966b14c0bca6c600fbe49fc7397f880b

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

• Loss: 2.3309
• Data Size: 1.0
• Epoch Runtime: 204.1448
• Bleu: 8.1907

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	6.9235	0	17.2948	0.4027
No log	1	806	5.2303	0.0078	18.9460	0.7872
No log	2	1612	4.3241	0.0156	22.1377	1.5498
No log	3	2418	3.8067	0.0312	25.2368	2.3395
0.1342	4	3224	3.3675	0.0625	30.9939	3.4374
3.1773	5	4030	2.9721	0.125	42.9696	4.5035
2.7058	6	4836	2.6135	0.25	67.1469	5.8561
2.3095	7	5642	2.3150	0.5	112.4183	6.6937
1.9767	8.0	6448	2.0900	1.0	205.0316	8.1492
1.6427	9.0	7254	2.0222	1.0	204.8774	9.0628
1.3888	10.0	8060	2.0285	1.0	204.3249	9.0692
1.1684	11.0	8866	2.1046	1.0	203.4881	9.2240
0.9556	12.0	9672	2.1836	1.0	204.7703	8.6930
0.7596	13.0	10478	2.3309	1.0	204.1448	8.1907

Framework versions

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