mBART-large-50-si-spelling-v2

This model is a fine-tuned version of facebook/mbart-large-50 on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.0309
• Bleu: 69.0972
• Exact Match: 0.5423

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: 32
• eval_batch_size: 16
• 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
• lr_scheduler_warmup_steps: 500
• num_epochs: 5

Training results

Training Loss	Epoch	Step	Validation Loss	Bleu	Exact Match
0.0403	0.4990	500	0.0408	58.0063	0.4076
0.0325	0.9980	1000	0.0317	63.4292	0.4707
0.0241	1.4970	1500	0.0292	66.1302	0.5019
0.0213	1.9960	2000	0.0269	68.4227	0.5305
0.0109	2.4950	2500	0.0302	67.7918	0.5243
0.0112	2.9940	3000	0.0285	68.6315	0.5367
0.0065	3.4930	3500	0.0309	69.0972	0.5423

Framework versions

• Transformers 5.0.0
• Pytorch 2.8.0+cu128
• Datasets 4.5.0
• Tokenizers 0.22.2