907dd443bbd98db036bbffcd473705bb

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

• Loss: 2.0561
• Data Size: 1.0
• Epoch Runtime: 585.6671
• Bleu: 9.3001

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	4.0412	0	48.6728	4.9704
No log	1	2336	2.5093	0.0078	53.8947	14.4456
0.0377	2	4672	2.3854	0.0156	58.8653	11.5857
0.0531	3	7008	2.2885	0.0312	68.3426	9.4640
2.19	4	9344	2.1915	0.0625	84.0403	9.5793
2.0981	5	11680	2.0762	0.125	116.1446	13.8308
1.9194	6	14016	1.9619	0.25	184.0402	11.5032
1.8001	7	16352	1.8519	0.5	319.6601	12.1335
1.6215	8.0	18688	1.7680	1.0	600.1598	9.4563
1.3544	9.0	21024	1.7784	1.0	589.7112	9.5562
1.1577	10.0	23360	1.8256	1.0	587.3730	9.5030
0.9461	11.0	25696	1.9410	1.0	589.6437	11.4156
0.7754	12.0	28032	2.0561	1.0	585.6671	9.3001

Framework versions

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