mt-sq-sv-finetuned

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

• Loss: 1.2250
• Bleu: 47.0111

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: 5e-06
• train_batch_size: 24
• eval_batch_size: 4
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 10
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Bleu
1.7042	1.0	4219	1.4806	41.9650
1.5537	2.0	8438	1.3955	43.1524
1.4352	3.0	12657	1.3142	44.4373
1.3346	4.0	16876	1.2793	45.2265
1.2847	5.0	21095	1.2597	45.8071
1.2821	6.0	25314	1.2454	46.3737
1.2342	7.0	29533	1.2363	46.6308
1.2092	8.0	33752	1.2301	46.8227
1.1766	9.0	37971	1.2260	46.9719
1.1836	10.0	42190	1.2250	47.0111

Framework versions

• Transformers 4.25.0.dev0
• Pytorch 1.13.0+cu117
• Datasets 2.6.1
• Tokenizers 0.13.1