Bert_TPF_v10

This model is a fine-tuned version of dccuchile/bert-base-spanish-wwm-uncased on the None dataset. It achieves the following results on the evaluation set:

• Accuracy@en: 0.8315
• F1@en: 0.8323
• Precision@en: 0.8373
• Recall@en: 0.8368
• Loss@en: 0.6173
• Loss: 0.6173

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: cosine
• lr_scheduler_warmup_steps: 500
• num_epochs: 20

Training results

Training Loss	Epoch	Step	Accuracy@en	F1@en	Precision@en	Recall@en	Loss@en	Validation Loss
3.2405	1.0	552	0.2037	0.1306	0.1465	0.2065	2.5934	2.5934
2.3891	2.0	1104	0.2992	0.2349	0.2586	0.3058	2.0876	2.0876
2.0117	3.0	1656	0.3765	0.3448	0.3683	0.3839	1.8638	1.8638
1.7804	4.0	2208	0.4619	0.4287	0.4433	0.4705	1.6337	1.6337
1.4913	5.0	2760	0.5228	0.4905	0.5357	0.5306	1.3950	1.3950
1.2177	6.0	3312	0.5696	0.5529	0.6054	0.5773	1.2562	1.2562
1.0274	7.0	3864	0.6278	0.6086	0.6598	0.6360	1.0466	1.0466
0.8372	8.0	4416	0.7050	0.7007	0.7254	0.7104	0.8734	0.8734
0.67	9.0	4968	0.7407	0.7373	0.7510	0.7463	0.8112	0.8112
0.5259	10.0	5520	0.8	0.7999	0.8069	0.8050	0.6594	0.6594
0.4333	11.0	6072	0.8095	0.8056	0.8219	0.8159	0.6305	0.6305
0.3503	12.0	6624	0.8019	0.7985	0.8132	0.8074	0.6698	0.6698
0.2961	13.0	7176	0.8315	0.8323	0.8373	0.8368	0.6173	0.6173
0.2441	14.0	7728	0.8450	0.8459	0.8482	0.8493	0.6287	0.6287
0.2078	15.0	8280	0.8471	0.8477	0.8508	0.8511	0.6280	0.6280
0.1857	16.0	8832	0.8463	0.8470	0.8513	0.8510	0.6293	0.6293
0.164	17.0	9384	0.8471	0.8480	0.8510	0.8512	0.6371	0.6371
0.1467	18.0	9936	0.8489	0.8497	0.8536	0.8532	0.6410	0.6410
0.1409	19.0	10488	0.8489	0.8496	0.8535	0.8528	0.6396	0.6396
0.1378	20.0	11040	0.8497	0.8505	0.8543	0.8537	0.6395	0.6395

Framework versions

• Transformers 4.44.2
• Pytorch 2.5.0+cu121
• Datasets 3.0.2
• Tokenizers 0.19.1