Youtube3kTTSModel

This model is a fine-tuned version of microsoft/speecht5_tts on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.4839

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: 4
• eval_batch_size: 2
• seed: 42
• gradient_accumulation_steps: 8
• total_train_batch_size: 32
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 100
• training_steps: 5000
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss
0.6301	0.2222	100	0.5639
0.6013	0.4444	200	0.5471
0.5666	0.6667	300	0.5315
0.5632	0.8889	400	0.5254
0.5547	1.1111	500	0.5184
0.5583	1.3333	600	0.5211
0.5527	1.5556	700	0.5150
0.5508	1.7778	800	0.5123
0.5432	2.0	900	0.5135
0.5478	2.2222	1000	0.5077
0.5419	2.4444	1100	0.5073
0.5439	2.6667	1200	0.5083
0.5381	2.8889	1300	0.5108
0.5355	3.1111	1400	0.5075
0.5317	3.3333	1500	0.5053
0.5345	3.5556	1600	0.5022
0.5329	3.7778	1700	0.5006
0.53	4.0	1800	0.4965
0.5261	4.2222	1900	0.4971
0.5272	4.4444	2000	0.4976
0.5272	4.6667	2100	0.4943
0.5282	4.8889	2200	0.4938
0.5188	5.1111	2300	0.4980
0.523	5.3333	2400	0.4894
0.5225	5.5556	2500	0.4915
0.5178	5.7778	2600	0.4960
0.5165	6.0	2700	0.4893
0.5098	6.2222	2800	0.4892
0.512	6.4444	2900	0.4868
0.5177	6.6667	3000	0.4868
0.5128	6.8889	3100	0.4883
0.5062	7.1111	3200	0.4852
0.5104	7.3333	3300	0.4898
0.5126	7.5556	3400	0.4887
0.5093	7.7778	3500	0.4908
0.5075	8.0	3600	0.4828
0.5029	8.2222	3700	0.4842
0.5079	8.4444	3800	0.4850
0.5049	8.6667	3900	0.4853
0.5034	8.8889	4000	0.4849
0.4984	9.1111	4100	0.4833
0.5079	9.3333	4200	0.4863
0.5023	9.5556	4300	0.4830
0.5023	9.7778	4400	0.4833
0.5037	10.0	4500	0.4825
0.5035	10.2222	4600	0.4822
0.5011	10.4444	4700	0.4826
0.4969	10.6667	4800	0.4815
0.4958	10.8889	4900	0.4839
0.4972	11.1111	5000	0.4839

Framework versions

• Transformers 4.44.2
• Pytorch 2.4.1+cu121
• Datasets 3.0.0
• Tokenizers 0.19.1