Hugging Face's logo

HueyNemud
/
icdar23-entrydetector_plaintext_breaks_indents_left_ref_right_ref

Token Classification
Transformers
PyTorch
camembert
Generated from Trainer
Inference Endpoints
Model card Files Community
1
HueyNemud commited on
Commit
d10c546
•
1 Parent(s): 5548fa7

End of training

Browse files
Files changed (5) hide show
  1. added_tokens.json +33 -32
  2. config.json +2 -2
  3. pytorch_model.bin +2 -2
  4. tokenizer.json +41 -32
  5. training_args.bin +1 -1
added_tokens.json CHANGED
@@ -1,35 +1,36 @@
1
  {
2
- "</ENTRY>": 32033,
3
- "<ENTRY>": 32021,
4
- "<break>": 32028,
5
- "<hspace-10>": 32006,
6
- "<hspace-1>": 32017,
7
- "<hspace-2>": 32012,
8
- "<hspace-3>": 32014,
9
- "<hspace-4>": 32025,
10
- "<hspace-5>": 32005,
11
  "<hspace-6>": 32026,
12
- "<hspace-7>": 32030,
13
- "<hspace-8>": 32013,
14
- "<hspace-9>": 32015,
15
- "<lhspace-10>": 32009,
16
- "<lhspace-1>": 32010,
17
- "<lhspace-2>": 32036,
18
- "<lhspace-3>": 32034,
19
- "<lhspace-4>": 32032,
20
- "<lhspace-5>": 32029,
21
- "<lhspace-6>": 32022,
22
- "<lhspace-7>": 32008,
23
- "<lhspace-8>": 32020,
24
- "<lhspace-9>": 32023,
25
- "<rhspace-10>": 32027,
26
- "<rhspace-1>": 32011,
27
- "<rhspace-2>": 32024,
28
- "<rhspace-3>": 32007,
29
- "<rhspace-4>": 32019,
30
- "<rhspace-5>": 32035,
31
- "<rhspace-6>": 32031,
32
- "<rhspace-7>": 32016,
33
- "<rhspace-8>": 32018,
34
- "<rhspace-9>": 32037
 
35
  }
 
1
  {
2
+ "</ENTRY>": 32014,
3
+ "<ENTRY>": 32035,
4
+ "<break>": 32036,
5
+ "<hspace-10>": 32030,
6
+ "<hspace-1>": 32038,
7
+ "<hspace-2>": 32020,
8
+ "<hspace-3>": 32009,
9
+ "<hspace-4>": 32021,
10
+ "<hspace-5>": 32015,
11
  "<hspace-6>": 32026,
12
+ "<hspace-7>": 32012,
13
+ "<hspace-8>": 32010,
14
+ "<hspace-9>": 32023,
15
+ "<lhspace-10>": 32024,
16
+ "<lhspace-1>": 32032,
17
+ "<lhspace-2>": 32025,
18
+ "<lhspace-3>": 32027,
19
+ "<lhspace-4>": 32028,
20
+ "<lhspace-5>": 32016,
21
+ "<lhspace-6>": 32008,
22
+ "<lhspace-7>": 32033,
23
+ "<lhspace-8>": 32007,
24
+ "<lhspace-9>": 32018,
25
+ "<rhspace-10>": 32013,
26
+ "<rhspace-1>": 32019,
27
+ "<rhspace-2>": 32017,
28
+ "<rhspace-3>": 32011,
29
+ "<rhspace-4>": 32006,
30
+ "<rhspace-5>": 32031,
31
+ "<rhspace-6>": 32037,
32
+ "<rhspace-7>": 32022,
33
+ "<rhspace-8>": 32029,
34
+ "<rhspace-9>": 32034,
35
+ "<textline>": 32005
36
  }
config.json CHANGED
@@ -48,8 +48,8 @@
48
  "pad_token_id": 1,
49
  "position_embedding_type": "absolute",
50
  "torch_dtype": "float32",
51
- "transformers_version": "4.26.0",
52
  "type_vocab_size": 1,
53
  "use_cache": true,
54
- "vocab_size": 32038
55
  }
 
48
  "pad_token_id": 1,
49
  "position_embedding_type": "absolute",
50
  "torch_dtype": "float32",
51
+ "transformers_version": "4.26.1",
52
  "type_vocab_size": 1,
53
  "use_cache": true,
54
+ "vocab_size": 32039
55
  }
pytorch_model.bin CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:ca56ea79a1304fbfba01a083d6be9c20f4cc5a3efb87ccdd26fbc3379e9feda3
3
- size 440333357
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:3067a1bcd5aa28b70a58178933ac308fb002a16a01d69c522bb36376d8a22fcc
3
+ size 440336429
tokenizer.json CHANGED
@@ -73,7 +73,7 @@
73
  },
74
  {
75
  "id": 32005,
76
- "content": "<hspace-5>",
77
  "single_word": false,
78
  "lstrip": false,
79
  "rstrip": false,
@@ -82,7 +82,7 @@
82
  },
83
  {
84
  "id": 32006,
85
- "content": "<hspace-10>",
86
  "single_word": false,
87
  "lstrip": false,
88
  "rstrip": false,
@@ -91,7 +91,7 @@
91
  },
92
  {
93
  "id": 32007,
94
- "content": "<rhspace-3>",
95
  "single_word": false,
96
  "lstrip": false,
97
  "rstrip": false,
@@ -100,7 +100,7 @@
100
  },
101
  {
102
  "id": 32008,
103
- "content": "<lhspace-7>",
104
  "single_word": false,
105
  "lstrip": false,
106
  "rstrip": false,
@@ -109,7 +109,7 @@
109
  },
110
  {
111
  "id": 32009,
112
- "content": "<lhspace-10>",
113
  "single_word": false,
114
  "lstrip": false,
115
  "rstrip": false,
@@ -118,7 +118,7 @@
118
  },
119
  {
120
  "id": 32010,
121
- "content": "<lhspace-1>",
122
  "single_word": false,
123
  "lstrip": false,
124
  "rstrip": false,
@@ -127,7 +127,7 @@
127
  },
128
  {
129
  "id": 32011,
130
- "content": "<rhspace-1>",
131
  "single_word": false,
132
  "lstrip": false,
133
  "rstrip": false,
@@ -136,7 +136,7 @@
136
  },
137
  {
138
  "id": 32012,
139
- "content": "<hspace-2>",
140
  "single_word": false,
141
  "lstrip": false,
142
  "rstrip": false,
@@ -145,7 +145,7 @@
145
  },
146
  {
147
  "id": 32013,
148
- "content": "<hspace-8>",
149
  "single_word": false,
150
  "lstrip": false,
151
  "rstrip": false,
@@ -154,7 +154,7 @@
154
  },
155
  {
156
  "id": 32014,
157
- "content": "<hspace-3>",
158
  "single_word": false,
159
  "lstrip": false,
160
  "rstrip": false,
@@ -163,7 +163,7 @@
163
  },
164
  {
165
  "id": 32015,
166
- "content": "<hspace-9>",
167
  "single_word": false,
168
  "lstrip": false,
169
  "rstrip": false,
@@ -172,7 +172,7 @@
172
  },
173
  {
174
  "id": 32016,
175
- "content": "<rhspace-7>",
176
  "single_word": false,
177
  "lstrip": false,
178
  "rstrip": false,
@@ -181,7 +181,7 @@
181
  },
182
  {
183
  "id": 32017,
184
- "content": "<hspace-1>",
185
  "single_word": false,
186
  "lstrip": false,
187
  "rstrip": false,
@@ -190,7 +190,7 @@
190
  },
191
  {
192
  "id": 32018,
193
- "content": "<rhspace-8>",
194
  "single_word": false,
195
  "lstrip": false,
196
  "rstrip": false,
@@ -199,7 +199,7 @@
199
  },
200
  {
201
  "id": 32019,
202
- "content": "<rhspace-4>",
203
  "single_word": false,
204
  "lstrip": false,
205
  "rstrip": false,
@@ -208,7 +208,7 @@
208
  },
209
  {
210
  "id": 32020,
211
- "content": "<lhspace-8>",
212
  "single_word": false,
213
  "lstrip": false,
214
  "rstrip": false,
@@ -217,7 +217,7 @@
217
  },
218
  {
219
  "id": 32021,
220
- "content": "<ENTRY>",
221
  "single_word": false,
222
  "lstrip": false,
223
  "rstrip": false,
@@ -226,7 +226,7 @@
226
  },
227
  {
228
  "id": 32022,
229
- "content": "<lhspace-6>",
230
  "single_word": false,
231
  "lstrip": false,
232
  "rstrip": false,
@@ -235,7 +235,7 @@
235
  },
236
  {
237
  "id": 32023,
238
- "content": "<lhspace-9>",
239
  "single_word": false,
240
  "lstrip": false,
241
  "rstrip": false,
@@ -244,7 +244,7 @@
244
  },
245
  {
246
  "id": 32024,
247
- "content": "<rhspace-2>",
248
  "single_word": false,
249
  "lstrip": false,
250
  "rstrip": false,
@@ -253,7 +253,7 @@
253
  },
254
  {
255
  "id": 32025,
256
- "content": "<hspace-4>",
257
  "single_word": false,
258
  "lstrip": false,
259
  "rstrip": false,
@@ -271,7 +271,7 @@
271
  },
272
  {
273
  "id": 32027,
274
- "content": "<rhspace-10>",
275
  "single_word": false,
276
  "lstrip": false,
277
  "rstrip": false,
@@ -280,7 +280,7 @@
280
  },
281
  {
282
  "id": 32028,
283
- "content": "<break>",
284
  "single_word": false,
285
  "lstrip": false,
286
  "rstrip": false,
@@ -289,7 +289,7 @@
289
  },
290
  {
291
  "id": 32029,
292
- "content": "<lhspace-5>",
293
  "single_word": false,
294
  "lstrip": false,
295
  "rstrip": false,
@@ -298,7 +298,7 @@
298
  },
299
  {
300
  "id": 32030,
301
- "content": "<hspace-7>",
302
  "single_word": false,
303
  "lstrip": false,
304
  "rstrip": false,
@@ -307,7 +307,7 @@
307
  },
308
  {
309
  "id": 32031,
310
- "content": "<rhspace-6>",
311
  "single_word": false,
312
  "lstrip": false,
313
  "rstrip": false,
@@ -316,7 +316,7 @@
316
  },
317
  {
318
  "id": 32032,
319
- "content": "<lhspace-4>",
320
  "single_word": false,
321
  "lstrip": false,
322
  "rstrip": false,
@@ -325,7 +325,7 @@
325
  },
326
  {
327
  "id": 32033,
328
- "content": "</ENTRY>",
329
  "single_word": false,
330
  "lstrip": false,
331
  "rstrip": false,
@@ -334,7 +334,7 @@
334
  },
335
  {
336
  "id": 32034,
337
- "content": "<lhspace-3>",
338
  "single_word": false,
339
  "lstrip": false,
340
  "rstrip": false,
@@ -343,7 +343,7 @@
343
  },
344
  {
345
  "id": 32035,
346
- "content": "<rhspace-5>",
347
  "single_word": false,
348
  "lstrip": false,
349
  "rstrip": false,
@@ -352,7 +352,7 @@
352
  },
353
  {
354
  "id": 32036,
355
- "content": "<lhspace-2>",
356
  "single_word": false,
357
  "lstrip": false,
358
  "rstrip": false,
@@ -361,7 +361,16 @@
361
  },
362
  {
363
  "id": 32037,
364
- "content": "<rhspace-9>",
 
 
 
 
 
 
 
 
 
365
  "single_word": false,
366
  "lstrip": false,
367
  "rstrip": false,
 
73
  },
74
  {
75
  "id": 32005,
76
+ "content": "<textline>",
77
  "single_word": false,
78
  "lstrip": false,
79
  "rstrip": false,
 
82
  },
83
  {
84
  "id": 32006,
85
+ "content": "<rhspace-4>",
86
  "single_word": false,
87
  "lstrip": false,
88
  "rstrip": false,
 
91
  },
92
  {
93
  "id": 32007,
94
+ "content": "<lhspace-8>",
95
  "single_word": false,
96
  "lstrip": false,
97
  "rstrip": false,
 
100
  },
101
  {
102
  "id": 32008,
103
+ "content": "<lhspace-6>",
104
  "single_word": false,
105
  "lstrip": false,
106
  "rstrip": false,
 
109
  },
110
  {
111
  "id": 32009,
112
+ "content": "<hspace-3>",
113
  "single_word": false,
114
  "lstrip": false,
115
  "rstrip": false,
 
118
  },
119
  {
120
  "id": 32010,
121
+ "content": "<hspace-8>",
122
  "single_word": false,
123
  "lstrip": false,
124
  "rstrip": false,
 
127
  },
128
  {
129
  "id": 32011,
130
+ "content": "<rhspace-3>",
131
  "single_word": false,
132
  "lstrip": false,
133
  "rstrip": false,
 
136
  },
137
  {
138
  "id": 32012,
139
+ "content": "<hspace-7>",
140
  "single_word": false,
141
  "lstrip": false,
142
  "rstrip": false,
 
145
  },
146
  {
147
  "id": 32013,
148
+ "content": "<rhspace-10>",
149
  "single_word": false,
150
  "lstrip": false,
151
  "rstrip": false,
 
154
  },
155
  {
156
  "id": 32014,
157
+ "content": "</ENTRY>",
158
  "single_word": false,
159
  "lstrip": false,
160
  "rstrip": false,
 
163
  },
164
  {
165
  "id": 32015,
166
+ "content": "<hspace-5>",
167
  "single_word": false,
168
  "lstrip": false,
169
  "rstrip": false,
 
172
  },
173
  {
174
  "id": 32016,
175
+ "content": "<lhspace-5>",
176
  "single_word": false,
177
  "lstrip": false,
178
  "rstrip": false,
 
181
  },
182
  {
183
  "id": 32017,
184
+ "content": "<rhspace-2>",
185
  "single_word": false,
186
  "lstrip": false,
187
  "rstrip": false,
 
190
  },
191
  {
192
  "id": 32018,
193
+ "content": "<lhspace-9>",
194
  "single_word": false,
195
  "lstrip": false,
196
  "rstrip": false,
 
199
  },
200
  {
201
  "id": 32019,
202
+ "content": "<rhspace-1>",
203
  "single_word": false,
204
  "lstrip": false,
205
  "rstrip": false,
 
208
  },
209
  {
210
  "id": 32020,
211
+ "content": "<hspace-2>",
212
  "single_word": false,
213
  "lstrip": false,
214
  "rstrip": false,
 
217
  },
218
  {
219
  "id": 32021,
220
+ "content": "<hspace-4>",
221
  "single_word": false,
222
  "lstrip": false,
223
  "rstrip": false,
 
226
  },
227
  {
228
  "id": 32022,
229
+ "content": "<rhspace-7>",
230
  "single_word": false,
231
  "lstrip": false,
232
  "rstrip": false,
 
235
  },
236
  {
237
  "id": 32023,
238
+ "content": "<hspace-9>",
239
  "single_word": false,
240
  "lstrip": false,
241
  "rstrip": false,
 
244
  },
245
  {
246
  "id": 32024,
247
+ "content": "<lhspace-10>",
248
  "single_word": false,
249
  "lstrip": false,
250
  "rstrip": false,
 
253
  },
254
  {
255
  "id": 32025,
256
+ "content": "<lhspace-2>",
257
  "single_word": false,
258
  "lstrip": false,
259
  "rstrip": false,
 
271
  },
272
  {
273
  "id": 32027,
274
+ "content": "<lhspace-3>",
275
  "single_word": false,
276
  "lstrip": false,
277
  "rstrip": false,
 
280
  },
281
  {
282
  "id": 32028,
283
+ "content": "<lhspace-4>",
284
  "single_word": false,
285
  "lstrip": false,
286
  "rstrip": false,
 
289
  },
290
  {
291
  "id": 32029,
292
+ "content": "<rhspace-8>",
293
  "single_word": false,
294
  "lstrip": false,
295
  "rstrip": false,
 
298
  },
299
  {
300
  "id": 32030,
301
+ "content": "<hspace-10>",
302
  "single_word": false,
303
  "lstrip": false,
304
  "rstrip": false,
 
307
  },
308
  {
309
  "id": 32031,
310
+ "content": "<rhspace-5>",
311
  "single_word": false,
312
  "lstrip": false,
313
  "rstrip": false,
 
316
  },
317
  {
318
  "id": 32032,
319
+ "content": "<lhspace-1>",
320
  "single_word": false,
321
  "lstrip": false,
322
  "rstrip": false,
 
325
  },
326
  {
327
  "id": 32033,
328
+ "content": "<lhspace-7>",
329
  "single_word": false,
330
  "lstrip": false,
331
  "rstrip": false,
 
334
  },
335
  {
336
  "id": 32034,
337
+ "content": "<rhspace-9>",
338
  "single_word": false,
339
  "lstrip": false,
340
  "rstrip": false,
 
343
  },
344
  {
345
  "id": 32035,
346
+ "content": "<ENTRY>",
347
  "single_word": false,
348
  "lstrip": false,
349
  "rstrip": false,
 
352
  },
353
  {
354
  "id": 32036,
355
+ "content": "<break>",
356
  "single_word": false,
357
  "lstrip": false,
358
  "rstrip": false,
 
361
  },
362
  {
363
  "id": 32037,
364
+ "content": "<rhspace-6>",
365
+ "single_word": false,
366
+ "lstrip": false,
367
+ "rstrip": false,
368
+ "normalized": true,
369
+ "special": false
370
+ },
371
+ {
372
+ "id": 32038,
373
+ "content": "<hspace-1>",
374
  "single_word": false,
375
  "lstrip": false,
376
  "rstrip": false,
training_args.bin CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:de43fae90ec0b425d398ec4ecd4c516b176e2be4d5a3e36f38d10f984e213018
3
  size 3515
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:a049de58137daf7f418ab88851d85a9688d766df8bcc27f0b8d9549f00e158d9
3
  size 3515