augmented

This model is a fine-tuned version of Helsinki-NLP/opus-mt-es-es on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.0293
• Bleu Msl: 95.6270
• Bleu Asl: 0
• Ter Msl: 2.0212
• Ter Asl: 100

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: 1e-05
• train_batch_size: 32
• eval_batch_size: 64
• seed: 42
• optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: linear
• num_epochs: 30
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Bleu Msl	Bleu Asl	Ter Msl	Ter Asl
No log	1.0	250	2.5165	14.8999	0	102.2714	100
3.1383	2.0	500	1.3557	6.4514	0	181.1590	100
3.1383	3.0	750	0.6870	40.7848	0	32.4694	100
0.9384	4.0	1000	0.4223	31.8139	0	39.1380	100
0.9384	5.0	1250	0.3244	59.5708	0	19.2778	100
0.393	6.0	1500	0.2693	79.8758	0	12.9878	100
0.393	7.0	1750	0.2303	80.9355	0	11.8812	100
0.2573	8.0	2000	0.2041	81.8443	0	11.1532	100
0.2573	9.0	2250	0.1894	81.3993	0	11.0367	100
0.1956	10.0	2500	0.1708	80.6953	0	10.9202	100
0.1956	11.0	2750	0.1551	83.8621	0	9.6680	100
0.1573	12.0	3000	0.1507	85.1923	0	9.7263	100
0.1573	13.0	3250	0.1416	84.8165	0	9.4351	100
0.1349	14.0	3500	0.1345	85.4052	0	9.4059	100
0.1349	15.0	3750	0.1306	85.7668	0	9.2312	100
0.1183	16.0	4000	0.1281	85.0627	0	9.0856	100
0.1183	17.0	4250	0.1211	86.3222	0	8.6779	100
0.1052	18.0	4500	0.1176	87.0423	0	8.7070	100
0.1052	19.0	4750	0.1154	86.6227	0	8.4741	100
0.0974	20.0	5000	0.1111	86.8667	0	8.1538	100
0.0974	21.0	5250	0.1120	87.1441	0	8.4450	100
0.0904	22.0	5500	0.1090	87.0076	0	8.2994	100
0.0904	23.0	5750	0.1065	87.5612	0	7.8626	100
0.0868	24.0	6000	0.1049	87.8206	0	7.8917	100
0.0868	25.0	6250	0.1039	87.9881	0	7.7169	100
0.0827	26.0	6500	0.1032	87.9361	0	7.8043	100
0.0827	27.0	6750	0.1030	87.6810	0	7.7752	100
0.0801	28.0	7000	0.1028	87.8956	0	7.6296	100
0.0801	29.0	7250	0.1021	87.8736	0	7.6878	100
0.0782	30.0	7500	0.1020	87.9235	0	7.6878	100

Framework versions

• Transformers 4.46.3
• Pytorch 2.5.1+cu121
• Datasets 3.1.0
• Tokenizers 0.20.3