DocQA / Model /tokenizer_config.json
likhithv's picture
initial commit
cc9f92c
raw
history blame
41.7 kB
{
"added_tokens_decoder": {
"0": {
"content": "<s>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": true
},
"1": {
"content": "<pad>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"2": {
"content": "</s>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"3": {
"content": "<unk>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"57521": {
"content": "<mask>",
"lstrip": true,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": true
},
"57522": {
"content": "<sep/>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57523": {
"content": "<s_iitcdip>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"57524": {
"content": "<s_synthdog>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"57525": {
"content": "<s_items>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57526": {
"content": "</s_items>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57527": {
"content": "<s_Street address (including apt: no.)>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57528": {
"content": "</s_Street address (including apt: no.)>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57529": {
"content": "<s_RECIPIENTS TIN>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57530": {
"content": "</s_RECIPIENTS TIN>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57531": {
"content": "<s_RECIPIENT'S name>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57532": {
"content": "</s_RECIPIENT'S name>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57533": {
"content": "<s_PAYERS TIN>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57534": {
"content": "</s_PAYERS TIN>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57535": {
"content": "<s_PAYER'S name, Street address, city or [OWI, Stale province, country, ZIP or foreign postal code; and telephone no_>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57536": {
"content": "</s_PAYER'S name, Street address, city or [OWI, Stale province, country, ZIP or foreign postal code; and telephone no_>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57537": {
"content": "<s_OMB No.>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57538": {
"content": "</s_OMB No.>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57539": {
"content": "<s_Form>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57540": {
"content": "</s_Form>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57541": {
"content": "<s_For calendar year>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57542": {
"content": "</s_For calendar year>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57543": {
"content": "<s_City or town, state Or province; country, and ZIP or foreign postal code>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57544": {
"content": "</s_City or town, state Or province; country, and ZIP or foreign postal code>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57545": {
"content": "<s_Accoung number (see instructions}>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57546": {
"content": "</s_Accoung number (see instructions}>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57547": {
"content": "<s_9 Cash liquidation distributions>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57548": {
"content": "</s_9 Cash liquidation distributions>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57549": {
"content": "<s_8 Foreign country or U.S. possession>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57550": {
"content": "</s_8 Foreign country or U.S. possession>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57551": {
"content": "<s_7 Foreign tax paid>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57552": {
"content": "</s_7 Foreign tax paid>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57553": {
"content": "<s_6 Investment expenses>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57554": {
"content": "</s_6 Investment expenses>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57555": {
"content": "<s_5 Section 199A dividends>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57556": {
"content": "</s_5 Section 199A dividends>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57557": {
"content": "<s_4 Federal income tax withheld>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57558": {
"content": "</s_4 Federal income tax withheld>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57559": {
"content": "<s_3 Nondividend distributions>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57560": {
"content": "</s_3 Nondividend distributions>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57561": {
"content": "<s_2f Section 897 capital gain>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57562": {
"content": "</s_2f Section 897 capital gain>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57563": {
"content": "<s_2e Section 897 ordinary dividends>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57564": {
"content": "</s_2e Section 897 ordinary dividends>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57565": {
"content": "<s_2d Collectibles (28%) gain>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57566": {
"content": "</s_2d Collectibles (28%) gain>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57567": {
"content": "<s_2c Section 1202 gain>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57568": {
"content": "</s_2c Section 1202 gain>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57569": {
"content": "<s_2b Unrecap. Sec: 1250 gain>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57570": {
"content": "</s_2b Unrecap. Sec: 1250 gain>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57571": {
"content": "<s_2a Total capital gain distr:>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57572": {
"content": "</s_2a Total capital gain distr:>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57573": {
"content": "<s_1b Qualified dividends>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57574": {
"content": "</s_1b Qualified dividends>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57575": {
"content": "<s_1a Total ordinary dividends>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57576": {
"content": "</s_1a Total ordinary dividends>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57577": {
"content": "<s_16 State tax withheld>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57578": {
"content": "</s_16 State tax withheld>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57579": {
"content": "<s_15 State identification no.>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57580": {
"content": "</s_15 State identification no.>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57581": {
"content": "<s_14 State>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57582": {
"content": "</s_14 State>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57583": {
"content": "<s_13 Specified private activity bond interest dividends>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57584": {
"content": "</s_13 Specified private activity bond interest dividends>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57585": {
"content": "<s_12 Exempt-interest dividends>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57586": {
"content": "</s_12 Exempt-interest dividends>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57587": {
"content": "<s_11 FATCA filing requirement>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57588": {
"content": "</s_11 FATCA filing requirement>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57589": {
"content": "<s_10 Noncash liquidation distributions>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57590": {
"content": "</s_10 Noncash liquidation distributions>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57591": {
"content": "<s_(Rev.>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57592": {
"content": "</s_(Rev.>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57593": {
"content": "<s_Street address (including apt. no.)>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57594": {
"content": "</s_Street address (including apt. no.)>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57595": {
"content": "<s_Payer’s RTN (optional)>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57596": {
"content": "</s_Payer’s RTN (optional)>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57597": {
"content": "<s_PAYER'S name, Street address, city or town, State province, country, ZIP or foreign postal code, and telephone no.>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57598": {
"content": "</s_PAYER'S name, Street address, city or town, State province, country, ZIP or foreign postal code, and telephone no.>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57599": {
"content": "<s_FATCA filing requirement>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57600": {
"content": "</s_FATCA filing requirement>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57601": {
"content": "<s_City or town, state Or province, country, and ZIP or foreign postal code>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57602": {
"content": "</s_City or town, state Or province, country, and ZIP or foreign postal code>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57603": {
"content": "<s_Accoung number (see instructions)>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57604": {
"content": "</s_Accoung number (see instructions)>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57605": {
"content": "<s_9 Specified private activity bond interest>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57606": {
"content": "</s_9 Specified private activity bond interest>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57607": {
"content": "<s_8 Tax-exempt interest>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57608": {
"content": "</s_8 Tax-exempt interest>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57609": {
"content": "<s_7 Foreign country or U.S. possession>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57610": {
"content": "</s_7 Foreign country or U.S. possession>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57611": {
"content": "<s_6 Foreign tax paid>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57612": {
"content": "</s_6 Foreign tax paid>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57613": {
"content": "<s_5 Investment expenses>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57614": {
"content": "</s_5 Investment expenses>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57615": {
"content": "<s_3 Interest on U.S. Savings Bonds and Treasury obligations>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57616": {
"content": "</s_3 Interest on U.S. Savings Bonds and Treasury obligations>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57617": {
"content": "<s_2 Early withdrawal penalty>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57618": {
"content": "</s_2 Early withdrawal penalty>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57619": {
"content": "<s_17 State tax withheld>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57620": {
"content": "</s_17 State tax withheld>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57621": {
"content": "<s_16 State identification no.>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57622": {
"content": "</s_16 State identification no.>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57623": {
"content": "<s_15 State>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57624": {
"content": "</s_15 State>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57625": {
"content": "<s_14 Tax-exempt and tax credit bond CUSIP no.>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57626": {
"content": "</s_14 Tax-exempt and tax credit bond CUSIP no.>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57627": {
"content": "<s_13 Bond premium on tax-exempt bond>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57628": {
"content": "</s_13 Bond premium on tax-exempt bond>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57629": {
"content": "<s_12 Bond premium on Treasury obligations>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57630": {
"content": "</s_12 Bond premium on Treasury obligations>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57631": {
"content": "<s_11 Bond premium>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57632": {
"content": "</s_11 Bond premium>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57633": {
"content": "<s_10 Market discount>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57634": {
"content": "</s_10 Market discount>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57635": {
"content": "<s_1 Interest income>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57636": {
"content": "</s_1 Interest income>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57637": {
"content": "<s_f Employee’s address and ZIP code>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57638": {
"content": "</s_f Employee’s address and ZIP code>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57639": {
"content": "<s_e Employee’s first name and initial>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57640": {
"content": "</s_e Employee’s first name and initial>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57641": {
"content": "<s_d Control number>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57642": {
"content": "</s_d Control number>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57643": {
"content": "<s_c Employer’s name, address, and ZIP code>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57644": {
"content": "</s_c Employer’s name, address, and ZIP code>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57645": {
"content": "<s_b Employer identification number (EIN)>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57646": {
"content": "</s_b Employer identification number (EIN)>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57647": {
"content": "<s_a Employee’s social security number>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57648": {
"content": "</s_a Employee’s social security number>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57649": {
"content": "<s_Last name>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57650": {
"content": "</s_Last name>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57651": {
"content": "<s_8 Allocated tips>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57652": {
"content": "</s_8 Allocated tips>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57653": {
"content": "<s_7 Social security tips>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57654": {
"content": "</s_7 Social security tips>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57655": {
"content": "<s_6 Medicare tax withheld>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57656": {
"content": "</s_6 Medicare tax withheld>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57657": {
"content": "<s_5 Medicare wages and tips>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57658": {
"content": "</s_5 Medicare wages and tips>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57659": {
"content": "<s_4 Social security tax withheld>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57660": {
"content": "</s_4 Social security tax withheld>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57661": {
"content": "<s_3 Social security wages>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57662": {
"content": "</s_3 Social security wages>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57663": {
"content": "<s_20 Locality name>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57664": {
"content": "</s_20 Locality name>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57665": {
"content": "<s_2 Federal income tax withheld>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57666": {
"content": "</s_2 Federal income tax withheld>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57667": {
"content": "<s_19 Local income tax>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57668": {
"content": "</s_19 Local income tax>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57669": {
"content": "<s_18 Local wages, tips, etc.>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57670": {
"content": "</s_18 Local wages, tips, etc.>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57671": {
"content": "<s_17 State income tax>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57672": {
"content": "</s_17 State income tax>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57673": {
"content": "<s_16 State wages, tips, etc. >",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57674": {
"content": "</s_16 State wages, tips, etc. >",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57675": {
"content": "<s_14 Other>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57676": {
"content": "</s_14 Other>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57677": {
"content": "<s_13 Statutory employee, Retirement plan, Third-party sick pay>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57678": {
"content": "</s_13 Statutory employee, Retirement plan, Third-party sick pay>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57679": {
"content": "<s_12d>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57680": {
"content": "</s_12d>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57681": {
"content": "<s_12c>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57682": {
"content": "</s_12c>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57683": {
"content": "<s_12b>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57684": {
"content": "</s_12b>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57685": {
"content": "<s_12a>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57686": {
"content": "</s_12a>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57687": {
"content": "<s_11 Nonqualified plans>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57688": {
"content": "</s_11 Nonqualified plans>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57689": {
"content": "<s_10 Dependent care benefits>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57690": {
"content": "</s_10 Dependent care benefits>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57691": {
"content": "<s_1 Wages, tips, other compensation>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57692": {
"content": "</s_1 Wages, tips, other compensation>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57693": {
"content": "<s_h Other EIN used this year>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57694": {
"content": "</s_h Other EIN used this year>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57695": {
"content": "<s_g Employer’s address and ZIP code>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57696": {
"content": "</s_g Employer’s address and ZIP code>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57697": {
"content": "<s_f Employer’s name>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57698": {
"content": "</s_f Employer’s name>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57699": {
"content": "<s_e Employer identification number (EIN)>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57700": {
"content": "</s_e Employer identification number (EIN)>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57701": {
"content": "<s_d Establishment number>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57702": {
"content": "</s_d Establishment number>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57703": {
"content": "<s_c Total number of Forms W-2>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57704": {
"content": "</s_c Total number of Forms W-2>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57705": {
"content": "<s_a Control number>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57706": {
"content": "</s_a Control number>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57707": {
"content": "<s_Employer’s telephone number>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57708": {
"content": "</s_Employer’s telephone number>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57709": {
"content": "<s_Employer’s state ID number>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57710": {
"content": "</s_Employer’s state ID number>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57711": {
"content": "<s_Employer’s fax number>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57712": {
"content": "</s_Employer’s fax number>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57713": {
"content": "<s_Employer’s email address>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57714": {
"content": "</s_Employer’s email address>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57715": {
"content": "<s_Employer’s contact person>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57716": {
"content": "</s_Employer’s contact person>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57717": {
"content": "<s_16 State wages, tips, etc.>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57718": {
"content": "</s_16 State wages, tips, etc.>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57719": {
"content": "<s_14 Income tax withheld by payer of third-party sick pay>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57720": {
"content": "</s_14 Income tax withheld by payer of third-party sick pay>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57721": {
"content": "<s_13 For third-party sick pay use only>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57722": {
"content": "</s_13 For third-party sick pay use only>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57723": {
"content": "<s_12a Deferred compensation>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
},
"57724": {
"content": "</s_12a Deferred compensation>",
"lstrip": false,
"normalized": true,
"rstrip": false,
"single_word": false,
"special": false
}
},
"additional_special_tokens": [
"<s_iitcdip>",
"<s_synthdog>"
],
"bos_token": "<s>",
"clean_up_tokenization_spaces": true,
"cls_token": "<s>",
"eos_token": "</s>",
"mask_token": "<mask>",
"max_length": 1000,
"model_max_length": 1000000000000000019884624838656,
"pad_to_multiple_of": null,
"pad_token": "<pad>",
"pad_token_type_id": 0,
"padding_side": "right",
"processor_class": "DonutProcessor",
"sep_token": "</s>",
"sp_model_kwargs": {},
"stride": 0,
"tokenizer_class": "XLMRobertaTokenizer",
"truncation_side": "right",
"truncation_strategy": "longest_first",
"unk_token": "<unk>"
}