english-to-darija-2

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

• Loss: 0.8514
• Bleu: 70.9947
• Gen Len: 9.092

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: 10
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Bleu	Gen Len
1.6032	1.0	4651	1.4540	25.9364	8.9697
1.1191	2.0	9302	1.0805	48.0549	9.0661
0.8048	3.0	13953	0.9419	61.3646	9.1018
0.5978	4.0	18604	0.8939	65.6846	9.1161
0.477	5.0	23255	0.8623	68.0005	9.1049
0.4228	6.0	27906	0.8540	69.1959	9.1276
0.3534	7.0	32557	0.8479	69.944	9.0744
0.305	8.0	37208	0.8473	70.55	9.0987
0.2678	9.0	41859	0.8489	70.8065	9.1166
0.243	10.0	46510	0.8514	70.9947	9.092

Framework versions

• Transformers 4.37.2
• Pytorch 2.2.0+cu121
• Datasets 2.17.0
• Tokenizers 0.15.1