jszot commited on
Commit
ebbec28
·
verified ·
1 Parent(s): bc2f349

Training in progress, step 420

Browse files
Files changed (3) hide show
  1. config.json +2 -2
  2. model.safetensors +2 -2
  3. tokenizer.json +47 -42
config.json CHANGED
@@ -40,7 +40,7 @@
40
  "tie_word_embeddings": true,
41
  "type_vocab_size": 2,
42
  "use_cache": true,
43
- "vocab_size": 52
44
  },
45
  "decoder_start_token_id": 2,
46
  "dtype": "float32",
@@ -82,7 +82,7 @@
82
  "tie_word_embeddings": true,
83
  "type_vocab_size": 2,
84
  "use_cache": true,
85
- "vocab_size": 52
86
  },
87
  "eos_token_id": 0,
88
  "is_encoder_decoder": true,
 
40
  "tie_word_embeddings": true,
41
  "type_vocab_size": 2,
42
  "use_cache": true,
43
+ "vocab_size": 53
44
  },
45
  "decoder_start_token_id": 2,
46
  "dtype": "float32",
 
82
  "tie_word_embeddings": true,
83
  "type_vocab_size": 2,
84
  "use_cache": true,
85
+ "vocab_size": 53
86
  },
87
  "eos_token_id": 0,
88
  "is_encoder_decoder": true,
model.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:2fb358809a9cfc728c7b289c48e4912bc94a4bd4c4f82f87d92423a518324650
3
- size 31205552
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c662ba763f2c49bad6623f08916e42f466895b21fcced01763ab0bd5e40ef34e
3
+ size 31207604
tokenizer.json CHANGED
@@ -116,39 +116,40 @@
116
  "=": 16,
117
  "99": 17,
118
  "10": 18,
119
- "11": 19,
120
- "98": 20,
121
- "12": 21,
122
- "97": 22,
123
- "13": 23,
124
- "96": 24,
125
  "95": 25,
126
  "14": 26,
127
  "94": 27,
128
  "15": 28,
129
- "16": 29,
130
- "93": 30,
131
  "17": 31,
132
  "92": 32,
133
  "18": 33,
134
  "91": 34,
135
- "19": 35,
136
- "90": 36,
137
- "89": 37,
138
- "20": 38,
139
  "21": 39,
140
  "88": 40,
141
- "22": 41,
142
- "87": 42,
143
- "86": 43,
144
- "23": 44,
145
  "85": 45,
146
  "24": 46,
147
  "25": 47,
148
  "84": 48,
149
- "26": 49,
150
- "83": 50,
151
- "82": 51
 
152
  },
153
  "merges": [
154
  [
@@ -159,17 +160,13 @@
159
  "1",
160
  "0"
161
  ],
162
- [
163
- "1",
164
- "1"
165
- ],
166
  [
167
  "9",
168
  "8"
169
  ],
170
  [
171
  "1",
172
- "2"
173
  ],
174
  [
175
  "9",
@@ -177,12 +174,16 @@
177
  ],
178
  [
179
  "1",
180
- "3"
181
  ],
182
  [
183
  "9",
184
  "6"
185
  ],
 
 
 
 
186
  [
187
  "9",
188
  "5"
@@ -199,14 +200,14 @@
199
  "1",
200
  "5"
201
  ],
202
- [
203
- "1",
204
- "6"
205
- ],
206
  [
207
  "9",
208
  "3"
209
  ],
 
 
 
 
210
  [
211
  "1",
212
  "7"
@@ -223,22 +224,22 @@
223
  "9",
224
  "1"
225
  ],
226
- [
227
- "1",
228
- "9"
229
- ],
230
  [
231
  "9",
232
  "0"
233
  ],
234
  [
235
- "8",
236
  "9"
237
  ],
238
  [
239
  "2",
240
  "0"
241
  ],
 
 
 
 
242
  [
243
  "2",
244
  "1"
@@ -247,22 +248,22 @@
247
  "8",
248
  "8"
249
  ],
250
- [
251
- "2",
252
- "2"
253
- ],
254
  [
255
  "8",
256
  "7"
257
  ],
258
  [
259
- "8",
260
- "6"
261
  ],
262
  [
263
  "2",
264
  "3"
265
  ],
 
 
 
 
266
  [
267
  "8",
268
  "5"
@@ -279,13 +280,17 @@
279
  "8",
280
  "4"
281
  ],
 
 
 
 
282
  [
283
  "2",
284
  "6"
285
  ],
286
  [
287
- "8",
288
- "3"
289
  ],
290
  [
291
  "8",
 
116
  "=": 16,
117
  "99": 17,
118
  "10": 18,
119
+ "98": 19,
120
+ "11": 20,
121
+ "97": 21,
122
+ "12": 22,
123
+ "96": 23,
124
+ "13": 24,
125
  "95": 25,
126
  "14": 26,
127
  "94": 27,
128
  "15": 28,
129
+ "93": 29,
130
+ "16": 30,
131
  "17": 31,
132
  "92": 32,
133
  "18": 33,
134
  "91": 34,
135
+ "90": 35,
136
+ "19": 36,
137
+ "20": 37,
138
+ "89": 38,
139
  "21": 39,
140
  "88": 40,
141
+ "87": 41,
142
+ "22": 42,
143
+ "23": 43,
144
+ "86": 44,
145
  "85": 45,
146
  "24": 46,
147
  "25": 47,
148
  "84": 48,
149
+ "83": 49,
150
+ "26": 50,
151
+ "27": 51,
152
+ "82": 52
153
  },
154
  "merges": [
155
  [
 
160
  "1",
161
  "0"
162
  ],
 
 
 
 
163
  [
164
  "9",
165
  "8"
166
  ],
167
  [
168
  "1",
169
+ "1"
170
  ],
171
  [
172
  "9",
 
174
  ],
175
  [
176
  "1",
177
+ "2"
178
  ],
179
  [
180
  "9",
181
  "6"
182
  ],
183
+ [
184
+ "1",
185
+ "3"
186
+ ],
187
  [
188
  "9",
189
  "5"
 
200
  "1",
201
  "5"
202
  ],
 
 
 
 
203
  [
204
  "9",
205
  "3"
206
  ],
207
+ [
208
+ "1",
209
+ "6"
210
+ ],
211
  [
212
  "1",
213
  "7"
 
224
  "9",
225
  "1"
226
  ],
 
 
 
 
227
  [
228
  "9",
229
  "0"
230
  ],
231
  [
232
+ "1",
233
  "9"
234
  ],
235
  [
236
  "2",
237
  "0"
238
  ],
239
+ [
240
+ "8",
241
+ "9"
242
+ ],
243
  [
244
  "2",
245
  "1"
 
248
  "8",
249
  "8"
250
  ],
 
 
 
 
251
  [
252
  "8",
253
  "7"
254
  ],
255
  [
256
+ "2",
257
+ "2"
258
  ],
259
  [
260
  "2",
261
  "3"
262
  ],
263
+ [
264
+ "8",
265
+ "6"
266
+ ],
267
  [
268
  "8",
269
  "5"
 
280
  "8",
281
  "4"
282
  ],
283
+ [
284
+ "8",
285
+ "3"
286
+ ],
287
  [
288
  "2",
289
  "6"
290
  ],
291
  [
292
+ "2",
293
+ "7"
294
  ],
295
  [
296
  "8",