esp_msl

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: 0.2048
• Model Preparation Time: 0.0048
• Bleu Msl: 87.1599
• Bleu Asl: 0
• Ter Msl: 7.6997
• 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	Model Preparation Time	Bleu Msl	Bleu Asl	Ter Msl	Ter Asl
No log	1.0	75	3.9416	0.0048	9.6418	0	95.7604	100
No log	2.0	150	3.0683	0.0048	11.3858	0	98.1567	100
No log	3.0	225	2.5362	0.0048	19.3789	0	85.5300	100
No log	4.0	300	2.1286	0.0048	20.3180	0	83.8710	100
No log	5.0	375	1.8211	0.0048	17.8046	0	89.4931	100
No log	6.0	450	1.5708	0.0048	58.5580	0	29.5853	100
2.865	7.0	525	1.3571	0.0048	63.7680	0	24.7005	100
2.865	8.0	600	1.1614	0.0048	65.7864	0	21.8433	100
2.865	9.0	675	0.9983	0.0048	57.8092	0	23.8710	100
2.865	10.0	750	0.8741	0.0048	65.5640	0	21.2903	100
2.865	11.0	825	0.7724	0.0048	69.4951	0	19.3548	100
2.865	12.0	900	0.6838	0.0048	74.3444	0	16.8664	100
2.865	13.0	975	0.6211	0.0048	71.7643	0	17.6959	100
0.8947	14.0	1050	0.5723	0.0048	75.2869	0	15.8525	100
0.8947	15.0	1125	0.5436	0.0048	75.9376	0	15.2995	100
0.8947	16.0	1200	0.5171	0.0048	60.9052	0	19.8157	100
0.8947	17.0	1275	0.4969	0.0048	76.2738	0	14.1935	100
0.8947	18.0	1350	0.4818	0.0048	76.5583	0	14.1935	100
0.8947	19.0	1425	0.4685	0.0048	76.8689	0	14.3779	100
0.3654	20.0	1500	0.4626	0.0048	77.2378	0	13.8249	100
0.3654	21.0	1575	0.4511	0.0048	76.4648	0	14.0092	100
0.3654	22.0	1650	0.4480	0.0048	76.3980	0	13.9171	100
0.3654	23.0	1725	0.4454	0.0048	77.1739	0	13.6406	100
0.3654	24.0	1800	0.4380	0.0048	77.3622	0	13.7327	100
0.3654	25.0	1875	0.4342	0.0048	75.8442	0	14.1935	100
0.3654	26.0	1950	0.4346	0.0048	77.4371	0	13.7327	100
0.2434	27.0	2025	0.4321	0.0048	78.0849	0	13.6406	100
0.2434	28.0	2100	0.4312	0.0048	77.8954	0	13.6406	100
0.2434	29.0	2175	0.4300	0.0048	77.7089	0	13.7327	100
0.2434	30.0	2250	0.4297	0.0048	77.7089	0	13.7327	100

Framework versions

• Transformers 4.46.2
• Pytorch 2.5.1+cu121
• Datasets 3.1.0
• Tokenizers 0.20.3