This model is a fine-tuned version of facebook/mms-1b-fl102 on the common_voice_17_0 dataset. It achieves the following results on the evaluation set:

Loss: 6.7720
Wer: 1.1976

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: 32
eval_batch_size: 8
seed: 42
optimizer: Use 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: 8
num_epochs: 4
mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
5.9826	0.4	16	8.2570	1.0439
4.2851	0.8	32	8.1419	1.0687
8.2118	1.2	48	7.8111	1.0730
4.362	1.6	64	7.3669	1.1766
7.9241	2.0	80	7.0512	1.2072
4.5478	2.4	96	6.9138	1.2007
3.9726	2.8	112	6.8219	1.2193
6.0029	3.2	128	6.8214	1.2069
4.328	3.6	144	6.8251	1.2014
6.5417	4.0	160	6.7720	1.1976

Framework versions

Transformers 4.46.0.dev0
Pytorch 2.4.0+cu121
Datasets 3.0.1
Tokenizers 0.20.1

abatpool/mal-mms-epoch4

mal-mms-epoch4

Model description

Intended uses & limitations

Training and evaluation data

Training procedure

Training hyperparameters

Training results

Framework versions

abatpool/
mal-mms-epoch4