speecht5_finetuned_uz

This model is a fine-tuned version of speecht5 on the google/fleurs and the mozilla-foundation/common_voice_17_0 datasets. It achieves the following results on the evaluation set:

• Loss: 0.3808

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-06
• train_batch_size: 32
• eval_batch_size: 32
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 128
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: cosine_with_restarts
• lr_scheduler_warmup_steps: 500
• num_epochs: 100

Training results

Training Loss	Epoch	Step	Validation Loss
No log	0	0	0.3819
0.4112	3.3333	40	0.3825
0.4122	6.6667	80	0.3815
0.4133	10.0	120	0.3814
0.4132	13.3333	160	0.3804
0.412	16.6667	200	0.3756
0.4128	20.0	240	0.3812
0.4157	23.3333	280	0.3808
0.4088	26.6667	320	0.3801
0.4107	30.0	360	0.3806
0.4085	33.3333	400	0.3822
0.413	36.6667	440	0.3811
0.4091	40.0	480	0.3821
0.4114	43.3333	520	0.3770
0.4064	46.6667	560	0.3844
0.4104	50.0	600	0.3804
0.4124	53.3333	640	0.3786
0.4068	56.6667	680	0.3773
0.4072	60.0	720	0.3813
0.4107	63.3333	760	0.3777
0.4109	66.6667	800	0.3759
0.4118	70.0	840	0.3828
0.408	73.3333	880	0.3789
0.4104	76.6667	920	0.3824
0.408	80.0	960	0.3790
0.4096	83.3333	1000	0.3802
0.4069	86.6667	1040	0.3771
0.4119	90.0	1080	0.3774
0.4063	93.3333	1120	0.3829
0.4061	96.6667	1160	0.3795
0.4073	100.0	1200	0.3808

Framework versions

• Transformers 4.45.2
• Pytorch 2.4.1
• Datasets 3.0.1
• Tokenizers 0.20.1