Modelpublic

Armandodelca/Prototipo_5_EMI

sourceHugging Faceupdated 2y agoView on Hugging Face

0likes9downloads

Model Card

Prototipo5EMI

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

—Loss: 1.4215
—Accuracy: 0.538

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
—trainbatchsize: 20
—evalbatchsize: 20
—seed: 42
—optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
—lrschedulertype: linear
—lrschedulerwarmup_steps: 500
—num_epochs: 5
—mixedprecisiontraining: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
1.2459	0.1481	200	1.2168	0.4493
1.1445	0.2963	400	1.0823	0.512
1.1117	0.4444	600	1.0979	0.5053
1.0618	0.5926	800	1.0457	0.5273
1.0343	0.7407	1000	1.0219	0.537
1.1239	0.8889	1200	1.0353	0.5257
0.9012	1.0370	1400	1.0637	0.5383
0.86	1.1852	1600	1.0682	0.5333
0.898	1.3333	1800	1.0341	0.5483
0.929	1.4815	2000	1.0437	0.5363
0.9921	1.6296	2200	0.9968	0.5473
0.9776	1.7778	2400	1.0418	0.5553
0.9166	1.9259	2600	0.9874	0.5573
0.703	2.0741	2800	1.0564	0.556
0.8123	2.2222	3000	1.0582	0.561
0.6727	2.3704	3200	1.0942	0.5483
0.6843	2.5185	3400	1.1128	0.558
0.7528	2.6667	3600	1.0823	0.5547
0.7747	2.8148	3800	1.0744	0.5497
0.7471	2.9630	4000	1.0749	0.5527
0.5774	3.1111	4200	1.1422	0.552
0.6105	3.2593	4400	1.2226	0.543
0.573	3.4074	4600	1.2427	0.5417
0.6047	3.5556	4800	1.2403	0.537
0.5334	3.7037	5000	1.2470	0.5413
0.5688	3.8519	5200	1.2585	0.5507
0.4928	4.0	5400	1.2653	0.5437
0.4314	4.1481	5600	1.3419	0.541
0.4556	4.2963	5800	1.3677	0.5413
0.4815	4.4444	6000	1.3912	0.5407
0.4431	4.5926	6200	1.4004	0.5347
0.4312	4.7407	6400	1.4161	0.5397
0.459	4.8889	6600	1.4215	0.538

Framework versions

—Transformers 4.40.1
—Pytorch 2.2.1+cu121
—Datasets 2.19.0
—Tokenizers 0.19.1