MRG_Donut / tokenizer_config.json
KMeera's picture
End of training
1a766bf
{
"added_tokens_decoder": {
"0": {
"content": "<s>",
"lstrip": false,
"normalized": false,
"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_pO2(A-a)e>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"57526": {
"content": "</s_TIME>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"57527": {
"content": "</s_TEST>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"57528": {
"content": "<s_Anion Gapc>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"57529": {
"content": "<s_TEST>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"57530": {
"content": "<s_Set RR>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"57531": {
"content": "</s_Blood Gas Values>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"57532": {
"content": "</s_Set RR>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"57533": {
"content": "<s_Calculated Values>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"57534": {
"content": "<s_LAB>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"57535": {
"content": "<s_SBEc>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"57536": {
"content": "<s_SYRINGE>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"57537": {
"content": "<s_DATE>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"57538": {
"content": "</s_FO2(1)>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"57539": {
"content": "</s_ABEc>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"57540": {
"content": "<s_ABEc>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"57541": {
"content": "<s_MODE>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"57542": {
"content": "</s_Anion Gapc>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"57543": {
"content": "<s_Identifications>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"57544": {
"content": "</s_Baro>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"57545": {
"content": "<s_FO2Hbe>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"57546": {
"content": "<s_cHOO3-(Past)c>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"57547": {
"content": "</s_cBase(Box)c>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"57548": {
"content": "<s_p5O6>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"57549": {
"content": "</s_cHOO3-(P)c>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"57550": {
"content": "</s_SAMPLE NO>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"57551": {
"content": "</s_cBase(Ecfox)c>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"57552": {
"content": "</s_SBEc>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"57553": {
"content": "<s_pO2>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"57554": {
"content": "<s_cBase(Box)c>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"57555": {
"content": "</s_FHHbe>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"57556": {
"content": "</s_pH(st)c>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"57557": {
"content": "</s_pH(T)>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"57558": {
"content": "</s_pO2(A,T)e>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"57559": {
"content": "<s_Oximetry Values>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"57560": {
"content": "<s_cHOO3-(P)c>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"57561": {
"content": "<s_T>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"57562": {
"content": "</s_DATE>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"57563": {
"content": "<s_cK+>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"57564": {
"content": "</s_LAB>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"57565": {
"content": "<s_SAMPLE NO>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"57566": {
"content": "<s_Sample Type>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"57567": {
"content": "<s_Header>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"57568": {
"content": "<s_cBase(Ecfox)c>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"57569": {
"content": "<s_SAMPLE>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"57570": {
"content": "<s_TIME>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"57571": {
"content": "</s_pO2(T)>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"57572": {
"content": "<s_Patient Last Name>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"57573": {
"content": "<s_pO2(a)/FO2(I)c>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"57574": {
"content": "</s_Mear Airway Press>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"57575": {
"content": "<s_Baro>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"57576": {
"content": "</s_pH>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"57577": {
"content": "</s_FO2Hbe>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"57578": {
"content": "</s_Hctc>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"57579": {
"content": "<s_Minute Volume>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"57580": {
"content": "</s_HFNC FLOW>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"57581": {
"content": "<s_pO2(A)e>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"57582": {
"content": "</s_SEX>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"57583": {
"content": "<s_pO2(a/A)e>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"57584": {
"content": "</s_Identifications>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"57585": {
"content": "<s_sO2>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"57586": {
"content": "<s_cLac>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"57587": {
"content": "<s_pH(T)>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"57588": {
"content": "</s_T>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"57589": {
"content": "<s_FO2(1)>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"57590": {
"content": "<s_ctHb>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"57591": {
"content": "</s_pO2(a/A)e>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"57592": {
"content": "</s_pO2>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"57593": {
"content": "</s_Temperature Corrected Values>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"57594": {
"content": "</s_Patient Last Name>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"57595": {
"content": "<s_Temperature Corrected Values>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"57596": {
"content": "</s_ctCO2(P)c>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"57597": {
"content": "<s_Hctc>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"57598": {
"content": "</s_Tidal Volume>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"57599": {
"content": "</s_pC02>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"57600": {
"content": "</s_cCl- >",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"57601": {
"content": "</s_Minute Volume>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"57602": {
"content": "</s_cCa2+>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"57603": {
"content": "<s_FHHbe>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"57604": {
"content": "</s_p5O6>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"57605": {
"content": "<s_TAG1>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"57606": {
"content": "</s_Header>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"57607": {
"content": "</s_cNa+>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"57608": {
"content": "<s_cCa2+>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"57609": {
"content": "</s_cK+>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"57610": {
"content": "</s_PEEP>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"57611": {
"content": "</s_pO2(A-a)e>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"57612": {
"content": "</s_SAMPLE>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"57613": {
"content": "<s_cH+c>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"57614": {
"content": "</s_pO2(a)/FO2(I)c>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"57615": {
"content": "<s_HFNC FLOW>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"57616": {
"content": "<s_pH(st)c>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"57617": {
"content": "</s_pC02(T)>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"57618": {
"content": "<s_Acid Base Status>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"57619": {
"content": "</s_cLac>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"57620": {
"content": "<s_ctCO2(P)c>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"57621": {
"content": "</s_ctHb>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"57622": {
"content": "<s_Tidal Volume>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"57623": {
"content": "<s_pC02(T)>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"57624": {
"content": "</s_sO2>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"57625": {
"content": "<s_pC02>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"57626": {
"content": "<s_PEEP>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"57627": {
"content": "</s_Electrolyte Values>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"57628": {
"content": "</s_cHOO3-(Past)c>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"57629": {
"content": "<s_SEX>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"57630": {
"content": "<s_Electrolyte Values>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"57631": {
"content": "</s_cH+c>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"57632": {
"content": "</s_MODE>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"57633": {
"content": "</s_Acid Base Status>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"57634": {
"content": "<s_Patient ID>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"57635": {
"content": "</s_SYRINGE>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"57636": {
"content": "</s_pO2(A)e>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"57637": {
"content": "<s_cNa+>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"57638": {
"content": "<s_Mear Airway Press>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"57639": {
"content": "</s_TAG1>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"57640": {
"content": "</s_Metabolite Values>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"57641": {
"content": "<s_pO2(T)>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"57642": {
"content": "</s_Oximetry Values>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"57643": {
"content": "<s_Metabolite Values>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"57644": {
"content": "<s_cCl- >",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"57645": {
"content": "</s_Calculated Values>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"57646": {
"content": "<s_pO2(A,T)e>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"57647": {
"content": "</s_Sample Type>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"57648": {
"content": "<s_pH>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"57649": {
"content": "</s_Patient ID>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
},
"57650": {
"content": "<s_Blood Gas Values>",
"lstrip": false,
"normalized": false,
"rstrip": false,
"single_word": false,
"special": true
}
},
"additional_special_tokens": [
"<s_pO2(A-a)e>",
"</s_TIME>",
"</s_TEST>",
"<s_Anion Gapc>",
"<s_TEST>",
"<s_Set RR>",
"</s_Blood Gas Values>",
"</s_Set RR>",
"<s_Calculated Values>",
"<s_LAB>",
"<s_SBEc>",
"<s_SYRINGE>",
"<s_DATE>",
"</s_FO2(1)>",
"</s_ABEc>",
"<s_ABEc>",
"<s_MODE>",
"</s_Anion Gapc>",
"<s_Identifications>",
"</s_Baro>",
"<s_FO2Hbe>",
"<s_cHOO3-(Past)c>",
"</s_cBase(Box)c>",
"<s_p5O6>",
"</s_cHOO3-(P)c>",
"</s_SAMPLE NO>",
"</s_cBase(Ecfox)c>",
"</s_SBEc>",
"<s_pO2>",
"<s_cBase(Box)c>",
"</s_FHHbe>",
"</s_pH(st)c>",
"</s_pH(T)>",
"</s_pO2(A,T)e>",
"<s_Oximetry Values>",
"<s_cHOO3-(P)c>",
"<s_T>",
"</s_DATE>",
"<s_cK+>",
"</s_LAB>",
"<s_SAMPLE NO>",
"<s_Sample Type>",
"<s_Header>",
"<s_cBase(Ecfox)c>",
"<s_SAMPLE>",
"<s_TIME>",
"</s_pO2(T)>",
"<s_Patient Last Name>",
"<s_pO2(a)/FO2(I)c>",
"</s_Mear Airway Press>",
"<s_Baro>",
"</s_pH>",
"</s_FO2Hbe>",
"</s_Hctc>",
"<s_Minute Volume>",
"</s_HFNC FLOW>",
"<s_pO2(A)e>",
"</s_SEX>",
"<s_pO2(a/A)e>",
"</s_Identifications>",
"<s_sO2>",
"<s_cLac>",
"</s>",
"<s_pH(T)>",
"</s_T>",
"<s_FO2(1)>",
"<s_ctHb>",
"</s_pO2(a/A)e>",
"</s_pO2>",
"</s_Temperature Corrected Values>",
"</s_Patient Last Name>",
"<s_Temperature Corrected Values>",
"</s_ctCO2(P)c>",
"<s_Hctc>",
"</s_Tidal Volume>",
"</s_pC02>",
"</s_cCl- >",
"</s_Minute Volume>",
"</s_cCa2+>",
"<s_FHHbe>",
"</s_p5O6>",
"<s_TAG1>",
"</s_Header>",
"</s_cNa+>",
"<s_cCa2+>",
"</s_cK+>",
"</s_PEEP>",
"</s_pO2(A-a)e>",
"</s_SAMPLE>",
"<s_cH+c>",
"</s_pO2(a)/FO2(I)c>",
"<s_HFNC FLOW>",
"<s_pH(st)c>",
"</s_pC02(T)>",
"<s_Acid Base Status>",
"</s_cLac>",
"<s_ctCO2(P)c>",
"</s_ctHb>",
"<s_Tidal Volume>",
"<s_pC02(T)>",
"</s_sO2>",
"<s_pC02>",
"<s_PEEP>",
"</s_Electrolyte Values>",
"</s_cHOO3-(Past)c>",
"<s_SEX>",
"<s_Electrolyte Values>",
"</s_cH+c>",
"</s_MODE>",
"</s_Acid Base Status>",
"<s_Patient ID>",
"</s_SYRINGE>",
"</s_pO2(A)e>",
"<s>",
"<s_cNa+>",
"<s_Mear Airway Press>",
"</s_TAG1>",
"</s_Metabolite Values>",
"<s_pO2(T)>",
"</s_Oximetry Values>",
"<s_Metabolite Values>",
"<s_cCl- >",
"</s_Calculated Values>",
"<s_pO2(A,T)e>",
"</s_Sample Type>",
"<s_pH>",
"</s_Patient ID>",
"<s_Blood Gas Values>"
],
"bos_token": "<s>",
"clean_up_tokenization_spaces": true,
"cls_token": "<s>",
"eos_token": "</s>",
"mask_token": "<mask>",
"model_max_length": 1000000000000000019884624838656,
"pad_token": "<pad>",
"processor_class": "DonutProcessor",
"sep_token": "</s>",
"sp_model_kwargs": {},
"tokenizer_class": "XLMRobertaTokenizer",
"unk_token": "<unk>"
}