ClasificadorV2

This model is a fine-tuned version of dccuchile/bert-base-spanish-wwm-cased on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 1.1411
• Accuracy: 0.5708
• Off By One Accuracy: 0.9434
• F1: 0.5724
• Recall: 0.5708

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: 50
• eval_batch_size: 50
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 500
• num_epochs: 5
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	Off By One Accuracy	F1	Recall
1.2177	0.3333	500	1.0309	0.5426	0.93	0.5414	0.5426
1.011	0.6667	1000	0.9836	0.565	0.9336	0.5485	0.565
0.9833	1.0	1500	0.9664	0.5752	0.9448	0.5704	0.5752
0.9004	1.3333	2000	0.9566	0.5728	0.9476	0.5743	0.5728
0.8974	1.6667	2500	0.9583	0.5782	0.9472	0.5784	0.5782
0.8912	2.0	3000	0.9480	0.5816	0.9498	0.5768	0.5816
0.7935	2.3333	3500	0.9768	0.582	0.9472	0.5800	0.582
0.7898	2.6667	4000	0.9831	0.5716	0.9426	0.5715	0.5716
0.7801	3.0	4500	0.9969	0.5736	0.9514	0.5759	0.5736
0.6714	3.3333	5000	1.0782	0.5826	0.9392	0.5795	0.5826
0.6783	3.6667	5500	1.0672	0.5724	0.9456	0.5752	0.5724
0.6764	4.0	6000	1.0762	0.567	0.9458	0.5708	0.567
0.5986	4.3333	6500	1.1349	0.5698	0.9412	0.5684	0.5698
0.5887	4.6667	7000	1.1335	0.5706	0.9398	0.5716	0.5706
0.5798	5.0	7500	1.1411	0.5708	0.9434	0.5724	0.5708

Framework versions

• Transformers 4.40.2
• Pytorch 2.2.1+cu121
• Datasets 2.19.1
• Tokenizers 0.19.1