opus-mt-ar-en-finetuned-ar-to-en-final2

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: 1.6824
• Bleu: 0.2641
• Meteor: 0.5714
• Gen Len: 38.5649

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: 1e-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: 15

Training results

Training Loss	Epoch	Step	Validation Loss	Bleu	Meteor	Gen Len
No log	1.0	360	1.8975	0.1623	0.5061	52.5552
2.0487	2.0	720	1.8039	0.1923	0.5345	47.5779
1.6565	3.0	1080	1.7538	0.2254	0.5493	41.7695
1.6565	4.0	1440	1.7228	0.2231	0.5539	45.3669
1.4299	5.0	1800	1.6990	0.2467	0.5579	39.5974
1.2508	6.0	2160	1.6865	0.2493	0.5611	39.8474
1.1161	7.0	2520	1.6754	0.2516	0.564	39.7597
1.1161	8.0	2880	1.6740	0.254	0.562	39.5942
1.0023	9.0	3240	1.6727	0.2584	0.5707	39.8896
0.9296	10.0	3600	1.6713	0.2644	0.5709	38.6688
0.9296	11.0	3960	1.6753	0.2717	0.5721	38.4481
0.8645	12.0	4320	1.6775	0.2687	0.5742	38.5779
0.8232	13.0	4680	1.6823	0.2542	0.5708	40.3019
0.7956	14.0	5040	1.6824	0.2641	0.5714	38.5649

Framework versions

• Transformers 4.50.3
• Pytorch 2.6.0+cu124
• Datasets 3.5.0
• Tokenizers 0.21.1