b0a191b341454d814ab8768b60b3dcbf

This model is a fine-tuned version of facebook/mbart-large-50-many-to-one-mmt on the Helsinki-NLP/opus_books [de-es] dataset. It achieves the following results on the evaluation set:

• Loss: 2.8865
• Data Size: 1.0
• Epoch Runtime: 174.4245
• Bleu: 5.2599

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	6.6781	0	14.7388	0.5201
No log	1	688	5.1289	0.0078	16.4687	0.8397
No log	2	1376	4.2997	0.0156	18.9362	1.2548
No log	3	2064	3.8249	0.0312	21.9662	1.8110
0.1485	4	2752	3.4606	0.0625	27.0649	2.4140
0.2697	5	3440	3.1336	0.125	37.4588	3.0243
2.8758	6	4128	2.8331	0.25	56.4544	3.9237
2.5106	7	4816	2.5732	0.5	96.6185	3.5624
2.2019	8.0	5504	2.3631	1.0	175.3382	5.4216
1.8291	9.0	6192	2.3291	1.0	173.1523	5.6092
1.5207	10.0	6880	2.3914	1.0	173.2789	5.5179
1.2361	11.0	7568	2.5498	1.0	173.9626	5.4698
1.0001	12.0	8256	2.7248	1.0	173.9492	5.3054
0.7843	13.0	8944	2.8865	1.0	174.4245	5.2599

Framework versions

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