1841cf27b603d72ea1c185f74f8e29f1

This model is a fine-tuned version of Helsinki-NLP/opus-mt-en-ru on the Helsinki-NLP/opus_books [fr-pt] dataset. It achieves the following results on the evaluation set:

• Loss: 2.7953
• Data Size: 1.0
• Epoch Runtime: 3.4190
• Bleu: 3.4204

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: 5e-05
• train_batch_size: 8
• eval_batch_size: 8
• seed: 42
• distributed_type: multi-GPU
• num_devices: 4
• total_train_batch_size: 32
• total_eval_batch_size: 32
• optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: constant
• num_epochs: 50

Training results

Training Loss	Epoch	Step	Validation Loss	Data Size	Epoch Runtime	Bleu
No log	0	0	7.9481	0	0.7836	0.0369
No log	1	31	7.0871	0.0078	1.3340	0.1365
No log	2	62	6.6978	0.0156	1.0362	0.1352
No log	3	93	6.4607	0.0312	1.2240	0.1282
No log	4	124	6.1688	0.0625	1.3449	0.1194
No log	5	155	5.7099	0.125	1.8259	0.1439
No log	6	186	5.1185	0.25	1.7490	0.1737
0.8675	7	217	4.4669	0.5	2.4597	1.1079
0.8675	8.0	248	3.9068	1.0	3.5947	1.2059
3.0058	9.0	279	3.5783	1.0	3.3976	1.6261
3.6555	10.0	310	3.3619	1.0	3.0747	1.9997
3.6555	11.0	341	3.2250	1.0	3.5968	2.2291
3.2322	12.0	372	3.1106	1.0	3.2125	2.3759
2.8921	13.0	403	3.0129	1.0	3.1930	2.5502
2.8921	14.0	434	2.9536	1.0	3.3047	2.5684
2.6366	15.0	465	2.8945	1.0	3.3760	2.7098
2.6366	16.0	496	2.8617	1.0	3.3897	2.7397
2.3965	17.0	527	2.8539	1.0	3.3920	2.8475
2.1999	18.0	558	2.7918	1.0	4.0243	2.8580
2.1999	19.0	589	2.8037	1.0	3.3992	3.0532
2.0315	20.0	620	2.7839	1.0	3.1250	3.1286
1.8727	21.0	651	2.7748	1.0	3.1799	3.0327
1.8727	22.0	682	2.7709	1.0	3.2287	3.1335
1.715	23.0	713	2.7748	1.0	3.2648	3.2209
1.715	24.0	744	2.7720	1.0	3.3691	3.3815
1.5815	25.0	775	2.7848	1.0	3.2767	3.3297
1.4563	26.0	806	2.7953	1.0	3.4190	3.4204

Framework versions

• Transformers 4.57.0
• Pytorch 2.8.0+cu128
• Datasets 4.2.0
• Tokenizers 0.22.1