distilbert-base-uncased-v7-finetuned-squad-es

Procedimiento de entrenamiento

Hiperparámetros de entrenamiento

Los siguientes hiperparámetros se usaron durante el entrenamiento:

optimizer: {'name': 'Adam', 'weight_decay': None, 'clipnorm': None, 'global_clipnorm': None, 'clipvalue': None, 'use_ema': False, 'ema_momentum': 0.99, 'ema_overwrite_frequency': None, 'jit_compile': True, 'is_legacy_optimizer': False, 'learning_rate': {'module': 'keras.optimizers.schedules', 'class_name': 'PolynomialDecay', 'config': {'initial_learning_rate': 2e-05, 'decay_steps': 14650, 'end_learning_rate': 0.0, 'power': 1.0, 'cycle': False, 'name': None}, 'registered_name': None}, 'beta_1': 0.9, 'beta_2': 0.999, 'epsilon': 1e-08, 'amsgrad': False}
training_precision: float32

Resultados de entrenamiento

Train Loss
Train End Logits Accuracy
Train Start Logits Accuracy
Validation Loss
Validation End Logits Accuracy
Validation Start Logits Accuracy
Epoch

3.6949
0.1984
0.1779
2.8564
0.2769
0.2538
0

2.5459
0.3648
0.3635
2.2433
0.4436
0.4231
1

1.8293
0.5205
0.5107
2.1826
0.5013
0.4654
2

1.3619
0.6135
0.6067
2.2598
0.5090
0.4795
3

1.0012
0.7018
0.6826
2.4600
0.5308
0.4808
4

0.7736
0.7555
0.7624
2.4332
0.5410
0.5077
5

0.6124
0.7914
0.8131
2.7033
0.5410
0.5141
6

0.4781
0.8439
0.8498
2.8627
0.5462
0.5205
7

0.3867
0.8673
0.8660
3.0181
0.5449
0.5269
8

0.3473
0.8759
0.8882
3.0705
0.5410
0.5154
9

0.2735
0.9096
0.9083
3.1680
0.5590
0.5359
10

0.2354
0.9189
0.9206
3.2071
0.5705
0.5231
11

0.1955
0.9360
0.9300
3.4207
0.5449
0.5141
12

0.2068
0.9283
0.9296
3.2288
0.5551
0.5333
13

0.1852
0.9364
0.9381
3.5434
0.5385
0.5218
14

0.1522
0.9509
0.9471
3.5845
0.5487
0.5256
15

0.1404
0.9548
0.9582
3.6228
0.5628
0.5192
16

0.1255
0.9501
0.9578
3.6708
0.5628
0.5295
17

0.1253
0.9608
0.9578
3.7048
0.5564
0.5154
18

0.1120
0.9539
0.9633
3.6301
0.5628
0.5295
19

0.1043
0.9578
0.9642
3.6380
0.5474
0.5295
20

0.0999
0.9612
0.9659
3.8969
0.5449
0.5321
21

0.0845
0.9710
0.9757
3.9082
0.5590
0.5321
22

0.0874
0.9735
0.9689
3.7159
0.5603
0.5436
23

0.0700
0.9731
0.9748
3.9612
0.5564
0.5462
24

0.0572
0.9787
0.9774
4.0000
0.5590
0.5333
25

0.0628
0.9761
0.9778
3.8762
0.5551
0.5372
26

0.0550
0.9804
0.9821
3.9125
0.5590
0.5397
27

0.0586
0.9817
0.9808
3.9667
0.5603
0.5346
28

0.0465
0.9812
0.9838
3.9716
0.5603
0.5295
29

0.0438
0.9842
0.9825
4.0324
0.5577
0.5333
30

0.0422
0.9834
0.9872
4.2007
0.5603
0.5397
31

0.0427
0.9868
0.9846
4.1012
0.5513
0.5372
32

0.0395
0.9885
0.9855
4.0936
0.5487
0.5308
33

0.0355
0.9863
0.9872
4.1443
0.5667
0.5321
34

0.0366
0.9876
0.9881
4.2423
0.5551
0.5410
35

0.0323
0.9881
0.9872
4.3990
0.5513
0.5295
36

0.0252
0.9915
0.9893
4.2288
0.5551
0.5359
37

0.0285
0.9863
0.9906
4.3026
0.5654
0.5346
38

0.0251
0.9906
0.9915
4.2990
0.5628
0.5346
39

0.0313
0.9868
0.9885
4.2994
0.5679
0.5359
40

0.0208
0.9932
0.9932
4.2457
0.5603
0.5372
41

0.0225
0.9927
0.9910
4.4447
0.5628
0.5295
42

0.0194
0.9932
0.9919
4.3625
0.5603
0.5359
43

0.0189
0.9906
0.9923
4.3148
0.5679
0.5410
44

0.0182
0.9949
0.9927
4.3577
0.5628
0.5385
45

0.0160
0.9923
0.9949
4.3897
0.5615
0.5346
46

0.0146
0.9949
0.9936
4.3823
0.5679
0.5385
47

0.0163
0.9936
0.9945
4.3764
0.5667
0.5397
48

0.0143
0.9966
0.9932
4.3715
0.5654
0.5385
49