\"}, {\"field\": \"medicare_tax\", \"value\": \"\"}, {\"field\": \"hire_date\", \"value\": \"\"}, {\"field\": \"direct_deposit_bank_type\", \"value\": \"\"}]}"}]} {"id": "paystub_fin_085_p1_classify", "task": "classify", "class": "paystub", "doc": "paystub_fin_085.pdf", "images": ["images/train/paystub/paystub_fin_085_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"paystub\", \"confidence\": 1.0, \"issuer\": \"PASCO SCHOOL DISTRICT #1\", \"holder\": \"CUSTODIAN, JOHNNY\", \"page_marker\": \"\"}]}"}]} {"id": "paystub_fin_085_extract", "task": "extract", "class": "paystub", "doc": "paystub_fin_085.pdf", "images": ["images/train/paystub/paystub_fin_085_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('paystub', 'The document class.')\nfields = (['pay_date', 'employee_name', 'employer_name', 'gross_pay', 'ytd_gross', 'pay_frequency', 'pay_period_start', 'pay_period_end', 'federal_tax', 'state_tax', 'social_security_tax', 'medicare_tax', 'hire_date', 'direct_deposit_bank_type'], 'The exact field names to extract.')\nhints = ({'pay_date': 'Date the paycheck was issued', 'employee_name': 'Name of Employee', 'employer_name': 'Name of Employer', 'gross_pay': 'Gross pay per period (current period, not YTD)', 'ytd_gross': 'Gross pay on paystub year to date (cumulative, not current period)', 'pay_frequency': 'The frequency if provided: weekly, biweekly, monthly, semi-monthly', 'pay_period_start': 'Pay period start date', 'pay_period_end': 'Pay period end date', 'hire_date': 'Job hire date', 'direct_deposit_bank_type': 'Direct deposit bank type'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"pay_date\", \"value\": \"2012-12-31\"}, {\"field\": \"employee_name\", \"value\": \"CUSTODIAN, JOHNNY\"}, {\"field\": \"employer_name\", \"value\": \"PASCO SCHOOL DISTRICT #1\"}, {\"field\": \"gross_pay\", \"value\": \"3003.87\"}, {\"field\": \"ytd_gross\", \"value\": \"36582.51\"}, {\"field\": \"pay_frequency\", \"value\": \"\"}, {\"field\": \"pay_period_start\", \"value\": \"\"}, {\"field\": \"pay_period_end\", \"value\": \"2012-12-31\"}, {\"field\": \"federal_tax\", \"value\": \"464.68\"}, {\"field\": \"state_tax\", \"value\": \"\"}, {\"field\": \"social_security_tax\", \"value\": \"126.1\"}, {\"field\": \"medicare_tax\", \"value\": \"43.53\"}, {\"field\": \"hire_date\", \"value\": \"\"}, {\"field\": \"direct_deposit_bank_type\", \"value\": \"FLINTSTONE BANK\"}]}"}]} {"id": "paystub_fin_086_p1_classify", "task": "classify", "class": "paystub", "doc": "paystub_fin_086.pdf", "images": ["images/train/paystub/paystub_fin_086_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"paystub\", \"confidence\": 1.0, \"issuer\": \"ACME SUPPLIES CORP.\", \"holder\": \"JANE HARPER\", \"page_marker\": \"\"}]}"}]} {"id": "paystub_fin_086_extract", "task": "extract", "class": "paystub", "doc": "paystub_fin_086.pdf", "images": ["images/train/paystub/paystub_fin_086_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('paystub', 'The document class.')\nfields = (['pay_date', 'employee_name', 'employer_name', 'gross_pay', 'ytd_gross', 'pay_frequency', 'pay_period_start', 'pay_period_end', 'federal_tax', 'state_tax', 'social_security_tax', 'medicare_tax', 'hire_date', 'direct_deposit_bank_type'], 'The exact field names to extract.')\nhints = ({'pay_date': 'Date the paycheck was issued', 'employee_name': 'Name of Employee', 'employer_name': 'Name of Employer', 'gross_pay': 'Gross pay per period (current period, not YTD)', 'ytd_gross': 'Gross pay on paystub year to date (cumulative, not current period)', 'pay_frequency': 'The frequency if provided: weekly, biweekly, monthly, semi-monthly', 'pay_period_start': 'Pay period start date', 'pay_period_end': 'Pay period end date', 'hire_date': 'Job hire date', 'direct_deposit_bank_type': 'Direct deposit bank type'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"pay_date\", \"value\": \"2008-07-25\"}, {\"field\": \"employee_name\", \"value\": \"JANE HARPER\"}, {\"field\": \"employer_name\", \"value\": \"ACME SUPPLIES CORP.\"}, {\"field\": \"gross_pay\", \"value\": \"452.43\"}, {\"field\": \"ytd_gross\", \"value\": \"23526.8\"}, {\"field\": \"pay_frequency\", \"value\": \"\"}, {\"field\": \"pay_period_start\", \"value\": \"\"}, {\"field\": \"pay_period_end\", \"value\": \"2008-07-18\"}, {\"field\": \"federal_tax\", \"value\": \"40.6\"}, {\"field\": \"state_tax\", \"value\": \"8.43\"}, {\"field\": \"social_security_tax\", \"value\": \"28.05\"}, {\"field\": \"medicare_tax\", \"value\": \"6.56\"}, {\"field\": \"hire_date\", \"value\": \"\"}, {\"field\": \"direct_deposit_bank_type\", \"value\": \"\"}]}"}]} {"id": "paystub_fin_087_p1_classify", "task": "classify", "class": "paystub", "doc": "paystub_fin_087.pdf", "images": ["images/train/paystub/paystub_fin_087_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"paystub\", \"confidence\": 1.0, \"issuer\": \"Sample Company Name\", \"holder\": \"Sample Employee Name\", \"page_marker\": \"\"}]}"}]} {"id": "paystub_fin_087_extract", "task": "extract", "class": "paystub", "doc": "paystub_fin_087.pdf", "images": ["images/train/paystub/paystub_fin_087_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('paystub', 'The document class.')\nfields = (['pay_date', 'employee_name', 'employer_name', 'gross_pay', 'ytd_gross', 'pay_frequency', 'pay_period_start', 'pay_period_end', 'federal_tax', 'state_tax', 'social_security_tax', 'medicare_tax', 'hire_date', 'direct_deposit_bank_type'], 'The exact field names to extract.')\nhints = ({'pay_date': 'Date the paycheck was issued', 'employee_name': 'Name of Employee', 'employer_name': 'Name of Employer', 'gross_pay': 'Gross pay per period (current period, not YTD)', 'ytd_gross': 'Gross pay on paystub year to date (cumulative, not current period)', 'pay_frequency': 'The frequency if provided: weekly, biweekly, monthly, semi-monthly', 'pay_period_start': 'Pay period start date', 'pay_period_end': 'Pay period end date', 'hire_date': 'Job hire date', 'direct_deposit_bank_type': 'Direct deposit bank type'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"pay_date\", \"value\": \"2023-12-01\"}, {\"field\": \"employee_name\", \"value\": \"Sample Employee Name\"}, {\"field\": \"employer_name\", \"value\": \"Sample Company Name\"}, {\"field\": \"gross_pay\", \"value\": \"3163.0\"}, {\"field\": \"ytd_gross\", \"value\": \"75912.0\"}, {\"field\": \"pay_frequency\", \"value\": \"\"}, {\"field\": \"pay_period_start\", \"value\": \"2023-11-15\"}, {\"field\": \"pay_period_end\", \"value\": \"2023-11-28\"}, {\"field\": \"federal_tax\", \"value\": \"434.57\"}, {\"field\": \"state_tax\", \"value\": \"69.35\"}, {\"field\": \"social_security_tax\", \"value\": \"196.11\"}, {\"field\": \"medicare_tax\", \"value\": \"45.86\"}, {\"field\": \"hire_date\", \"value\": \"\"}, {\"field\": \"direct_deposit_bank_type\", \"value\": \"\"}]}"}]} {"id": "paystub_fin_088_p1_classify", "task": "classify", "class": "paystub", "doc": "paystub_fin_088.pdf", "images": ["images/train/paystub/paystub_fin_088_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"paystub\", \"confidence\": 1.0, \"issuer\": \"\", \"holder\": \"[YOUR NAME]\", \"page_marker\": \"\"}]}"}]} {"id": "paystub_fin_088_extract", "task": "extract", "class": "paystub", "doc": "paystub_fin_088.pdf", "images": ["images/train/paystub/paystub_fin_088_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('paystub', 'The document class.')\nfields = (['pay_date', 'employee_name', 'employer_name', 'gross_pay', 'ytd_gross', 'pay_frequency', 'pay_period_start', 'pay_period_end', 'federal_tax', 'state_tax', 'social_security_tax', 'medicare_tax', 'hire_date', 'direct_deposit_bank_type'], 'The exact field names to extract.')\nhints = ({'pay_date': 'Date the paycheck was issued', 'employee_name': 'Name of Employee', 'employer_name': 'Name of Employer', 'gross_pay': 'Gross pay per period (current period, not YTD)', 'ytd_gross': 'Gross pay on paystub year to date (cumulative, not current period)', 'pay_frequency': 'The frequency if provided: weekly, biweekly, monthly, semi-monthly', 'pay_period_start': 'Pay period start date', 'pay_period_end': 'Pay period end date', 'hire_date': 'Job hire date', 'direct_deposit_bank_type': 'Direct deposit bank type'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"pay_date\", \"value\": \"\"}, {\"field\": \"employee_name\", \"value\": \"[YOUR NAME]\"}, {\"field\": \"employer_name\", \"value\": \"\"}, {\"field\": \"gross_pay\", \"value\": \"5500.0\"}, {\"field\": \"ytd_gross\", \"value\": \"\"}, {\"field\": \"pay_frequency\", \"value\": \"\"}, {\"field\": \"pay_period_start\", \"value\": \"2050-01-01\"}, {\"field\": \"pay_period_end\", \"value\": \"2050-01-31\"}, {\"field\": \"federal_tax\", \"value\": \"1200.0\"}, {\"field\": \"state_tax\", \"value\": \"300.0\"}, {\"field\": \"social_security_tax\", \"value\": \"\"}, {\"field\": \"medicare_tax\", \"value\": \"\"}, {\"field\": \"hire_date\", \"value\": \"\"}, {\"field\": \"direct_deposit_bank_type\", \"value\": \"\"}]}"}]} {"id": "paystub_fin_089_p1_classify", "task": "classify", "class": "paystub", "doc": "paystub_fin_089.pdf", "images": ["images/train/paystub/paystub_fin_089_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"paystub\", \"confidence\": 1.0, \"issuer\": \"GlaxoSmithKline\", \"holder\": \"Alamsyah Hidayat\", \"page_marker\": \"\"}]}"}]} {"id": "paystub_fin_089_extract", "task": "extract", "class": "paystub", "doc": "paystub_fin_089.pdf", "images": ["images/train/paystub/paystub_fin_089_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('paystub', 'The document class.')\nfields = (['pay_date', 'employee_name', 'employer_name', 'gross_pay', 'ytd_gross', 'pay_frequency', 'pay_period_start', 'pay_period_end', 'federal_tax', 'state_tax', 'social_security_tax', 'medicare_tax', 'hire_date', 'direct_deposit_bank_type'], 'The exact field names to extract.')\nhints = ({'pay_date': 'Date the paycheck was issued', 'employee_name': 'Name of Employee', 'employer_name': 'Name of Employer', 'gross_pay': 'Gross pay per period (current period, not YTD)', 'ytd_gross': 'Gross pay on paystub year to date (cumulative, not current period)', 'pay_frequency': 'The frequency if provided: weekly, biweekly, monthly, semi-monthly', 'pay_period_start': 'Pay period start date', 'pay_period_end': 'Pay period end date', 'hire_date': 'Job hire date', 'direct_deposit_bank_type': 'Direct deposit bank type'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"pay_date\", \"value\": \"2024-12-21\"}, {\"field\": \"employee_name\", \"value\": \"Alamsyah Hidayat\"}, {\"field\": \"employer_name\", \"value\": \"GlaxoSmithKline\"}, {\"field\": \"gross_pay\", \"value\": \"3033.75\"}, {\"field\": \"ytd_gross\", \"value\": \"3033.75\"}, {\"field\": \"pay_frequency\", \"value\": \"monthly\"}, {\"field\": \"pay_period_start\", \"value\": \"\"}, {\"field\": \"pay_period_end\", \"value\": \"\"}, {\"field\": \"federal_tax\", \"value\": \"250.0\"}, {\"field\": \"state_tax\", \"value\": \"\"}, {\"field\": \"social_security_tax\", \"value\": \"\"}, {\"field\": \"medicare_tax\", \"value\": \"\"}, {\"field\": \"hire_date\", \"value\": \"\"}, {\"field\": \"direct_deposit_bank_type\", \"value\": \"\"}]}"}]} {"id": "paystub_fin_090_p1_classify", "task": "classify", "class": "paystub", "doc": "paystub_fin_090.pdf", "images": ["images/train/paystub/paystub_fin_090_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"paystub\", \"confidence\": 1.0, \"issuer\": \"YOUR COMPANY NAME\", \"holder\": \"Sarah Wilson\", \"page_marker\": \"\"}]}"}]} {"id": "paystub_fin_090_extract", "task": "extract", "class": "paystub", "doc": "paystub_fin_090.pdf", "images": ["images/train/paystub/paystub_fin_090_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('paystub', 'The document class.')\nfields = (['pay_date', 'employee_name', 'employer_name', 'gross_pay', 'ytd_gross', 'pay_frequency', 'pay_period_start', 'pay_period_end', 'federal_tax', 'state_tax', 'social_security_tax', 'medicare_tax', 'hire_date', 'direct_deposit_bank_type'], 'The exact field names to extract.')\nhints = ({'pay_date': 'Date the paycheck was issued', 'employee_name': 'Name of Employee', 'employer_name': 'Name of Employer', 'gross_pay': 'Gross pay per period (current period, not YTD)', 'ytd_gross': 'Gross pay on paystub year to date (cumulative, not current period)', 'pay_frequency': 'The frequency if provided: weekly, biweekly, monthly, semi-monthly', 'pay_period_start': 'Pay period start date', 'pay_period_end': 'Pay period end date', 'hire_date': 'Job hire date', 'direct_deposit_bank_type': 'Direct deposit bank type'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"pay_date\", \"value\": \"2023-01-10\"}, {\"field\": \"employee_name\", \"value\": \"Sarah Wilson\"}, {\"field\": \"employer_name\", \"value\": \"YOUR COMPANY NAME\"}, {\"field\": \"gross_pay\", \"value\": \"6354.0\"}, {\"field\": \"ytd_gross\", \"value\": \"\"}, {\"field\": \"pay_frequency\", \"value\": \"\"}, {\"field\": \"pay_period_start\", \"value\": \"\"}, {\"field\": \"pay_period_end\", \"value\": \"\"}, {\"field\": \"federal_tax\", \"value\": \"158.0\"}, {\"field\": \"state_tax\", \"value\": \"127.08\"}, {\"field\": \"social_security_tax\", \"value\": \"\"}, {\"field\": \"medicare_tax\", \"value\": \"\"}, {\"field\": \"hire_date\", \"value\": \"\"}, {\"field\": \"direct_deposit_bank_type\", \"value\": \"\"}]}"}]} {"id": "paystub_fin_091_p1_classify", "task": "classify", "class": "paystub", "doc": "paystub_fin_091.pdf", "images": ["images/train/paystub/paystub_fin_091_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"paystub\", \"confidence\": 1.0, \"issuer\": \"Company Name Ltd.\", \"holder\": \"Mr. James Parker\", \"page_marker\": \"\"}]}"}]} {"id": "paystub_fin_091_extract", "task": "extract", "class": "paystub", "doc": "paystub_fin_091.pdf", "images": ["images/train/paystub/paystub_fin_091_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('paystub', 'The document class.')\nfields = (['pay_date', 'employee_name', 'employer_name', 'gross_pay', 'ytd_gross', 'pay_frequency', 'pay_period_start', 'pay_period_end', 'federal_tax', 'state_tax', 'social_security_tax', 'medicare_tax', 'hire_date', 'direct_deposit_bank_type'], 'The exact field names to extract.')\nhints = ({'pay_date': 'Date the paycheck was issued', 'employee_name': 'Name of Employee', 'employer_name': 'Name of Employer', 'gross_pay': 'Gross pay per period (current period, not YTD)', 'ytd_gross': 'Gross pay on paystub year to date (cumulative, not current period)', 'pay_frequency': 'The frequency if provided: weekly, biweekly, monthly, semi-monthly', 'pay_period_start': 'Pay period start date', 'pay_period_end': 'Pay period end date', 'hire_date': 'Job hire date', 'direct_deposit_bank_type': 'Direct deposit bank type'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"pay_date\", \"value\": \"2025-02-28\"}, {\"field\": \"employee_name\", \"value\": \"Mr. James Parker\"}, {\"field\": \"employer_name\", \"value\": \"Company Name Ltd.\"}, {\"field\": \"gross_pay\", \"value\": \"2193.71\"}, {\"field\": \"ytd_gross\", \"value\": \"23934.74\"}, {\"field\": \"pay_frequency\", \"value\": \"monthly\"}, {\"field\": \"pay_period_start\", \"value\": \"\"}, {\"field\": \"pay_period_end\", \"value\": \"\"}, {\"field\": \"federal_tax\", \"value\": \"207.2\"}, {\"field\": \"state_tax\", \"value\": \"\"}, {\"field\": \"social_security_tax\", \"value\": \"\"}, {\"field\": \"medicare_tax\", \"value\": \"\"}, {\"field\": \"hire_date\", \"value\": \"\"}, {\"field\": \"direct_deposit_bank_type\", \"value\": \"BACS\"}]}"}]} {"id": "paystub_fin_092_p1_classify", "task": "classify", "class": "paystub", "doc": "paystub_fin_092.pdf", "images": ["images/train/paystub/paystub_fin_092_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"paystub\", \"confidence\": 1.0, \"issuer\": \"PERFECT PAYROLL LTD\", \"holder\": \"Dwayne Johnson\", \"page_marker\": \"\"}]}"}]} {"id": "paystub_fin_092_extract", "task": "extract", "class": "paystub", "doc": "paystub_fin_092.pdf", "images": ["images/train/paystub/paystub_fin_092_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('paystub', 'The document class.')\nfields = (['pay_date', 'employee_name', 'employer_name', 'gross_pay', 'ytd_gross', 'pay_frequency', 'pay_period_start', 'pay_period_end', 'federal_tax', 'state_tax', 'social_security_tax', 'medicare_tax', 'hire_date', 'direct_deposit_bank_type'], 'The exact field names to extract.')\nhints = ({'pay_date': 'Date the paycheck was issued', 'employee_name': 'Name of Employee', 'employer_name': 'Name of Employer', 'gross_pay': 'Gross pay per period (current period, not YTD)', 'ytd_gross': 'Gross pay on paystub year to date (cumulative, not current period)', 'pay_frequency': 'The frequency if provided: weekly, biweekly, monthly, semi-monthly', 'pay_period_start': 'Pay period start date', 'pay_period_end': 'Pay period end date', 'hire_date': 'Job hire date', 'direct_deposit_bank_type': 'Direct deposit bank type'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"pay_date\", \"value\": \"2020-06-30\"}, {\"field\": \"employee_name\", \"value\": \"Dwayne Johnson\"}, {\"field\": \"employer_name\", \"value\": \"PERFECT PAYROLL LTD\"}, {\"field\": \"gross_pay\", \"value\": \"5000.0\"}, {\"field\": \"ytd_gross\", \"value\": \"15000.0\"}, {\"field\": \"pay_frequency\", \"value\": \"monthly\"}, {\"field\": \"pay_period_start\", \"value\": \"\"}, {\"field\": \"pay_period_end\", \"value\": \"\"}, {\"field\": \"federal_tax\", \"value\": \"957.8\"}, {\"field\": \"state_tax\", \"value\": \"\"}, {\"field\": \"social_security_tax\", \"value\": \"\"}, {\"field\": \"medicare_tax\", \"value\": \"\"}, {\"field\": \"hire_date\", \"value\": \"\"}, {\"field\": \"direct_deposit_bank_type\", \"value\": \"\"}]}"}]} {"id": "paystub_fin_093_p1_classify", "task": "classify", "class": "paystub", "doc": "paystub_fin_093.pdf", "images": ["images/train/paystub/paystub_fin_093_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"paystub\", \"confidence\": 1.0, \"issuer\": \"Design LLC\", \"holder\": \"George Mathew\", \"page_marker\": \"\"}]}"}]} {"id": "paystub_fin_093_extract", "task": "extract", "class": "paystub", "doc": "paystub_fin_093.pdf", "images": ["images/train/paystub/paystub_fin_093_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('paystub', 'The document class.')\nfields = (['pay_date', 'employee_name', 'employer_name', 'gross_pay', 'ytd_gross', 'pay_frequency', 'pay_period_start', 'pay_period_end', 'federal_tax', 'state_tax', 'social_security_tax', 'medicare_tax', 'hire_date', 'direct_deposit_bank_type'], 'The exact field names to extract.')\nhints = ({'pay_date': 'Date the paycheck was issued', 'employee_name': 'Name of Employee', 'employer_name': 'Name of Employer', 'gross_pay': 'Gross pay per period (current period, not YTD)', 'ytd_gross': 'Gross pay on paystub year to date (cumulative, not current period)', 'pay_frequency': 'The frequency if provided: weekly, biweekly, monthly, semi-monthly', 'pay_period_start': 'Pay period start date', 'pay_period_end': 'Pay period end date', 'hire_date': 'Job hire date', 'direct_deposit_bank_type': 'Direct deposit bank type'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"pay_date\", \"value\": \"2023-01-06\"}, {\"field\": \"employee_name\", \"value\": \"George Mathew\"}, {\"field\": \"employer_name\", \"value\": \"Design LLC\"}, {\"field\": \"gross_pay\", \"value\": \"600.0\"}, {\"field\": \"ytd_gross\", \"value\": \"600.0\"}, {\"field\": \"pay_frequency\", \"value\": \"weekly\"}, {\"field\": \"pay_period_start\", \"value\": \"2023-01-01\"}, {\"field\": \"pay_period_end\", \"value\": \"2023-01-07\"}, {\"field\": \"federal_tax\", \"value\": \"35.81\"}, {\"field\": \"state_tax\", \"value\": \"9.73\"}, {\"field\": \"social_security_tax\", \"value\": \"37.2\"}, {\"field\": \"medicare_tax\", \"value\": \"8.7\"}, {\"field\": \"hire_date\", \"value\": \"\"}, {\"field\": \"direct_deposit_bank_type\", \"value\": \"\"}]}"}]} {"id": "state_id_fin_001_p1_classify", "task": "classify", "class": "state_id", "doc": "state_id_fin_001.pdf", "images": ["images/train/state_id/state_id_fin_001_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"state_id\", \"confidence\": 1.0, \"issuer\": \"Connecticut\", \"holder\": \"PABON WILLIAM EDGARDO\", \"page_marker\": \"\"}]}"}]} {"id": "state_id_fin_001_extract", "task": "extract", "class": "state_id", "doc": "state_id_fin_001.pdf", "images": ["images/train/state_id/state_id_fin_001_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('state_id', 'The document class.')\nfields = (['full_name', 'date_of_birth', 'address', 'issue_date', 'expiration_date', 'state'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name', 'date_of_birth': 'Date of Birth', 'address': 'Address', 'issue_date': 'Issued Date', 'expiration_date': 'Valid Until Date', 'state': 'Issuing state'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"PABON WILLIAM EDGARDO\"}, {\"field\": \"date_of_birth\", \"value\": \"1964-11-13\"}, {\"field\": \"address\", \"value\": \"30 HICKORY ST FL 3S, WATERBURY, CT 06706-1406\"}, {\"field\": \"issue_date\", \"value\": \"2022-11-09\"}, {\"field\": \"expiration_date\", \"value\": \"2030-11-13\"}, {\"field\": \"state\", \"value\": \"Connecticut\"}]}"}]} {"id": "state_id_fin_002_p1_classify", "task": "classify", "class": "state_id", "doc": "state_id_fin_002.pdf", "images": ["images/train/state_id/state_id_fin_002_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"state_id\", \"confidence\": 1.0, \"issuer\": \"Republic of Turkey\", \"holder\": \"MUSTAFA ALI YILMAZ\", \"page_marker\": \"\"}]}"}]} {"id": "state_id_fin_002_extract", "task": "extract", "class": "state_id", "doc": "state_id_fin_002.pdf", "images": ["images/train/state_id/state_id_fin_002_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('state_id', 'The document class.')\nfields = (['full_name', 'date_of_birth', 'address', 'issue_date', 'expiration_date', 'state'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name', 'date_of_birth': 'Date of Birth', 'address': 'Address', 'issue_date': 'Issued Date', 'expiration_date': 'Valid Until Date', 'state': 'Issuing state'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"MUSTAFA ALI YILMAZ\"}, {\"field\": \"date_of_birth\", \"value\": \"1999-07-07\"}, {\"field\": \"address\", \"value\": \"\"}, {\"field\": \"issue_date\", \"value\": \"\"}, {\"field\": \"expiration_date\", \"value\": \"2018-12-12\"}, {\"field\": \"state\", \"value\": \"Republic of Turkey\"}]}"}]} {"id": "state_id_fin_003_p1_classify", "task": "classify", "class": "state_id", "doc": "state_id_fin_003.pdf", "images": ["images/train/state_id/state_id_fin_003_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"state_id\", \"confidence\": 1.0, \"issuer\": \"OKLAHOMA\", \"holder\": \"LOWE RAVEN NICOLE\", \"page_marker\": \"\"}]}"}]} {"id": "state_id_fin_003_extract", "task": "extract", "class": "state_id", "doc": "state_id_fin_003.pdf", "images": ["images/train/state_id/state_id_fin_003_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('state_id', 'The document class.')\nfields = (['full_name', 'date_of_birth', 'address', 'issue_date', 'expiration_date', 'state'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name', 'date_of_birth': 'Date of Birth', 'address': 'Address', 'issue_date': 'Issued Date', 'expiration_date': 'Valid Until Date', 'state': 'Issuing state'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"LOWE RAVEN NICOLE\"}, {\"field\": \"date_of_birth\", \"value\": \"1986-02-23\"}, {\"field\": \"address\", \"value\": \"5337 W 8TH ST, TULSA, OK 74127-7557\"}, {\"field\": \"issue_date\", \"value\": \"2022-11-01\"}, {\"field\": \"expiration_date\", \"value\": \"2025-12-31\"}, {\"field\": \"state\", \"value\": \"OKLAHOMA\"}]}"}]} {"id": "state_id_fin_004_p1_classify", "task": "classify", "class": "state_id", "doc": "state_id_fin_004.pdf", "images": ["images/train/state_id/state_id_fin_004_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"state_id\", \"confidence\": 1.0, \"issuer\": \"Texas\", \"holder\": \"DE SANTIAGO JORGE\", \"page_marker\": \"\"}]}"}]} {"id": "state_id_fin_004_extract", "task": "extract", "class": "state_id", "doc": "state_id_fin_004.pdf", "images": ["images/train/state_id/state_id_fin_004_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('state_id', 'The document class.')\nfields = (['full_name', 'date_of_birth', 'address', 'issue_date', 'expiration_date', 'state'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name', 'date_of_birth': 'Date of Birth', 'address': 'Address', 'issue_date': 'Issued Date', 'expiration_date': 'Valid Until Date', 'state': 'Issuing state'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"DE SANTIAGO JORGE\"}, {\"field\": \"date_of_birth\", \"value\": \"1979-01-26\"}, {\"field\": \"address\", \"value\": \"1717 COKE ST FRONT, LAREDO, TX 78040\"}, {\"field\": \"issue_date\", \"value\": \"2024-04-04\"}, {\"field\": \"expiration_date\", \"value\": \"2030-01-26\"}, {\"field\": \"state\", \"value\": \"Texas\"}]}"}]} {"id": "state_id_fin_005_p1_classify", "task": "classify", "class": "state_id", "doc": "state_id_fin_005.pdf", "images": ["images/train/state_id/state_id_fin_005_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"state_id\", \"confidence\": 1.0, \"issuer\": \"WASHINGTON\", \"holder\": \"SAMPLE JOHN A\", \"page_marker\": \"\"}]}"}]} {"id": "state_id_fin_005_extract", "task": "extract", "class": "state_id", "doc": "state_id_fin_005.pdf", "images": ["images/train/state_id/state_id_fin_005_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('state_id', 'The document class.')\nfields = (['full_name', 'date_of_birth', 'address', 'issue_date', 'expiration_date', 'state'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name', 'date_of_birth': 'Date of Birth', 'address': 'Address', 'issue_date': 'Issued Date', 'expiration_date': 'Valid Until Date', 'state': 'Issuing state'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"SAMPLE JOHN A\"}, {\"field\": \"date_of_birth\", \"value\": \"1958-01-06\"}, {\"field\": \"address\", \"value\": \"123 STREET ADDRESS YOUR CITY WA 99999-0000\"}, {\"field\": \"issue_date\", \"value\": \"2015-01-06\"}, {\"field\": \"expiration_date\", \"value\": \"2020-08-12\"}, {\"field\": \"state\", \"value\": \"WASHINGTON\"}]}"}]} {"id": "state_id_fin_007_p1_classify", "task": "classify", "class": "state_id", "doc": "state_id_fin_007.pdf", "images": ["images/train/state_id/state_id_fin_007_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"state_id\", \"confidence\": 1.0, \"issuer\": \"Republic of Turkey\", \"holder\": \"HAKAN ACAR\", \"page_marker\": \"\"}]}"}]} {"id": "state_id_fin_007_extract", "task": "extract", "class": "state_id", "doc": "state_id_fin_007.pdf", "images": ["images/train/state_id/state_id_fin_007_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('state_id', 'The document class.')\nfields = (['full_name', 'date_of_birth', 'address', 'issue_date', 'expiration_date', 'state'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name', 'date_of_birth': 'Date of Birth', 'address': 'Address', 'issue_date': 'Issued Date', 'expiration_date': 'Valid Until Date', 'state': 'Issuing state'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"HAKAN ACAR\"}, {\"field\": \"date_of_birth\", \"value\": \"1975-05-22\"}, {\"field\": \"address\", \"value\": \"\"}, {\"field\": \"issue_date\", \"value\": \"\"}, {\"field\": \"expiration_date\", \"value\": \"2017-07-20\"}, {\"field\": \"state\", \"value\": \"Republic of Turkey\"}]}"}]} {"id": "state_id_fin_008_p1_classify", "task": "classify", "class": "state_id", "doc": "state_id_fin_008.pdf", "images": ["images/train/state_id/state_id_fin_008_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"state_id\", \"confidence\": 1.0, \"issuer\": \"\", \"holder\": \"MARTIN Ma\\u00eblys-Ga\\u00eblle, Marie\", \"page_marker\": \"\"}]}"}]} {"id": "state_id_fin_008_extract", "task": "extract", "class": "state_id", "doc": "state_id_fin_008.pdf", "images": ["images/train/state_id/state_id_fin_008_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('state_id', 'The document class.')\nfields = (['full_name', 'date_of_birth', 'address', 'issue_date', 'expiration_date', 'state'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name', 'date_of_birth': 'Date of Birth', 'address': 'Address', 'issue_date': 'Issued Date', 'expiration_date': 'Valid Until Date', 'state': 'Issuing state'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"MARTIN Ma\\u00eblys-Ga\\u00eblle, Marie\"}, {\"field\": \"date_of_birth\", \"value\": \"1990-07-13\"}, {\"field\": \"address\", \"value\": \"\"}, {\"field\": \"issue_date\", \"value\": \"\"}, {\"field\": \"expiration_date\", \"value\": \"2030-02-11\"}, {\"field\": \"state\", \"value\": \"\"}]}"}]} {"id": "state_id_fin_009_p1_classify", "task": "classify", "class": "state_id", "doc": "state_id_fin_009.pdf", "images": ["images/train/state_id/state_id_fin_009_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"state_id\", \"confidence\": 1.0, \"issuer\": \"Republic of Lithuania\", \"holder\": \"BRU\\u017dAIT\\u0116 VIGILIJA\", \"page_marker\": \"\"}]}"}]} {"id": "state_id_fin_009_extract", "task": "extract", "class": "state_id", "doc": "state_id_fin_009.pdf", "images": ["images/train/state_id/state_id_fin_009_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('state_id', 'The document class.')\nfields = (['full_name', 'date_of_birth', 'address', 'issue_date', 'expiration_date', 'state'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name', 'date_of_birth': 'Date of Birth', 'address': 'Address', 'issue_date': 'Issued Date', 'expiration_date': 'Valid Until Date', 'state': 'Issuing state'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"BRU\\u017dAIT\\u0116 VIGILIJA\"}, {\"field\": \"date_of_birth\", \"value\": \"1978-03-11\"}, {\"field\": \"address\", \"value\": \"\"}, {\"field\": \"issue_date\", \"value\": \"\"}, {\"field\": \"expiration_date\", \"value\": \"2022-07-04\"}, {\"field\": \"state\", \"value\": \"Republic of Lithuania\"}]}"}]} {"id": "state_id_fin_012_p1_classify", "task": "classify", "class": "state_id", "doc": "state_id_fin_012.pdf", "images": ["images/train/state_id/state_id_fin_012_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"state_id\", \"confidence\": 1.0, \"issuer\": \"Wyoming\", \"holder\": \"SAMPLE TINA A\", \"page_marker\": \"\"}]}"}]} {"id": "state_id_fin_012_extract", "task": "extract", "class": "state_id", "doc": "state_id_fin_012.pdf", "images": ["images/train/state_id/state_id_fin_012_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('state_id', 'The document class.')\nfields = (['full_name', 'date_of_birth', 'address', 'issue_date', 'expiration_date', 'state'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name', 'date_of_birth': 'Date of Birth', 'address': 'Address', 'issue_date': 'Issued Date', 'expiration_date': 'Valid Until Date', 'state': 'Issuing state'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"SAMPLE TINA A\"}, {\"field\": \"date_of_birth\", \"value\": \"1984-11-24\"}, {\"field\": \"address\", \"value\": \"123 ANYSTREET ANYTOWN, WY 82000\"}, {\"field\": \"issue_date\", \"value\": \"2020-07-21\"}, {\"field\": \"expiration_date\", \"value\": \"2028-07-21\"}, {\"field\": \"state\", \"value\": \"Wyoming\"}]}"}]} {"id": "state_id_fin_013_p1_classify", "task": "classify", "class": "state_id", "doc": "state_id_fin_013.pdf", "images": ["images/train/state_id/state_id_fin_013_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"state_id\", \"confidence\": 1.0, \"issuer\": \"WASHINGTON\", \"holder\": \"SAMPLE JOHN A\", \"page_marker\": \"\"}]}"}]} {"id": "state_id_fin_013_extract", "task": "extract", "class": "state_id", "doc": "state_id_fin_013.pdf", "images": ["images/train/state_id/state_id_fin_013_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('state_id', 'The document class.')\nfields = (['full_name', 'date_of_birth', 'address', 'issue_date', 'expiration_date', 'state'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name', 'date_of_birth': 'Date of Birth', 'address': 'Address', 'issue_date': 'Issued Date', 'expiration_date': 'Valid Until Date', 'state': 'Issuing state'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"SAMPLE JOHN A\"}, {\"field\": \"date_of_birth\", \"value\": \"1958-01-06\"}, {\"field\": \"address\", \"value\": \"123 STREET ADDRESS YOUR CITY WA 99999-1234\"}, {\"field\": \"issue_date\", \"value\": \"2015-01-06\"}, {\"field\": \"expiration_date\", \"value\": \"2020-08-12\"}, {\"field\": \"state\", \"value\": \"WASHINGTON\"}]}"}]} {"id": "state_id_fin_014_p1_classify", "task": "classify", "class": "state_id", "doc": "state_id_fin_014.pdf", "images": ["images/train/state_id/state_id_fin_014_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"state_id\", \"confidence\": 1.0, \"issuer\": \"WASHINGTON\", \"holder\": \"CHEMOSH SECOTHBENOTH HAURES\", \"page_marker\": \"\"}]}"}]} {"id": "state_id_fin_014_extract", "task": "extract", "class": "state_id", "doc": "state_id_fin_014.pdf", "images": ["images/train/state_id/state_id_fin_014_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('state_id', 'The document class.')\nfields = (['full_name', 'date_of_birth', 'address', 'issue_date', 'expiration_date', 'state'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name', 'date_of_birth': 'Date of Birth', 'address': 'Address', 'issue_date': 'Issued Date', 'expiration_date': 'Valid Until Date', 'state': 'Issuing state'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"CHEMOSH SECOTHBENOTH HAURES\"}, {\"field\": \"date_of_birth\", \"value\": \"1977-11-11\"}, {\"field\": \"address\", \"value\": \"4220 6TH AVE SE, LACEY WA 98503-1024\"}, {\"field\": \"issue_date\", \"value\": \"2017-01-17\"}, {\"field\": \"expiration_date\", \"value\": \"2022-11-11\"}, {\"field\": \"state\", \"value\": \"WASHINGTON\"}]}"}]} {"id": "state_id_fin_016_p1_classify", "task": "classify", "class": "state_id", "doc": "state_id_fin_016.pdf", "images": ["images/train/state_id/state_id_fin_016_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"state_id\", \"confidence\": 1.0, \"issuer\": \"WASHINGTON\", \"holder\": \"DOE, MARY JANE\", \"page_marker\": \"\"}]}"}]} {"id": "state_id_fin_016_extract", "task": "extract", "class": "state_id", "doc": "state_id_fin_016.pdf", "images": ["images/train/state_id/state_id_fin_016_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('state_id', 'The document class.')\nfields = (['full_name', 'date_of_birth', 'address', 'issue_date', 'expiration_date', 'state'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name', 'date_of_birth': 'Date of Birth', 'address': 'Address', 'issue_date': 'Issued Date', 'expiration_date': 'Valid Until Date', 'state': 'Issuing state'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"DOE, MARY JANE\"}, {\"field\": \"date_of_birth\", \"value\": \"1993-06-06\"}, {\"field\": \"address\", \"value\": \"2604 MARTIN WAY, OLYMPIA WA 98516-0000\"}, {\"field\": \"issue_date\", \"value\": \"2010-09-14\"}, {\"field\": \"expiration_date\", \"value\": \"2015-06-06\"}, {\"field\": \"state\", \"value\": \"WASHINGTON\"}]}"}]} {"id": "state_id_fin_017_p1_classify", "task": "classify", "class": "state_id", "doc": "state_id_fin_017.pdf", "images": ["images/train/state_id/state_id_fin_017_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"state_id\", \"confidence\": 1.0, \"issuer\": \"CALIFORNIA\", \"holder\": \"SEIJI SAMPLE\", \"page_marker\": \"\"}]}"}]} {"id": "state_id_fin_017_extract", "task": "extract", "class": "state_id", "doc": "state_id_fin_017.pdf", "images": ["images/train/state_id/state_id_fin_017_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('state_id', 'The document class.')\nfields = (['full_name', 'date_of_birth', 'address', 'issue_date', 'expiration_date', 'state'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name', 'date_of_birth': 'Date of Birth', 'address': 'Address', 'issue_date': 'Issued Date', 'expiration_date': 'Valid Until Date', 'state': 'Issuing state'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"SEIJI SAMPLE\"}, {\"field\": \"date_of_birth\", \"value\": \"1977-08-31\"}, {\"field\": \"address\", \"value\": \"2570 24TH STREET, SACRAMENTO, CA 95818\"}, {\"field\": \"issue_date\", \"value\": \"2010-09-30\"}, {\"field\": \"expiration_date\", \"value\": \"2015-08-31\"}, {\"field\": \"state\", \"value\": \"CALIFORNIA\"}]}"}]} {"id": "state_id_fin_018_p1_classify", "task": "classify", "class": "state_id", "doc": "state_id_fin_018.pdf", "images": ["images/train/state_id/state_id_fin_018_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"state_id\", \"confidence\": 1.0, \"issuer\": \"Colorado\", \"holder\": \"LEROY HAROLD SHAW\", \"page_marker\": \"\"}]}"}]} {"id": "state_id_fin_018_extract", "task": "extract", "class": "state_id", "doc": "state_id_fin_018.pdf", "images": ["images/train/state_id/state_id_fin_018_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('state_id', 'The document class.')\nfields = (['full_name', 'date_of_birth', 'address', 'issue_date', 'expiration_date', 'state'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name', 'date_of_birth': 'Date of Birth', 'address': 'Address', 'issue_date': 'Issued Date', 'expiration_date': 'Valid Until Date', 'state': 'Issuing state'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"LEROY HAROLD SHAW\"}, {\"field\": \"date_of_birth\", \"value\": \"1952-09-30\"}, {\"field\": \"address\", \"value\": \"1934 DALLAS ST, DENVER, CO 80205\"}, {\"field\": \"issue_date\", \"value\": \"2015-11-03\"}, {\"field\": \"expiration_date\", \"value\": \"2020-09-30\"}, {\"field\": \"state\", \"value\": \"Colorado\"}]}"}]} {"id": "state_id_fin_019_p1_classify", "task": "classify", "class": "state_id", "doc": "state_id_fin_019.pdf", "images": ["images/train/state_id/state_id_fin_019_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"state_id\", \"confidence\": 1.0, \"issuer\": \"MARYLAND\", \"holder\": \"JASON THOMAS SAMPLE\", \"page_marker\": \"\"}]}"}]} {"id": "state_id_fin_019_extract", "task": "extract", "class": "state_id", "doc": "state_id_fin_019.pdf", "images": ["images/train/state_id/state_id_fin_019_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('state_id', 'The document class.')\nfields = (['full_name', 'date_of_birth', 'address', 'issue_date', 'expiration_date', 'state'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name', 'date_of_birth': 'Date of Birth', 'address': 'Address', 'issue_date': 'Issued Date', 'expiration_date': 'Valid Until Date', 'state': 'Issuing state'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"JASON THOMAS SAMPLE\"}, {\"field\": \"date_of_birth\", \"value\": \"1992-02-18\"}, {\"field\": \"address\", \"value\": \"123 ANY STREET GLEN BURNIE MD 21062\"}, {\"field\": \"issue_date\", \"value\": \"2016-03-25\"}, {\"field\": \"expiration_date\", \"value\": \"2023-02-18\"}, {\"field\": \"state\", \"value\": \"MARYLAND\"}]}"}]} {"id": "state_id_fin_020_p1_classify", "task": "classify", "class": "state_id", "doc": "state_id_fin_020.pdf", "images": ["images/train/state_id/state_id_fin_020_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"state_id\", \"confidence\": 1.0, \"issuer\": \"Republica Moldova\", \"holder\": \"MOR\\u0102RU\\u015e TATIANA\", \"page_marker\": \"\"}]}"}]} {"id": "state_id_fin_020_extract", "task": "extract", "class": "state_id", "doc": "state_id_fin_020.pdf", "images": ["images/train/state_id/state_id_fin_020_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('state_id', 'The document class.')\nfields = (['full_name', 'date_of_birth', 'address', 'issue_date', 'expiration_date', 'state'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name', 'date_of_birth': 'Date of Birth', 'address': 'Address', 'issue_date': 'Issued Date', 'expiration_date': 'Valid Until Date', 'state': 'Issuing state'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"MOR\\u0102RU\\u015e TATIANA\"}, {\"field\": \"date_of_birth\", \"value\": \"1992-08-09\"}, {\"field\": \"address\", \"value\": \"\"}, {\"field\": \"issue_date\", \"value\": \"\"}, {\"field\": \"expiration_date\", \"value\": \"2034-07-03\"}, {\"field\": \"state\", \"value\": \"Republica Moldova\"}]}"}]} {"id": "state_id_fin_021_p1_classify", "task": "classify", "class": "state_id", "doc": "state_id_fin_021.pdf", "images": ["images/train/state_id/state_id_fin_021_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"state_id\", \"confidence\": 1.0, \"issuer\": \"VERMONT\", \"holder\": \"SAMPLE JOHN A\", \"page_marker\": \"\"}]}"}]} {"id": "state_id_fin_021_extract", "task": "extract", "class": "state_id", "doc": "state_id_fin_021.pdf", "images": ["images/train/state_id/state_id_fin_021_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('state_id', 'The document class.')\nfields = (['full_name', 'date_of_birth', 'address', 'issue_date', 'expiration_date', 'state'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name', 'date_of_birth': 'Date of Birth', 'address': 'Address', 'issue_date': 'Issued Date', 'expiration_date': 'Valid Until Date', 'state': 'Issuing state'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"SAMPLE JOHN A\"}, {\"field\": \"date_of_birth\", \"value\": \"1968-01-06\"}, {\"field\": \"address\", \"value\": \"123 STREET ADDRESS ADDRESS LINE 2 MONTPELIER, VT 12345-1234\"}, {\"field\": \"issue_date\", \"value\": \"2019-01-05\"}, {\"field\": \"expiration_date\", \"value\": \"2023-01-06\"}, {\"field\": \"state\", \"value\": \"VERMONT\"}]}"}]} {"id": "state_id_fin_022_p1_classify", "task": "classify", "class": "state_id", "doc": "state_id_fin_022.pdf", "images": ["images/train/state_id/state_id_fin_022_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"state_id\", \"confidence\": 1.0, \"issuer\": \"NY\", \"holder\": \"SAMPLE WENDY, S\", \"page_marker\": \"\"}]}"}]} {"id": "state_id_fin_022_extract", "task": "extract", "class": "state_id", "doc": "state_id_fin_022.pdf", "images": ["images/train/state_id/state_id_fin_022_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('state_id', 'The document class.')\nfields = (['full_name', 'date_of_birth', 'address', 'issue_date', 'expiration_date', 'state'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name', 'date_of_birth': 'Date of Birth', 'address': 'Address', 'issue_date': 'Issued Date', 'expiration_date': 'Valid Until Date', 'state': 'Issuing state'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"SAMPLE WENDY, S\"}, {\"field\": \"date_of_birth\", \"value\": \"2001-12-24\"}, {\"field\": \"address\", \"value\": \"3300 ACORN STREET APARTMENT 1 BROOKLYN, NY 11201\"}, {\"field\": \"issue_date\", \"value\": \"2025-03-11\"}, {\"field\": \"expiration_date\", \"value\": \"2030-03-11\"}, {\"field\": \"state\", \"value\": \"NY\"}]}"}]} {"id": "state_id_fin_023_p1_classify", "task": "classify", "class": "state_id", "doc": "state_id_fin_023.pdf", "images": ["images/train/state_id/state_id_fin_023_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"state_id\", \"confidence\": 1.0, \"issuer\": \"Nebraska\", \"holder\": \"JOEL SAMPLE\", \"page_marker\": \"\"}]}"}]} {"id": "state_id_fin_023_extract", "task": "extract", "class": "state_id", "doc": "state_id_fin_023.pdf", "images": ["images/train/state_id/state_id_fin_023_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('state_id', 'The document class.')\nfields = (['full_name', 'date_of_birth', 'address', 'issue_date', 'expiration_date', 'state'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name', 'date_of_birth': 'Date of Birth', 'address': 'Address', 'issue_date': 'Issued Date', 'expiration_date': 'Valid Until Date', 'state': 'Issuing state'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"JOEL SAMPLE\"}, {\"field\": \"date_of_birth\", \"value\": \"1963-03-21\"}, {\"field\": \"address\", \"value\": \"762 ALDERMAN RD, KEARNEY, NE 87717\"}, {\"field\": \"issue_date\", \"value\": \"2009-10-20\"}, {\"field\": \"expiration_date\", \"value\": \"2014-03-21\"}, {\"field\": \"state\", \"value\": \"Nebraska\"}]}"}]} {"id": "state_id_fin_024_p1_classify", "task": "classify", "class": "state_id", "doc": "state_id_fin_024.pdf", "images": ["images/train/state_id/state_id_fin_024_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"state_id\", \"confidence\": 1.0, \"issuer\": \"NEBRASKA\", \"holder\": \"HEIDI SAMPLE\", \"page_marker\": \"\"}]}"}]} {"id": "state_id_fin_024_extract", "task": "extract", "class": "state_id", "doc": "state_id_fin_024.pdf", "images": ["images/train/state_id/state_id_fin_024_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('state_id', 'The document class.')\nfields = (['full_name', 'date_of_birth', 'address', 'issue_date', 'expiration_date', 'state'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name', 'date_of_birth': 'Date of Birth', 'address': 'Address', 'issue_date': 'Issued Date', 'expiration_date': 'Valid Until Date', 'state': 'Issuing state'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"HEIDI SAMPLE\"}, {\"field\": \"date_of_birth\", \"value\": \"2005-01-12\"}, {\"field\": \"address\", \"value\": \"123 MAIN STREET, ANYTOWN, NE 12345\"}, {\"field\": \"issue_date\", \"value\": \"2023-05-01\"}, {\"field\": \"expiration_date\", \"value\": \"2028-01-12\"}, {\"field\": \"state\", \"value\": \"NEBRASKA\"}]}"}]} {"id": "state_id_fin_025_p1_classify", "task": "classify", "class": "state_id", "doc": "state_id_fin_025.pdf", "images": ["images/train/state_id/state_id_fin_025_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"state_id\", \"confidence\": 1.0, \"issuer\": \"Arizona\", \"holder\": \"SAMPLE CITIZEN\", \"page_marker\": \"\"}]}"}]} {"id": "state_id_fin_025_extract", "task": "extract", "class": "state_id", "doc": "state_id_fin_025.pdf", "images": ["images/train/state_id/state_id_fin_025_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('state_id', 'The document class.')\nfields = (['full_name', 'date_of_birth', 'address', 'issue_date', 'expiration_date', 'state'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name', 'date_of_birth': 'Date of Birth', 'address': 'Address', 'issue_date': 'Issued Date', 'expiration_date': 'Valid Until Date', 'state': 'Issuing state'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"SAMPLE CITIZEN\"}, {\"field\": \"date_of_birth\", \"value\": \"1978-02-14\"}, {\"field\": \"address\", \"value\": \"1234 EXAMPLE ST PHOENIX, AZ 85007\"}, {\"field\": \"issue_date\", \"value\": \"2023-02-01\"}, {\"field\": \"expiration_date\", \"value\": \"2028-02-01\"}, {\"field\": \"state\", \"value\": \"Arizona\"}]}"}]} {"id": "state_id_fin_027_p1_classify", "task": "classify", "class": "state_id", "doc": "state_id_fin_027.pdf", "images": ["images/train/state_id/state_id_fin_027_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"state_id\", \"confidence\": 1.0, \"issuer\": \"Indiana\", \"holder\": \"SAMPLE JELANI\", \"page_marker\": \"\"}]}"}]} {"id": "state_id_fin_027_extract", "task": "extract", "class": "state_id", "doc": "state_id_fin_027.pdf", "images": ["images/train/state_id/state_id_fin_027_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('state_id', 'The document class.')\nfields = (['full_name', 'date_of_birth', 'address', 'issue_date', 'expiration_date', 'state'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name', 'date_of_birth': 'Date of Birth', 'address': 'Address', 'issue_date': 'Issued Date', 'expiration_date': 'Valid Until Date', 'state': 'Issuing state'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"SAMPLE JELANI\"}, {\"field\": \"date_of_birth\", \"value\": \"1967-11-17\"}, {\"field\": \"address\", \"value\": \"123 MAIN STREET, INDIANAPOLIS, IN 46204-0000\"}, {\"field\": \"issue_date\", \"value\": \"2017-11-17\"}, {\"field\": \"expiration_date\", \"value\": \"2023-11-17\"}, {\"field\": \"state\", \"value\": \"Indiana\"}]}"}]} {"id": "state_id_fin_029_p1_classify", "task": "classify", "class": "state_id", "doc": "state_id_fin_029.pdf", "images": ["images/train/state_id/state_id_fin_029_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"state_id\", \"confidence\": 1.0, \"issuer\": \"Republic of Turkey\", \"holder\": \"MELEK NUR T\\u00dcRKO\\u011eLU\", \"page_marker\": \"\"}]}"}]} {"id": "state_id_fin_029_extract", "task": "extract", "class": "state_id", "doc": "state_id_fin_029.pdf", "images": ["images/train/state_id/state_id_fin_029_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('state_id', 'The document class.')\nfields = (['full_name', 'date_of_birth', 'address', 'issue_date', 'expiration_date', 'state'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name', 'date_of_birth': 'Date of Birth', 'address': 'Address', 'issue_date': 'Issued Date', 'expiration_date': 'Valid Until Date', 'state': 'Issuing state'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"MELEK NUR T\\u00dcRKO\\u011eLU\"}, {\"field\": \"date_of_birth\", \"value\": \"1993-05-29\"}, {\"field\": \"address\", \"value\": \"\"}, {\"field\": \"issue_date\", \"value\": \"\"}, {\"field\": \"expiration_date\", \"value\": \"2024-07-27\"}, {\"field\": \"state\", \"value\": \"Republic of Turkey\"}]}"}]} {"id": "state_id_fin_030_p1_classify", "task": "classify", "class": "state_id", "doc": "state_id_fin_030.pdf", "images": ["images/train/state_id/state_id_fin_030_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"state_id\", \"confidence\": 1.0, \"issuer\": \"Wisconsin\", \"holder\": \"SAMPLE DEBRA A\", \"page_marker\": \"\"}]}"}]} {"id": "state_id_fin_030_extract", "task": "extract", "class": "state_id", "doc": "state_id_fin_030.pdf", "images": ["images/train/state_id/state_id_fin_030_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('state_id', 'The document class.')\nfields = (['full_name', 'date_of_birth', 'address', 'issue_date', 'expiration_date', 'state'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name', 'date_of_birth': 'Date of Birth', 'address': 'Address', 'issue_date': 'Issued Date', 'expiration_date': 'Valid Until Date', 'state': 'Issuing state'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"SAMPLE DEBRA A\"}, {\"field\": \"date_of_birth\", \"value\": \"1949-06-24\"}, {\"field\": \"address\", \"value\": \"4802 SHEBOYGAN AVE, MADISON, WI 53705\"}, {\"field\": \"issue_date\", \"value\": \"2016-01-06\"}, {\"field\": \"expiration_date\", \"value\": \"\"}, {\"field\": \"state\", \"value\": \"Wisconsin\"}]}"}]} {"id": "state_id_fin_031_p1_classify", "task": "classify", "class": "state_id", "doc": "state_id_fin_031.pdf", "images": ["images/train/state_id/state_id_fin_031_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"state_id\", \"confidence\": 1.0, \"issuer\": \"OKLAHOMA\", \"holder\": \"SAMPLE NONDRIVER\", \"page_marker\": \"\"}]}"}]} {"id": "state_id_fin_031_extract", "task": "extract", "class": "state_id", "doc": "state_id_fin_031.pdf", "images": ["images/train/state_id/state_id_fin_031_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('state_id', 'The document class.')\nfields = (['full_name', 'date_of_birth', 'address', 'issue_date', 'expiration_date', 'state'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name', 'date_of_birth': 'Date of Birth', 'address': 'Address', 'issue_date': 'Issued Date', 'expiration_date': 'Valid Until Date', 'state': 'Issuing state'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"SAMPLE NONDRIVER\"}, {\"field\": \"date_of_birth\", \"value\": \"1951-10-15\"}, {\"field\": \"address\", \"value\": \"123 ANYSTREET, ANYCITY, OK 12345-0000\"}, {\"field\": \"issue_date\", \"value\": \"2026-12-03\"}, {\"field\": \"expiration_date\", \"value\": \"2030-10-31\"}, {\"field\": \"state\", \"value\": \"OKLAHOMA\"}]}"}]} {"id": "state_id_fin_032_p1_classify", "task": "classify", "class": "state_id", "doc": "state_id_fin_032.pdf", "images": ["images/train/state_id/state_id_fin_032_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"state_id\", \"confidence\": 1.0, \"issuer\": \"MISSOURI\", \"holder\": \"MALCOLM DUNN\", \"page_marker\": \"\"}]}"}]} {"id": "state_id_fin_032_extract", "task": "extract", "class": "state_id", "doc": "state_id_fin_032.pdf", "images": ["images/train/state_id/state_id_fin_032_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('state_id', 'The document class.')\nfields = (['full_name', 'date_of_birth', 'address', 'issue_date', 'expiration_date', 'state'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name', 'date_of_birth': 'Date of Birth', 'address': 'Address', 'issue_date': 'Issued Date', 'expiration_date': 'Valid Until Date', 'state': 'Issuing state'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"MALCOLM DUNN\"}, {\"field\": \"date_of_birth\", \"value\": \"2017-05-01\"}, {\"field\": \"address\", \"value\": \"123 YOUR STREET KANSAS CITY, MO 64116\"}, {\"field\": \"issue_date\", \"value\": \"2022-12-25\"}, {\"field\": \"expiration_date\", \"value\": \"2037-05-01\"}, {\"field\": \"state\", \"value\": \"MISSOURI\"}]}"}]} {"id": "state_id_fin_033_p1_classify", "task": "classify", "class": "state_id", "doc": "state_id_fin_033.pdf", "images": ["images/train/state_id/state_id_fin_033_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"state_id\", \"confidence\": 1.0, \"issuer\": \"\", \"holder\": \"JOSH WELLINGTON\", \"page_marker\": \"\"}]}"}]} {"id": "state_id_fin_033_extract", "task": "extract", "class": "state_id", "doc": "state_id_fin_033.pdf", "images": ["images/train/state_id/state_id_fin_033_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('state_id', 'The document class.')\nfields = (['full_name', 'date_of_birth', 'address', 'issue_date', 'expiration_date', 'state'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name', 'date_of_birth': 'Date of Birth', 'address': 'Address', 'issue_date': 'Issued Date', 'expiration_date': 'Valid Until Date', 'state': 'Issuing state'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"JOSH WELLINGTON\"}, {\"field\": \"date_of_birth\", \"value\": \"1988-05-10\"}, {\"field\": \"address\", \"value\": \"1944 BILMORE ST APT 5674, NEW YORK, NEW YORK 10005, UNITED STATES\"}, {\"field\": \"issue_date\", \"value\": \"2017-06-01\"}, {\"field\": \"expiration_date\", \"value\": \"2022-06-01\"}, {\"field\": \"state\", \"value\": \"\"}]}"}]} {"id": "state_id_fin_034_p1_classify", "task": "classify", "class": "state_id", "doc": "state_id_fin_034.pdf", "images": ["images/train/state_id/state_id_fin_034_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"state_id\", \"confidence\": 1.0, \"issuer\": \"ALABAMA\", \"holder\": \"BENJAMIN TAYLOR\", \"page_marker\": \"\"}]}"}]} {"id": "state_id_fin_034_extract", "task": "extract", "class": "state_id", "doc": "state_id_fin_034.pdf", "images": ["images/train/state_id/state_id_fin_034_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('state_id', 'The document class.')\nfields = (['full_name', 'date_of_birth', 'address', 'issue_date', 'expiration_date', 'state'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name', 'date_of_birth': 'Date of Birth', 'address': 'Address', 'issue_date': 'Issued Date', 'expiration_date': 'Valid Until Date', 'state': 'Issuing state'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"BENJAMIN TAYLOR\"}, {\"field\": \"date_of_birth\", \"value\": \"2017-04-24\"}, {\"field\": \"address\", \"value\": \"1234 YOUR STREET, AUBURN, AL 36832\"}, {\"field\": \"issue_date\", \"value\": \"2022-12-25\"}, {\"field\": \"expiration_date\", \"value\": \"2027-04-24\"}, {\"field\": \"state\", \"value\": \"ALABAMA\"}]}"}]} {"id": "state_id_fin_037_p1_classify", "task": "classify", "class": "state_id", "doc": "state_id_fin_037.pdf", "images": ["images/train/state_id/state_id_fin_037_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"state_id\", \"confidence\": 1.0, \"issuer\": \"Texas\", \"holder\": \"JOHNSON ROBERT III ESQ.\", \"page_marker\": \"\"}]}"}]} {"id": "state_id_fin_037_extract", "task": "extract", "class": "state_id", "doc": "state_id_fin_037.pdf", "images": ["images/train/state_id/state_id_fin_037_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('state_id', 'The document class.')\nfields = (['full_name', 'date_of_birth', 'address', 'issue_date', 'expiration_date', 'state'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name', 'date_of_birth': 'Date of Birth', 'address': 'Address', 'issue_date': 'Issued Date', 'expiration_date': 'Valid Until Date', 'state': 'Issuing state'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"JOHNSON ROBERT III ESQ.\"}, {\"field\": \"date_of_birth\", \"value\": \"1987-06-08\"}, {\"field\": \"address\", \"value\": \"100 FOREST HILLS AVE, CHILDRESS, TX 12345-0000\"}, {\"field\": \"issue_date\", \"value\": \"2017-03-06\"}, {\"field\": \"expiration_date\", \"value\": \"2022-03-06\"}, {\"field\": \"state\", \"value\": \"Texas\"}]}"}]} {"id": "state_id_fin_038_p1_classify", "task": "classify", "class": "state_id", "doc": "state_id_fin_038.pdf", "images": ["images/train/state_id/state_id_fin_038_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"state_id\", \"confidence\": 1.0, \"issuer\": \"WASHINGTON\", \"holder\": \"RAYMOND RUSHER EDL\", \"page_marker\": \"\"}]}"}]} {"id": "state_id_fin_038_extract", "task": "extract", "class": "state_id", "doc": "state_id_fin_038.pdf", "images": ["images/train/state_id/state_id_fin_038_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('state_id', 'The document class.')\nfields = (['full_name', 'date_of_birth', 'address', 'issue_date', 'expiration_date', 'state'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name', 'date_of_birth': 'Date of Birth', 'address': 'Address', 'issue_date': 'Issued Date', 'expiration_date': 'Valid Until Date', 'state': 'Issuing state'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"RAYMOND RUSHER EDL\"}, {\"field\": \"date_of_birth\", \"value\": \"1950-11-01\"}, {\"field\": \"address\", \"value\": \"9318 ALEXANDER DR NE, LACEY WA 98516-7113\"}, {\"field\": \"issue_date\", \"value\": \"2008-09-15\"}, {\"field\": \"expiration_date\", \"value\": \"2012-11-01\"}, {\"field\": \"state\", \"value\": \"WASHINGTON\"}]}"}]} {"id": "state_id_fin_039_p1_classify", "task": "classify", "class": "state_id", "doc": "state_id_fin_039.pdf", "images": ["images/train/state_id/state_id_fin_039_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"state_id\", \"confidence\": 1.0, \"issuer\": \"WASHINGTON\", \"holder\": \"TEST PERSON HUMAN\", \"page_marker\": \"\"}]}"}]} {"id": "state_id_fin_039_extract", "task": "extract", "class": "state_id", "doc": "state_id_fin_039.pdf", "images": ["images/train/state_id/state_id_fin_039_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('state_id', 'The document class.')\nfields = (['full_name', 'date_of_birth', 'address', 'issue_date', 'expiration_date', 'state'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name', 'date_of_birth': 'Date of Birth', 'address': 'Address', 'issue_date': 'Issued Date', 'expiration_date': 'Valid Until Date', 'state': 'Issuing state'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"TEST PERSON HUMAN\"}, {\"field\": \"date_of_birth\", \"value\": \"1989-08-15\"}, {\"field\": \"address\", \"value\": \"405 BLACK LAKE BLVD SW, OLYMPIA WA 98502-5046\"}, {\"field\": \"issue_date\", \"value\": \"2017-01-20\"}, {\"field\": \"expiration_date\", \"value\": \"2022-08-15\"}, {\"field\": \"state\", \"value\": \"WASHINGTON\"}]}"}]} {"id": "state_id_fin_040_p1_classify", "task": "classify", "class": "state_id", "doc": "state_id_fin_040.pdf", "images": ["images/train/state_id/state_id_fin_040_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"state_id\", \"confidence\": 1.0, \"issuer\": \"HUN\", \"holder\": \"SZ\\u00c9PEN\\u00c9 KISS ROZ\\u00c1LIA\", \"page_marker\": \"\"}]}"}]} {"id": "state_id_fin_040_extract", "task": "extract", "class": "state_id", "doc": "state_id_fin_040.pdf", "images": ["images/train/state_id/state_id_fin_040_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('state_id', 'The document class.')\nfields = (['full_name', 'date_of_birth', 'address', 'issue_date', 'expiration_date', 'state'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name', 'date_of_birth': 'Date of Birth', 'address': 'Address', 'issue_date': 'Issued Date', 'expiration_date': 'Valid Until Date', 'state': 'Issuing state'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"SZ\\u00c9PEN\\u00c9 KISS ROZ\\u00c1LIA\"}, {\"field\": \"date_of_birth\", \"value\": \"1979-06-30\"}, {\"field\": \"address\", \"value\": \"\"}, {\"field\": \"issue_date\", \"value\": \"\"}, {\"field\": \"expiration_date\", \"value\": \"2028-06-30\"}, {\"field\": \"state\", \"value\": \"HUN\"}]}"}]} {"id": "state_id_fin_041_p1_classify", "task": "classify", "class": "state_id", "doc": "state_id_fin_041.pdf", "images": ["images/train/state_id/state_id_fin_041_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"state_id\", \"confidence\": 1.0, \"issuer\": \"CALIFORNIA\", \"holder\": \"\", \"page_marker\": \"\"}]}"}]} {"id": "state_id_fin_041_extract", "task": "extract", "class": "state_id", "doc": "state_id_fin_041.pdf", "images": ["images/train/state_id/state_id_fin_041_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('state_id', 'The document class.')\nfields = (['full_name', 'date_of_birth', 'address', 'issue_date', 'expiration_date', 'state'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name', 'date_of_birth': 'Date of Birth', 'address': 'Address', 'issue_date': 'Issued Date', 'expiration_date': 'Valid Until Date', 'state': 'Issuing state'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"\"}, {\"field\": \"date_of_birth\", \"value\": \"\"}, {\"field\": \"address\", \"value\": \"\"}, {\"field\": \"issue_date\", \"value\": \"\"}, {\"field\": \"expiration_date\", \"value\": \"\"}, {\"field\": \"state\", \"value\": \"CALIFORNIA\"}]}"}]} {"id": "state_id_fin_042_p1_classify", "task": "classify", "class": "state_id", "doc": "state_id_fin_042.pdf", "images": ["images/train/state_id/state_id_fin_042_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"state_id\", \"confidence\": 1.0, \"issuer\": \"Colorado\", \"holder\": \"SAMPLE JANE IDENTIFICATIONCARD\", \"page_marker\": \"\"}]}"}]} {"id": "state_id_fin_042_extract", "task": "extract", "class": "state_id", "doc": "state_id_fin_042.pdf", "images": ["images/train/state_id/state_id_fin_042_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('state_id', 'The document class.')\nfields = (['full_name', 'date_of_birth', 'address', 'issue_date', 'expiration_date', 'state'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name', 'date_of_birth': 'Date of Birth', 'address': 'Address', 'issue_date': 'Issued Date', 'expiration_date': 'Valid Until Date', 'state': 'Issuing state'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"SAMPLE JANE IDENTIFICATIONCARD\"}, {\"field\": \"date_of_birth\", \"value\": \"2005-01-01\"}, {\"field\": \"address\", \"value\": \"1881 PIERCE ST LAKEWOOD, CO 80214\"}, {\"field\": \"issue_date\", \"value\": \"2020-03-04\"}, {\"field\": \"expiration_date\", \"value\": \"2026-01-01\"}, {\"field\": \"state\", \"value\": \"Colorado\"}]}"}]} {"id": "state_id_fin_044_p1_classify", "task": "classify", "class": "state_id", "doc": "state_id_fin_044.pdf", "images": ["images/train/state_id/state_id_fin_044_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"state_id\", \"confidence\": 1.0, \"issuer\": \"MAGYARORSZ\\u00c1G / HUNGARY\", \"holder\": \"SZEKER MIRKO ISTV\\u00c1N\", \"page_marker\": \"\"}]}"}]} {"id": "state_id_fin_044_extract", "task": "extract", "class": "state_id", "doc": "state_id_fin_044.pdf", "images": ["images/train/state_id/state_id_fin_044_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('state_id', 'The document class.')\nfields = (['full_name', 'date_of_birth', 'address', 'issue_date', 'expiration_date', 'state'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name', 'date_of_birth': 'Date of Birth', 'address': 'Address', 'issue_date': 'Issued Date', 'expiration_date': 'Valid Until Date', 'state': 'Issuing state'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"SZEKER MIRKO ISTV\\u00c1N\"}, {\"field\": \"date_of_birth\", \"value\": \"1945-10-16\"}, {\"field\": \"address\", \"value\": \"\"}, {\"field\": \"issue_date\", \"value\": \"\"}, {\"field\": \"expiration_date\", \"value\": \"2076-01-01\"}, {\"field\": \"state\", \"value\": \"MAGYARORSZ\\u00c1G / HUNGARY\"}]}"}]} {"id": "state_id_fin_045_p1_classify", "task": "classify", "class": "state_id", "doc": "state_id_fin_045.pdf", "images": ["images/train/state_id/state_id_fin_045_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"state_id\", \"confidence\": 1.0, \"issuer\": \"SOUTH CAROLINA\", \"holder\": \"SAMPLE WILLIAM\", \"page_marker\": \"\"}]}"}]} {"id": "state_id_fin_045_extract", "task": "extract", "class": "state_id", "doc": "state_id_fin_045.pdf", "images": ["images/train/state_id/state_id_fin_045_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('state_id', 'The document class.')\nfields = (['full_name', 'date_of_birth', 'address', 'issue_date', 'expiration_date', 'state'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name', 'date_of_birth': 'Date of Birth', 'address': 'Address', 'issue_date': 'Issued Date', 'expiration_date': 'Valid Until Date', 'state': 'Issuing state'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"SAMPLE WILLIAM\"}, {\"field\": \"date_of_birth\", \"value\": \"1980-01-01\"}, {\"field\": \"address\", \"value\": \"789 MAIN STREET, COLUMBIA, SC 29201\"}, {\"field\": \"issue_date\", \"value\": \"2025-08-27\"}, {\"field\": \"expiration_date\", \"value\": \"2033-08-27\"}, {\"field\": \"state\", \"value\": \"SOUTH CAROLINA\"}]}"}]} {"id": "state_id_fin_046_p1_classify", "task": "classify", "class": "state_id", "doc": "state_id_fin_046.pdf", "images": ["images/train/state_id/state_id_fin_046_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"state_id\", \"confidence\": 1.0, \"issuer\": \"Virginia\", \"holder\": \"SARKO JAMES ALAN\", \"page_marker\": \"\"}]}"}]} {"id": "state_id_fin_046_extract", "task": "extract", "class": "state_id", "doc": "state_id_fin_046.pdf", "images": ["images/train/state_id/state_id_fin_046_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('state_id', 'The document class.')\nfields = (['full_name', 'date_of_birth', 'address', 'issue_date', 'expiration_date', 'state'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name', 'date_of_birth': 'Date of Birth', 'address': 'Address', 'issue_date': 'Issued Date', 'expiration_date': 'Valid Until Date', 'state': 'Issuing state'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"SARKO JAMES ALAN\"}, {\"field\": \"date_of_birth\", \"value\": \"1950-03-23\"}, {\"field\": \"address\", \"value\": \"110 EAST MIDLAND TRL, LEXINGTON, VA 24450\"}, {\"field\": \"issue_date\", \"value\": \"2023-05-19\"}, {\"field\": \"expiration_date\", \"value\": \"2031-03-23\"}, {\"field\": \"state\", \"value\": \"Virginia\"}]}"}]} {"id": "state_id_fin_047_p1_classify", "task": "classify", "class": "state_id", "doc": "state_id_fin_047.pdf", "images": ["images/train/state_id/state_id_fin_047_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"state_id\", \"confidence\": 1.0, \"issuer\": \"New York\", \"holder\": \"NONDRIVER MICHELLE, MARIE\", \"page_marker\": \"\"}]}"}]} {"id": "state_id_fin_047_extract", "task": "extract", "class": "state_id", "doc": "state_id_fin_047.pdf", "images": ["images/train/state_id/state_id_fin_047_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('state_id', 'The document class.')\nfields = (['full_name', 'date_of_birth', 'address', 'issue_date', 'expiration_date', 'state'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name', 'date_of_birth': 'Date of Birth', 'address': 'Address', 'issue_date': 'Issued Date', 'expiration_date': 'Valid Until Date', 'state': 'Issuing state'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"NONDRIVER MICHELLE, MARIE\"}, {\"field\": \"date_of_birth\", \"value\": \"1993-10-31\"}, {\"field\": \"address\", \"value\": \"2345 ANYWHERE STREET YOUR CITY, NY 12345\"}, {\"field\": \"issue_date\", \"value\": \"2019-10-31\"}, {\"field\": \"expiration_date\", \"value\": \"2027-10-31\"}, {\"field\": \"state\", \"value\": \"New York\"}]}"}]} {"id": "state_id_fin_048_p1_classify", "task": "classify", "class": "state_id", "doc": "state_id_fin_048.pdf", "images": ["images/train/state_id/state_id_fin_048_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"state_id\", \"confidence\": 1.0, \"issuer\": \"Arizona\", \"holder\": \"SAMPLE JANICE\", \"page_marker\": \"\"}]}"}]} {"id": "state_id_fin_048_extract", "task": "extract", "class": "state_id", "doc": "state_id_fin_048.pdf", "images": ["images/train/state_id/state_id_fin_048_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('state_id', 'The document class.')\nfields = (['full_name', 'date_of_birth', 'address', 'issue_date', 'expiration_date', 'state'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name', 'date_of_birth': 'Date of Birth', 'address': 'Address', 'issue_date': 'Issued Date', 'expiration_date': 'Valid Until Date', 'state': 'Issuing state'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"SAMPLE JANICE\"}, {\"field\": \"date_of_birth\", \"value\": \"2008-01-01\"}, {\"field\": \"address\", \"value\": \"123 MAIN ST, PHOENIX, AZ 85007\"}, {\"field\": \"issue_date\", \"value\": \"2019-03-01\"}, {\"field\": \"expiration_date\", \"value\": \"\"}, {\"field\": \"state\", \"value\": \"Arizona\"}]}"}]} {"id": "state_id_fin_049_p1_classify", "task": "classify", "class": "state_id", "doc": "state_id_fin_049.pdf", "images": ["images/train/state_id/state_id_fin_049_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"state_id\", \"confidence\": 1.0, \"issuer\": \"Arizona\", \"holder\": \"SAMPLE JANICE\", \"page_marker\": \"\"}]}"}]} {"id": "state_id_fin_049_extract", "task": "extract", "class": "state_id", "doc": "state_id_fin_049.pdf", "images": ["images/train/state_id/state_id_fin_049_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('state_id', 'The document class.')\nfields = (['full_name', 'date_of_birth', 'address', 'issue_date', 'expiration_date', 'state'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name', 'date_of_birth': 'Date of Birth', 'address': 'Address', 'issue_date': 'Issued Date', 'expiration_date': 'Valid Until Date', 'state': 'Issuing state'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"SAMPLE JANICE\"}, {\"field\": \"date_of_birth\", \"value\": \"1998-01-01\"}, {\"field\": \"address\", \"value\": \"123 MAIN ST, PHOENIX, AZ 85007\"}, {\"field\": \"issue_date\", \"value\": \"2016-03-01\"}, {\"field\": \"expiration_date\", \"value\": \"\"}, {\"field\": \"state\", \"value\": \"Arizona\"}]}"}]} {"id": "state_id_fin_050_p1_classify", "task": "classify", "class": "state_id", "doc": "state_id_fin_050.pdf", "images": ["images/train/state_id/state_id_fin_050_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"state_id\", \"confidence\": 1.0, \"issuer\": \"WASHINGTON\", \"holder\": \"SAMPLE SUZY A\", \"page_marker\": \"\"}]}"}]} {"id": "state_id_fin_050_extract", "task": "extract", "class": "state_id", "doc": "state_id_fin_050.pdf", "images": ["images/train/state_id/state_id_fin_050_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('state_id', 'The document class.')\nfields = (['full_name', 'date_of_birth', 'address', 'issue_date', 'expiration_date', 'state'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name', 'date_of_birth': 'Date of Birth', 'address': 'Address', 'issue_date': 'Issued Date', 'expiration_date': 'Valid Until Date', 'state': 'Issuing state'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"SAMPLE SUZY A\"}, {\"field\": \"date_of_birth\", \"value\": \"1997-01-06\"}, {\"field\": \"address\", \"value\": \"123 STREET ADDRESS YOUR CITY WA 99999-1234\"}, {\"field\": \"issue_date\", \"value\": \"2015-08-12\"}, {\"field\": \"expiration_date\", \"value\": \"2020-08-12\"}, {\"field\": \"state\", \"value\": \"WASHINGTON\"}]}"}]} {"id": "state_id_fin_051_p1_classify", "task": "classify", "class": "state_id", "doc": "state_id_fin_051.pdf", "images": ["images/train/state_id/state_id_fin_051_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"state_id\", \"confidence\": 1.0, \"issuer\": \"Massachusetts\", \"holder\": \"SAMPLE ABIGAIL K.\", \"page_marker\": \"\"}]}"}]} {"id": "state_id_fin_051_extract", "task": "extract", "class": "state_id", "doc": "state_id_fin_051.pdf", "images": ["images/train/state_id/state_id_fin_051_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('state_id', 'The document class.')\nfields = (['full_name', 'date_of_birth', 'address', 'issue_date', 'expiration_date', 'state'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name', 'date_of_birth': 'Date of Birth', 'address': 'Address', 'issue_date': 'Issued Date', 'expiration_date': 'Valid Until Date', 'state': 'Issuing state'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"SAMPLE ABIGAIL K.\"}, {\"field\": \"date_of_birth\", \"value\": \"1989-10-22\"}, {\"field\": \"address\", \"value\": \"25 NEWPORT AVENUE, QUINCY, MA 02171-0000\"}, {\"field\": \"issue_date\", \"value\": \"2009-10-22\"}, {\"field\": \"expiration_date\", \"value\": \"2014-10-22\"}, {\"field\": \"state\", \"value\": \"Massachusetts\"}]}"}]} {"id": "state_id_fin_052_p1_classify", "task": "classify", "class": "state_id", "doc": "state_id_fin_052.pdf", "images": ["images/train/state_id/state_id_fin_052_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"state_id\", \"confidence\": 1.0, \"issuer\": \"Indiana\", \"holder\": \"BRENDA SAMPLE\", \"page_marker\": \"\"}]}"}]} {"id": "state_id_fin_052_extract", "task": "extract", "class": "state_id", "doc": "state_id_fin_052.pdf", "images": ["images/train/state_id/state_id_fin_052_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('state_id', 'The document class.')\nfields = (['full_name', 'date_of_birth', 'address', 'issue_date', 'expiration_date', 'state'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name', 'date_of_birth': 'Date of Birth', 'address': 'Address', 'issue_date': 'Issued Date', 'expiration_date': 'Valid Until Date', 'state': 'Issuing state'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"BRENDA SAMPLE\"}, {\"field\": \"date_of_birth\", \"value\": \"1970-11-17\"}, {\"field\": \"address\", \"value\": \"123 MAIN STREET, INDIANAPOLIS, IN 46204-0000\"}, {\"field\": \"issue_date\", \"value\": \"2017-11-17\"}, {\"field\": \"expiration_date\", \"value\": \"2023-11-17\"}, {\"field\": \"state\", \"value\": \"Indiana\"}]}"}]} {"id": "state_id_fin_053_p1_classify", "task": "classify", "class": "state_id", "doc": "state_id_fin_053.pdf", "images": ["images/train/state_id/state_id_fin_053_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"state_id\", \"confidence\": 1.0, \"issuer\": \"MICHIGAN\", \"holder\": \"JUNE MARIE SAMPLE\", \"page_marker\": \"\"}]}"}]} {"id": "state_id_fin_053_extract", "task": "extract", "class": "state_id", "doc": "state_id_fin_053.pdf", "images": ["images/train/state_id/state_id_fin_053_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('state_id', 'The document class.')\nfields = (['full_name', 'date_of_birth', 'address', 'issue_date', 'expiration_date', 'state'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name', 'date_of_birth': 'Date of Birth', 'address': 'Address', 'issue_date': 'Issued Date', 'expiration_date': 'Valid Until Date', 'state': 'Issuing state'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"JUNE MARIE SAMPLE\"}, {\"field\": \"date_of_birth\", \"value\": \"2006-04-14\"}, {\"field\": \"address\", \"value\": \"123 NORTH STATE ST. LANSING, MI 48918-0000\"}, {\"field\": \"issue_date\", \"value\": \"2014-05-01\"}, {\"field\": \"expiration_date\", \"value\": \"2018-04-14\"}, {\"field\": \"state\", \"value\": \"MICHIGAN\"}]}"}]} {"id": "state_id_fin_055_p1_classify", "task": "classify", "class": "state_id", "doc": "state_id_fin_055.pdf", "images": ["images/train/state_id/state_id_fin_055_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"state_id\", \"confidence\": 1.0, \"issuer\": \"New York\", \"holder\": \"NONDRIVER MICHAEL, MATTHEW\", \"page_marker\": \"\"}]}"}]} {"id": "state_id_fin_055_extract", "task": "extract", "class": "state_id", "doc": "state_id_fin_055.pdf", "images": ["images/train/state_id/state_id_fin_055_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('state_id', 'The document class.')\nfields = (['full_name', 'date_of_birth', 'address', 'issue_date', 'expiration_date', 'state'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name', 'date_of_birth': 'Date of Birth', 'address': 'Address', 'issue_date': 'Issued Date', 'expiration_date': 'Valid Until Date', 'state': 'Issuing state'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"NONDRIVER MICHAEL, MATTHEW\"}, {\"field\": \"date_of_birth\", \"value\": \"2003-10-31\"}, {\"field\": \"address\", \"value\": \"2345 ANYWHERE STREET, ALBANY, NY 12222\"}, {\"field\": \"issue_date\", \"value\": \"2022-03-07\"}, {\"field\": \"expiration_date\", \"value\": \"2026-10-31\"}, {\"field\": \"state\", \"value\": \"New York\"}]}"}]} {"id": "state_id_fin_056_p1_classify", "task": "classify", "class": "state_id", "doc": "state_id_fin_056.pdf", "images": ["images/train/state_id/state_id_fin_056_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"state_id\", \"confidence\": 1.0, \"issuer\": \"ALABAMA\", \"holder\": \"SUE WIYGUL MARTIN\", \"page_marker\": \"\"}]}"}]} {"id": "state_id_fin_056_extract", "task": "extract", "class": "state_id", "doc": "state_id_fin_056.pdf", "images": ["images/train/state_id/state_id_fin_056_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('state_id', 'The document class.')\nfields = (['full_name', 'date_of_birth', 'address', 'issue_date', 'expiration_date', 'state'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name', 'date_of_birth': 'Date of Birth', 'address': 'Address', 'issue_date': 'Issued Date', 'expiration_date': 'Valid Until Date', 'state': 'Issuing state'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"SUE WIYGUL MARTIN\"}, {\"field\": \"date_of_birth\", \"value\": \"\"}, {\"field\": \"address\", \"value\": \"\"}, {\"field\": \"issue_date\", \"value\": \"2013-09-27\"}, {\"field\": \"expiration_date\", \"value\": \"2013-10-27\"}, {\"field\": \"state\", \"value\": \"ALABAMA\"}]}"}]} {"id": "state_id_fin_057_p1_classify", "task": "classify", "class": "state_id", "doc": "state_id_fin_057.pdf", "images": ["images/train/state_id/state_id_fin_057_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"state_id\", \"confidence\": 1.0, \"issuer\": \"Ohio\", \"holder\": \"DOE JANE\", \"page_marker\": \"\"}]}"}]} {"id": "state_id_fin_057_extract", "task": "extract", "class": "state_id", "doc": "state_id_fin_057.pdf", "images": ["images/train/state_id/state_id_fin_057_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('state_id', 'The document class.')\nfields = (['full_name', 'date_of_birth', 'address', 'issue_date', 'expiration_date', 'state'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name', 'date_of_birth': 'Date of Birth', 'address': 'Address', 'issue_date': 'Issued Date', 'expiration_date': 'Valid Until Date', 'state': 'Issuing state'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"DOE JANE\"}, {\"field\": \"date_of_birth\", \"value\": \"1990-09-16\"}, {\"field\": \"address\", \"value\": \"123 ANYWHERE ST, ANYTOWN, OH 12345\"}, {\"field\": \"issue_date\", \"value\": \"2018-07-02\"}, {\"field\": \"expiration_date\", \"value\": \"2025-01-21\"}, {\"field\": \"state\", \"value\": \"Ohio\"}]}"}]} {"id": "state_id_fin_058_p1_classify", "task": "classify", "class": "state_id", "doc": "state_id_fin_058.pdf", "images": ["images/train/state_id/state_id_fin_058_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"state_id\", \"confidence\": 1.0, \"issuer\": \"West Virginia\", \"holder\": \"SAMPLE ANDREW\", \"page_marker\": \"\"}]}"}]} {"id": "state_id_fin_058_extract", "task": "extract", "class": "state_id", "doc": "state_id_fin_058.pdf", "images": ["images/train/state_id/state_id_fin_058_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('state_id', 'The document class.')\nfields = (['full_name', 'date_of_birth', 'address', 'issue_date', 'expiration_date', 'state'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name', 'date_of_birth': 'Date of Birth', 'address': 'Address', 'issue_date': 'Issued Date', 'expiration_date': 'Valid Until Date', 'state': 'Issuing state'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"SAMPLE ANDREW\"}, {\"field\": \"date_of_birth\", \"value\": \"1958-05-30\"}, {\"field\": \"address\", \"value\": \"123 CARDINAL STREET, SCENIC VIEW, WV 29999\"}, {\"field\": \"issue_date\", \"value\": \"2023-04-20\"}, {\"field\": \"expiration_date\", \"value\": \"2028-05-30\"}, {\"field\": \"state\", \"value\": \"West Virginia\"}]}"}]} {"id": "state_id_fin_059_p1_classify", "task": "classify", "class": "state_id", "doc": "state_id_fin_059.pdf", "images": ["images/train/state_id/state_id_fin_059_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"state_id\", \"confidence\": 1.0, \"issuer\": \"Ohio\", \"holder\": \"JONES ALICIA GRABLE\", \"page_marker\": \"\"}]}"}]} {"id": "state_id_fin_059_extract", "task": "extract", "class": "state_id", "doc": "state_id_fin_059.pdf", "images": ["images/train/state_id/state_id_fin_059_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('state_id', 'The document class.')\nfields = (['full_name', 'date_of_birth', 'address', 'issue_date', 'expiration_date', 'state'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name', 'date_of_birth': 'Date of Birth', 'address': 'Address', 'issue_date': 'Issued Date', 'expiration_date': 'Valid Until Date', 'state': 'Issuing state'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"JONES ALICIA GRABLE\"}, {\"field\": \"date_of_birth\", \"value\": \"1990-02-04\"}, {\"field\": \"address\", \"value\": \"70 W BROAD ST COLUMBUS, OH 43215-1234\"}, {\"field\": \"issue_date\", \"value\": \"2009-02-03\"}, {\"field\": \"expiration_date\", \"value\": \"2017-02-03\"}, {\"field\": \"state\", \"value\": \"Ohio\"}]}"}]} {"id": "state_id_fin_060_p1_classify", "task": "classify", "class": "state_id", "doc": "state_id_fin_060.pdf", "images": ["images/train/state_id/state_id_fin_060_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"state_id\", \"confidence\": 1.0, \"issuer\": \"North Dakota\", \"holder\": \"SAMPLE MARK\", \"page_marker\": \"\"}]}"}]} {"id": "state_id_fin_060_extract", "task": "extract", "class": "state_id", "doc": "state_id_fin_060.pdf", "images": ["images/train/state_id/state_id_fin_060_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('state_id', 'The document class.')\nfields = (['full_name', 'date_of_birth', 'address', 'issue_date', 'expiration_date', 'state'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name', 'date_of_birth': 'Date of Birth', 'address': 'Address', 'issue_date': 'Issued Date', 'expiration_date': 'Valid Until Date', 'state': 'Issuing state'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"SAMPLE MARK\"}, {\"field\": \"date_of_birth\", \"value\": \"1973-01-07\"}, {\"field\": \"address\", \"value\": \"123 MAIN STREET, BISMARCK, ND 58505-0700\"}, {\"field\": \"issue_date\", \"value\": \"2023-06-07\"}, {\"field\": \"expiration_date\", \"value\": \"2031-01-07\"}, {\"field\": \"state\", \"value\": \"North Dakota\"}]}"}]} {"id": "state_id_fin_063_p1_classify", "task": "classify", "class": "state_id", "doc": "state_id_fin_063.pdf", "images": ["images/train/state_id/state_id_fin_063_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"state_id\", \"confidence\": 1.0, \"issuer\": \"Virginia\", \"holder\": \"ANDREWS TAYLOR LYNN\", \"page_marker\": \"\"}]}"}]} {"id": "state_id_fin_063_extract", "task": "extract", "class": "state_id", "doc": "state_id_fin_063.pdf", "images": ["images/train/state_id/state_id_fin_063_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('state_id', 'The document class.')\nfields = (['full_name', 'date_of_birth', 'address', 'issue_date', 'expiration_date', 'state'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name', 'date_of_birth': 'Date of Birth', 'address': 'Address', 'issue_date': 'Issued Date', 'expiration_date': 'Valid Until Date', 'state': 'Issuing state'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"ANDREWS TAYLOR LYNN\"}, {\"field\": \"date_of_birth\", \"value\": \"2012-01-08\"}, {\"field\": \"address\", \"value\": \"206 SOUTH BRUNSWICK AVE, SOUTH HILL, VA 23970\"}, {\"field\": \"issue_date\", \"value\": \"2023-05-11\"}, {\"field\": \"expiration_date\", \"value\": \"2027-01-08\"}, {\"field\": \"state\", \"value\": \"Virginia\"}]}"}]} {"id": "state_id_fin_065_p1_classify", "task": "classify", "class": "state_id", "doc": "state_id_fin_065.pdf", "images": ["images/train/state_id/state_id_fin_065_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"state_id\", \"confidence\": 1.0, \"issuer\": \"ALASKA\", \"holder\": \"MEGAN ADAMS\", \"page_marker\": \"\"}]}"}]} {"id": "state_id_fin_065_extract", "task": "extract", "class": "state_id", "doc": "state_id_fin_065.pdf", "images": ["images/train/state_id/state_id_fin_065_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('state_id', 'The document class.')\nfields = (['full_name', 'date_of_birth', 'address', 'issue_date', 'expiration_date', 'state'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name', 'date_of_birth': 'Date of Birth', 'address': 'Address', 'issue_date': 'Issued Date', 'expiration_date': 'Valid Until Date', 'state': 'Issuing state'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"MEGAN ADAMS\"}, {\"field\": \"date_of_birth\", \"value\": \"2014-09-19\"}, {\"field\": \"address\", \"value\": \"1234 YOUR STREET, FAIRBANKS, AK 99708\"}, {\"field\": \"issue_date\", \"value\": \"2022-12-25\"}, {\"field\": \"expiration_date\", \"value\": \"2034-09-19\"}, {\"field\": \"state\", \"value\": \"ALASKA\"}]}"}]} {"id": "state_id_fin_066_p1_classify", "task": "classify", "class": "state_id", "doc": "state_id_fin_066.pdf", "images": ["images/train/state_id/state_id_fin_066_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"state_id\", \"confidence\": 1.0, \"issuer\": \"ALASKA\", \"holder\": \"SAMPLE AM D\", \"page_marker\": \"\"}]}"}]} {"id": "state_id_fin_066_extract", "task": "extract", "class": "state_id", "doc": "state_id_fin_066.pdf", "images": ["images/train/state_id/state_id_fin_066_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('state_id', 'The document class.')\nfields = (['full_name', 'date_of_birth', 'address', 'issue_date', 'expiration_date', 'state'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name', 'date_of_birth': 'Date of Birth', 'address': 'Address', 'issue_date': 'Issued Date', 'expiration_date': 'Valid Until Date', 'state': 'Issuing state'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"SAMPLE AM D\"}, {\"field\": \"date_of_birth\", \"value\": \"1980-10-03\"}, {\"field\": \"address\", \"value\": \"12345 TEST SAMPLE, JUNEAU, AK 99801\"}, {\"field\": \"issue_date\", \"value\": \"2018-08-03\"}, {\"field\": \"expiration_date\", \"value\": \"2022-03-13\"}, {\"field\": \"state\", \"value\": \"ALASKA\"}]}"}]} {"id": "state_id_fin_067_p1_classify", "task": "classify", "class": "state_id", "doc": "state_id_fin_067.pdf", "images": ["images/train/state_id/state_id_fin_067_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"state_id\", \"confidence\": 1.0, \"issuer\": \"ALASKA\", \"holder\": \"MISTER TESTER\", \"page_marker\": \"\"}]}"}]} {"id": "state_id_fin_067_extract", "task": "extract", "class": "state_id", "doc": "state_id_fin_067.pdf", "images": ["images/train/state_id/state_id_fin_067_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('state_id', 'The document class.')\nfields = (['full_name', 'date_of_birth', 'address', 'issue_date', 'expiration_date', 'state'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name', 'date_of_birth': 'Date of Birth', 'address': 'Address', 'issue_date': 'Issued Date', 'expiration_date': 'Valid Until Date', 'state': 'Issuing state'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"MISTER TESTER\"}, {\"field\": \"date_of_birth\", \"value\": \"2001-01-01\"}, {\"field\": \"address\", \"value\": \"123 SAMPLE STREET, FORT RICHARDSON, AK 99505\"}, {\"field\": \"issue_date\", \"value\": \"2018-08-03\"}, {\"field\": \"expiration_date\", \"value\": \"2019-01-01\"}, {\"field\": \"state\", \"value\": \"ALASKA\"}]}"}]} {"id": "state_id_fin_068_p1_classify", "task": "classify", "class": "state_id", "doc": "state_id_fin_068.pdf", "images": ["images/train/state_id/state_id_fin_068_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"state_id\", \"confidence\": 1.0, \"issuer\": \"WASHINGTON\", \"holder\": \"SAMPLE JOHN A\", \"page_marker\": \"\"}]}"}]} {"id": "state_id_fin_068_extract", "task": "extract", "class": "state_id", "doc": "state_id_fin_068.pdf", "images": ["images/train/state_id/state_id_fin_068_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('state_id', 'The document class.')\nfields = (['full_name', 'date_of_birth', 'address', 'issue_date', 'expiration_date', 'state'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name', 'date_of_birth': 'Date of Birth', 'address': 'Address', 'issue_date': 'Issued Date', 'expiration_date': 'Valid Until Date', 'state': 'Issuing state'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"SAMPLE JOHN A\"}, {\"field\": \"date_of_birth\", \"value\": \"1960-01-06\"}, {\"field\": \"address\", \"value\": \"2424 BRISTOL CT SW OLYMPIA WA 98502-6004\"}, {\"field\": \"issue_date\", \"value\": \"2026-08-15\"}, {\"field\": \"expiration_date\", \"value\": \"2034-01-06\"}, {\"field\": \"state\", \"value\": \"WASHINGTON\"}]}"}]} {"id": "state_id_fin_070_p1_classify", "task": "classify", "class": "state_id", "doc": "state_id_fin_070.pdf", "images": ["images/train/state_id/state_id_fin_070_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"state_id\", \"confidence\": 1.0, \"issuer\": \"Florida\", \"holder\": \"DRIVER FLORIDA SAMPLE\", \"page_marker\": \"\"}]}"}]} {"id": "state_id_fin_070_extract", "task": "extract", "class": "state_id", "doc": "state_id_fin_070.pdf", "images": ["images/train/state_id/state_id_fin_070_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('state_id', 'The document class.')\nfields = (['full_name', 'date_of_birth', 'address', 'issue_date', 'expiration_date', 'state'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name', 'date_of_birth': 'Date of Birth', 'address': 'Address', 'issue_date': 'Issued Date', 'expiration_date': 'Valid Until Date', 'state': 'Issuing state'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"DRIVER FLORIDA SAMPLE\"}, {\"field\": \"date_of_birth\", \"value\": \"1987-09-28\"}, {\"field\": \"address\", \"value\": \"2900 APALACHEE PARKWAY TALLAHASSEE, FL 32399\"}, {\"field\": \"issue_date\", \"value\": \"2020-03-15\"}, {\"field\": \"expiration_date\", \"value\": \"2028-09-28\"}, {\"field\": \"state\", \"value\": \"Florida\"}]}"}]} {"id": "state_id_fin_071_p1_classify", "task": "classify", "class": "state_id", "doc": "state_id_fin_071.pdf", "images": ["images/train/state_id/state_id_fin_071_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"state_id\", \"confidence\": 1.0, \"issuer\": \"Florida\", \"holder\": \"CARLOS DIAZ\", \"page_marker\": \"\"}]}"}]} {"id": "state_id_fin_071_extract", "task": "extract", "class": "state_id", "doc": "state_id_fin_071.pdf", "images": ["images/train/state_id/state_id_fin_071_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('state_id', 'The document class.')\nfields = (['full_name', 'date_of_birth', 'address', 'issue_date', 'expiration_date', 'state'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name', 'date_of_birth': 'Date of Birth', 'address': 'Address', 'issue_date': 'Issued Date', 'expiration_date': 'Valid Until Date', 'state': 'Issuing state'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"CARLOS DIAZ\"}, {\"field\": \"date_of_birth\", \"value\": \"2012-03-20\"}, {\"field\": \"address\", \"value\": \"123 YOUR STREET, MIAMI, FL 33132\"}, {\"field\": \"issue_date\", \"value\": \"2023-09-01\"}, {\"field\": \"expiration_date\", \"value\": \"2028-03-20\"}, {\"field\": \"state\", \"value\": \"Florida\"}]}"}]} {"id": "state_id_fin_073_p1_classify", "task": "classify", "class": "state_id", "doc": "state_id_fin_073.pdf", "images": ["images/train/state_id/state_id_fin_073_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"state_id\", \"confidence\": 1.0, \"issuer\": \"Ohio\", \"holder\": \"SMITH EMMA\", \"page_marker\": \"\"}]}"}]} {"id": "state_id_fin_073_extract", "task": "extract", "class": "state_id", "doc": "state_id_fin_073.pdf", "images": ["images/train/state_id/state_id_fin_073_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('state_id', 'The document class.')\nfields = (['full_name', 'date_of_birth', 'address', 'issue_date', 'expiration_date', 'state'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name', 'date_of_birth': 'Date of Birth', 'address': 'Address', 'issue_date': 'Issued Date', 'expiration_date': 'Valid Until Date', 'state': 'Issuing state'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"SMITH EMMA\"}, {\"field\": \"date_of_birth\", \"value\": \"1990-09-16\"}, {\"field\": \"address\", \"value\": \"123 ANYWHERE STREET ANYTOWN, OH 12345\"}, {\"field\": \"issue_date\", \"value\": \"2018-07-02\"}, {\"field\": \"expiration_date\", \"value\": \"2021-01-21\"}, {\"field\": \"state\", \"value\": \"Ohio\"}]}"}]} {"id": "state_id_fin_074_p1_classify", "task": "classify", "class": "state_id", "doc": "state_id_fin_074.pdf", "images": ["images/train/state_id/state_id_fin_074_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"state_id\", \"confidence\": 1.0, \"issuer\": \"SOUTH CAROLINA\", \"holder\": \"MARY SAMPLE\", \"page_marker\": \"\"}]}"}]} {"id": "state_id_fin_074_extract", "task": "extract", "class": "state_id", "doc": "state_id_fin_074.pdf", "images": ["images/train/state_id/state_id_fin_074_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('state_id', 'The document class.')\nfields = (['full_name', 'date_of_birth', 'address', 'issue_date', 'expiration_date', 'state'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name', 'date_of_birth': 'Date of Birth', 'address': 'Address', 'issue_date': 'Issued Date', 'expiration_date': 'Valid Until Date', 'state': 'Issuing state'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"MARY SAMPLE\"}, {\"field\": \"date_of_birth\", \"value\": \"2010-05-01\"}, {\"field\": \"address\", \"value\": \"456 MAIN STREET, COLUMBIA, SC 29201\"}, {\"field\": \"issue_date\", \"value\": \"2025-09-01\"}, {\"field\": \"expiration_date\", \"value\": \"2033-09-01\"}, {\"field\": \"state\", \"value\": \"SOUTH CAROLINA\"}]}"}]} {"id": "state_id_fin_075_p1_classify", "task": "classify", "class": "state_id", "doc": "state_id_fin_075.pdf", "images": ["images/train/state_id/state_id_fin_075_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"state_id\", \"confidence\": 1.0, \"issuer\": \"Texas\", \"holder\": \"JOHN SAMPLE\", \"page_marker\": \"\"}]}"}]} {"id": "state_id_fin_075_extract", "task": "extract", "class": "state_id", "doc": "state_id_fin_075.pdf", "images": ["images/train/state_id/state_id_fin_075_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('state_id', 'The document class.')\nfields = (['full_name', 'date_of_birth', 'address', 'issue_date', 'expiration_date', 'state'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name', 'date_of_birth': 'Date of Birth', 'address': 'Address', 'issue_date': 'Issued Date', 'expiration_date': 'Valid Until Date', 'state': 'Issuing state'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"JOHN SAMPLE\"}, {\"field\": \"date_of_birth\", \"value\": \"2007-01-01\"}, {\"field\": \"address\", \"value\": \"123 SAMPLE STREET, SAMPLE CITY, TX 12345\"}, {\"field\": \"issue_date\", \"value\": \"2025-01-04\"}, {\"field\": \"expiration_date\", \"value\": \"2028-01-01\"}, {\"field\": \"state\", \"value\": \"Texas\"}]}"}]} {"id": "state_id_fin_076_p1_classify", "task": "classify", "class": "state_id", "doc": "state_id_fin_076.pdf", "images": ["images/train/state_id/state_id_fin_076_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"state_id\", \"confidence\": 1.0, \"issuer\": \"CALIFORNIA\", \"holder\": \"CYD CHARISSE MARTIN\", \"page_marker\": \"\"}]}"}]} {"id": "state_id_fin_076_extract", "task": "extract", "class": "state_id", "doc": "state_id_fin_076.pdf", "images": ["images/train/state_id/state_id_fin_076_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('state_id', 'The document class.')\nfields = (['full_name', 'date_of_birth', 'address', 'issue_date', 'expiration_date', 'state'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name', 'date_of_birth': 'Date of Birth', 'address': 'Address', 'issue_date': 'Issued Date', 'expiration_date': 'Valid Until Date', 'state': 'Issuing state'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"CYD CHARISSE MARTIN\"}, {\"field\": \"date_of_birth\", \"value\": \"1922-03-08\"}, {\"field\": \"address\", \"value\": \"10724 WILSHIRE BLVD STE 1406, LOS ANGELES CA 90024\"}, {\"field\": \"issue_date\", \"value\": \"2007-09-28\"}, {\"field\": \"expiration_date\", \"value\": \"2017-03-08\"}, {\"field\": \"state\", \"value\": \"CALIFORNIA\"}]}"}]} {"id": "state_id_fin_078_p1_classify", "task": "classify", "class": "state_id", "doc": "state_id_fin_078.pdf", "images": ["images/train/state_id/state_id_fin_078_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"state_id\", \"confidence\": 1.0, \"issuer\": \"California\", \"holder\": \"MARY RUTH MELLI\", \"page_marker\": \"\"}]}"}]} {"id": "state_id_fin_078_extract", "task": "extract", "class": "state_id", "doc": "state_id_fin_078.pdf", "images": ["images/train/state_id/state_id_fin_078_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('state_id', 'The document class.')\nfields = (['full_name', 'date_of_birth', 'address', 'issue_date', 'expiration_date', 'state'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name', 'date_of_birth': 'Date of Birth', 'address': 'Address', 'issue_date': 'Issued Date', 'expiration_date': 'Valid Until Date', 'state': 'Issuing state'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"MARY RUTH MELLI\"}, {\"field\": \"date_of_birth\", \"value\": \"1926-02-16\"}, {\"field\": \"address\", \"value\": \"245 W BOBIER DR SP 43, VISTA CA 92083\"}, {\"field\": \"issue_date\", \"value\": \"1996-08-20\"}, {\"field\": \"expiration_date\", \"value\": \"2006-02-16\"}, {\"field\": \"state\", \"value\": \"California\"}]}"}]} {"id": "state_id_fin_079_p1_classify", "task": "classify", "class": "state_id", "doc": "state_id_fin_079.pdf", "images": ["images/train/state_id/state_id_fin_079_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"state_id\", \"confidence\": 1.0, \"issuer\": \"ALABAMA\", \"holder\": \"BENJAMIN TAYLOR\", \"page_marker\": \"\"}]}"}]} {"id": "state_id_fin_079_extract", "task": "extract", "class": "state_id", "doc": "state_id_fin_079.pdf", "images": ["images/train/state_id/state_id_fin_079_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('state_id', 'The document class.')\nfields = (['full_name', 'date_of_birth', 'address', 'issue_date', 'expiration_date', 'state'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name', 'date_of_birth': 'Date of Birth', 'address': 'Address', 'issue_date': 'Issued Date', 'expiration_date': 'Valid Until Date', 'state': 'Issuing state'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"BENJAMIN TAYLOR\"}, {\"field\": \"date_of_birth\", \"value\": \"2017-04-24\"}, {\"field\": \"address\", \"value\": \"1234 YOUR STREET, AUBURN, AL 36832\"}, {\"field\": \"issue_date\", \"value\": \"2022-12-25\"}, {\"field\": \"expiration_date\", \"value\": \"2027-04-24\"}, {\"field\": \"state\", \"value\": \"ALABAMA\"}]}"}]} {"id": "state_id_fin_080_p1_classify", "task": "classify", "class": "state_id", "doc": "state_id_fin_080.pdf", "images": ["images/train/state_id/state_id_fin_080_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"state_id\", \"confidence\": 1.0, \"issuer\": \"CALIFORNIA\", \"holder\": \"MOTEN LAWRENCE PAUL\", \"page_marker\": \"\"}]}"}]} {"id": "state_id_fin_080_extract", "task": "extract", "class": "state_id", "doc": "state_id_fin_080.pdf", "images": ["images/train/state_id/state_id_fin_080_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('state_id', 'The document class.')\nfields = (['full_name', 'date_of_birth', 'address', 'issue_date', 'expiration_date', 'state'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name', 'date_of_birth': 'Date of Birth', 'address': 'Address', 'issue_date': 'Issued Date', 'expiration_date': 'Valid Until Date', 'state': 'Issuing state'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"MOTEN LAWRENCE PAUL\"}, {\"field\": \"date_of_birth\", \"value\": \"1949-01-02\"}, {\"field\": \"address\", \"value\": \"150 W COLORADO BLVD 105 MONROVIA, CA 91016\"}, {\"field\": \"issue_date\", \"value\": \"2011-05-23\"}, {\"field\": \"expiration_date\", \"value\": \"2021-01-02\"}, {\"field\": \"state\", \"value\": \"CALIFORNIA\"}]}"}]} {"id": "state_id_fin_082_p1_classify", "task": "classify", "class": "state_id", "doc": "state_id_fin_082.pdf", "images": ["images/train/state_id/state_id_fin_082_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"state_id\", \"confidence\": 1.0, \"issuer\": \"Ohio\", \"holder\": \"PUBLIC STEVEN Q\", \"page_marker\": \"\"}]}"}]} {"id": "state_id_fin_082_extract", "task": "extract", "class": "state_id", "doc": "state_id_fin_082.pdf", "images": ["images/train/state_id/state_id_fin_082_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('state_id', 'The document class.')\nfields = (['full_name', 'date_of_birth', 'address', 'issue_date', 'expiration_date', 'state'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name', 'date_of_birth': 'Date of Birth', 'address': 'Address', 'issue_date': 'Issued Date', 'expiration_date': 'Valid Until Date', 'state': 'Issuing state'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"PUBLIC STEVEN Q\"}, {\"field\": \"date_of_birth\", \"value\": \"2011-01-01\"}, {\"field\": \"address\", \"value\": \"123 ANYWHERE STREET COLUMBUS, OH 43223\"}, {\"field\": \"issue_date\", \"value\": \"2018-07-05\"}, {\"field\": \"expiration_date\", \"value\": \"2022-01-01\"}, {\"field\": \"state\", \"value\": \"Ohio\"}]}"}]} {"id": "state_id_fin_083_p1_classify", "task": "classify", "class": "state_id", "doc": "state_id_fin_083.pdf", "images": ["images/train/state_id/state_id_fin_083_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"state_id\", \"confidence\": 1.0, \"issuer\": \"South Carolina\", \"holder\": \"SAMPLE, REGULAR ID\", \"page_marker\": \"\"}]}"}]} {"id": "state_id_fin_083_extract", "task": "extract", "class": "state_id", "doc": "state_id_fin_083.pdf", "images": ["images/train/state_id/state_id_fin_083_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('state_id', 'The document class.')\nfields = (['full_name', 'date_of_birth', 'address', 'issue_date', 'expiration_date', 'state'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name', 'date_of_birth': 'Date of Birth', 'address': 'Address', 'issue_date': 'Issued Date', 'expiration_date': 'Valid Until Date', 'state': 'Issuing state'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"SAMPLE, REGULAR ID\"}, {\"field\": \"date_of_birth\", \"value\": \"1960-05-10\"}, {\"field\": \"address\", \"value\": \"110 MAPLE STREET, ANYTOWN, SC 292221111\"}, {\"field\": \"issue_date\", \"value\": \"2006-04-27\"}, {\"field\": \"expiration_date\", \"value\": \"2016-05-10\"}, {\"field\": \"state\", \"value\": \"South Carolina\"}]}"}]} {"id": "state_id_fin_084_p1_classify", "task": "classify", "class": "state_id", "doc": "state_id_fin_084.pdf", "images": ["images/train/state_id/state_id_fin_084_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"state_id\", \"confidence\": 1.0, \"issuer\": \"South Carolina\", \"holder\": \"SAMPLE, IDENTIFICATION\", \"page_marker\": \"\"}]}"}]} {"id": "state_id_fin_084_extract", "task": "extract", "class": "state_id", "doc": "state_id_fin_084.pdf", "images": ["images/train/state_id/state_id_fin_084_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('state_id', 'The document class.')\nfields = (['full_name', 'date_of_birth', 'address', 'issue_date', 'expiration_date', 'state'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name', 'date_of_birth': 'Date of Birth', 'address': 'Address', 'issue_date': 'Issued Date', 'expiration_date': 'Valid Until Date', 'state': 'Issuing state'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"SAMPLE, IDENTIFICATION\"}, {\"field\": \"date_of_birth\", \"value\": \"1960-05-10\"}, {\"field\": \"address\", \"value\": \"110 MAPLE STREET, ANYTOWN, SC 292221111\"}, {\"field\": \"issue_date\", \"value\": \"2008-04-27\"}, {\"field\": \"expiration_date\", \"value\": \"2013-05-10\"}, {\"field\": \"state\", \"value\": \"South Carolina\"}]}"}]} {"id": "state_id_fin_086_p1_classify", "task": "classify", "class": "state_id", "doc": "state_id_fin_086.pdf", "images": ["images/train/state_id/state_id_fin_086_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"state_id\", \"confidence\": 1.0, \"issuer\": \"Virginia\", \"holder\": \"RIVERS BRIDGETT, NICOLE\", \"page_marker\": \"\"}]}"}]} {"id": "state_id_fin_086_extract", "task": "extract", "class": "state_id", "doc": "state_id_fin_086.pdf", "images": ["images/train/state_id/state_id_fin_086_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('state_id', 'The document class.')\nfields = (['full_name', 'date_of_birth', 'address', 'issue_date', 'expiration_date', 'state'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name', 'date_of_birth': 'Date of Birth', 'address': 'Address', 'issue_date': 'Issued Date', 'expiration_date': 'Valid Until Date', 'state': 'Issuing state'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"RIVERS BRIDGETT, NICOLE\"}, {\"field\": \"date_of_birth\", \"value\": \"1987-06-12\"}, {\"field\": \"address\", \"value\": \"2300 WEST BROAD STREET, RICHMOND, VA 23220\"}, {\"field\": \"issue_date\", \"value\": \"2010-05-14\"}, {\"field\": \"expiration_date\", \"value\": \"2018-06-30\"}, {\"field\": \"state\", \"value\": \"Virginia\"}]}"}]} {"id": "state_id_fin_087_p1_classify", "task": "classify", "class": "state_id", "doc": "state_id_fin_087.pdf", "images": ["images/train/state_id/state_id_fin_087_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"state_id\", \"confidence\": 1.0, \"issuer\": \"WISCONSIN\", \"holder\": \"JANE QUINCY SAMPLE WISCONSIN DRIVER\", \"page_marker\": \"\"}]}"}]} {"id": "state_id_fin_087_extract", "task": "extract", "class": "state_id", "doc": "state_id_fin_087.pdf", "images": ["images/train/state_id/state_id_fin_087_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('state_id', 'The document class.')\nfields = (['full_name', 'date_of_birth', 'address', 'issue_date', 'expiration_date', 'state'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name', 'date_of_birth': 'Date of Birth', 'address': 'Address', 'issue_date': 'Issued Date', 'expiration_date': 'Valid Until Date', 'state': 'Issuing state'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"JANE QUINCY SAMPLE WISCONSIN DRIVER\"}, {\"field\": \"date_of_birth\", \"value\": \"1970-12-28\"}, {\"field\": \"address\", \"value\": \"12345 NORTH LAKE WISCONSIN PARKWAY SUITE 266 BLACK RIVER FALLS WI 54615-0000\"}, {\"field\": \"issue_date\", \"value\": \"2015-12-28\"}, {\"field\": \"expiration_date\", \"value\": \"2020-12-28\"}, {\"field\": \"state\", \"value\": \"WISCONSIN\"}]}"}]} {"id": "state_id_fin_089_p1_classify", "task": "classify", "class": "state_id", "doc": "state_id_fin_089.pdf", "images": ["images/train/state_id/state_id_fin_089_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"state_id\", \"confidence\": 1.0, \"issuer\": \"Vermont\", \"holder\": \"LASTNAME, FIRSTNAME, MIDDLE\", \"page_marker\": \"\"}]}"}]} {"id": "state_id_fin_089_extract", "task": "extract", "class": "state_id", "doc": "state_id_fin_089.pdf", "images": ["images/train/state_id/state_id_fin_089_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('state_id', 'The document class.')\nfields = (['full_name', 'date_of_birth', 'address', 'issue_date', 'expiration_date', 'state'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name', 'date_of_birth': 'Date of Birth', 'address': 'Address', 'issue_date': 'Issued Date', 'expiration_date': 'Valid Until Date', 'state': 'Issuing state'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"LASTNAME, FIRSTNAME, MIDDLE\"}, {\"field\": \"date_of_birth\", \"value\": \"1977-08-16\"}, {\"field\": \"address\", \"value\": \"120 STATE STREET, MONTPELIER, VT 05603\"}, {\"field\": \"issue_date\", \"value\": \"2003-08-16\"}, {\"field\": \"expiration_date\", \"value\": \"2007-08-16\"}, {\"field\": \"state\", \"value\": \"Vermont\"}]}"}]} {"id": "state_id_fin_090_p1_classify", "task": "classify", "class": "state_id", "doc": "state_id_fin_090.pdf", "images": ["images/train/state_id/state_id_fin_090_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"state_id\", \"confidence\": 1.0, \"issuer\": \"NEBRASKA\", \"holder\": \"SAMPLECARD JOHN ALEXANDER\", \"page_marker\": \"\"}]}"}]} {"id": "state_id_fin_090_extract", "task": "extract", "class": "state_id", "doc": "state_id_fin_090.pdf", "images": ["images/train/state_id/state_id_fin_090_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('state_id', 'The document class.')\nfields = (['full_name', 'date_of_birth', 'address', 'issue_date', 'expiration_date', 'state'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name', 'date_of_birth': 'Date of Birth', 'address': 'Address', 'issue_date': 'Issued Date', 'expiration_date': 'Valid Until Date', 'state': 'Issuing state'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"SAMPLECARD JOHN ALEXANDER\"}, {\"field\": \"date_of_birth\", \"value\": \"1990-05-25\"}, {\"field\": \"address\", \"value\": \"1234 EXAMPLE ST, SOUTH SIOUX CITY, SA 12345-0000\"}, {\"field\": \"issue_date\", \"value\": \"2026-06-01\"}, {\"field\": \"expiration_date\", \"value\": \"2031-05-25\"}, {\"field\": \"state\", \"value\": \"NEBRASKA\"}]}"}]} {"id": "state_id_fin_091_p1_classify", "task": "classify", "class": "state_id", "doc": "state_id_fin_091.pdf", "images": ["images/train/state_id/state_id_fin_091_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"state_id\", \"confidence\": 1.0, \"issuer\": \"NEBRASKA\", \"holder\": \"FIRSTNAME MIDDLENAME LASTNAME, SUFFIXXX\", \"page_marker\": \"\"}]}"}]} {"id": "state_id_fin_091_extract", "task": "extract", "class": "state_id", "doc": "state_id_fin_091.pdf", "images": ["images/train/state_id/state_id_fin_091_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('state_id', 'The document class.')\nfields = (['full_name', 'date_of_birth', 'address', 'issue_date', 'expiration_date', 'state'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name', 'date_of_birth': 'Date of Birth', 'address': 'Address', 'issue_date': 'Issued Date', 'expiration_date': 'Valid Until Date', 'state': 'Issuing state'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"FIRSTNAME MIDDLENAME LASTNAME, SUFFIXXX\"}, {\"field\": \"date_of_birth\", \"value\": \"1977-02-01\"}, {\"field\": \"address\", \"value\": \"ADDRESS LINE 1UEXXUUEXXUUEXXUUEXXUU CITYUEXXUUEXXUUEXXUU, NE 00000\"}, {\"field\": \"issue_date\", \"value\": \"2008-02-01\"}, {\"field\": \"expiration_date\", \"value\": \"2013-02-01\"}, {\"field\": \"state\", \"value\": \"NEBRASKA\"}]}"}]} {"id": "state_id_fin_092_p1_classify", "task": "classify", "class": "state_id", "doc": "state_id_fin_092.pdf", "images": ["images/train/state_id/state_id_fin_092_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"state_id\", \"confidence\": 1.0, \"issuer\": \"Wisconsin\", \"holder\": \"SAMPLE HERMAN L\", \"page_marker\": \"\"}]}"}]} {"id": "state_id_fin_092_extract", "task": "extract", "class": "state_id", "doc": "state_id_fin_092.pdf", "images": ["images/train/state_id/state_id_fin_092_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('state_id', 'The document class.')\nfields = (['full_name', 'date_of_birth', 'address', 'issue_date', 'expiration_date', 'state'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name', 'date_of_birth': 'Date of Birth', 'address': 'Address', 'issue_date': 'Issued Date', 'expiration_date': 'Valid Until Date', 'state': 'Issuing state'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"SAMPLE HERMAN L\"}, {\"field\": \"date_of_birth\", \"value\": \"1947-03-18\"}, {\"field\": \"address\", \"value\": \"4822 MADISON YARDS WAY SUITE 1205 MADISON, WI 53705\"}, {\"field\": \"issue_date\", \"value\": \"2023-05-23\"}, {\"field\": \"expiration_date\", \"value\": \"\"}, {\"field\": \"state\", \"value\": \"Wisconsin\"}]}"}]} {"id": "state_id_fin_093_p1_classify", "task": "classify", "class": "state_id", "doc": "state_id_fin_093.pdf", "images": ["images/train/state_id/state_id_fin_093_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"state_id\", \"confidence\": 1.0, \"issuer\": \"\", \"holder\": \"Sample Wendy, D\", \"page_marker\": \"\"}]}"}]} {"id": "state_id_fin_093_extract", "task": "extract", "class": "state_id", "doc": "state_id_fin_093.pdf", "images": ["images/train/state_id/state_id_fin_093_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('state_id', 'The document class.')\nfields = (['full_name', 'date_of_birth', 'address', 'issue_date', 'expiration_date', 'state'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name', 'date_of_birth': 'Date of Birth', 'address': 'Address', 'issue_date': 'Issued Date', 'expiration_date': 'Valid Until Date', 'state': 'Issuing state'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"Sample Wendy, D\"}, {\"field\": \"date_of_birth\", \"value\": \"1988-03-16\"}, {\"field\": \"address\", \"value\": \"1628 Cabrini Boulevard New York, NY 10007\"}, {\"field\": \"issue_date\", \"value\": \"\"}, {\"field\": \"expiration_date\", \"value\": \"2019-04-04\"}, {\"field\": \"state\", \"value\": \"\"}]}"}]} {"id": "state_id_fin_094_p1_classify", "task": "classify", "class": "state_id", "doc": "state_id_fin_094.pdf", "images": ["images/train/state_id/state_id_fin_094_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"state_id\", \"confidence\": 1.0, \"issuer\": \"CALIFORNIA\", \"holder\": \"MORTON PATRICIA CHRISTINE\", \"page_marker\": \"\"}]}"}]} {"id": "state_id_fin_094_extract", "task": "extract", "class": "state_id", "doc": "state_id_fin_094.pdf", "images": ["images/train/state_id/state_id_fin_094_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('state_id', 'The document class.')\nfields = (['full_name', 'date_of_birth', 'address', 'issue_date', 'expiration_date', 'state'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name', 'date_of_birth': 'Date of Birth', 'address': 'Address', 'issue_date': 'Issued Date', 'expiration_date': 'Valid Until Date', 'state': 'Issuing state'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"MORTON PATRICIA CHRISTINE\"}, {\"field\": \"date_of_birth\", \"value\": \"1941-09-07\"}, {\"field\": \"address\", \"value\": \"1115 ELM DR, NOVATO, CA 94945\"}, {\"field\": \"issue_date\", \"value\": \"2016-09-19\"}, {\"field\": \"expiration_date\", \"value\": \"2026-09-07\"}, {\"field\": \"state\", \"value\": \"CALIFORNIA\"}]}"}]} {"id": "state_id_fin_095_p1_classify", "task": "classify", "class": "state_id", "doc": "state_id_fin_095.pdf", "images": ["images/train/state_id/state_id_fin_095_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"state_id\", \"confidence\": 1.0, \"issuer\": \"Vermont\", \"holder\": \"SAMPLE, NON-DRIVER ID\", \"page_marker\": \"\"}]}"}]} {"id": "state_id_fin_095_extract", "task": "extract", "class": "state_id", "doc": "state_id_fin_095.pdf", "images": ["images/train/state_id/state_id_fin_095_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('state_id', 'The document class.')\nfields = (['full_name', 'date_of_birth', 'address', 'issue_date', 'expiration_date', 'state'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name', 'date_of_birth': 'Date of Birth', 'address': 'Address', 'issue_date': 'Issued Date', 'expiration_date': 'Valid Until Date', 'state': 'Issuing state'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"SAMPLE, NON-DRIVER ID\"}, {\"field\": \"date_of_birth\", \"value\": \"1953-10-03\"}, {\"field\": \"address\", \"value\": \"120 STATE STREET, MONTPELIER, VT 05603\"}, {\"field\": \"issue_date\", \"value\": \"2007-06-29\"}, {\"field\": \"expiration_date\", \"value\": \"2010-10-03\"}, {\"field\": \"state\", \"value\": \"Vermont\"}]}"}]} {"id": "state_id_fin_096_p1_classify", "task": "classify", "class": "state_id", "doc": "state_id_fin_096.pdf", "images": ["images/train/state_id/state_id_fin_096_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"state_id\", \"confidence\": 1.0, \"issuer\": \"Connecticut\", \"holder\": \"ORTIZ, HECTOR, E\", \"page_marker\": \"\"}]}"}]} {"id": "state_id_fin_096_extract", "task": "extract", "class": "state_id", "doc": "state_id_fin_096.pdf", "images": ["images/train/state_id/state_id_fin_096_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('state_id', 'The document class.')\nfields = (['full_name', 'date_of_birth', 'address', 'issue_date', 'expiration_date', 'state'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name', 'date_of_birth': 'Date of Birth', 'address': 'Address', 'issue_date': 'Issued Date', 'expiration_date': 'Valid Until Date', 'state': 'Issuing state'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"ORTIZ, HECTOR, E\"}, {\"field\": \"date_of_birth\", \"value\": \"\"}, {\"field\": \"address\", \"value\": \"692 S PACIFIC ST, STAMFORD, CT 06902\"}, {\"field\": \"issue_date\", \"value\": \"1997-01-29\"}, {\"field\": \"expiration_date\", \"value\": \"\"}, {\"field\": \"state\", \"value\": \"Connecticut\"}]}"}]} {"id": "state_id_fin_098_p1_classify", "task": "classify", "class": "state_id", "doc": "state_id_fin_098.pdf", "images": ["images/train/state_id/state_id_fin_098_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Classify the document image into exactly one of the given class names (unknown if none fit), and read its identity fields (issuer, holder, page marker) as printed on the page.\n\nclasses = (['bank_statement', 'drivers_license', 'monthly_bill', 'paystub', 'state_id', 'unknown'], 'Allowed class names to choose from.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"cls\": \"state_id\", \"confidence\": 1.0, \"issuer\": \"Republic of Utopia\", \"holder\": \"AL\\u0130 \\u0130HSAN KAHRAMAN\", \"page_marker\": \"\"}]}"}]} {"id": "state_id_fin_098_extract", "task": "extract", "class": "state_id", "doc": "state_id_fin_098.pdf", "images": ["images/train/state_id/state_id_fin_098_p1.png"], "messages": [{"role": "system", "content": " This is a task you must complete by returning only the output.\n Do not include explanations, code, or extra text\u2014only the result.\n "}, {"role": "user", "content": [{"type": "text", "text": "Read the financial document image and extract the requested fields from what is actually printed on it. Return exactly one entry per requested field, using an empty string when a field is absent. Values are short verbatim strings from the document - never sentences or repeated text. Dates as ISO 8601 (YYYY-MM-DD). Amounts as plain numbers without currency symbols or thousands separators.\n\ncls = ('state_id', 'The document class.')\nfields = (['full_name', 'date_of_birth', 'address', 'issue_date', 'expiration_date', 'state'], 'The exact field names to extract.')\nhints = ({'full_name': 'Name', 'date_of_birth': 'Date of Birth', 'address': 'Address', 'issue_date': 'Issued Date', 'expiration_date': 'Valid Until Date', 'state': 'Issuing state'}, 'Per-field guidance mapping a field name to a short description.')"}, {"type": "image"}, {"type": "text", "text": "Schema requirements:\n- schema_object_wrapper (array)"}]}, {"role": "assistant", "content": "{\"schema_object_wrapper\": [{\"field\": \"full_name\", \"value\": \"AL\\u0130 \\u0130HSAN KAHRAMAN\"}, {\"field\": \"date_of_birth\", \"value\": \"1978-10-18\"}, {\"field\": \"address\", \"value\": \"\"}, {\"field\": \"issue_date\", \"value\": \"\"}, {\"field\": \"expiration_date\", \"value\": \"2024-02-15\"}, {\"field\": \"state\", \"value\": \"Republic of Utopia\"}]}"}]}