Training done

added_tokens.json

@@ -1,46 +1,32 @@
 {
-  "</s_None>": 57564,
-  "</s_client>": 57534,
-  "</s_client_tax_id>": 57538,
-  "</s_header>": 57526,
-  "</s_iban>": 57540,
-  "</s_invoice_date>": 57530,
-  "</s_invoice_no>": 57528,
-  "</s_item_desc>": 57544,
-  "</s_item_gross_worth>": 57554,
-  "</s_item_net_price>": 57548,
-  "</s_item_net_worth>": 57550,
-  "</s_item_qty>": 57546,
-  "</s_item_vat>": 57552,
-  "</s_items>": 57542,
-  "</s_seller>": 57532,
-  "</s_seller_tax_id>": 57536,
-  "</s_summary>": 57556,
-  "</s_total_gross_worth>": 57562,
-  "</s_total_net_worth>": 57558,
-  "</s_total_vat>": 57560,
-  "<s_None>": 57563,
-  "<s_client>": 57533,
-  "<s_client_tax_id>": 57537,
-  "<s_cord-v2>": 57565,
-  "<s_header>": 57525,
-  "<s_iban>": 57539,
+  "</s_authority>": 57548,
+  "</s_city>": 57544,
+  "</s_country>": 57526,
+  "</s_countryCode>": 57530,
+  "</s_dateOfExpiry>": 57550,
+  "</s_dateOfIssue>": 57546,
+  "</s_dob>": 57540,
+  "</s_firstName>": 57536,
+  "</s_gender>": 57542,
+  "</s_lastName>": 57534,
+  "</s_nationality>": 57538,
+  "</s_passportNumber>": 57532,
+  "</s_passportType>": 57528,
+  "<s_authority>": 57547,
+  "<s_city>": 57543,
+  "<s_cord-v2>": 57551,
+  "<s_country>": 57525,
+  "<s_countryCode>": 57529,
+  "<s_dateOfExpiry>": 57549,
+  "<s_dateOfIssue>": 57545,
+  "<s_dob>": 57539,
+  "<s_firstName>": 57535,
+  "<s_gender>": 57541,
   "<s_iitcdip>": 57523,
-  "<s_invoice_date>": 57529,
-  "<s_invoice_no>": 57527,
-  "<s_item_desc>": 57543,
-  "<s_item_gross_worth>": 57553,
-  "<s_item_net_price>": 57547,
-  "<s_item_net_worth>": 57549,
-  "<s_item_qty>": 57545,
-  "<s_item_vat>": 57551,
-  "<s_items>": 57541,
-  "<s_seller>": 57531,
-  "<s_seller_tax_id>": 57535,
-  "<s_summary>": 57555,
+  "<s_lastName>": 57533,
+  "<s_nationality>": 57537,
+  "<s_passportNumber>": 57531,
+  "<s_passportType>": 57527,
   "<s_synthdog>": 57524,
-  "<s_total_gross_worth>": 57561,
-  "<s_total_net_worth>": 57557,
-  "<s_total_vat>": 57559,
   "<sep/>": 57522
 }

tokenizer.json

@@ -91,7 +91,7 @@
     },
     {
       "id": 57525,
-      "content": "<s_header>",
+      "content": "<s_country>",
       "single_word": false,
       "lstrip": false,
       "rstrip": false,
@@ -100,7 +100,7 @@
     },
     {
       "id": 57526,
-      "content": "</s_header>",
+      "content": "</s_country>",
       "single_word": false,
       "lstrip": false,
       "rstrip": false,
@@ -109,7 +109,7 @@
     },
     {
       "id": 57527,
-      "content": "<s_invoice_no>",
+      "content": "<s_passportType>",
       "single_word": false,
       "lstrip": false,
       "rstrip": false,
@@ -118,7 +118,7 @@
     },
     {
       "id": 57528,
-      "content": "</s_invoice_no>",
+      "content": "</s_passportType>",
       "single_word": false,
       "lstrip": false,
       "rstrip": false,
@@ -127,7 +127,7 @@
     },
     {
       "id": 57529,
-      "content": "<s_invoice_date>",
+      "content": "<s_countryCode>",
       "single_word": false,
       "lstrip": false,
       "rstrip": false,
@@ -136,7 +136,7 @@
     },
     {
       "id": 57530,
-      "content": "</s_invoice_date>",
+      "content": "</s_countryCode>",
       "single_word": false,
       "lstrip": false,
       "rstrip": false,
@@ -145,7 +145,7 @@
     },
     {
       "id": 57531,
-      "content": "<s_seller>",
+      "content": "<s_passportNumber>",
       "single_word": false,
       "lstrip": false,
       "rstrip": false,
@@ -154,7 +154,7 @@
     },
     {
       "id": 57532,
-      "content": "</s_seller>",
+      "content": "</s_passportNumber>",
       "single_word": false,
       "lstrip": false,
       "rstrip": false,
@@ -163,7 +163,7 @@
     },
     {
       "id": 57533,
-      "content": "<s_client>",
+      "content": "<s_lastName>",
       "single_word": false,
       "lstrip": false,
       "rstrip": false,
@@ -172,7 +172,7 @@
     },
     {
       "id": 57534,
-      "content": "</s_client>",
+      "content": "</s_lastName>",
       "single_word": false,
       "lstrip": false,
       "rstrip": false,
@@ -181,7 +181,7 @@
     },
     {
       "id": 57535,
-      "content": "<s_seller_tax_id>",
+      "content": "<s_firstName>",
       "single_word": false,
       "lstrip": false,
       "rstrip": false,
@@ -190,7 +190,7 @@
     },
     {
       "id": 57536,
-      "content": "</s_seller_tax_id>",
+      "content": "</s_firstName>",
       "single_word": false,
       "lstrip": false,
       "rstrip": false,
@@ -199,7 +199,7 @@
     },
     {
       "id": 57537,
-      "content": "<s_client_tax_id>",
+      "content": "<s_nationality>",
       "single_word": false,
       "lstrip": false,
       "rstrip": false,
@@ -208,7 +208,7 @@
     },
     {
       "id": 57538,
-      "content": "</s_client_tax_id>",
+      "content": "</s_nationality>",
       "single_word": false,
       "lstrip": false,
       "rstrip": false,
@@ -217,7 +217,7 @@
     },
     {
       "id": 57539,
-      "content": "<s_iban>",
+      "content": "<s_dob>",
       "single_word": false,
       "lstrip": false,
       "rstrip": false,
@@ -226,7 +226,7 @@
     },
     {
       "id": 57540,
-      "content": "</s_iban>",
+      "content": "</s_dob>",
       "single_word": false,
       "lstrip": false,
       "rstrip": false,
@@ -235,7 +235,7 @@
     },
     {
       "id": 57541,
-      "content": "<s_items>",
+      "content": "<s_gender>",
       "single_word": false,
       "lstrip": false,
       "rstrip": false,
@@ -244,7 +244,7 @@
     },
     {
       "id": 57542,
-      "content": "</s_items>",
+      "content": "</s_gender>",
       "single_word": false,
       "lstrip": false,
       "rstrip": false,
@@ -253,7 +253,7 @@
     },
     {
       "id": 57543,
-      "content": "<s_item_desc>",
+      "content": "<s_city>",
       "single_word": false,
       "lstrip": false,
       "rstrip": false,
@@ -262,7 +262,7 @@
     },
     {
       "id": 57544,
-      "content": "</s_item_desc>",
+      "content": "</s_city>",
       "single_word": false,
       "lstrip": false,
       "rstrip": false,
@@ -271,7 +271,7 @@
     },
     {
       "id": 57545,
-      "content": "<s_item_qty>",
+      "content": "<s_dateOfIssue>",
       "single_word": false,
       "lstrip": false,
       "rstrip": false,
@@ -280,7 +280,7 @@
     },
     {
       "id": 57546,
-      "content": "</s_item_qty>",
+      "content": "</s_dateOfIssue>",
       "single_word": false,
       "lstrip": false,
       "rstrip": false,
@@ -289,7 +289,7 @@
     },
     {
       "id": 57547,
-      "content": "<s_item_net_price>",
+      "content": "<s_authority>",
       "single_word": false,
       "lstrip": false,
       "rstrip": false,
@@ -298,7 +298,7 @@
     },
     {
       "id": 57548,
-      "content": "</s_item_net_price>",
+      "content": "</s_authority>",
       "single_word": false,
       "lstrip": false,
       "rstrip": false,
@@ -307,7 +307,7 @@
     },
     {
       "id": 57549,
-      "content": "<s_item_net_worth>",
+      "content": "<s_dateOfExpiry>",
       "single_word": false,
       "lstrip": false,
       "rstrip": false,
@@ -316,7 +316,7 @@
     },
     {
       "id": 57550,
-      "content": "</s_item_net_worth>",
+      "content": "</s_dateOfExpiry>",
       "single_word": false,
       "lstrip": false,
       "rstrip": false,
@@ -325,132 +325,6 @@
     },
     {
       "id": 57551,
-      "content": "<s_item_vat>",
-      "single_word": false,
-      "lstrip": false,
-      "rstrip": false,
-      "normalized": true,
-      "special": false
-    },
-    {
-      "id": 57552,
-      "content": "</s_item_vat>",
-      "single_word": false,
-      "lstrip": false,
-      "rstrip": false,
-      "normalized": true,
-      "special": false
-    },
-    {
-      "id": 57553,
-      "content": "<s_item_gross_worth>",
-      "single_word": false,
-      "lstrip": false,
-      "rstrip": false,
-      "normalized": true,
-      "special": false
-    },
-    {
-      "id": 57554,
-      "content": "</s_item_gross_worth>",
-      "single_word": false,
-      "lstrip": false,
-      "rstrip": false,
-      "normalized": true,
-      "special": false
-    },
-    {
-      "id": 57555,
-      "content": "<s_summary>",
-      "single_word": false,
-      "lstrip": false,
-      "rstrip": false,
-      "normalized": true,
-      "special": false
-    },
-    {
-      "id": 57556,
-      "content": "</s_summary>",
-      "single_word": false,
-      "lstrip": false,
-      "rstrip": false,
-      "normalized": true,
-      "special": false
-    },
-    {
-      "id": 57557,
-      "content": "<s_total_net_worth>",
-      "single_word": false,
-      "lstrip": false,
-      "rstrip": false,
-      "normalized": true,
-      "special": false
-    },
-    {
-      "id": 57558,
-      "content": "</s_total_net_worth>",
-      "single_word": false,
-      "lstrip": false,
-      "rstrip": false,
-      "normalized": true,
-      "special": false
-    },
-    {
-      "id": 57559,
-      "content": "<s_total_vat>",
-      "single_word": false,
-      "lstrip": false,
-      "rstrip": false,
-      "normalized": true,
-      "special": false
-    },
-    {
-      "id": 57560,
-      "content": "</s_total_vat>",
-      "single_word": false,
-      "lstrip": false,
-      "rstrip": false,
-      "normalized": true,
-      "special": false
-    },
-    {
-      "id": 57561,
-      "content": "<s_total_gross_worth>",
-      "single_word": false,
-      "lstrip": false,
-      "rstrip": false,
-      "normalized": true,
-      "special": false
-    },
-    {
-      "id": 57562,
-      "content": "</s_total_gross_worth>",
-      "single_word": false,
-      "lstrip": false,
-      "rstrip": false,
-      "normalized": true,
-      "special": false
-    },
-    {
-      "id": 57563,
-      "content": "<s_None>",
-      "single_word": false,
-      "lstrip": false,
-      "rstrip": false,
-      "normalized": true,
-      "special": false
-    },
-    {
-      "id": 57564,
-      "content": "</s_None>",
-      "single_word": false,
-      "lstrip": false,
-      "rstrip": false,
-      "normalized": true,
-      "special": false
-    },
-    {
-      "id": 57565,
       "content": "<s_cord-v2>",
       "single_word": false,
       "lstrip": false,

tokenizer_config.json

@@ -65,7 +65,7 @@
       "special": true
     },
     "57525": {
-      "content": "<s_header>",
+      "content": "<s_country>",
       "lstrip": false,
       "normalized": true,
       "rstrip": false,
@@ -73,7 +73,7 @@
       "special": false
     },
     "57526": {
-      "content": "</s_header>",
+      "content": "</s_country>",
       "lstrip": false,
       "normalized": true,
       "rstrip": false,
@@ -81,7 +81,7 @@
       "special": false
     },
     "57527": {
-      "content": "<s_invoice_no>",
+      "content": "<s_passportType>",
       "lstrip": false,
       "normalized": true,
       "rstrip": false,
@@ -89,7 +89,7 @@
       "special": false
     },
     "57528": {
-      "content": "</s_invoice_no>",
+      "content": "</s_passportType>",
       "lstrip": false,
       "normalized": true,
       "rstrip": false,
@@ -97,7 +97,7 @@
       "special": false
     },
     "57529": {
-      "content": "<s_invoice_date>",
+      "content": "<s_countryCode>",
       "lstrip": false,
       "normalized": true,
       "rstrip": false,
@@ -105,7 +105,7 @@
       "special": false
     },
     "57530": {
-      "content": "</s_invoice_date>",
+      "content": "</s_countryCode>",
       "lstrip": false,
       "normalized": true,
       "rstrip": false,
@@ -113,7 +113,7 @@
       "special": false
     },
     "57531": {
-      "content": "<s_seller>",
+      "content": "<s_passportNumber>",
       "lstrip": false,
       "normalized": true,
       "rstrip": false,
@@ -121,7 +121,7 @@
       "special": false
     },
     "57532": {
-      "content": "</s_seller>",
+      "content": "</s_passportNumber>",
       "lstrip": false,
       "normalized": true,
       "rstrip": false,
@@ -129,7 +129,7 @@
       "special": false
     },
     "57533": {
-      "content": "<s_client>",
+      "content": "<s_lastName>",
       "lstrip": false,
       "normalized": true,
       "rstrip": false,
@@ -137,7 +137,7 @@
       "special": false
     },
     "57534": {
-      "content": "</s_client>",
+      "content": "</s_lastName>",
       "lstrip": false,
       "normalized": true,
       "rstrip": false,
@@ -145,7 +145,7 @@
       "special": false
     },
     "57535": {
-      "content": "<s_seller_tax_id>",
+      "content": "<s_firstName>",
       "lstrip": false,
       "normalized": true,
       "rstrip": false,
@@ -153,7 +153,7 @@
       "special": false
     },
     "57536": {
-      "content": "</s_seller_tax_id>",
+      "content": "</s_firstName>",
       "lstrip": false,
       "normalized": true,
       "rstrip": false,
@@ -161,7 +161,7 @@
       "special": false
     },
     "57537": {
-      "content": "<s_client_tax_id>",
+      "content": "<s_nationality>",
       "lstrip": false,
       "normalized": true,
       "rstrip": false,
@@ -169,7 +169,7 @@
       "special": false
     },
     "57538": {
-      "content": "</s_client_tax_id>",
+      "content": "</s_nationality>",
       "lstrip": false,
       "normalized": true,
       "rstrip": false,
@@ -177,7 +177,7 @@
       "special": false
     },
     "57539": {
-      "content": "<s_iban>",
+      "content": "<s_dob>",
       "lstrip": false,
       "normalized": true,
       "rstrip": false,
@@ -185,7 +185,7 @@
       "special": false
     },
     "57540": {
-      "content": "</s_iban>",
+      "content": "</s_dob>",
       "lstrip": false,
       "normalized": true,
       "rstrip": false,
@@ -193,7 +193,7 @@
       "special": false
     },
     "57541": {
-      "content": "<s_items>",
+      "content": "<s_gender>",
       "lstrip": false,
       "normalized": true,
       "rstrip": false,
@@ -201,7 +201,7 @@
       "special": false
     },
     "57542": {
-      "content": "</s_items>",
+      "content": "</s_gender>",
       "lstrip": false,
       "normalized": true,
       "rstrip": false,
@@ -209,7 +209,7 @@
       "special": false
     },
     "57543": {
-      "content": "<s_item_desc>",
+      "content": "<s_city>",
       "lstrip": false,
       "normalized": true,
       "rstrip": false,
@@ -217,7 +217,7 @@
       "special": false
     },
     "57544": {
-      "content": "</s_item_desc>",
+      "content": "</s_city>",
       "lstrip": false,
       "normalized": true,
       "rstrip": false,
@@ -225,7 +225,7 @@
       "special": false
     },
     "57545": {
-      "content": "<s_item_qty>",
+      "content": "<s_dateOfIssue>",
       "lstrip": false,
       "normalized": true,
       "rstrip": false,
@@ -233,7 +233,7 @@
       "special": false
     },
     "57546": {
-      "content": "</s_item_qty>",
+      "content": "</s_dateOfIssue>",
       "lstrip": false,
       "normalized": true,
       "rstrip": false,
@@ -241,7 +241,7 @@
       "special": false
     },
     "57547": {
-      "content": "<s_item_net_price>",
+      "content": "<s_authority>",
       "lstrip": false,
       "normalized": true,
       "rstrip": false,
@@ -249,7 +249,7 @@
       "special": false
     },
     "57548": {
-      "content": "</s_item_net_price>",
+      "content": "</s_authority>",
       "lstrip": false,
       "normalized": true,
       "rstrip": false,
@@ -257,7 +257,7 @@
       "special": false
     },
     "57549": {
-      "content": "<s_item_net_worth>",
+      "content": "<s_dateOfExpiry>",
       "lstrip": false,
       "normalized": true,
       "rstrip": false,
@@ -265,7 +265,7 @@
       "special": false
     },
     "57550": {
-      "content": "</s_item_net_worth>",
+      "content": "</s_dateOfExpiry>",
       "lstrip": false,
       "normalized": true,
       "rstrip": false,
@@ -273,118 +273,6 @@
       "special": false
     },
     "57551": {
-      "content": "<s_item_vat>",
-      "lstrip": false,
-      "normalized": true,
-      "rstrip": false,
-      "single_word": false,
-      "special": false
-    },
-    "57552": {
-      "content": "</s_item_vat>",
-      "lstrip": false,
-      "normalized": true,
-      "rstrip": false,
-      "single_word": false,
-      "special": false
-    },
-    "57553": {
-      "content": "<s_item_gross_worth>",
-      "lstrip": false,
-      "normalized": true,
-      "rstrip": false,
-      "single_word": false,
-      "special": false
-    },
-    "57554": {
-      "content": "</s_item_gross_worth>",
-      "lstrip": false,
-      "normalized": true,
-      "rstrip": false,
-      "single_word": false,
-      "special": false
-    },
-    "57555": {
-      "content": "<s_summary>",
-      "lstrip": false,
-      "normalized": true,
-      "rstrip": false,
-      "single_word": false,
-      "special": false
-    },
-    "57556": {
-      "content": "</s_summary>",
-      "lstrip": false,
-      "normalized": true,
-      "rstrip": false,
-      "single_word": false,
-      "special": false
-    },
-    "57557": {
-      "content": "<s_total_net_worth>",
-      "lstrip": false,
-      "normalized": true,
-      "rstrip": false,
-      "single_word": false,
-      "special": false
-    },
-    "57558": {
-      "content": "</s_total_net_worth>",
-      "lstrip": false,
-      "normalized": true,
-      "rstrip": false,
-      "single_word": false,
-      "special": false
-    },
-    "57559": {
-      "content": "<s_total_vat>",
-      "lstrip": false,
-      "normalized": true,
-      "rstrip": false,
-      "single_word": false,
-      "special": false
-    },
-    "57560": {
-      "content": "</s_total_vat>",
-      "lstrip": false,
-      "normalized": true,
-      "rstrip": false,
-      "single_word": false,
-      "special": false
-    },
-    "57561": {
-      "content": "<s_total_gross_worth>",
-      "lstrip": false,
-      "normalized": true,
-      "rstrip": false,
-      "single_word": false,
-      "special": false
-    },
-    "57562": {
-      "content": "</s_total_gross_worth>",
-      "lstrip": false,
-      "normalized": true,
-      "rstrip": false,
-      "single_word": false,
-      "special": false
-    },
-    "57563": {
-      "content": "<s_None>",
-      "lstrip": false,
-      "normalized": true,
-      "rstrip": false,
-      "single_word": false,
-      "special": false
-    },
-    "57564": {
-      "content": "</s_None>",
-      "lstrip": false,
-      "normalized": true,
-      "rstrip": false,
-      "single_word": false,
-      "special": false
-    },
-    "57565": {
       "content": "<s_cord-v2>",
       "lstrip": false,
       "normalized": true,