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
a6d6284
•
1 Parent(s): 2cf603b

End of training

Browse files
Files changed (4) hide show
  1. added_tokens.json +32 -32
  2. pytorch_model.bin +1 -1
  3. tokenizer.json +32 -32
  4. training_args.bin +1 -1
added_tokens.json CHANGED
@@ -1,36 +1,36 @@
1
  {
2
- "</ENTRY>": 32033,
3
- "<ENTRY>": 32025,
4
- "<break>": 32007,
5
- "<hspace-10>": 32022,
6
- "<hspace-1>": 32026,
7
- "<hspace-2>": 32006,
8
  "<hspace-3>": 32014,
9
- "<hspace-4>": 32011,
10
- "<hspace-5>": 32036,
11
- "<hspace-6>": 32005,
12
- "<hspace-7>": 32016,
13
- "<hspace-8>": 32031,
14
- "<hspace-9>": 32037,
15
- "<lhspace-10>": 32019,
16
- "<lhspace-1>": 32012,
17
- "<lhspace-2>": 32034,
18
- "<lhspace-3>": 32008,
19
- "<lhspace-4>": 32032,
20
- "<lhspace-5>": 32010,
21
- "<lhspace-6>": 32029,
22
- "<lhspace-7>": 32015,
23
- "<lhspace-8>": 32035,
24
  "<lhspace-9>": 32024,
25
- "<rhspace-10>": 32023,
26
- "<rhspace-1>": 32028,
27
- "<rhspace-2>": 32021,
28
- "<rhspace-3>": 32009,
29
- "<rhspace-4>": 32013,
30
- "<rhspace-5>": 32038,
31
- "<rhspace-6>": 32017,
32
- "<rhspace-7>": 32018,
33
- "<rhspace-8>": 32020,
34
- "<rhspace-9>": 32027,
35
- "<textline>": 32030
36
  }
 
1
  {
2
+ "</ENTRY>": 32005,
3
+ "<ENTRY>": 32012,
4
+ "<break>": 32035,
5
+ "<hspace-10>": 32010,
6
+ "<hspace-1>": 32027,
7
+ "<hspace-2>": 32029,
8
  "<hspace-3>": 32014,
9
+ "<hspace-4>": 32031,
10
+ "<hspace-5>": 32008,
11
+ "<hspace-6>": 32022,
12
+ "<hspace-7>": 32019,
13
+ "<hspace-8>": 32028,
14
+ "<hspace-9>": 32033,
15
+ "<lhspace-10>": 32009,
16
+ "<lhspace-1>": 32018,
17
+ "<lhspace-2>": 32037,
18
+ "<lhspace-3>": 32006,
19
+ "<lhspace-4>": 32038,
20
+ "<lhspace-5>": 32023,
21
+ "<lhspace-6>": 32011,
22
+ "<lhspace-7>": 32013,
23
+ "<lhspace-8>": 32032,
24
  "<lhspace-9>": 32024,
25
+ "<rhspace-10>": 32030,
26
+ "<rhspace-1>": 32034,
27
+ "<rhspace-2>": 32017,
28
+ "<rhspace-3>": 32016,
29
+ "<rhspace-4>": 32025,
30
+ "<rhspace-5>": 32026,
31
+ "<rhspace-6>": 32015,
32
+ "<rhspace-7>": 32007,
33
+ "<rhspace-8>": 32036,
34
+ "<rhspace-9>": 32020,
35
+ "<textline>": 32021
36
  }
pytorch_model.bin CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:2df5a7a334b6bea086b7c7a0ed513536fc086fa6ed9603604a61d4bd38d2165e
3
  size 440336429
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ac77ab93d2f2d7e7bfae91ac2fd55b91c6d33b310a43b611c366eebef3fb56b2
3
  size 440336429
tokenizer.json CHANGED
@@ -73,7 +73,7 @@
73
  },
74
  {
75
  "id": 32005,
76
- "content": "<hspace-6>",
77
  "single_word": false,
78
  "lstrip": false,
79
  "rstrip": false,
@@ -82,7 +82,7 @@
82
  },
83
  {
84
  "id": 32006,
85
- "content": "<hspace-2>",
86
  "single_word": false,
87
  "lstrip": false,
88
  "rstrip": false,
@@ -91,7 +91,7 @@
91
  },
92
  {
93
  "id": 32007,
94
- "content": "<break>",
95
  "single_word": false,
96
  "lstrip": false,
97
  "rstrip": false,
@@ -100,7 +100,7 @@
100
  },
101
  {
102
  "id": 32008,
103
- "content": "<lhspace-3>",
104
  "single_word": false,
105
  "lstrip": false,
106
  "rstrip": false,
@@ -109,7 +109,7 @@
109
  },
110
  {
111
  "id": 32009,
112
- "content": "<rhspace-3>",
113
  "single_word": false,
114
  "lstrip": false,
115
  "rstrip": false,
@@ -118,7 +118,7 @@
118
  },
119
  {
120
  "id": 32010,
121
- "content": "<lhspace-5>",
122
  "single_word": false,
123
  "lstrip": false,
124
  "rstrip": false,
@@ -127,7 +127,7 @@
127
  },
128
  {
129
  "id": 32011,
130
- "content": "<hspace-4>",
131
  "single_word": false,
132
  "lstrip": false,
133
  "rstrip": false,
@@ -136,7 +136,7 @@
136
  },
137
  {
138
  "id": 32012,
139
- "content": "<lhspace-1>",
140
  "single_word": false,
141
  "lstrip": false,
142
  "rstrip": false,
@@ -145,7 +145,7 @@
145
  },
146
  {
147
  "id": 32013,
148
- "content": "<rhspace-4>",
149
  "single_word": false,
150
  "lstrip": false,
151
  "rstrip": false,
@@ -163,7 +163,7 @@
163
  },
164
  {
165
  "id": 32015,
166
- "content": "<lhspace-7>",
167
  "single_word": false,
168
  "lstrip": false,
169
  "rstrip": false,
@@ -172,7 +172,7 @@
172
  },
173
  {
174
  "id": 32016,
175
- "content": "<hspace-7>",
176
  "single_word": false,
177
  "lstrip": false,
178
  "rstrip": false,
@@ -181,7 +181,7 @@
181
  },
182
  {
183
  "id": 32017,
184
- "content": "<rhspace-6>",
185
  "single_word": false,
186
  "lstrip": false,
187
  "rstrip": false,
@@ -190,7 +190,7 @@
190
  },
191
  {
192
  "id": 32018,
193
- "content": "<rhspace-7>",
194
  "single_word": false,
195
  "lstrip": false,
196
  "rstrip": false,
@@ -199,7 +199,7 @@
199
  },
200
  {
201
  "id": 32019,
202
- "content": "<lhspace-10>",
203
  "single_word": false,
204
  "lstrip": false,
205
  "rstrip": false,
@@ -208,7 +208,7 @@
208
  },
209
  {
210
  "id": 32020,
211
- "content": "<rhspace-8>",
212
  "single_word": false,
213
  "lstrip": false,
214
  "rstrip": false,
@@ -217,7 +217,7 @@
217
  },
218
  {
219
  "id": 32021,
220
- "content": "<rhspace-2>",
221
  "single_word": false,
222
  "lstrip": false,
223
  "rstrip": false,
@@ -226,7 +226,7 @@
226
  },
227
  {
228
  "id": 32022,
229
- "content": "<hspace-10>",
230
  "single_word": false,
231
  "lstrip": false,
232
  "rstrip": false,
@@ -235,7 +235,7 @@
235
  },
236
  {
237
  "id": 32023,
238
- "content": "<rhspace-10>",
239
  "single_word": false,
240
  "lstrip": false,
241
  "rstrip": false,
@@ -253,7 +253,7 @@
253
  },
254
  {
255
  "id": 32025,
256
- "content": "<ENTRY>",
257
  "single_word": false,
258
  "lstrip": false,
259
  "rstrip": false,
@@ -262,7 +262,7 @@
262
  },
263
  {
264
  "id": 32026,
265
- "content": "<hspace-1>",
266
  "single_word": false,
267
  "lstrip": false,
268
  "rstrip": false,
@@ -271,7 +271,7 @@
271
  },
272
  {
273
  "id": 32027,
274
- "content": "<rhspace-9>",
275
  "single_word": false,
276
  "lstrip": false,
277
  "rstrip": false,
@@ -280,7 +280,7 @@
280
  },
281
  {
282
  "id": 32028,
283
- "content": "<rhspace-1>",
284
  "single_word": false,
285
  "lstrip": false,
286
  "rstrip": false,
@@ -289,7 +289,7 @@
289
  },
290
  {
291
  "id": 32029,
292
- "content": "<lhspace-6>",
293
  "single_word": false,
294
  "lstrip": false,
295
  "rstrip": false,
@@ -298,7 +298,7 @@
298
  },
299
  {
300
  "id": 32030,
301
- "content": "<textline>",
302
  "single_word": false,
303
  "lstrip": false,
304
  "rstrip": false,
@@ -307,7 +307,7 @@
307
  },
308
  {
309
  "id": 32031,
310
- "content": "<hspace-8>",
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-2>",
338
  "single_word": false,
339
  "lstrip": false,
340
  "rstrip": false,
@@ -343,7 +343,7 @@
343
  },
344
  {
345
  "id": 32035,
346
- "content": "<lhspace-8>",
347
  "single_word": false,
348
  "lstrip": false,
349
  "rstrip": false,
@@ -352,7 +352,7 @@
352
  },
353
  {
354
  "id": 32036,
355
- "content": "<hspace-5>",
356
  "single_word": false,
357
  "lstrip": false,
358
  "rstrip": false,
@@ -361,7 +361,7 @@
361
  },
362
  {
363
  "id": 32037,
364
- "content": "<hspace-9>",
365
  "single_word": false,
366
  "lstrip": false,
367
  "rstrip": false,
@@ -370,7 +370,7 @@
370
  },
371
  {
372
  "id": 32038,
373
- "content": "<rhspace-5>",
374
  "single_word": false,
375
  "lstrip": false,
376
  "rstrip": false,
 
73
  },
74
  {
75
  "id": 32005,
76
+ "content": "</ENTRY>",
77
  "single_word": false,
78
  "lstrip": false,
79
  "rstrip": false,
 
82
  },
83
  {
84
  "id": 32006,
85
+ "content": "<lhspace-3>",
86
  "single_word": false,
87
  "lstrip": false,
88
  "rstrip": false,
 
91
  },
92
  {
93
  "id": 32007,
94
+ "content": "<rhspace-7>",
95
  "single_word": false,
96
  "lstrip": false,
97
  "rstrip": false,
 
100
  },
101
  {
102
  "id": 32008,
103
+ "content": "<hspace-5>",
104
  "single_word": false,
105
  "lstrip": false,
106
  "rstrip": false,
 
109
  },
110
  {
111
  "id": 32009,
112
+ "content": "<lhspace-10>",
113
  "single_word": false,
114
  "lstrip": false,
115
  "rstrip": false,
 
118
  },
119
  {
120
  "id": 32010,
121
+ "content": "<hspace-10>",
122
  "single_word": false,
123
  "lstrip": false,
124
  "rstrip": false,
 
127
  },
128
  {
129
  "id": 32011,
130
+ "content": "<lhspace-6>",
131
  "single_word": false,
132
  "lstrip": false,
133
  "rstrip": false,
 
136
  },
137
  {
138
  "id": 32012,
139
+ "content": "<ENTRY>",
140
  "single_word": false,
141
  "lstrip": false,
142
  "rstrip": false,
 
145
  },
146
  {
147
  "id": 32013,
148
+ "content": "<lhspace-7>",
149
  "single_word": false,
150
  "lstrip": false,
151
  "rstrip": false,
 
163
  },
164
  {
165
  "id": 32015,
166
+ "content": "<rhspace-6>",
167
  "single_word": false,
168
  "lstrip": false,
169
  "rstrip": false,
 
172
  },
173
  {
174
  "id": 32016,
175
+ "content": "<rhspace-3>",
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-1>",
194
  "single_word": false,
195
  "lstrip": false,
196
  "rstrip": false,
 
199
  },
200
  {
201
  "id": 32019,
202
+ "content": "<hspace-7>",
203
  "single_word": false,
204
  "lstrip": false,
205
  "rstrip": false,
 
208
  },
209
  {
210
  "id": 32020,
211
+ "content": "<rhspace-9>",
212
  "single_word": false,
213
  "lstrip": false,
214
  "rstrip": false,
 
217
  },
218
  {
219
  "id": 32021,
220
+ "content": "<textline>",
221
  "single_word": false,
222
  "lstrip": false,
223
  "rstrip": false,
 
226
  },
227
  {
228
  "id": 32022,
229
+ "content": "<hspace-6>",
230
  "single_word": false,
231
  "lstrip": false,
232
  "rstrip": false,
 
235
  },
236
  {
237
  "id": 32023,
238
+ "content": "<lhspace-5>",
239
  "single_word": false,
240
  "lstrip": false,
241
  "rstrip": false,
 
253
  },
254
  {
255
  "id": 32025,
256
+ "content": "<rhspace-4>",
257
  "single_word": false,
258
  "lstrip": false,
259
  "rstrip": false,
 
262
  },
263
  {
264
  "id": 32026,
265
+ "content": "<rhspace-5>",
266
  "single_word": false,
267
  "lstrip": false,
268
  "rstrip": false,
 
271
  },
272
  {
273
  "id": 32027,
274
+ "content": "<hspace-1>",
275
  "single_word": false,
276
  "lstrip": false,
277
  "rstrip": false,
 
280
  },
281
  {
282
  "id": 32028,
283
+ "content": "<hspace-8>",
284
  "single_word": false,
285
  "lstrip": false,
286
  "rstrip": false,
 
289
  },
290
  {
291
  "id": 32029,
292
+ "content": "<hspace-2>",
293
  "single_word": false,
294
  "lstrip": false,
295
  "rstrip": false,
 
298
  },
299
  {
300
  "id": 32030,
301
+ "content": "<rhspace-10>",
302
  "single_word": false,
303
  "lstrip": false,
304
  "rstrip": false,
 
307
  },
308
  {
309
  "id": 32031,
310
+ "content": "<hspace-4>",
311
  "single_word": false,
312
  "lstrip": false,
313
  "rstrip": false,
 
316
  },
317
  {
318
  "id": 32032,
319
+ "content": "<lhspace-8>",
320
  "single_word": false,
321
  "lstrip": false,
322
  "rstrip": false,
 
325
  },
326
  {
327
  "id": 32033,
328
+ "content": "<hspace-9>",
329
  "single_word": false,
330
  "lstrip": false,
331
  "rstrip": false,
 
334
  },
335
  {
336
  "id": 32034,
337
+ "content": "<rhspace-1>",
338
  "single_word": false,
339
  "lstrip": false,
340
  "rstrip": false,
 
343
  },
344
  {
345
  "id": 32035,
346
+ "content": "<break>",
347
  "single_word": false,
348
  "lstrip": false,
349
  "rstrip": false,
 
352
  },
353
  {
354
  "id": 32036,
355
+ "content": "<rhspace-8>",
356
  "single_word": false,
357
  "lstrip": false,
358
  "rstrip": false,
 
361
  },
362
  {
363
  "id": 32037,
364
+ "content": "<lhspace-2>",
365
  "single_word": false,
366
  "lstrip": false,
367
  "rstrip": false,
 
370
  },
371
  {
372
  "id": 32038,
373
+ "content": "<lhspace-4>",
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 sha256:959421a1290c4504f6eadef91a4229c3c841497cc0a8ad1c5d3b7b1f045501e9
3
  size 3515
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:cf1eb3bd6e18f5ec2f847400b99a3ef046ac0e963b9fbb78878dc5d7bbfb4576
3
  size 3515