Commit
•
0dd4e7e
1
Parent(s):
0f774d3
End of training
Browse files- added_tokens.json +10 -8
- special_tokens_map.json +2 -0
- tokenizer.json +25 -7
- tokenizer_config.json +25 -7
added_tokens.json
CHANGED
@@ -1,16 +1,18 @@
|
|
1 |
{
|
2 |
"</s>": 2,
|
3 |
-
"</s_First>":
|
4 |
-
"</s_Last>":
|
5 |
-
"</s_Middle>":
|
6 |
-
"</
|
7 |
-
"</
|
|
|
8 |
"<mask>": 57521,
|
9 |
"<pad>": 1,
|
10 |
"<s>": 0,
|
11 |
-
"<s_First>":
|
12 |
-
"<s_Last>":
|
13 |
-
"<s_Middle>":
|
|
|
14 |
"<s_iitcdip>": 57523,
|
15 |
"<s_no>": 57526,
|
16 |
"<s_province>": 57525,
|
|
|
1 |
{
|
2 |
"</s>": 2,
|
3 |
+
"</s_First>": 57531,
|
4 |
+
"</s_Last>": 57533,
|
5 |
+
"</s_Middle>": 57536,
|
6 |
+
"</s_ignore>": 57527,
|
7 |
+
"</s_no>": 57530,
|
8 |
+
"</s_province>": 57534,
|
9 |
"<mask>": 57521,
|
10 |
"<pad>": 1,
|
11 |
"<s>": 0,
|
12 |
+
"<s_First>": 57529,
|
13 |
+
"<s_Last>": 57532,
|
14 |
+
"<s_Middle>": 57535,
|
15 |
+
"<s_ignore>": 57528,
|
16 |
"<s_iitcdip>": 57523,
|
17 |
"<s_no>": 57526,
|
18 |
"<s_province>": 57525,
|
special_tokens_map.json
CHANGED
@@ -5,6 +5,8 @@
|
|
5 |
"</s>",
|
6 |
"<s_province>",
|
7 |
"<s_no>",
|
|
|
|
|
8 |
"<s_First>",
|
9 |
"</s_no>",
|
10 |
"</s_First>",
|
|
|
5 |
"</s>",
|
6 |
"<s_province>",
|
7 |
"<s_no>",
|
8 |
+
"</s_ignore>",
|
9 |
+
"<s_ignore>",
|
10 |
"<s_First>",
|
11 |
"</s_no>",
|
12 |
"</s_First>",
|
tokenizer.json
CHANGED
@@ -109,7 +109,7 @@
|
|
109 |
},
|
110 |
{
|
111 |
"id": 57527,
|
112 |
-
"content": "
|
113 |
"single_word": false,
|
114 |
"lstrip": true,
|
115 |
"rstrip": true,
|
@@ -118,7 +118,7 @@
|
|
118 |
},
|
119 |
{
|
120 |
"id": 57528,
|
121 |
-
"content": "
|
122 |
"single_word": false,
|
123 |
"lstrip": true,
|
124 |
"rstrip": true,
|
@@ -127,7 +127,7 @@
|
|
127 |
},
|
128 |
{
|
129 |
"id": 57529,
|
130 |
-
"content": "
|
131 |
"single_word": false,
|
132 |
"lstrip": true,
|
133 |
"rstrip": true,
|
@@ -136,7 +136,7 @@
|
|
136 |
},
|
137 |
{
|
138 |
"id": 57530,
|
139 |
-
"content": "
|
140 |
"single_word": false,
|
141 |
"lstrip": true,
|
142 |
"rstrip": true,
|
@@ -145,7 +145,7 @@
|
|
145 |
},
|
146 |
{
|
147 |
"id": 57531,
|
148 |
-
"content": "</
|
149 |
"single_word": false,
|
150 |
"lstrip": true,
|
151 |
"rstrip": true,
|
@@ -154,7 +154,7 @@
|
|
154 |
},
|
155 |
{
|
156 |
"id": 57532,
|
157 |
-
"content": "
|
158 |
"single_word": false,
|
159 |
"lstrip": true,
|
160 |
"rstrip": true,
|
@@ -163,7 +163,7 @@
|
|
163 |
},
|
164 |
{
|
165 |
"id": 57533,
|
166 |
-
"content": "
|
167 |
"single_word": false,
|
168 |
"lstrip": true,
|
169 |
"rstrip": true,
|
@@ -172,6 +172,24 @@
|
|
172 |
},
|
173 |
{
|
174 |
"id": 57534,
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
175 |
"content": "</s_Middle>",
|
176 |
"single_word": false,
|
177 |
"lstrip": true,
|
|
|
109 |
},
|
110 |
{
|
111 |
"id": 57527,
|
112 |
+
"content": "</s_ignore>",
|
113 |
"single_word": false,
|
114 |
"lstrip": true,
|
115 |
"rstrip": true,
|
|
|
118 |
},
|
119 |
{
|
120 |
"id": 57528,
|
121 |
+
"content": "<s_ignore>",
|
122 |
"single_word": false,
|
123 |
"lstrip": true,
|
124 |
"rstrip": true,
|
|
|
127 |
},
|
128 |
{
|
129 |
"id": 57529,
|
130 |
+
"content": "<s_First>",
|
131 |
"single_word": false,
|
132 |
"lstrip": true,
|
133 |
"rstrip": true,
|
|
|
136 |
},
|
137 |
{
|
138 |
"id": 57530,
|
139 |
+
"content": "</s_no>",
|
140 |
"single_word": false,
|
141 |
"lstrip": true,
|
142 |
"rstrip": true,
|
|
|
145 |
},
|
146 |
{
|
147 |
"id": 57531,
|
148 |
+
"content": "</s_First>",
|
149 |
"single_word": false,
|
150 |
"lstrip": true,
|
151 |
"rstrip": true,
|
|
|
154 |
},
|
155 |
{
|
156 |
"id": 57532,
|
157 |
+
"content": "<s_Last>",
|
158 |
"single_word": false,
|
159 |
"lstrip": true,
|
160 |
"rstrip": true,
|
|
|
163 |
},
|
164 |
{
|
165 |
"id": 57533,
|
166 |
+
"content": "</s_Last>",
|
167 |
"single_word": false,
|
168 |
"lstrip": true,
|
169 |
"rstrip": true,
|
|
|
172 |
},
|
173 |
{
|
174 |
"id": 57534,
|
175 |
+
"content": "</s_province>",
|
176 |
+
"single_word": false,
|
177 |
+
"lstrip": true,
|
178 |
+
"rstrip": true,
|
179 |
+
"normalized": false,
|
180 |
+
"special": true
|
181 |
+
},
|
182 |
+
{
|
183 |
+
"id": 57535,
|
184 |
+
"content": "<s_Middle>",
|
185 |
+
"single_word": false,
|
186 |
+
"lstrip": true,
|
187 |
+
"rstrip": true,
|
188 |
+
"normalized": false,
|
189 |
+
"special": true
|
190 |
+
},
|
191 |
+
{
|
192 |
+
"id": 57536,
|
193 |
"content": "</s_Middle>",
|
194 |
"single_word": false,
|
195 |
"lstrip": true,
|
tokenizer_config.json
CHANGED
@@ -81,7 +81,7 @@
|
|
81 |
"special": true
|
82 |
},
|
83 |
"57527": {
|
84 |
-
"content": "
|
85 |
"lstrip": true,
|
86 |
"normalized": false,
|
87 |
"rstrip": true,
|
@@ -89,7 +89,7 @@
|
|
89 |
"special": true
|
90 |
},
|
91 |
"57528": {
|
92 |
-
"content": "
|
93 |
"lstrip": true,
|
94 |
"normalized": false,
|
95 |
"rstrip": true,
|
@@ -97,7 +97,7 @@
|
|
97 |
"special": true
|
98 |
},
|
99 |
"57529": {
|
100 |
-
"content": "
|
101 |
"lstrip": true,
|
102 |
"normalized": false,
|
103 |
"rstrip": true,
|
@@ -105,7 +105,7 @@
|
|
105 |
"special": true
|
106 |
},
|
107 |
"57530": {
|
108 |
-
"content": "
|
109 |
"lstrip": true,
|
110 |
"normalized": false,
|
111 |
"rstrip": true,
|
@@ -113,7 +113,7 @@
|
|
113 |
"special": true
|
114 |
},
|
115 |
"57531": {
|
116 |
-
"content": "</
|
117 |
"lstrip": true,
|
118 |
"normalized": false,
|
119 |
"rstrip": true,
|
@@ -121,7 +121,7 @@
|
|
121 |
"special": true
|
122 |
},
|
123 |
"57532": {
|
124 |
-
"content": "
|
125 |
"lstrip": true,
|
126 |
"normalized": false,
|
127 |
"rstrip": true,
|
@@ -129,7 +129,7 @@
|
|
129 |
"special": true
|
130 |
},
|
131 |
"57533": {
|
132 |
-
"content": "
|
133 |
"lstrip": true,
|
134 |
"normalized": false,
|
135 |
"rstrip": true,
|
@@ -137,6 +137,22 @@
|
|
137 |
"special": true
|
138 |
},
|
139 |
"57534": {
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
140 |
"content": "</s_Middle>",
|
141 |
"lstrip": true,
|
142 |
"normalized": false,
|
@@ -151,6 +167,8 @@
|
|
151 |
"</s>",
|
152 |
"<s_province>",
|
153 |
"<s_no>",
|
|
|
|
|
154 |
"<s_First>",
|
155 |
"</s_no>",
|
156 |
"</s_First>",
|
|
|
81 |
"special": true
|
82 |
},
|
83 |
"57527": {
|
84 |
+
"content": "</s_ignore>",
|
85 |
"lstrip": true,
|
86 |
"normalized": false,
|
87 |
"rstrip": true,
|
|
|
89 |
"special": true
|
90 |
},
|
91 |
"57528": {
|
92 |
+
"content": "<s_ignore>",
|
93 |
"lstrip": true,
|
94 |
"normalized": false,
|
95 |
"rstrip": true,
|
|
|
97 |
"special": true
|
98 |
},
|
99 |
"57529": {
|
100 |
+
"content": "<s_First>",
|
101 |
"lstrip": true,
|
102 |
"normalized": false,
|
103 |
"rstrip": true,
|
|
|
105 |
"special": true
|
106 |
},
|
107 |
"57530": {
|
108 |
+
"content": "</s_no>",
|
109 |
"lstrip": true,
|
110 |
"normalized": false,
|
111 |
"rstrip": true,
|
|
|
113 |
"special": true
|
114 |
},
|
115 |
"57531": {
|
116 |
+
"content": "</s_First>",
|
117 |
"lstrip": true,
|
118 |
"normalized": false,
|
119 |
"rstrip": true,
|
|
|
121 |
"special": true
|
122 |
},
|
123 |
"57532": {
|
124 |
+
"content": "<s_Last>",
|
125 |
"lstrip": true,
|
126 |
"normalized": false,
|
127 |
"rstrip": true,
|
|
|
129 |
"special": true
|
130 |
},
|
131 |
"57533": {
|
132 |
+
"content": "</s_Last>",
|
133 |
"lstrip": true,
|
134 |
"normalized": false,
|
135 |
"rstrip": true,
|
|
|
137 |
"special": true
|
138 |
},
|
139 |
"57534": {
|
140 |
+
"content": "</s_province>",
|
141 |
+
"lstrip": true,
|
142 |
+
"normalized": false,
|
143 |
+
"rstrip": true,
|
144 |
+
"single_word": false,
|
145 |
+
"special": true
|
146 |
+
},
|
147 |
+
"57535": {
|
148 |
+
"content": "<s_Middle>",
|
149 |
+
"lstrip": true,
|
150 |
+
"normalized": false,
|
151 |
+
"rstrip": true,
|
152 |
+
"single_word": false,
|
153 |
+
"special": true
|
154 |
+
},
|
155 |
+
"57536": {
|
156 |
"content": "</s_Middle>",
|
157 |
"lstrip": true,
|
158 |
"normalized": false,
|
|
|
167 |
"</s>",
|
168 |
"<s_province>",
|
169 |
"<s_no>",
|
170 |
+
"</s_ignore>",
|
171 |
+
"<s_ignore>",
|
172 |
"<s_First>",
|
173 |
"</s_no>",
|
174 |
"</s_First>",
|