opus-mt-ar-en-finetuned-ar-to-en-4

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

• Loss: 1.7059
• Bleu: 0.2701
• Meteor: 0.5767
• Gen Len: 38.9481

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: 4
• eval_batch_size: 4
• 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
• lr_scheduler_warmup_steps: 20
• num_epochs: 10

Training results

Training Loss	Epoch	Step	Validation Loss	Bleu	Meteor	Gen Len
No log	1.0	360	1.8146	0.2219	0.5321	40.1753
1.9496	2.0	720	1.7198	0.2276	0.5554	42.513
1.4434	3.0	1080	1.6687	0.2294	0.5609	44.2987
1.4434	4.0	1440	1.6512	0.2433	0.558	41.5877
1.1113	5.0	1800	1.6547	0.2421	0.5591	40.9805
0.9037	6.0	2160	1.6629	0.2484	0.5612	39.5877
0.7819	7.0	2520	1.6840	0.2565	0.5717	40.1136
0.7819	8.0	2880	1.6961	0.2667	0.5711	38.6299
0.6853	9.0	3240	1.7043	0.267	0.5743	38.974
0.6277	10.0	3600	1.7059	0.2701	0.5767	38.9481

Framework versions

• Transformers 4.48.3
• Pytorch 2.5.1+cu124
• Datasets 3.3.1
• Tokenizers 0.21.0