2da5d24fae71677d9db77e9b424e425e

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

• Loss: 2.3215
• Data Size: 1.0
• Epoch Runtime: 204.8652
• Bleu: 8.6303

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	7.5263	0	17.4051	0.2929
No log	1	806	3.7052	0.0078	18.8974	2.0878
No log	2	1612	3.1681	0.0156	22.1537	3.2049
No log	3	2418	2.8621	0.0312	26.1644	4.1265
0.1013	4	3224	2.6206	0.0625	31.9874	4.8944
2.524	5	4030	2.4120	0.125	43.9786	5.8341
2.2544	6	4836	2.2292	0.25	66.6121	6.8878
2.0167	7	5642	2.0622	0.5	110.9904	7.6778
1.7706	8.0	6448	1.9228	1.0	203.4298	8.7388
1.4787	9.0	7254	1.9013	1.0	202.0192	8.7366
1.2406	10.0	8060	1.9539	1.0	203.9049	9.4070
1.043	11.0	8866	2.0609	1.0	205.5862	8.9680
0.845	12.0	9672	2.1800	1.0	204.3644	9.0829
0.6557	13.0	10478	2.3215	1.0	204.8652	8.6303

Framework versions

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