senga-nt-nse-tts

This model is a fine-tuned version of microsoft/speecht5_tts on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.0879

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: 8
• eval_batch_size: 8
• seed: 3407
• gradient_accumulation_steps: 4
• 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: cosine
• lr_scheduler_warmup_steps: 4000
• training_steps: 40000
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss
0.1688	9.1324	2000	0.1078
0.1536	18.2648	4000	0.0979
0.1427	27.3973	6000	0.0946
0.1347	36.5297	8000	0.0987
0.1271	45.6621	10000	0.0925
0.1218	54.7945	12000	0.0911
0.117	63.9269	14000	0.0906
0.1088	73.0594	16000	0.0913
0.1126	82.1918	18000	0.0901
0.1064	91.3242	20000	0.0891
0.104	100.4566	22000	0.0898
0.0996	109.5890	24000	0.0898
0.0974	118.7215	26000	0.0888
0.0918	127.8539	28000	0.0886
0.0945	136.9863	30000	0.0883
0.0893	146.1187	32000	0.0891
0.0899	155.2511	34000	0.0885
0.0886	164.3836	36000	0.0879
0.0855	173.5160	38000	0.0886
0.0881	182.6484	40000	0.0879

Framework versions

• Transformers 4.57.1
• Pytorch 2.8.0+cu128
• Datasets 4.2.0
• Tokenizers 0.22.2