speecht5_finetuned_ar

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: 0.4771

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: 5e-06
• train_batch_size: 16
• eval_batch_size: 8
• 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: linear
• lr_scheduler_warmup_steps: 1000
• training_steps: 20000
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss
1.1297	20.0	1000	0.5249
1.0651	40.0	2000	0.5058
1.0377	60.0	3000	0.4958
1.024	80.0	4000	0.4904
1.0095	100.0	5000	0.4870
1.0052	120.0	6000	0.4832
0.9984	140.0	7000	0.4830
0.9908	160.0	8000	0.4818
0.99	180.0	9000	0.4810
0.98	200.0	10000	0.4814
0.9765	220.0	11000	0.4793
0.971	240.0	12000	0.4786
0.9756	260.0	13000	0.4823
0.9703	280.0	14000	0.4792
0.9718	300.0	15000	0.4780
0.959	320.0	16000	0.4792
0.963	340.0	17000	0.4789
0.9639	360.0	18000	0.4789
0.9613	380.0	19000	0.4786
0.9577	400.0	20000	0.4771

Framework versions

• Transformers 4.46.1
• Pytorch 2.7.1+cu118
• Datasets 4.0.0
• Tokenizers 0.20.3