mbart-50-spa-shp

This model is a fine-tuned version of facebook/mbart-large-50-many-to-many-mmt on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.5238
• Chrf: 44.1438
• Bleu: 11.7655

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: 2e-05
• train_batch_size: 4
• eval_batch_size: 4
• seed: 42
• optimizer: Use OptimizerNames.ADAMW_TORCH_FUSED with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: linear
• num_epochs: 50
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Chrf	Bleu
0.6263	1.0	5524	0.5003	37.1850	7.3508
0.391	2.0	11048	0.4422	40.3175	10.2132
0.3138	3.0	16572	0.4228	42.6570	10.8193
0.2206	4.0	22096	0.4260	43.7274	11.6145
0.1767	5.0	27620	0.4408	43.5213	11.9067
0.1522	6.0	33144	0.4654	44.6322	12.3956
0.1382	7.0	38668	0.4957	43.9743	11.7023
0.1086	8.0	44192	0.5238	44.1438	11.7655

Framework versions

• Transformers 4.57.0
• Pytorch 2.8.0+cu129
• Datasets 4.2.0
• Tokenizers 0.22.1