mms-300m-ewe-cmu

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: 0.4662
• Wer: 0.3375
• Cer: 0.1085

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: 0.0003
• 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
7.6970	0.5313	250	3.0084	1.0	1.0
3.9096	1.0616	500	1.5903	0.9570	0.4932
1.2426	1.5930	750	0.5749	0.4546	0.1441
1.0259	2.1233	1000	0.5037	0.3836	0.1223
0.9771	2.6546	1250	0.4789	0.3595	0.1161
1.6200	3.1849	1500	0.4634	0.3711	0.1176
0.8915	3.7163	1750	0.4558	0.3480	0.1101
0.8420	4.2465	2000	0.4475	0.3474	0.1113
0.7808	4.7779	2250	0.4404	0.3404	0.1087
0.6902	5.3082	2500	0.4533	0.3491	0.1122
0.7328	5.8395	2750	0.4560	0.3353	0.1074
0.6970	6.3698	3000	0.4662	0.3375	0.1085

Framework versions

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