speecht5_fine_tuned_dhivehi_tts_V005

This model was trained from scratch on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.3451

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: 16
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 32
• 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: cosine_with_restarts
• lr_scheduler_warmup_steps: 500
• training_steps: 15000
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss
0.4463	1.4514	1000	0.4008
0.4319	2.9028	2000	0.3937
0.432	4.3541	3000	0.3926
0.4258	5.8055	4000	0.3834
0.4242	7.2569	5000	0.3828
0.4104	8.7083	6000	0.3761
0.4046	10.1597	7000	0.3702
0.3947	11.6110	8000	0.3635
0.3941	13.0624	9000	0.3593
0.3899	14.5138	10000	0.3547
0.3817	15.9652	11000	0.3509
0.383	17.4165	12000	0.3481
0.3766	18.8679	13000	0.3456
0.3793	20.3193	14000	0.3449
0.3783	21.7707	15000	0.3451

Framework versions

• Transformers 4.49.0.dev0
• Pytorch 2.5.1+cu124
• Datasets 3.2.0
• Tokenizers 0.21.0