finetuned_helsinki_peft_model__en_to_ar

This model is a fine-tuned version of Helsinki-NLP/opus-mt-en-ar on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 1.7705
• Bleu: 26.8378
• Gen Len: 13.5315

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: 24
• eval_batch_size: 24
• seed: 42
• optimizer: Use 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: 500
• training_steps: 10000
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Bleu	Gen Len
2.1194	0.2	500	1.8368	26.7617	13.59
1.9481	0.4	1000	1.8082	26.9303	13.562
1.9398	0.6	1500	1.7980	26.9162	13.467
1.9521	0.8	2000	1.7925	26.9365	13.5355
1.9478	1.0	2500	1.7894	27.1345	13.5425
1.9272	1.2	3000	1.7871	27.4766	13.501
1.8803	1.4	3500	1.7809	27.106	13.5995
1.911	1.6	4000	1.7781	27.1273	13.5465
1.9052	1.8	4500	1.7756	26.9803	13.554
1.9043	2.0	5000	1.7742	27.0919	13.5305
1.8904	2.2	5500	1.7742	26.9754	13.4635
1.8951	2.4	6000	1.7727	26.9621	13.529
1.8741	2.6	6500	1.7727	26.991	13.4795
1.9037	2.8	7000	1.7716	26.9663	13.5065
1.8931	3.0	7500	1.7716	26.8411	13.54
1.925	3.2	8000	1.7707	26.8684	13.5635
1.8847	3.4	8500	1.7708	26.892	13.5185
1.8693	3.6	9000	1.7706	26.8667	13.5075
1.9089	3.8	9500	1.7705	26.8321	13.5285
1.8638	4.0	10000	1.7705	26.8378	13.5315

Framework versions

• PEFT 0.14.1.dev0
• Transformers 4.47.1
• Pytorch 2.5.1+cpu
• Datasets 3.2.0
• Tokenizers 0.21.0