Hugging Face's logo

nlpso
/
m0_flat_ner_ref_ptrn_cmbert_io

Token Classification
Transformers
PyTorch
French
camembert
Model card Files
xet
Community
1
nlpso commited on
Commit
481efee
·
1 Parent(s): 35ce58a

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 +508 -0
  11. training_args.bin +3 -0
config.json ADDED
@@ -0,0 +1,47 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_name_or_path": "HueyNemud/das22-10-camembert_pretrained",
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:ec01f5ffae3fc004ef26544f84a2e984e43be4a2a77b353f2923fc299c0697c9
3
+ size 880409317
pytorch_model.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:5d9b24316ff69fd59a5afedc5bf51e9a764968fe9dbd223f3a5ebdb13b8d6cc1
3
+ size 440218033
rng_state.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ea0cd7e8eefcd5c6d560934a44a8ff0652a44a00edb31dedd9ec9975612a4c31
3
+ size 14503
scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ec6331a2d4f82e15c898be3d09fa0a39ccbc9d6fd132b0505ce16e06776b6d8e
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": "HueyNemud/das22-10-camembert_pretrained",
19
+ "pad_token": "<pad>",
20
+ "sep_token": "</s>",
21
+ "sp_model_kwargs": {},
22
+ "special_tokens_map_file": "/lrde/home2/jchazalo/.cache/huggingface/transformers/fe0e213c44079a9ee091098f81fff2941484006e9ba3001a9bf1ee9f87537599.cb3ec3a6c1200d181228d8825ae9767572abca54efa1bbb37fd83d721b2ef323",
23
+ "tokenizer_class": "CamembertTokenizer",
24
+ "unk_token": "<unk>"
25
+ }
trainer_state.json ADDED
@@ -0,0 +1,508 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": 0.9786205614426474,
3
+ "best_model_checkpoint": "/work/stual/res_ICDAR/method_0/tmp/22-flat-ner-ref-pretrained_camembert_ner/checkpoint-1000",
4
+ "epoch": 2.6246719160104988,
5
+ "global_step": 1000,
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": 0.9162011173184358,
14
+ "number": 519,
15
+ "precision": 0.8864864864864865,
16
+ "recall": 0.9479768786127167
17
+ },
18
+ "eval_CARDINAL": {
19
+ "f1": 0.9919177075679647,
20
+ "number": 678,
21
+ "precision": 0.9882869692532943,
22
+ "recall": 0.995575221238938
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": 0.944300518134715,
32
+ "number": 761,
33
+ "precision": 0.9310344827586207,
34
+ "recall": 0.9579500657030223
35
+ },
36
+ "eval_PER": {
37
+ "f1": 0.9288334556126192,
38
+ "number": 679,
39
+ "precision": 0.9254385964912281,
40
+ "recall": 0.9322533136966127
41
+ },
42
+ "eval_TITRE": {
43
+ "f1": 0.32000000000000006,
44
+ "number": 40,
45
+ "precision": 0.8,
46
+ "recall": 0.2
47
+ },
48
+ "eval_accuracy": 0.9653674280039722,
49
+ "eval_f1": 0.9398036673458048,
50
+ "eval_loss": 0.20965276658535004,
51
+ "eval_precision": 0.934438305709024,
52
+ "eval_recall": 0.9452309985096871,
53
+ "eval_runtime": 2.353,
54
+ "eval_samples_per_second": 287.287,
55
+ "eval_steps_per_second": 18.274,
56
+ "step": 100
57
+ },
58
+ {
59
+ "epoch": 0.52,
60
+ "eval_ACT": {
61
+ "f1": 0.9487666034155597,
62
+ "number": 519,
63
+ "precision": 0.9345794392523364,
64
+ "recall": 0.9633911368015414
65
+ },
66
+ "eval_CARDINAL": {
67
+ "f1": 0.9919177075679647,
68
+ "number": 678,
69
+ "precision": 0.9882869692532943,
70
+ "recall": 0.995575221238938
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": 0.9441558441558441,
80
+ "number": 761,
81
+ "precision": 0.9332477535301669,
82
+ "recall": 0.9553219448094612
83
+ },
84
+ "eval_PER": {
85
+ "f1": 0.9618768328445747,
86
+ "number": 679,
87
+ "precision": 0.9576642335766423,
88
+ "recall": 0.9661266568483063
89
+ },
90
+ "eval_TITRE": {
91
+ "f1": 0.525,
92
+ "number": 40,
93
+ "precision": 0.525,
94
+ "recall": 0.525
95
+ },
96
+ "eval_accuracy": 0.9766633565044687,
97
+ "eval_f1": 0.9541250462449131,
98
+ "eval_loss": 0.12336099147796631,
99
+ "eval_precision": 0.9474650991917708,
100
+ "eval_recall": 0.9608792846497765,
101
+ "eval_runtime": 2.3033,
102
+ "eval_samples_per_second": 293.494,
103
+ "eval_steps_per_second": 18.669,
104
+ "step": 200
105
+ },
106
+ {
107
+ "epoch": 0.79,
108
+ "eval_ACT": {
109
+ "f1": 0.9493791786055396,
110
+ "number": 519,
111
+ "precision": 0.9412878787878788,
112
+ "recall": 0.9576107899807321
113
+ },
114
+ "eval_CARDINAL": {
115
+ "f1": 0.9933774834437086,
116
+ "number": 678,
117
+ "precision": 0.9911894273127754,
118
+ "recall": 0.995575221238938
119
+ },
120
+ "eval_FT": {
121
+ "f1": 0.5,
122
+ "number": 7,
123
+ "precision": 0.6,
124
+ "recall": 0.42857142857142855
125
+ },
126
+ "eval_LOC": {
127
+ "f1": 0.973718791064389,
128
+ "number": 761,
129
+ "precision": 0.973718791064389,
130
+ "recall": 0.973718791064389
131
+ },
132
+ "eval_PER": {
133
+ "f1": 0.9778761061946903,
134
+ "number": 679,
135
+ "precision": 0.9793205317577548,
136
+ "recall": 0.9764359351988218
137
+ },
138
+ "eval_TITRE": {
139
+ "f1": 0.7999999999999999,
140
+ "number": 40,
141
+ "precision": 0.8571428571428571,
142
+ "recall": 0.75
143
+ },
144
+ "eval_accuracy": 0.9800148957298908,
145
+ "eval_f1": 0.9715136846024949,
146
+ "eval_loss": 0.0935482531785965,
147
+ "eval_precision": 0.9709713435057685,
148
+ "eval_recall": 0.9720566318926974,
149
+ "eval_runtime": 2.0792,
150
+ "eval_samples_per_second": 325.122,
151
+ "eval_steps_per_second": 20.681,
152
+ "step": 300
153
+ },
154
+ {
155
+ "epoch": 1.05,
156
+ "eval_ACT": {
157
+ "f1": 0.9456625357483317,
158
+ "number": 519,
159
+ "precision": 0.9358490566037736,
160
+ "recall": 0.9556840077071291
161
+ },
162
+ "eval_CARDINAL": {
163
+ "f1": 0.9926470588235294,
164
+ "number": 678,
165
+ "precision": 0.9897360703812317,
166
+ "recall": 0.995575221238938
167
+ },
168
+ "eval_FT": {
169
+ "f1": 0.5454545454545454,
170
+ "number": 7,
171
+ "precision": 0.75,
172
+ "recall": 0.42857142857142855
173
+ },
174
+ "eval_LOC": {
175
+ "f1": 0.9677843523997371,
176
+ "number": 761,
177
+ "precision": 0.968421052631579,
178
+ "recall": 0.9671484888304862
179
+ },
180
+ "eval_PER": {
181
+ "f1": 0.9882352941176471,
182
+ "number": 679,
183
+ "precision": 0.986784140969163,
184
+ "recall": 0.9896907216494846
185
+ },
186
+ "eval_TITRE": {
187
+ "f1": 0.8192771084337349,
188
+ "number": 40,
189
+ "precision": 0.7906976744186046,
190
+ "recall": 0.85
191
+ },
192
+ "eval_accuracy": 0.9784011916583912,
193
+ "eval_f1": 0.9717682020802376,
194
+ "eval_loss": 0.08334896713495255,
195
+ "eval_precision": 0.9688888888888889,
196
+ "eval_recall": 0.9746646795827124,
197
+ "eval_runtime": 2.0578,
198
+ "eval_samples_per_second": 328.512,
199
+ "eval_steps_per_second": 20.896,
200
+ "step": 400
201
+ },
202
+ {
203
+ "epoch": 1.31,
204
+ "learning_rate": 9e-05,
205
+ "loss": 0.2168,
206
+ "step": 500
207
+ },
208
+ {
209
+ "epoch": 1.31,
210
+ "eval_ACT": {
211
+ "f1": 0.954110898661568,
212
+ "number": 519,
213
+ "precision": 0.9468690702087287,
214
+ "recall": 0.9614643545279383
215
+ },
216
+ "eval_CARDINAL": {
217
+ "f1": 0.9889298892988929,
218
+ "number": 678,
219
+ "precision": 0.9896602658788775,
220
+ "recall": 0.9882005899705014
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": 0.9556135770234986,
230
+ "number": 761,
231
+ "precision": 0.9494163424124513,
232
+ "recall": 0.961892247043364
233
+ },
234
+ "eval_PER": {
235
+ "f1": 0.9772893772893771,
236
+ "number": 679,
237
+ "precision": 0.9723032069970845,
238
+ "recall": 0.9823269513991163
239
+ },
240
+ "eval_TITRE": {
241
+ "f1": 0.85,
242
+ "number": 40,
243
+ "precision": 0.85,
244
+ "recall": 0.85
245
+ },
246
+ "eval_accuracy": 0.977656405163853,
247
+ "eval_f1": 0.9660293298682012,
248
+ "eval_loss": 0.08229256421327591,
249
+ "eval_precision": 0.9626341102478727,
250
+ "eval_recall": 0.9694485842026825,
251
+ "eval_runtime": 2.0948,
252
+ "eval_samples_per_second": 322.711,
253
+ "eval_steps_per_second": 20.527,
254
+ "step": 500
255
+ },
256
+ {
257
+ "epoch": 1.57,
258
+ "eval_ACT": {
259
+ "f1": 0.9524714828897339,
260
+ "number": 519,
261
+ "precision": 0.9399624765478424,
262
+ "recall": 0.9653179190751445
263
+ },
264
+ "eval_CARDINAL": {
265
+ "f1": 0.9933774834437086,
266
+ "number": 678,
267
+ "precision": 0.9911894273127754,
268
+ "recall": 0.995575221238938
269
+ },
270
+ "eval_FT": {
271
+ "f1": 0.5454545454545454,
272
+ "number": 7,
273
+ "precision": 0.75,
274
+ "recall": 0.42857142857142855
275
+ },
276
+ "eval_LOC": {
277
+ "f1": 0.9652459016393442,
278
+ "number": 761,
279
+ "precision": 0.9633507853403142,
280
+ "recall": 0.9671484888304862
281
+ },
282
+ "eval_PER": {
283
+ "f1": 0.986764705882353,
284
+ "number": 679,
285
+ "precision": 0.9853157121879589,
286
+ "recall": 0.9882179675994109
287
+ },
288
+ "eval_TITRE": {
289
+ "f1": 0.8780487804878048,
290
+ "number": 40,
291
+ "precision": 0.8571428571428571,
292
+ "recall": 0.9
293
+ },
294
+ "eval_accuracy": 0.9829940417080437,
295
+ "eval_f1": 0.9730933382816849,
296
+ "eval_loss": 0.07296791672706604,
297
+ "eval_precision": 0.9693160813308688,
298
+ "eval_recall": 0.9769001490312966,
299
+ "eval_runtime": 2.0927,
300
+ "eval_samples_per_second": 323.022,
301
+ "eval_steps_per_second": 20.547,
302
+ "step": 600
303
+ },
304
+ {
305
+ "epoch": 1.84,
306
+ "eval_ACT": {
307
+ "f1": 0.9654510556621881,
308
+ "number": 519,
309
+ "precision": 0.9617590822179732,
310
+ "recall": 0.9691714836223507
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": 0.5454545454545454,
320
+ "number": 7,
321
+ "precision": 0.75,
322
+ "recall": 0.42857142857142855
323
+ },
324
+ "eval_LOC": {
325
+ "f1": 0.968586387434555,
326
+ "number": 761,
327
+ "precision": 0.9647979139504563,
328
+ "recall": 0.9724047306176085
329
+ },
330
+ "eval_PER": {
331
+ "f1": 0.986019131714496,
332
+ "number": 679,
333
+ "precision": 0.9852941176470589,
334
+ "recall": 0.9867452135493373
335
+ },
336
+ "eval_TITRE": {
337
+ "f1": 0.9,
338
+ "number": 40,
339
+ "precision": 0.9,
340
+ "recall": 0.9
341
+ },
342
+ "eval_accuracy": 0.9829940417080437,
343
+ "eval_f1": 0.9763896635062279,
344
+ "eval_loss": 0.06497417390346527,
345
+ "eval_precision": 0.9743970315398887,
346
+ "eval_recall": 0.9783904619970194,
347
+ "eval_runtime": 2.0484,
348
+ "eval_samples_per_second": 330.007,
349
+ "eval_steps_per_second": 20.992,
350
+ "step": 700
351
+ },
352
+ {
353
+ "epoch": 2.1,
354
+ "eval_ACT": {
355
+ "f1": 0.9521988527724664,
356
+ "number": 519,
357
+ "precision": 0.9449715370018975,
358
+ "recall": 0.9595375722543352
359
+ },
360
+ "eval_CARDINAL": {
361
+ "f1": 0.9933774834437086,
362
+ "number": 678,
363
+ "precision": 0.9911894273127754,
364
+ "recall": 0.995575221238938
365
+ },
366
+ "eval_FT": {
367
+ "f1": 0.4,
368
+ "number": 7,
369
+ "precision": 0.6666666666666666,
370
+ "recall": 0.2857142857142857
371
+ },
372
+ "eval_LOC": {
373
+ "f1": 0.9653368214519294,
374
+ "number": 761,
375
+ "precision": 0.9609375,
376
+ "recall": 0.9697766097240473
377
+ },
378
+ "eval_PER": {
379
+ "f1": 0.9867452135493373,
380
+ "number": 679,
381
+ "precision": 0.9867452135493373,
382
+ "recall": 0.9867452135493373
383
+ },
384
+ "eval_TITRE": {
385
+ "f1": 0.925,
386
+ "number": 40,
387
+ "precision": 0.925,
388
+ "recall": 0.925
389
+ },
390
+ "eval_accuracy": 0.9829940417080437,
391
+ "eval_f1": 0.9736157562244518,
392
+ "eval_loss": 0.06101168692111969,
393
+ "eval_precision": 0.9710896960711638,
394
+ "eval_recall": 0.9761549925484352,
395
+ "eval_runtime": 2.0382,
396
+ "eval_samples_per_second": 331.668,
397
+ "eval_steps_per_second": 21.097,
398
+ "step": 800
399
+ },
400
+ {
401
+ "epoch": 2.36,
402
+ "eval_ACT": {
403
+ "f1": 0.9645254074784276,
404
+ "number": 519,
405
+ "precision": 0.9599236641221374,
406
+ "recall": 0.9691714836223507
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": 0.4,
416
+ "number": 7,
417
+ "precision": 0.6666666666666666,
418
+ "recall": 0.2857142857142857
419
+ },
420
+ "eval_LOC": {
421
+ "f1": 0.9652914210870989,
422
+ "number": 761,
423
+ "precision": 0.9621409921671018,
424
+ "recall": 0.9684625492772667
425
+ },
426
+ "eval_PER": {
427
+ "f1": 0.9852724594992637,
428
+ "number": 679,
429
+ "precision": 0.9852724594992637,
430
+ "recall": 0.9852724594992637
431
+ },
432
+ "eval_TITRE": {
433
+ "f1": 0.925,
434
+ "number": 40,
435
+ "precision": 0.925,
436
+ "recall": 0.925
437
+ },
438
+ "eval_accuracy": 0.9842353525322741,
439
+ "eval_f1": 0.9756369722893807,
440
+ "eval_loss": 0.06042904779314995,
441
+ "eval_precision": 0.974006683995544,
442
+ "eval_recall": 0.9772727272727273,
443
+ "eval_runtime": 2.0867,
444
+ "eval_samples_per_second": 323.95,
445
+ "eval_steps_per_second": 20.606,
446
+ "step": 900
447
+ },
448
+ {
449
+ "epoch": 2.62,
450
+ "learning_rate": 8e-05,
451
+ "loss": 0.0621,
452
+ "step": 1000
453
+ },
454
+ {
455
+ "epoch": 2.62,
456
+ "eval_ACT": {
457
+ "f1": 0.9647283126787417,
458
+ "number": 519,
459
+ "precision": 0.9547169811320755,
460
+ "recall": 0.9749518304431599
461
+ },
462
+ "eval_CARDINAL": {
463
+ "f1": 0.9933774834437086,
464
+ "number": 678,
465
+ "precision": 0.9911894273127754,
466
+ "recall": 0.995575221238938
467
+ },
468
+ "eval_FT": {
469
+ "f1": 0.4,
470
+ "number": 7,
471
+ "precision": 0.6666666666666666,
472
+ "recall": 0.2857142857142857
473
+ },
474
+ "eval_LOC": {
475
+ "f1": 0.9730794484569928,
476
+ "number": 761,
477
+ "precision": 0.9724409448818898,
478
+ "recall": 0.973718791064389
479
+ },
480
+ "eval_PER": {
481
+ "f1": 0.9882179675994109,
482
+ "number": 679,
483
+ "precision": 0.9882179675994109,
484
+ "recall": 0.9882179675994109
485
+ },
486
+ "eval_TITRE": {
487
+ "f1": 0.925,
488
+ "number": 40,
489
+ "precision": 0.925,
490
+ "recall": 0.925
491
+ },
492
+ "eval_accuracy": 0.9846077457795432,
493
+ "eval_f1": 0.9786205614426474,
494
+ "eval_loss": 0.06579616665840149,
495
+ "eval_precision": 0.9766233766233766,
496
+ "eval_recall": 0.9806259314456036,
497
+ "eval_runtime": 2.1397,
498
+ "eval_samples_per_second": 315.934,
499
+ "eval_steps_per_second": 20.096,
500
+ "step": 1000
501
+ }
502
+ ],
503
+ "max_steps": 5000,
504
+ "num_train_epochs": 14,
505
+ "total_flos": 406892146317096.0,
506
+ "trial_name": null,
507
+ "trial_params": null
508
+ }
training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:2c2dac3cb81648f330557a480beef0e9e0e88d34e95f751a64b643dd4e486ad9
3
+ size 3503