mms-300m-sog-emmilly

This model is a fine-tuned version of facebook/mms-300m on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.3946
• Wer: 0.4609
• Cer: 0.0913

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: 2
• total_train_batch_size: 32
• 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: 30
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer	Cer
5.7260	3.0120	500	2.8483	1.0	1.0
5.6272	6.0241	1000	2.8084	1.0	1.0
2.1636	9.0361	1500	0.8853	0.8830	0.2618
0.6961	12.0482	2000	0.3460	0.4921	0.1025
0.4720	15.0602	2500	0.3198	0.4547	0.0922
0.3624	18.0723	3000	0.3302	0.4538	0.0907
0.2997	21.0843	3500	0.3589	0.4584	0.0909
0.2411	24.0964	4000	0.3946	0.4609	0.0913

Framework versions

• Transformers 5.0.0
• Pytorch 2.10.0+cu128
• Datasets 4.0.0
• Tokenizers 0.22.2