addff7d285444a28a2e84bbf2541c25d

This model is a fine-tuned version of facebook/mbart-large-cc25 on the Helsinki-NLP/opus_books [en-es] dataset. It achieves the following results on the evaluation set:

• Loss: 2.2050
• Data Size: 1.0
• Epoch Runtime: 606.3494
• Bleu: 9.1844

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	9.4701	0	51.3156	0.1536
No log	1	2336	3.3513	0.0078	55.7935	3.5318
0.056	2	4672	2.8710	0.0156	61.6633	4.4855
0.0673	3	7008	2.5496	0.0312	71.2139	5.9108
2.4937	4	9344	2.3791	0.0625	88.2310	7.0117
2.3671	5	11680	2.2512	0.125	122.0304	7.1711
2.1738	6	14016	2.1369	0.25	191.3973	7.5971
2.0215	7	16352	1.9901	0.5	327.3715	8.5435
1.8497	8.0	18688	1.8931	1.0	603.6647	8.5576
1.5951	9.0	21024	1.8656	1.0	600.8735	9.8045
1.4013	10.0	23360	1.9068	1.0	600.9132	9.3170
1.2045	11.0	25696	1.9717	1.0	602.5358	9.7082
0.9987	12.0	28032	2.0725	1.0	612.5475	10.0370
0.8425	13.0	30368	2.2050	1.0	606.3494	9.1844

Framework versions

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