Hugging Face's logo

nlpso
/
m0_flat_ner_ref_cmbert_io

Token Classification
Transformers
PyTorch
French
camembert
Inference Endpoints
Model card Files Community
1
nlpso commited on
Commit
e44045f
·
1 Parent(s): ef59781

Load model and tokenizer

Browse files
Files changed (11) hide show
  1. config.json +47 -0
  2. optimizer.pt +3 -0
  3. pytorch_model.bin +3 -0
  4. rng_state.pth +3 -0
  5. scheduler.pt +3 -0
  6. sentencepiece.bpe.model +3 -0
  7. special_tokens_map.json +13 -0
  8. tokenizer.json +0 -0
  9. tokenizer_config.json +25 -0
  10. trainer_state.json +604 -0
  11. training_args.bin +3 -0
config.json ADDED
@@ -0,0 +1,47 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_name_or_path": "Jean-Baptiste/camembert-ner",
3
+ "architectures": [
4
+ "CamembertForTokenClassification"
5
+ ],
6
+ "attention_probs_dropout_prob": 0.1,
7
+ "bos_token_id": 5,
8
+ "classifier_dropout": null,
9
+ "eos_token_id": 6,
10
+ "gradient_checkpointing": false,
11
+ "hidden_act": "gelu",
12
+ "hidden_dropout_prob": 0.1,
13
+ "hidden_size": 768,
14
+ "id2label": {
15
+ "0": "O",
16
+ "1": "I-PER",
17
+ "2": "I-TITRE",
18
+ "3": "I-ACT",
19
+ "4": "I-LOC",
20
+ "5": "I-CARDINAL",
21
+ "6": "I-FT"
22
+ },
23
+ "initializer_range": 0.02,
24
+ "intermediate_size": 3072,
25
+ "label2id": {
26
+ "I-ACT": 3,
27
+ "I-CARDINAL": 5,
28
+ "I-FT": 6,
29
+ "I-LOC": 4,
30
+ "I-PER": 1,
31
+ "I-TITRE": 2,
32
+ "O": 0
33
+ },
34
+ "layer_norm_eps": 1e-05,
35
+ "max_position_embeddings": 514,
36
+ "model_type": "camembert",
37
+ "num_attention_heads": 12,
38
+ "num_hidden_layers": 12,
39
+ "output_past": true,
40
+ "pad_token_id": 1,
41
+ "position_embedding_type": "absolute",
42
+ "torch_dtype": "float32",
43
+ "transformers_version": "4.25.1",
44
+ "type_vocab_size": 1,
45
+ "use_cache": true,
46
+ "vocab_size": 32005
47
+ }
optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:7ebf4b3810d016aca7ea35be2a7e7598778b92b45149084928092bb99b71bc2a
3
+ size 880409317
pytorch_model.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:225f4728a501e8d8143aff21d07d89191d4d82d309b215ba76dc8786cb898a91
3
+ size 440218033
rng_state.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:10e0deb8984910703030ca3f179066fb3f496f3fa4c860d5804407f8c72066f2
3
+ size 14503
scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:48f2a75f3ed3f16767915866d392dbc6d954dd1374475cfd20580f7c483a9e55
3
+ size 623
sentencepiece.bpe.model ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:988bc5a00281c6d210a5d34bd143d0363741a432fefe741bf71e61b1869d4314
3
+ size 810912
special_tokens_map.json ADDED
@@ -0,0 +1,13 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "additional_special_tokens": [
3
+ "<s>NOTUSED",
4
+ "</s>NOTUSED"
5
+ ],
6
+ "bos_token": "<s>",
7
+ "cls_token": "<s>",
8
+ "eos_token": "</s>",
9
+ "mask_token": "<mask>",
10
+ "pad_token": "<pad>",
11
+ "sep_token": "</s>",
12
+ "unk_token": "<unk>"
13
+ }
tokenizer.json ADDED
The diff for this file is too large to render. See raw diff
 
tokenizer_config.json ADDED
@@ -0,0 +1,25 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "additional_special_tokens": [
3
+ "<s>NOTUSED",
4
+ "</s>NOTUSED"
5
+ ],
6
+ "bos_token": "<s>",
7
+ "cls_token": "<s>",
8
+ "eos_token": "</s>",
9
+ "mask_token": {
10
+ "__type": "AddedToken",
11
+ "content": "<mask>",
12
+ "lstrip": true,
13
+ "normalized": true,
14
+ "rstrip": false,
15
+ "single_word": false
16
+ },
17
+ "model_max_length": 512,
18
+ "name_or_path": "Jean-Baptiste/camembert-ner",
19
+ "pad_token": "<pad>",
20
+ "sep_token": "</s>",
21
+ "sp_model_kwargs": {},
22
+ "special_tokens_map_file": "/lrde/home2/stual/.cache/huggingface/hub/models--Jean-Baptiste--camembert-ner/snapshots/9f8b2203b6a2daba1ad279ac2adb822518caa167/special_tokens_map.json",
23
+ "tokenizer_class": "CamembertTokenizer",
24
+ "unk_token": "<unk>"
25
+ }
trainer_state.json ADDED
@@ -0,0 +1,604 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": 0.9778851514588367,
3
+ "best_model_checkpoint": "/work/stual/res_ICDAR/method_0/tmp/21-flat-ner-ref-camembert_ner/checkpoint-1200",
4
+ "epoch": 3.1496062992125986,
5
+ "global_step": 1200,
6
+ "is_hyper_param_search": false,
7
+ "is_local_process_zero": true,
8
+ "is_world_process_zero": true,
9
+ "log_history": [
10
+ {
11
+ "epoch": 0.26,
12
+ "eval_ACT": {
13
+ "f1": 0.9104618284637135,
14
+ "number": 519,
15
+ "precision": 0.8911439114391144,
16
+ "recall": 0.930635838150289
17
+ },
18
+ "eval_CARDINAL": {
19
+ "f1": 0.9919058130978661,
20
+ "number": 678,
21
+ "precision": 0.9897209985315712,
22
+ "recall": 0.9941002949852508
23
+ },
24
+ "eval_FT": {
25
+ "f1": 0.0,
26
+ "number": 7,
27
+ "precision": 0.0,
28
+ "recall": 0.0
29
+ },
30
+ "eval_LOC": {
31
+ "f1": 0.939820742637644,
32
+ "number": 761,
33
+ "precision": 0.916354556803995,
34
+ "recall": 0.9645203679369251
35
+ },
36
+ "eval_PER": {
37
+ "f1": 0.9195906432748537,
38
+ "number": 679,
39
+ "precision": 0.9129172714078374,
40
+ "recall": 0.9263622974963182
41
+ },
42
+ "eval_TITRE": {
43
+ "f1": 0.0,
44
+ "number": 40,
45
+ "precision": 0.0,
46
+ "recall": 0.0
47
+ },
48
+ "eval_accuracy": 0.9574230387288977,
49
+ "eval_f1": 0.933852140077821,
50
+ "eval_loss": 0.20205183327198029,
51
+ "eval_precision": 0.9288610394397346,
52
+ "eval_recall": 0.9388971684053651,
53
+ "eval_runtime": 2.1249,
54
+ "eval_samples_per_second": 318.138,
55
+ "eval_steps_per_second": 20.237,
56
+ "step": 100
57
+ },
58
+ {
59
+ "epoch": 0.52,
60
+ "eval_ACT": {
61
+ "f1": 0.9073900841908326,
62
+ "number": 519,
63
+ "precision": 0.8818181818181818,
64
+ "recall": 0.9344894026974951
65
+ },
66
+ "eval_CARDINAL": {
67
+ "f1": 0.9911764705882352,
68
+ "number": 678,
69
+ "precision": 0.9882697947214076,
70
+ "recall": 0.9941002949852508
71
+ },
72
+ "eval_FT": {
73
+ "f1": 0.0,
74
+ "number": 7,
75
+ "precision": 0.0,
76
+ "recall": 0.0
77
+ },
78
+ "eval_LOC": {
79
+ "f1": 0.943058221369162,
80
+ "number": 761,
81
+ "precision": 0.9189526184538653,
82
+ "recall": 0.9684625492772667
83
+ },
84
+ "eval_PER": {
85
+ "f1": 0.9545454545454546,
86
+ "number": 679,
87
+ "precision": 0.9503649635036496,
88
+ "recall": 0.9587628865979382
89
+ },
90
+ "eval_TITRE": {
91
+ "f1": 0.53125,
92
+ "number": 40,
93
+ "precision": 0.7083333333333334,
94
+ "recall": 0.425
95
+ },
96
+ "eval_accuracy": 0.961643495531281,
97
+ "eval_f1": 0.9449051041090842,
98
+ "eval_loss": 0.16001582145690918,
99
+ "eval_precision": 0.9347429821363471,
100
+ "eval_recall": 0.9552906110283159,
101
+ "eval_runtime": 2.1373,
102
+ "eval_samples_per_second": 316.284,
103
+ "eval_steps_per_second": 20.119,
104
+ "step": 200
105
+ },
106
+ {
107
+ "epoch": 0.79,
108
+ "eval_ACT": {
109
+ "f1": 0.9244019138755981,
110
+ "number": 519,
111
+ "precision": 0.9182509505703422,
112
+ "recall": 0.930635838150289
113
+ },
114
+ "eval_CARDINAL": {
115
+ "f1": 0.9926362297496317,
116
+ "number": 678,
117
+ "precision": 0.9911764705882353,
118
+ "recall": 0.9941002949852508
119
+ },
120
+ "eval_FT": {
121
+ "f1": 0.0,
122
+ "number": 7,
123
+ "precision": 0.0,
124
+ "recall": 0.0
125
+ },
126
+ "eval_LOC": {
127
+ "f1": 0.9435695538057742,
128
+ "number": 761,
129
+ "precision": 0.9423328964613368,
130
+ "recall": 0.9448094612352168
131
+ },
132
+ "eval_PER": {
133
+ "f1": 0.979381443298969,
134
+ "number": 679,
135
+ "precision": 0.979381443298969,
136
+ "recall": 0.979381443298969
137
+ },
138
+ "eval_TITRE": {
139
+ "f1": 0.8684210526315789,
140
+ "number": 40,
141
+ "precision": 0.9166666666666666,
142
+ "recall": 0.825
143
+ },
144
+ "eval_accuracy": 0.9690913604766633,
145
+ "eval_f1": 0.9590163934426229,
146
+ "eval_loss": 0.1148175373673439,
147
+ "eval_precision": 0.9590163934426229,
148
+ "eval_recall": 0.9590163934426229,
149
+ "eval_runtime": 2.1137,
150
+ "eval_samples_per_second": 319.821,
151
+ "eval_steps_per_second": 20.344,
152
+ "step": 300
153
+ },
154
+ {
155
+ "epoch": 1.05,
156
+ "eval_ACT": {
157
+ "f1": 0.9422894985808893,
158
+ "number": 519,
159
+ "precision": 0.9256505576208178,
160
+ "recall": 0.9595375722543352
161
+ },
162
+ "eval_CARDINAL": {
163
+ "f1": 0.9941089837997055,
164
+ "number": 678,
165
+ "precision": 0.9926470588235294,
166
+ "recall": 0.995575221238938
167
+ },
168
+ "eval_FT": {
169
+ "f1": 0.0,
170
+ "number": 7,
171
+ "precision": 0.0,
172
+ "recall": 0.0
173
+ },
174
+ "eval_LOC": {
175
+ "f1": 0.9652002626395274,
176
+ "number": 761,
177
+ "precision": 0.9645669291338582,
178
+ "recall": 0.9658344283837057
179
+ },
180
+ "eval_PER": {
181
+ "f1": 0.986784140969163,
182
+ "number": 679,
183
+ "precision": 0.9838945827232797,
184
+ "recall": 0.9896907216494846
185
+ },
186
+ "eval_TITRE": {
187
+ "f1": 0.8433734939759036,
188
+ "number": 40,
189
+ "precision": 0.813953488372093,
190
+ "recall": 0.875
191
+ },
192
+ "eval_accuracy": 0.977780536246276,
193
+ "eval_f1": 0.9703153988868275,
194
+ "eval_loss": 0.08161204308271408,
195
+ "eval_precision": 0.9663710273466372,
196
+ "eval_recall": 0.9742921013412816,
197
+ "eval_runtime": 2.1472,
198
+ "eval_samples_per_second": 314.828,
199
+ "eval_steps_per_second": 20.026,
200
+ "step": 400
201
+ },
202
+ {
203
+ "epoch": 1.31,
204
+ "learning_rate": 9e-05,
205
+ "loss": 0.2031,
206
+ "step": 500
207
+ },
208
+ {
209
+ "epoch": 1.31,
210
+ "eval_ACT": {
211
+ "f1": 0.9473684210526315,
212
+ "number": 519,
213
+ "precision": 0.94106463878327,
214
+ "recall": 0.953757225433526
215
+ },
216
+ "eval_CARDINAL": {
217
+ "f1": 0.9911634756995582,
218
+ "number": 678,
219
+ "precision": 0.9897058823529412,
220
+ "recall": 0.9926253687315634
221
+ },
222
+ "eval_FT": {
223
+ "f1": 0.0,
224
+ "number": 7,
225
+ "precision": 0.0,
226
+ "recall": 0.0
227
+ },
228
+ "eval_LOC": {
229
+ "f1": 0.9559014267185473,
230
+ "number": 761,
231
+ "precision": 0.9436619718309859,
232
+ "recall": 0.9684625492772667
233
+ },
234
+ "eval_PER": {
235
+ "f1": 0.9730517115804806,
236
+ "number": 679,
237
+ "precision": 0.962536023054755,
238
+ "recall": 0.9837997054491899
239
+ },
240
+ "eval_TITRE": {
241
+ "f1": 0.8536585365853658,
242
+ "number": 40,
243
+ "precision": 0.8333333333333334,
244
+ "recall": 0.875
245
+ },
246
+ "eval_accuracy": 0.9728152929493545,
247
+ "eval_f1": 0.9646754207508785,
248
+ "eval_loss": 0.10748306661844254,
249
+ "eval_precision": 0.9577671685640837,
250
+ "eval_recall": 0.9716840536512668,
251
+ "eval_runtime": 2.1738,
252
+ "eval_samples_per_second": 310.97,
253
+ "eval_steps_per_second": 19.781,
254
+ "step": 500
255
+ },
256
+ {
257
+ "epoch": 1.57,
258
+ "eval_ACT": {
259
+ "f1": 0.9438629876308278,
260
+ "number": 519,
261
+ "precision": 0.9323308270676691,
262
+ "recall": 0.9556840077071291
263
+ },
264
+ "eval_CARDINAL": {
265
+ "f1": 0.9919177075679647,
266
+ "number": 678,
267
+ "precision": 0.9882869692532943,
268
+ "recall": 0.995575221238938
269
+ },
270
+ "eval_FT": {
271
+ "f1": 0.0,
272
+ "number": 7,
273
+ "precision": 0.0,
274
+ "recall": 0.0
275
+ },
276
+ "eval_LOC": {
277
+ "f1": 0.9646596858638743,
278
+ "number": 761,
279
+ "precision": 0.9608865710560626,
280
+ "recall": 0.9684625492772667
281
+ },
282
+ "eval_PER": {
283
+ "f1": 0.9816311535635561,
284
+ "number": 679,
285
+ "precision": 0.9794721407624634,
286
+ "recall": 0.9837997054491899
287
+ },
288
+ "eval_TITRE": {
289
+ "f1": 0.8533333333333333,
290
+ "number": 40,
291
+ "precision": 0.9142857142857143,
292
+ "recall": 0.8
293
+ },
294
+ "eval_accuracy": 0.977780536246276,
295
+ "eval_f1": 0.9689764072078766,
296
+ "eval_loss": 0.08254563808441162,
297
+ "eval_precision": 0.9662838088180807,
298
+ "eval_recall": 0.9716840536512668,
299
+ "eval_runtime": 3.17,
300
+ "eval_samples_per_second": 213.252,
301
+ "eval_steps_per_second": 13.565,
302
+ "step": 600
303
+ },
304
+ {
305
+ "epoch": 1.84,
306
+ "eval_ACT": {
307
+ "f1": 0.958930276981853,
308
+ "number": 519,
309
+ "precision": 0.9507575757575758,
310
+ "recall": 0.9672447013487476
311
+ },
312
+ "eval_CARDINAL": {
313
+ "f1": 0.9919058130978661,
314
+ "number": 678,
315
+ "precision": 0.9897209985315712,
316
+ "recall": 0.9941002949852508
317
+ },
318
+ "eval_FT": {
319
+ "f1": 0.0,
320
+ "number": 7,
321
+ "precision": 0.0,
322
+ "recall": 0.0
323
+ },
324
+ "eval_LOC": {
325
+ "f1": 0.9653368214519294,
326
+ "number": 761,
327
+ "precision": 0.9609375,
328
+ "recall": 0.9697766097240473
329
+ },
330
+ "eval_PER": {
331
+ "f1": 0.9801616458486406,
332
+ "number": 679,
333
+ "precision": 0.9780058651026393,
334
+ "recall": 0.9823269513991163
335
+ },
336
+ "eval_TITRE": {
337
+ "f1": 0.7999999999999998,
338
+ "number": 40,
339
+ "precision": 0.7555555555555555,
340
+ "recall": 0.85
341
+ },
342
+ "eval_accuracy": 0.9781529294935452,
343
+ "eval_f1": 0.9706755753526355,
344
+ "eval_loss": 0.07011305540800095,
345
+ "eval_precision": 0.9670857988165681,
346
+ "eval_recall": 0.9742921013412816,
347
+ "eval_runtime": 2.1365,
348
+ "eval_samples_per_second": 316.402,
349
+ "eval_steps_per_second": 20.126,
350
+ "step": 700
351
+ },
352
+ {
353
+ "epoch": 2.1,
354
+ "eval_ACT": {
355
+ "f1": 0.9465648854961832,
356
+ "number": 519,
357
+ "precision": 0.9376181474480151,
358
+ "recall": 0.9556840077071291
359
+ },
360
+ "eval_CARDINAL": {
361
+ "f1": 0.9919058130978661,
362
+ "number": 678,
363
+ "precision": 0.9897209985315712,
364
+ "recall": 0.9941002949852508
365
+ },
366
+ "eval_FT": {
367
+ "f1": 0.0,
368
+ "number": 7,
369
+ "precision": 0.0,
370
+ "recall": 0.0
371
+ },
372
+ "eval_LOC": {
373
+ "f1": 0.9744931327665142,
374
+ "number": 761,
375
+ "precision": 0.9700520833333334,
376
+ "recall": 0.9789750328515112
377
+ },
378
+ "eval_PER": {
379
+ "f1": 0.9714285714285714,
380
+ "number": 679,
381
+ "precision": 0.9664723032069971,
382
+ "recall": 0.9764359351988218
383
+ },
384
+ "eval_TITRE": {
385
+ "f1": 0.8705882352941177,
386
+ "number": 40,
387
+ "precision": 0.8222222222222222,
388
+ "recall": 0.925
389
+ },
390
+ "eval_accuracy": 0.9767874875868917,
391
+ "eval_f1": 0.9697756350825143,
392
+ "eval_loss": 0.09748248755931854,
393
+ "eval_precision": 0.9653008490217793,
394
+ "eval_recall": 0.9742921013412816,
395
+ "eval_runtime": 2.1373,
396
+ "eval_samples_per_second": 316.283,
397
+ "eval_steps_per_second": 20.119,
398
+ "step": 800
399
+ },
400
+ {
401
+ "epoch": 2.36,
402
+ "eval_ACT": {
403
+ "f1": 0.9579349904397705,
404
+ "number": 519,
405
+ "precision": 0.9506641366223909,
406
+ "recall": 0.9653179190751445
407
+ },
408
+ "eval_CARDINAL": {
409
+ "f1": 0.9933774834437086,
410
+ "number": 678,
411
+ "precision": 0.9911894273127754,
412
+ "recall": 0.995575221238938
413
+ },
414
+ "eval_FT": {
415
+ "f1": 0.36363636363636365,
416
+ "number": 7,
417
+ "precision": 0.5,
418
+ "recall": 0.2857142857142857
419
+ },
420
+ "eval_LOC": {
421
+ "f1": 0.958957654723127,
422
+ "number": 761,
423
+ "precision": 0.9509043927648578,
424
+ "recall": 0.9671484888304862
425
+ },
426
+ "eval_PER": {
427
+ "f1": 0.9809663250366032,
428
+ "number": 679,
429
+ "precision": 0.975254730713246,
430
+ "recall": 0.9867452135493373
431
+ },
432
+ "eval_TITRE": {
433
+ "f1": 0.9135802469135802,
434
+ "number": 40,
435
+ "precision": 0.9024390243902439,
436
+ "recall": 0.925
437
+ },
438
+ "eval_accuracy": 0.9793942403177756,
439
+ "eval_f1": 0.9711004075583549,
440
+ "eval_loss": 0.07152657955884933,
441
+ "eval_precision": 0.9657332350773765,
442
+ "eval_recall": 0.9765275707898659,
443
+ "eval_runtime": 2.2181,
444
+ "eval_samples_per_second": 304.768,
445
+ "eval_steps_per_second": 19.386,
446
+ "step": 900
447
+ },
448
+ {
449
+ "epoch": 2.62,
450
+ "learning_rate": 8e-05,
451
+ "loss": 0.0707,
452
+ "step": 1000
453
+ },
454
+ {
455
+ "epoch": 2.62,
456
+ "eval_ACT": {
457
+ "f1": 0.9502868068833653,
458
+ "number": 519,
459
+ "precision": 0.9430740037950665,
460
+ "recall": 0.9576107899807321
461
+ },
462
+ "eval_CARDINAL": {
463
+ "f1": 0.9918938835666913,
464
+ "number": 678,
465
+ "precision": 0.9911634756995582,
466
+ "recall": 0.9926253687315634
467
+ },
468
+ "eval_FT": {
469
+ "f1": 0.0,
470
+ "number": 7,
471
+ "precision": 0.0,
472
+ "recall": 0.0
473
+ },
474
+ "eval_LOC": {
475
+ "f1": 0.9687499999999999,
476
+ "number": 761,
477
+ "precision": 0.96,
478
+ "recall": 0.9776609724047306
479
+ },
480
+ "eval_PER": {
481
+ "f1": 0.9889624724061811,
482
+ "number": 679,
483
+ "precision": 0.9882352941176471,
484
+ "recall": 0.9896907216494846
485
+ },
486
+ "eval_TITRE": {
487
+ "f1": 0.9,
488
+ "number": 40,
489
+ "precision": 0.9,
490
+ "recall": 0.9
491
+ },
492
+ "eval_accuracy": 0.9766633565044687,
493
+ "eval_f1": 0.9736353509097662,
494
+ "eval_loss": 0.08077961951494217,
495
+ "eval_precision": 0.9703923019985197,
496
+ "eval_recall": 0.9769001490312966,
497
+ "eval_runtime": 2.0842,
498
+ "eval_samples_per_second": 324.34,
499
+ "eval_steps_per_second": 20.631,
500
+ "step": 1000
501
+ },
502
+ {
503
+ "epoch": 2.89,
504
+ "eval_ACT": {
505
+ "f1": 0.9429657794676807,
506
+ "number": 519,
507
+ "precision": 0.9305816135084428,
508
+ "recall": 0.9556840077071291
509
+ },
510
+ "eval_CARDINAL": {
511
+ "f1": 0.9941089837997055,
512
+ "number": 678,
513
+ "precision": 0.9926470588235294,
514
+ "recall": 0.995575221238938
515
+ },
516
+ "eval_FT": {
517
+ "f1": 0.30769230769230765,
518
+ "number": 7,
519
+ "precision": 0.3333333333333333,
520
+ "recall": 0.2857142857142857
521
+ },
522
+ "eval_LOC": {
523
+ "f1": 0.9596354166666666,
524
+ "number": 761,
525
+ "precision": 0.9509677419354838,
526
+ "recall": 0.9684625492772667
527
+ },
528
+ "eval_PER": {
529
+ "f1": 0.9904341427520236,
530
+ "number": 679,
531
+ "precision": 0.9897058823529412,
532
+ "recall": 0.9911634756995582
533
+ },
534
+ "eval_TITRE": {
535
+ "f1": 0.9,
536
+ "number": 40,
537
+ "precision": 0.9,
538
+ "recall": 0.9
539
+ },
540
+ "eval_accuracy": 0.9740566037735849,
541
+ "eval_f1": 0.9703593923675434,
542
+ "eval_loss": 0.08018206059932709,
543
+ "eval_precision": 0.9649963154016212,
544
+ "eval_recall": 0.9757824143070045,
545
+ "eval_runtime": 2.2649,
546
+ "eval_samples_per_second": 298.463,
547
+ "eval_steps_per_second": 18.985,
548
+ "step": 1100
549
+ },
550
+ {
551
+ "epoch": 3.15,
552
+ "eval_ACT": {
553
+ "f1": 0.9636711281070746,
554
+ "number": 519,
555
+ "precision": 0.9563567362428842,
556
+ "recall": 0.9710982658959537
557
+ },
558
+ "eval_CARDINAL": {
559
+ "f1": 0.9941089837997055,
560
+ "number": 678,
561
+ "precision": 0.9926470588235294,
562
+ "recall": 0.995575221238938
563
+ },
564
+ "eval_FT": {
565
+ "f1": 0.4615384615384615,
566
+ "number": 7,
567
+ "precision": 0.5,
568
+ "recall": 0.42857142857142855
569
+ },
570
+ "eval_LOC": {
571
+ "f1": 0.9711286089238845,
572
+ "number": 761,
573
+ "precision": 0.9698558322411533,
574
+ "recall": 0.9724047306176085
575
+ },
576
+ "eval_PER": {
577
+ "f1": 0.9897058823529413,
578
+ "number": 679,
579
+ "precision": 0.9882525697503671,
580
+ "recall": 0.9911634756995582
581
+ },
582
+ "eval_TITRE": {
583
+ "f1": 0.9,
584
+ "number": 40,
585
+ "precision": 0.9,
586
+ "recall": 0.9
587
+ },
588
+ "eval_accuracy": 0.9822492552135055,
589
+ "eval_f1": 0.9778851514588367,
590
+ "eval_loss": 0.07905712723731995,
591
+ "eval_precision": 0.9755283648498332,
592
+ "eval_recall": 0.9802533532041728,
593
+ "eval_runtime": 2.1617,
594
+ "eval_samples_per_second": 312.718,
595
+ "eval_steps_per_second": 19.892,
596
+ "step": 1200
597
+ }
598
+ ],
599
+ "max_steps": 5000,
600
+ "num_train_epochs": 14,
601
+ "total_flos": 486838277989008.0,
602
+ "trial_name": null,
603
+ "trial_params": null
604
+ }
training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:90906da5dd107cb517c00b97f0487ab17b793287fab57b7d29528cd7df8065b0
3
+ size 3503