Best processor with eval_loss 4.6618

added_tokens.json

@@ -1,11 +1,11 @@
 {
-  "</s_agama>": 57554,
-  "</s_alamat>": 57552,
-  "</s_berlaku>": 57558,
-  "</s_gol_darah>": 57550,
-  "</s_jkel>": 57548,
-  "</s_kabkota_issuer>": 57546,
-  "</s_kecamatan>": 57556,
+  "</s_agama>": 57558,
+  "</s_alamat>": 57556,
+  "</s_berlaku>": 57554,
+  "</s_gol_darah>": 57552,
+  "</s_jkel>": 57550,
+  "</s_kabkota_issuer>": 57548,
+  "</s_kecamatan>": 57546,
   "</s_kelurahan>": 57544,
   "</s_nama>": 57542,
   "</s_nik>": 57540,
@@ -16,14 +16,14 @@
   "</s_tgl_terbit>": 57530,
   "</s_ttl>": 57528,
   "</s_wn>": 57526,
-  "<s_agama>": 57553,
-  "<s_alamat>": 57551,
-  "<s_berlaku>": 57557,
-  "<s_gol_darah>": 57549,
+  "<s_agama>": 57557,
+  "<s_alamat>": 57555,
+  "<s_berlaku>": 57553,
+  "<s_gol_darah>": 57551,
   "<s_iitcdip>": 57523,
-  "<s_jkel>": 57547,
-  "<s_kabkota_issuer>": 57545,
-  "<s_kecamatan>": 57555,
+  "<s_jkel>": 57549,
+  "<s_kabkota_issuer>": 57547,
+  "<s_kecamatan>": 57545,
   "<s_kelurahan>": 57543,
   "<s_nama>": 57541,
   "<s_nik>": 57539,

special_tokens_map.json

@@ -141,98 +141,98 @@
       "single_word": false
     },
     {
-      "content": "<s_kabkota_issuer>",
+      "content": "<s_kecamatan>",
       "lstrip": false,
       "normalized": false,
       "rstrip": false,
       "single_word": false
     },
     {
-      "content": "</s_kabkota_issuer>",
+      "content": "</s_kecamatan>",
       "lstrip": false,
       "normalized": false,
       "rstrip": false,
       "single_word": false
     },
     {
-      "content": "<s_jkel>",
+      "content": "<s_kabkota_issuer>",
       "lstrip": false,
       "normalized": false,
       "rstrip": false,
       "single_word": false
     },
     {
-      "content": "</s_jkel>",
+      "content": "</s_kabkota_issuer>",
       "lstrip": false,
       "normalized": false,
       "rstrip": false,
       "single_word": false
     },
     {
-      "content": "<s_gol_darah>",
+      "content": "<s_jkel>",
       "lstrip": false,
       "normalized": false,
       "rstrip": false,
       "single_word": false
     },
     {
-      "content": "</s_gol_darah>",
+      "content": "</s_jkel>",
       "lstrip": false,
       "normalized": false,
       "rstrip": false,
       "single_word": false
     },
     {
-      "content": "<s_alamat>",
+      "content": "<s_gol_darah>",
       "lstrip": false,
       "normalized": false,
       "rstrip": false,
       "single_word": false
     },
     {
-      "content": "</s_alamat>",
+      "content": "</s_gol_darah>",
       "lstrip": false,
       "normalized": false,
       "rstrip": false,
       "single_word": false
     },
     {
-      "content": "<s_agama>",
+      "content": "<s_berlaku>",
       "lstrip": false,
       "normalized": false,
       "rstrip": false,
       "single_word": false
     },
     {
-      "content": "</s_agama>",
+      "content": "</s_berlaku>",
       "lstrip": false,
       "normalized": false,
       "rstrip": false,
       "single_word": false
     },
     {
-      "content": "<s_kecamatan>",
+      "content": "<s_alamat>",
       "lstrip": false,
       "normalized": false,
       "rstrip": false,
       "single_word": false
     },
     {
-      "content": "</s_kecamatan>",
+      "content": "</s_alamat>",
       "lstrip": false,
       "normalized": false,
       "rstrip": false,
       "single_word": false
     },
     {
-      "content": "<s_berlaku>",
+      "content": "<s_agama>",
       "lstrip": false,
       "normalized": false,
       "rstrip": false,
       "single_word": false
     },
     {
-      "content": "</s_berlaku>",
+      "content": "</s_agama>",
       "lstrip": false,
       "normalized": false,
       "rstrip": false,

tokenizer.json

@@ -271,7 +271,7 @@
     },
     {
       "id": 57545,
-      "content": "<s_kabkota_issuer>",
+      "content": "<s_kecamatan>",
       "single_word": false,
       "lstrip": false,
       "rstrip": false,
@@ -280,7 +280,7 @@
     },
     {
       "id": 57546,
-      "content": "</s_kabkota_issuer>",
+      "content": "</s_kecamatan>",
       "single_word": false,
       "lstrip": false,
       "rstrip": false,
@@ -289,7 +289,7 @@
     },
     {
       "id": 57547,
-      "content": "<s_jkel>",
+      "content": "<s_kabkota_issuer>",
       "single_word": false,
       "lstrip": false,
       "rstrip": false,
@@ -298,7 +298,7 @@
     },
     {
       "id": 57548,
-      "content": "</s_jkel>",
+      "content": "</s_kabkota_issuer>",
       "single_word": false,
       "lstrip": false,
       "rstrip": false,
@@ -307,7 +307,7 @@
     },
     {
       "id": 57549,
-      "content": "<s_gol_darah>",
+      "content": "<s_jkel>",
       "single_word": false,
       "lstrip": false,
       "rstrip": false,
@@ -316,7 +316,7 @@
     },
     {
       "id": 57550,
-      "content": "</s_gol_darah>",
+      "content": "</s_jkel>",
       "single_word": false,
       "lstrip": false,
       "rstrip": false,
@@ -325,7 +325,7 @@
     },
     {
       "id": 57551,
-      "content": "<s_alamat>",
+      "content": "<s_gol_darah>",
       "single_word": false,
       "lstrip": false,
       "rstrip": false,
@@ -334,7 +334,7 @@
     },
     {
       "id": 57552,
-      "content": "</s_alamat>",
+      "content": "</s_gol_darah>",
       "single_word": false,
       "lstrip": false,
       "rstrip": false,
@@ -343,7 +343,7 @@
     },
     {
       "id": 57553,
-      "content": "<s_agama>",
+      "content": "<s_berlaku>",
       "single_word": false,
       "lstrip": false,
       "rstrip": false,
@@ -352,7 +352,7 @@
     },
     {
       "id": 57554,
-      "content": "</s_agama>",
+      "content": "</s_berlaku>",
       "single_word": false,
       "lstrip": false,
       "rstrip": false,
@@ -361,7 +361,7 @@
     },
     {
       "id": 57555,
-      "content": "<s_kecamatan>",
+      "content": "<s_alamat>",
       "single_word": false,
       "lstrip": false,
       "rstrip": false,
@@ -370,7 +370,7 @@
     },
     {
       "id": 57556,
-      "content": "</s_kecamatan>",
+      "content": "</s_alamat>",
       "single_word": false,
       "lstrip": false,
       "rstrip": false,
@@ -379,7 +379,7 @@
     },
     {
       "id": 57557,
-      "content": "<s_berlaku>",
+      "content": "<s_agama>",
       "single_word": false,
       "lstrip": false,
       "rstrip": false,
@@ -388,7 +388,7 @@
     },
     {
       "id": 57558,
-      "content": "</s_berlaku>",
+      "content": "</s_agama>",
       "single_word": false,
       "lstrip": false,
       "rstrip": false,

tokenizer_config.json

@@ -225,7 +225,7 @@
       "special": true
     },
     "57545": {
-      "content": "<s_kabkota_issuer>",
+      "content": "<s_kecamatan>",
       "lstrip": false,
       "normalized": false,
       "rstrip": false,
@@ -233,7 +233,7 @@
       "special": true
     },
     "57546": {
-      "content": "</s_kabkota_issuer>",
+      "content": "</s_kecamatan>",
       "lstrip": false,
       "normalized": false,
       "rstrip": false,
@@ -241,7 +241,7 @@
       "special": true
     },
     "57547": {
-      "content": "<s_jkel>",
+      "content": "<s_kabkota_issuer>",
       "lstrip": false,
       "normalized": false,
       "rstrip": false,
@@ -249,7 +249,7 @@
       "special": true
     },
     "57548": {
-      "content": "</s_jkel>",
+      "content": "</s_kabkota_issuer>",
       "lstrip": false,
       "normalized": false,
       "rstrip": false,
@@ -257,7 +257,7 @@
       "special": true
     },
     "57549": {
-      "content": "<s_gol_darah>",
+      "content": "<s_jkel>",
       "lstrip": false,
       "normalized": false,
       "rstrip": false,
@@ -265,7 +265,7 @@
       "special": true
     },
     "57550": {
-      "content": "</s_gol_darah>",
+      "content": "</s_jkel>",
       "lstrip": false,
       "normalized": false,
       "rstrip": false,
@@ -273,7 +273,7 @@
       "special": true
     },
     "57551": {
-      "content": "<s_alamat>",
+      "content": "<s_gol_darah>",
       "lstrip": false,
       "normalized": false,
       "rstrip": false,
@@ -281,7 +281,7 @@
       "special": true
     },
     "57552": {
-      "content": "</s_alamat>",
+      "content": "</s_gol_darah>",
       "lstrip": false,
       "normalized": false,
       "rstrip": false,
@@ -289,7 +289,7 @@
       "special": true
     },
     "57553": {
-      "content": "<s_agama>",
+      "content": "<s_berlaku>",
       "lstrip": false,
       "normalized": false,
       "rstrip": false,
@@ -297,7 +297,7 @@
       "special": true
     },
     "57554": {
-      "content": "</s_agama>",
+      "content": "</s_berlaku>",
       "lstrip": false,
       "normalized": false,
       "rstrip": false,
@@ -305,7 +305,7 @@
       "special": true
     },
     "57555": {
-      "content": "<s_kecamatan>",
+      "content": "<s_alamat>",
       "lstrip": false,
       "normalized": false,
       "rstrip": false,
@@ -313,7 +313,7 @@
       "special": true
     },
     "57556": {
-      "content": "</s_kecamatan>",
+      "content": "</s_alamat>",
       "lstrip": false,
       "normalized": false,
       "rstrip": false,
@@ -321,7 +321,7 @@
       "special": true
     },
     "57557": {
-      "content": "<s_berlaku>",
+      "content": "<s_agama>",
       "lstrip": false,
       "normalized": false,
       "rstrip": false,
@@ -329,7 +329,7 @@
       "special": true
     },
     "57558": {
-      "content": "</s_berlaku>",
+      "content": "</s_agama>",
       "lstrip": false,
       "normalized": false,
       "rstrip": false,
@@ -358,20 +358,20 @@
     "</s_nama>",
     "<s_kelurahan>",
     "</s_kelurahan>",
+    "<s_kecamatan>",
+    "</s_kecamatan>",
     "<s_kabkota_issuer>",
     "</s_kabkota_issuer>",
     "<s_jkel>",
     "</s_jkel>",
     "<s_gol_darah>",
     "</s_gol_darah>",
+    "<s_berlaku>",
+    "</s_berlaku>",
     "<s_alamat>",
     "</s_alamat>",
     "<s_agama>",
     "</s_agama>",
-    "<s_kecamatan>",
-    "</s_kecamatan>",
-    "<s_berlaku>",
-    "</s_berlaku>",
     "<s>",
     "</s>"
   ],