opus-mt-ar-en-finetuned-ar-to-en-5

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: 1.7828
• Bleu: 0.2786
• Meteor: 0.5622
• Gen Len: 33.3799

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	1.9032	0.2061	0.518	37.5325
2.0984	2.0	720	1.8059	0.242	0.5414	35.3799
1.5237	3.0	1080	1.7625	0.2572	0.5601	35.7532
1.5237	4.0	1440	1.7364	0.2542	0.5644	36.4968
1.1827	5.0	1800	1.7339	0.2713	0.5602	33.7338
0.967	6.0	2160	1.7407	0.2489	0.5582	36.4448
0.8214	7.0	2520	1.7554	0.262	0.5571	34.6688
0.8214	8.0	2880	1.7708	0.2773	0.5647	33.3279
0.7299	9.0	3240	1.7803	0.2748	0.5638	33.7695
0.6722	10.0	3600	1.7828	0.2786	0.5622	33.3799

Framework versions

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