opus-mt-en-ru-finetuned-en-to-ru-Legal

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

• Loss: 0.8561
• Bleu: 46.7284
• Gen Len: 23.1317

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: 16
• eval_batch_size: 16
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 10

Training results

Training Loss	Epoch	Step	Validation Loss	Bleu	Gen Len
No log	1.0	387	1.1719	34.0562	22.991
1.524	2.0	774	1.0342	37.7233	23.0052
1.0226	3.0	1161	0.9595	40.0983	22.9755
0.8066	4.0	1548	0.9188	41.9634	23.1162
0.8066	5.0	1935	0.8907	43.6537	23.0923
0.6637	6.0	2322	0.8771	44.5208	23.1097
0.5697	7.0	2709	0.8669	45.5589	23.1388
0.5175	8.0	3096	0.8603	46.2211	23.2356
0.5175	9.0	3483	0.8566	46.7201	23.1375
0.4768	10.0	3870	0.8561	46.7284	23.1317

Framework versions

• Transformers 4.13.0
• Pytorch 1.12.0
• Datasets 2.4.0
• Tokenizers 0.10.3