Helsinki_lg_inf_en

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

• Loss: 0.1764
• Bleu: 22.8149
• Gen Len: 17.7776

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: linear
• num_epochs: 30

Training results

Training Loss	Epoch	Step	Validation Loss	Bleu	Gen Len
No log	1.0	153	0.5662	0.6031	20.9541
No log	2.0	306	0.5118	0.8533	20.1433
No log	3.0	459	0.4754	1.1179	19.9124
0.6777	4.0	612	0.4452	1.4213	20.2326
0.6777	5.0	765	0.4181	1.7245	19.2424
0.6777	6.0	918	0.3940	2.0655	19.5872
0.463	7.0	1071	0.3722	2.6043	19.2969
0.463	8.0	1224	0.3512	3.4014	18.864
0.463	9.0	1377	0.3323	4.0558	19.0541
0.3973	10.0	1530	0.3150	4.9264	18.878
0.3973	11.0	1683	0.2989	6.1751	18.1102
0.3973	12.0	1836	0.2845	6.909	18.405
0.3973	13.0	1989	0.2708	8.2081	18.1388
0.3476	14.0	2142	0.2589	9.0267	18.1527
0.3476	15.0	2295	0.2477	9.8007	18.1826
0.3476	16.0	2448	0.2374	11.2825	17.9705
0.309	17.0	2601	0.2282	12.38	17.9427
0.309	18.0	2754	0.2200	13.1971	18.2629
0.309	19.0	2907	0.2127	14.6993	18.0356
0.278	20.0	3060	0.2058	15.8696	17.7944
0.278	21.0	3213	0.2001	17.2214	17.656
0.278	22.0	3366	0.1951	18.3989	17.6769
0.2597	23.0	3519	0.1906	19.6026	17.7543
0.2597	24.0	3672	0.1869	20.6405	17.817
0.2597	25.0	3825	0.1835	20.7913	17.7273
0.2597	26.0	3978	0.1809	21.5904	17.7518
0.2452	27.0	4131	0.1789	21.9249	17.69
0.2452	28.0	4284	0.1775	22.3964	17.6953
0.2452	29.0	4437	0.1767	22.6803	17.7547
0.2379	30.0	4590	0.1764	22.8149	17.7776

Framework versions

• Transformers 4.42.3
• Pytorch 2.1.2
• Datasets 2.20.0
• Tokenizers 0.19.1