bert-base-squad-v1.1-pt-IBAMA_v0.420240914224146

This model is a fine-tuned version of pierreguillou/bert-base-cased-squad-v1.1-portuguese on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 8.7428

Model description

More information needed

Intended uses & limitations

More information needed

Training and evaluation data

More information needed

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

• learning_rate: 2e-05
• train_batch_size: 16
• eval_batch_size: 16
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 50
• mixed_precision_training: Native AMP

Training results {'exact_match': 0.9689922480620154, 'f1': 30.759731618600995}

Training Loss	Epoch	Step	Validation Loss
No log	1.0	33	4.6990
No log	2.0	66	4.3280
No log	3.0	99	4.2582
No log	4.0	132	4.2636
No log	5.0	165	4.5098
No log	6.0	198	4.5878
No log	7.0	231	4.8520
No log	8.0	264	5.0870
No log	9.0	297	5.4772
No log	10.0	330	5.5812
No log	11.0	363	5.9264
No log	12.0	396	6.3056
No log	13.0	429	6.4991
No log	14.0	462	6.5704
No log	15.0	495	6.7604
2.3536	16.0	528	6.9113
2.3536	17.0	561	7.1457
2.3536	18.0	594	7.2865
2.3536	19.0	627	7.3866
2.3536	20.0	660	7.2945
2.3536	21.0	693	7.6477
2.3536	22.0	726	7.6016
2.3536	23.0	759	7.7623
2.3536	24.0	792	7.8580
2.3536	25.0	825	7.9034
2.3536	26.0	858	8.0633
2.3536	27.0	891	7.8441
2.3536	28.0	924	8.3173
2.3536	29.0	957	8.2840
2.3536	30.0	990	8.2639
0.5153	31.0	1023	8.4088
0.5153	32.0	1056	8.3245
0.5153	33.0	1089	8.1930
0.5153	34.0	1122	8.3888
0.5153	35.0	1155	8.3297
0.5153	36.0	1188	8.6224
0.5153	37.0	1221	8.5107
0.5153	38.0	1254	8.2904
0.5153	39.0	1287	8.5842
0.5153	40.0	1320	8.4732
0.5153	41.0	1353	8.5637
0.5153	42.0	1386	8.5919
0.5153	43.0	1419	8.6802
0.5153	44.0	1452	8.7390
0.5153	45.0	1485	8.6280
0.2321	46.0	1518	8.6732
0.2321	47.0	1551	8.6873
0.2321	48.0	1584	8.7090
0.2321	49.0	1617	8.7294
0.2321	50.0	1650	8.7428

Framework versions

• Transformers 4.44.2
• Pytorch 2.4.0+cu121
• Datasets 3.0.0
• Tokenizers 0.19.1