opus-mt-en-ro-finetuned-src-to-tgt

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

• Loss: 2.4552

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: 32
• eval_batch_size: 32
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 30
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss
No log	1.0	35	2.8624
No log	2.0	70	2.6600
No log	3.0	105	2.5610
No log	4.0	140	2.5369
No log	5.0	175	2.4976
No log	6.0	210	2.4731
No log	7.0	245	2.4463
No log	8.0	280	2.4558
No log	9.0	315	2.4386
No log	10.0	350	2.4377
No log	11.0	385	2.4342
No log	12.0	420	2.4332
No log	13.0	455	2.4401
No log	14.0	490	2.4453
2.3749	15.0	525	2.4237
2.3749	16.0	560	2.4336
2.3749	17.0	595	2.4262
2.3749	18.0	630	2.4340
2.3749	19.0	665	2.4358
2.3749	20.0	700	2.4423
2.3749	21.0	735	2.4370
2.3749	22.0	770	2.4404
2.3749	23.0	805	2.4451
2.3749	24.0	840	2.4482
2.3749	25.0	875	2.4477
2.3749	26.0	910	2.4503
2.3749	27.0	945	2.4533
2.3749	28.0	980	2.4570
2.1075	29.0	1015	2.4564
2.1075	30.0	1050	2.4552

Framework versions

• Transformers 4.36.2
• Pytorch 2.1.0+cu121
• Datasets 2.16.1
• Tokenizers 0.15.0