mms-300m-mlg-onitsikix

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: 0.1234
• Wer: 0.1347
• Cer: 0.0304

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
5.0400	0.8993	500	2.3270	1.0	0.8235
0.4158	1.7986	1000	0.1677	0.2040	0.0455
0.3160	2.6978	1500	0.1356	0.1629	0.0367
0.2501	3.5971	2000	0.1270	0.1508	0.0337
0.3604	4.4964	2500	0.1235	0.1457	0.0328
0.1967	5.3957	3000	0.1209	0.1361	0.0307
0.1873	6.2950	3500	0.1317	0.1369	0.0312
0.1551	7.1942	4000	0.1223	0.1324	0.0300
0.1464	8.0935	4500	0.1234	0.1347	0.0304

Framework versions

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