speecht5_dhivehi_tts_v2

This model is a fine-tuned version of ahmedhassan7030/speecht5_dhivehi_tts_v1 on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.3872

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: 1e-05
• train_batch_size: 128
• eval_batch_size: 16
• 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: cosine_with_restarts
• lr_scheduler_warmup_steps: 500
• training_steps: 20000
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss
0.4307	5.2632	500	0.3856
0.4248	10.5263	1000	0.3862
0.4247	15.7895	1500	0.3873
0.4184	21.0526	2000	0.3870
0.4238	26.3158	2500	0.3897
0.4217	31.5789	3000	0.3846
0.4168	36.8421	3500	0.3824
0.4119	42.1053	4000	0.3846
0.4149	47.3684	4500	0.3870
0.4116	52.6316	5000	0.3861
0.4103	57.8947	5500	0.3846
0.4132	63.1579	6000	0.3872

Framework versions

• Transformers 4.47.1
• Pytorch 2.5.1+cu121
• Datasets 3.2.0
• Tokenizers 0.21.0