mms-amh-finetuned

This model is a fine-tuned version of facebook/mms-1b-all on the generator dataset. It achieves the following results on the evaluation set:

• Loss: 0.2463
• Wer: 0.1681
• Cer: 0.0487

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: 0.0001
• train_batch_size: 16
• eval_batch_size: 8
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 64
• optimizer: Use OptimizerNames.ADAMW_TORCH 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: 99
• num_epochs: 3
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer	Cer
2.0536	0.5964	198	0.4753	0.3800	0.1057
1.7597	1.1928	396	0.3566	0.2586	0.0729
1.4389	1.7892	594	0.3160	0.2259	0.0634
1.2412	2.3855	792	0.2711	0.1872	0.0538
0.9865	2.9819	990	0.2463	0.1681	0.0487
0.9865	3.0	996	0.2459	0.1682	0.0487

Framework versions

• Transformers 5.5.4
• Pytorch 2.7.1+cu118
• Datasets 4.8.4
• Tokenizers 0.22.2