11_6kmslsamples

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: 0.0714
• Bleu Msl: 82.3301
• Bleu Asl: 97.1735
• Ter Msl: 11.2456
• Ter Asl: 1.4638

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: 0.0001
• train_batch_size: 32
• eval_batch_size: 64
• seed: 42
• optimizer: Use OptimizerNames.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
0.3888	1.0	567	0.1194	52.4711	92.5612	26.0716	3.7483
0.1178	2.0	1134	0.0928	12.3994	94.4382	95.6631	2.8878
0.0785	3.0	1701	0.0791	27.7561	95.2400	14.7756	2.4407
0.0607	4.0	2268	0.0784	76.2031	95.4984	12.9602	2.2967
0.048	5.0	2835	0.0745	69.5670	95.6702	18.6586	2.2539
0.0392	6.0	3402	0.0718	78.5639	96.0612	12.9097	1.9966
0.0324	7.0	3969	0.0697	80.1654	96.0441	11.2960	2.0548
0.0217	8.0	4536	0.0714	78.6078	96.3918	13.1114	1.9568
0.0177	9.0	5103	0.0732	80.7119	96.0813	11.8507	2.0916
0.0178	10.0	5670	0.0717	79.8444	96.2730	11.9516	1.9783
0.0142	11.0	6237	0.0692	68.7637	96.4883	20.9279	1.8251
0.0144	12.0	6804	0.0700	80.5359	94.8578	12.2542	3.4420
0.0119	13.0	7371	0.0718	80.6068	96.3883	11.7499	1.8650
0.0099	14.0	7938	0.0710	79.7237	96.5317	12.3046	1.7915
0.0089	15.0	8505	0.0725	79.6613	96.4755	11.9012	1.8986
0.008	16.0	9072	0.0720	82.1357	96.8505	10.9430	1.6169
0.0062	17.0	9639	0.0702	83.0745	96.7340	10.6909	1.6904
0.0059	18.0	10206	0.0698	79.5220	96.8042	11.8507	1.6628
0.0072	19.0	10773	0.0713	80.4408	96.8836	11.4473	1.6322
0.0049	20.0	11340	0.0716	81.6404	96.8682	11.2456	1.6292
0.0044	21.0	11907	0.0713	81.7670	96.9006	11.1952	1.6873
0.0035	22.0	12474	0.0718	80.5118	96.9222	11.4977	1.6230
0.0031	23.0	13041	0.0720	81.2657	97.0022	11.0439	1.5924
0.0031	24.0	13608	0.0723	82.0082	96.9102	11.0439	1.5955
0.0022	25.0	14175	0.0721	82.1310	97.0355	10.7917	1.5403
0.0022	26.0	14742	0.0726	83.0873	97.0320	10.8926	1.5373
0.002	27.0	15309	0.0711	82.5284	97.0864	10.9430	1.4944
0.002	28.0	15876	0.0710	83.1909	97.1182	10.6909	1.4852
0.0015	29.0	16443	0.0713	82.5500	97.0790	11.0943	1.5005
0.0013	30.0	17010	0.0714	82.3301	97.1735	11.2456	1.4638

Framework versions

• Transformers 4.48.3
• Pytorch 2.5.1+cu124
• Datasets 3.3.2
• Tokenizers 0.21.0