Training done

added_tokens.json

@@ -1,36 +1,46 @@
 {
-  "</s_height>": 57538,
-  "</s_image_id>": 57532,
-  "</s_image_size>": 57534,
-  "</s_label>": 57554,
-  "</s_meta>": 57526,
-  "</s_rect>": 57542,
-  "</s_split>": 57530,
-  "</s_value>": 57552,
-  "</s_version>": 57528,
-  "</s_width>": 57536,
-  "</s_words>": 57540,
-  "</s_x1>": 57544,
-  "</s_x2>": 57548,
-  "</s_y1>": 57546,
-  "</s_y2>": 57550,
-  "<s_cord-v2>": 57555,
-  "<s_height>": 57537,
+  "</s_ADDITION>": 57538,
+  "</s_BCCB BANK A/C NO. 1464>": 57560,
+  "</s_BCCB BANK A/C NO. 531>": 57562,
+  "</s_BIKE LOAN FROM KOTAK MAHINDRA>": 57558,
+  "</s_CAPITAL ACCOUNT>": 57530,
+  "</s_CARLOAN FROM HERO FINANCE>": 57556,
+  "</s_CASH AND BANK BALANCES>": 57548,
+  "</s_CASH IN HAND>": 57564,
+  "</s_CLOSING STOCK>": 57546,
+  "</s_CURRENT ASSETS>": 57542,
+  "</s_FIXED ASSETS 'As per schedule A'>": 57528,
+  "</s_FIXED ASSETS>": 57526,
+  "</s_INVESTMENTS>": 57534,
+  "</s_JEWELLERY>": 57536,
+  "</s_LOANS LIABILITY>": 57554,
+  "</s_PROPRIETOR>": 57532,
+  "</s_SUNDRY DEBITORS>": 57544,
+  "</s_TOTAL ASSETS>": 57552,
+  "</s_TOTAL LIABILITIES>": 57550,
+  "</s_UNSECURED LOANS>": 57540,
+  "<s_ADDITION>": 57537,
+  "<s_BCCB BANK A/C NO. 1464>": 57559,
+  "<s_BCCB BANK A/C NO. 531>": 57561,
+  "<s_BIKE LOAN FROM KOTAK MAHINDRA>": 57557,
+  "<s_CAPITAL ACCOUNT>": 57529,
+  "<s_CARLOAN FROM HERO FINANCE>": 57555,
+  "<s_CASH AND BANK BALANCES>": 57547,
+  "<s_CASH IN HAND>": 57563,
+  "<s_CLOSING STOCK>": 57545,
+  "<s_CURRENT ASSETS>": 57541,
+  "<s_FIXED ASSETS 'As per schedule A'>": 57527,
+  "<s_FIXED ASSETS>": 57525,
+  "<s_INVESTMENTS>": 57533,
+  "<s_JEWELLERY>": 57535,
+  "<s_LOANS LIABILITY>": 57553,
+  "<s_PROPRIETOR>": 57531,
+  "<s_SUNDRY DEBITORS>": 57543,
+  "<s_TOTAL ASSETS>": 57551,
+  "<s_TOTAL LIABILITIES>": 57549,
+  "<s_UNSECURED LOANS>": 57539,
+  "<s_cord-v2>": 57565,
   "<s_iitcdip>": 57523,
-  "<s_image_id>": 57531,
-  "<s_image_size>": 57533,
-  "<s_label>": 57553,
-  "<s_meta>": 57525,
-  "<s_rect>": 57541,
-  "<s_split>": 57529,
   "<s_synthdog>": 57524,
-  "<s_value>": 57551,
-  "<s_version>": 57527,
-  "<s_width>": 57535,
-  "<s_words>": 57539,
-  "<s_x1>": 57543,
-  "<s_x2>": 57547,
-  "<s_y1>": 57545,
-  "<s_y2>": 57549,
   "<sep/>": 57522
 }

tokenizer.json

@@ -91,7 +91,7 @@
     },
     {
       "id": 57525,
-      "content": "<s_meta>",
+      "content": "<s_FIXED ASSETS>",
       "single_word": false,
       "lstrip": false,
       "rstrip": false,
@@ -100,7 +100,7 @@
     },
     {
       "id": 57526,
-      "content": "</s_meta>",
+      "content": "</s_FIXED ASSETS>",
       "single_word": false,
       "lstrip": false,
       "rstrip": false,
@@ -109,7 +109,7 @@
     },
     {
       "id": 57527,
-      "content": "<s_version>",
+      "content": "<s_FIXED ASSETS 'As per schedule A'>",
       "single_word": false,
       "lstrip": false,
       "rstrip": false,
@@ -118,7 +118,7 @@
     },
     {
       "id": 57528,
-      "content": "</s_version>",
+      "content": "</s_FIXED ASSETS 'As per schedule A'>",
       "single_word": false,
       "lstrip": false,
       "rstrip": false,
@@ -127,7 +127,7 @@
     },
     {
       "id": 57529,
-      "content": "<s_split>",
+      "content": "<s_CAPITAL ACCOUNT>",
       "single_word": false,
       "lstrip": false,
       "rstrip": false,
@@ -136,7 +136,7 @@
     },
     {
       "id": 57530,
-      "content": "</s_split>",
+      "content": "</s_CAPITAL ACCOUNT>",
       "single_word": false,
       "lstrip": false,
       "rstrip": false,
@@ -145,7 +145,7 @@
     },
     {
       "id": 57531,
-      "content": "<s_image_id>",
+      "content": "<s_PROPRIETOR>",
       "single_word": false,
       "lstrip": false,
       "rstrip": false,
@@ -154,7 +154,7 @@
     },
     {
       "id": 57532,
-      "content": "</s_image_id>",
+      "content": "</s_PROPRIETOR>",
       "single_word": false,
       "lstrip": false,
       "rstrip": false,
@@ -163,7 +163,7 @@
     },
     {
       "id": 57533,
-      "content": "<s_image_size>",
+      "content": "<s_INVESTMENTS>",
       "single_word": false,
       "lstrip": false,
       "rstrip": false,
@@ -172,7 +172,7 @@
     },
     {
       "id": 57534,
-      "content": "</s_image_size>",
+      "content": "</s_INVESTMENTS>",
       "single_word": false,
       "lstrip": false,
       "rstrip": false,
@@ -181,7 +181,7 @@
     },
     {
       "id": 57535,
-      "content": "<s_width>",
+      "content": "<s_JEWELLERY>",
       "single_word": false,
       "lstrip": false,
       "rstrip": false,
@@ -190,7 +190,7 @@
     },
     {
       "id": 57536,
-      "content": "</s_width>",
+      "content": "</s_JEWELLERY>",
       "single_word": false,
       "lstrip": false,
       "rstrip": false,
@@ -199,7 +199,7 @@
     },
     {
       "id": 57537,
-      "content": "<s_height>",
+      "content": "<s_ADDITION>",
       "single_word": false,
       "lstrip": false,
       "rstrip": false,
@@ -208,7 +208,7 @@
     },
     {
       "id": 57538,
-      "content": "</s_height>",
+      "content": "</s_ADDITION>",
       "single_word": false,
       "lstrip": false,
       "rstrip": false,
@@ -217,7 +217,7 @@
     },
     {
       "id": 57539,
-      "content": "<s_words>",
+      "content": "<s_UNSECURED LOANS>",
       "single_word": false,
       "lstrip": false,
       "rstrip": false,
@@ -226,7 +226,7 @@
     },
     {
       "id": 57540,
-      "content": "</s_words>",
+      "content": "</s_UNSECURED LOANS>",
       "single_word": false,
       "lstrip": false,
       "rstrip": false,
@@ -235,7 +235,7 @@
     },
     {
       "id": 57541,
-      "content": "<s_rect>",
+      "content": "<s_CURRENT ASSETS>",
       "single_word": false,
       "lstrip": false,
       "rstrip": false,
@@ -244,7 +244,7 @@
     },
     {
       "id": 57542,
-      "content": "</s_rect>",
+      "content": "</s_CURRENT ASSETS>",
       "single_word": false,
       "lstrip": false,
       "rstrip": false,
@@ -253,7 +253,7 @@
     },
     {
       "id": 57543,
-      "content": "<s_x1>",
+      "content": "<s_SUNDRY DEBITORS>",
       "single_word": false,
       "lstrip": false,
       "rstrip": false,
@@ -262,7 +262,7 @@
     },
     {
       "id": 57544,
-      "content": "</s_x1>",
+      "content": "</s_SUNDRY DEBITORS>",
       "single_word": false,
       "lstrip": false,
       "rstrip": false,
@@ -271,7 +271,7 @@
     },
     {
       "id": 57545,
-      "content": "<s_y1>",
+      "content": "<s_CLOSING STOCK>",
       "single_word": false,
       "lstrip": false,
       "rstrip": false,
@@ -280,7 +280,7 @@
     },
     {
       "id": 57546,
-      "content": "</s_y1>",
+      "content": "</s_CLOSING STOCK>",
       "single_word": false,
       "lstrip": false,
       "rstrip": false,
@@ -289,7 +289,7 @@
     },
     {
       "id": 57547,
-      "content": "<s_x2>",
+      "content": "<s_CASH AND BANK BALANCES>",
       "single_word": false,
       "lstrip": false,
       "rstrip": false,
@@ -298,7 +298,7 @@
     },
     {
       "id": 57548,
-      "content": "</s_x2>",
+      "content": "</s_CASH AND BANK BALANCES>",
       "single_word": false,
       "lstrip": false,
       "rstrip": false,
@@ -307,7 +307,7 @@
     },
     {
       "id": 57549,
-      "content": "<s_y2>",
+      "content": "<s_TOTAL LIABILITIES>",
       "single_word": false,
       "lstrip": false,
       "rstrip": false,
@@ -316,7 +316,7 @@
     },
     {
       "id": 57550,
-      "content": "</s_y2>",
+      "content": "</s_TOTAL LIABILITIES>",
       "single_word": false,
       "lstrip": false,
       "rstrip": false,
@@ -325,7 +325,7 @@
     },
     {
       "id": 57551,
-      "content": "<s_value>",
+      "content": "<s_TOTAL ASSETS>",
       "single_word": false,
       "lstrip": false,
       "rstrip": false,
@@ -334,7 +334,7 @@
     },
     {
       "id": 57552,
-      "content": "</s_value>",
+      "content": "</s_TOTAL ASSETS>",
       "single_word": false,
       "lstrip": false,
       "rstrip": false,
@@ -343,7 +343,7 @@
     },
     {
       "id": 57553,
-      "content": "<s_label>",
+      "content": "<s_LOANS LIABILITY>",
       "single_word": false,
       "lstrip": false,
       "rstrip": false,
@@ -352,7 +352,7 @@
     },
     {
       "id": 57554,
-      "content": "</s_label>",
+      "content": "</s_LOANS LIABILITY>",
       "single_word": false,
       "lstrip": false,
       "rstrip": false,
@@ -361,6 +361,96 @@
     },
     {
       "id": 57555,
+      "content": "<s_CARLOAN FROM HERO FINANCE>",
+      "single_word": false,
+      "lstrip": false,
+      "rstrip": false,
+      "normalized": true,
+      "special": false
+    },
+    {
+      "id": 57556,
+      "content": "</s_CARLOAN FROM HERO FINANCE>",
+      "single_word": false,
+      "lstrip": false,
+      "rstrip": false,
+      "normalized": true,
+      "special": false
+    },
+    {
+      "id": 57557,
+      "content": "<s_BIKE LOAN FROM KOTAK MAHINDRA>",
+      "single_word": false,
+      "lstrip": false,
+      "rstrip": false,
+      "normalized": true,
+      "special": false
+    },
+    {
+      "id": 57558,
+      "content": "</s_BIKE LOAN FROM KOTAK MAHINDRA>",
+      "single_word": false,
+      "lstrip": false,
+      "rstrip": false,
+      "normalized": true,
+      "special": false
+    },
+    {
+      "id": 57559,
+      "content": "<s_BCCB BANK A/C NO. 1464>",
+      "single_word": false,
+      "lstrip": false,
+      "rstrip": false,
+      "normalized": true,
+      "special": false
+    },
+    {
+      "id": 57560,
+      "content": "</s_BCCB BANK A/C NO. 1464>",
+      "single_word": false,
+      "lstrip": false,
+      "rstrip": false,
+      "normalized": true,
+      "special": false
+    },
+    {
+      "id": 57561,
+      "content": "<s_BCCB BANK A/C NO. 531>",
+      "single_word": false,
+      "lstrip": false,
+      "rstrip": false,
+      "normalized": true,
+      "special": false
+    },
+    {
+      "id": 57562,
+      "content": "</s_BCCB BANK A/C NO. 531>",
+      "single_word": false,
+      "lstrip": false,
+      "rstrip": false,
+      "normalized": true,
+      "special": false
+    },
+    {
+      "id": 57563,
+      "content": "<s_CASH IN HAND>",
+      "single_word": false,
+      "lstrip": false,
+      "rstrip": false,
+      "normalized": true,
+      "special": false
+    },
+    {
+      "id": 57564,
+      "content": "</s_CASH IN HAND>",
+      "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_meta>",
+      "content": "<s_FIXED ASSETS>",
       "lstrip": false,
       "normalized": true,
       "rstrip": false,
@@ -73,7 +73,7 @@
       "special": false
     },
     "57526": {
-      "content": "</s_meta>",
+      "content": "</s_FIXED ASSETS>",
       "lstrip": false,
       "normalized": true,
       "rstrip": false,
@@ -81,7 +81,7 @@
       "special": false
     },
     "57527": {
-      "content": "<s_version>",
+      "content": "<s_FIXED ASSETS 'As per schedule A'>",
       "lstrip": false,
       "normalized": true,
       "rstrip": false,
@@ -89,7 +89,7 @@
       "special": false
     },
     "57528": {
-      "content": "</s_version>",
+      "content": "</s_FIXED ASSETS 'As per schedule A'>",
       "lstrip": false,
       "normalized": true,
       "rstrip": false,
@@ -97,7 +97,7 @@
       "special": false
     },
     "57529": {
-      "content": "<s_split>",
+      "content": "<s_CAPITAL ACCOUNT>",
       "lstrip": false,
       "normalized": true,
       "rstrip": false,
@@ -105,7 +105,7 @@
       "special": false
     },
     "57530": {
-      "content": "</s_split>",
+      "content": "</s_CAPITAL ACCOUNT>",
       "lstrip": false,
       "normalized": true,
       "rstrip": false,
@@ -113,7 +113,7 @@
       "special": false
     },
     "57531": {
-      "content": "<s_image_id>",
+      "content": "<s_PROPRIETOR>",
       "lstrip": false,
       "normalized": true,
       "rstrip": false,
@@ -121,7 +121,7 @@
       "special": false
     },
     "57532": {
-      "content": "</s_image_id>",
+      "content": "</s_PROPRIETOR>",
       "lstrip": false,
       "normalized": true,
       "rstrip": false,
@@ -129,7 +129,7 @@
       "special": false
     },
     "57533": {
-      "content": "<s_image_size>",
+      "content": "<s_INVESTMENTS>",
       "lstrip": false,
       "normalized": true,
       "rstrip": false,
@@ -137,7 +137,7 @@
       "special": false
     },
     "57534": {
-      "content": "</s_image_size>",
+      "content": "</s_INVESTMENTS>",
       "lstrip": false,
       "normalized": true,
       "rstrip": false,
@@ -145,7 +145,7 @@
       "special": false
     },
     "57535": {
-      "content": "<s_width>",
+      "content": "<s_JEWELLERY>",
       "lstrip": false,
       "normalized": true,
       "rstrip": false,
@@ -153,7 +153,7 @@
       "special": false
     },
     "57536": {
-      "content": "</s_width>",
+      "content": "</s_JEWELLERY>",
       "lstrip": false,
       "normalized": true,
       "rstrip": false,
@@ -161,7 +161,7 @@
       "special": false
     },
     "57537": {
-      "content": "<s_height>",
+      "content": "<s_ADDITION>",
       "lstrip": false,
       "normalized": true,
       "rstrip": false,
@@ -169,7 +169,7 @@
       "special": false
     },
     "57538": {
-      "content": "</s_height>",
+      "content": "</s_ADDITION>",
       "lstrip": false,
       "normalized": true,
       "rstrip": false,
@@ -177,7 +177,7 @@
       "special": false
     },
     "57539": {
-      "content": "<s_words>",
+      "content": "<s_UNSECURED LOANS>",
       "lstrip": false,
       "normalized": true,
       "rstrip": false,
@@ -185,7 +185,7 @@
       "special": false
     },
     "57540": {
-      "content": "</s_words>",
+      "content": "</s_UNSECURED LOANS>",
       "lstrip": false,
       "normalized": true,
       "rstrip": false,
@@ -193,7 +193,7 @@
       "special": false
     },
     "57541": {
-      "content": "<s_rect>",
+      "content": "<s_CURRENT ASSETS>",
       "lstrip": false,
       "normalized": true,
       "rstrip": false,
@@ -201,7 +201,7 @@
       "special": false
     },
     "57542": {
-      "content": "</s_rect>",
+      "content": "</s_CURRENT ASSETS>",
       "lstrip": false,
       "normalized": true,
       "rstrip": false,
@@ -209,7 +209,7 @@
       "special": false
     },
     "57543": {
-      "content": "<s_x1>",
+      "content": "<s_SUNDRY DEBITORS>",
       "lstrip": false,
       "normalized": true,
       "rstrip": false,
@@ -217,7 +217,7 @@
       "special": false
     },
     "57544": {
-      "content": "</s_x1>",
+      "content": "</s_SUNDRY DEBITORS>",
       "lstrip": false,
       "normalized": true,
       "rstrip": false,
@@ -225,7 +225,7 @@
       "special": false
     },
     "57545": {
-      "content": "<s_y1>",
+      "content": "<s_CLOSING STOCK>",
       "lstrip": false,
       "normalized": true,
       "rstrip": false,
@@ -233,7 +233,7 @@
       "special": false
     },
     "57546": {
-      "content": "</s_y1>",
+      "content": "</s_CLOSING STOCK>",
       "lstrip": false,
       "normalized": true,
       "rstrip": false,
@@ -241,7 +241,7 @@
       "special": false
     },
     "57547": {
-      "content": "<s_x2>",
+      "content": "<s_CASH AND BANK BALANCES>",
       "lstrip": false,
       "normalized": true,
       "rstrip": false,
@@ -249,7 +249,7 @@
       "special": false
     },
     "57548": {
-      "content": "</s_x2>",
+      "content": "</s_CASH AND BANK BALANCES>",
       "lstrip": false,
       "normalized": true,
       "rstrip": false,
@@ -257,7 +257,7 @@
       "special": false
     },
     "57549": {
-      "content": "<s_y2>",
+      "content": "<s_TOTAL LIABILITIES>",
       "lstrip": false,
       "normalized": true,
       "rstrip": false,
@@ -265,7 +265,7 @@
       "special": false
     },
     "57550": {
-      "content": "</s_y2>",
+      "content": "</s_TOTAL LIABILITIES>",
       "lstrip": false,
       "normalized": true,
       "rstrip": false,
@@ -273,7 +273,7 @@
       "special": false
     },
     "57551": {
-      "content": "<s_value>",
+      "content": "<s_TOTAL ASSETS>",
       "lstrip": false,
       "normalized": true,
       "rstrip": false,
@@ -281,7 +281,7 @@
       "special": false
     },
     "57552": {
-      "content": "</s_value>",
+      "content": "</s_TOTAL ASSETS>",
       "lstrip": false,
       "normalized": true,
       "rstrip": false,
@@ -289,7 +289,7 @@
       "special": false
     },
     "57553": {
-      "content": "<s_label>",
+      "content": "<s_LOANS LIABILITY>",
       "lstrip": false,
       "normalized": true,
       "rstrip": false,
@@ -297,7 +297,7 @@
       "special": false
     },
     "57554": {
-      "content": "</s_label>",
+      "content": "</s_LOANS LIABILITY>",
       "lstrip": false,
       "normalized": true,
       "rstrip": false,
@@ -305,6 +305,86 @@
       "special": false
     },
     "57555": {
+      "content": "<s_CARLOAN FROM HERO FINANCE>",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "57556": {
+      "content": "</s_CARLOAN FROM HERO FINANCE>",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "57557": {
+      "content": "<s_BIKE LOAN FROM KOTAK MAHINDRA>",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "57558": {
+      "content": "</s_BIKE LOAN FROM KOTAK MAHINDRA>",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "57559": {
+      "content": "<s_BCCB BANK A/C NO. 1464>",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "57560": {
+      "content": "</s_BCCB BANK A/C NO. 1464>",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "57561": {
+      "content": "<s_BCCB BANK A/C NO. 531>",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "57562": {
+      "content": "</s_BCCB BANK A/C NO. 531>",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "57563": {
+      "content": "<s_CASH IN HAND>",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "57564": {
+      "content": "</s_CASH IN HAND>",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "57565": {
       "content": "<s_cord-v2>",
       "lstrip": false,
       "normalized": true,