README.md

metadata

license: mit
base_model: microsoft/speecht5_tts
tags:
  - generated_from_trainer
model-index:
  - name: speecht5_tts
    results: []

speecht5_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: 0.5139

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: 2
• eval_batch_size: 2
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 500
• training_steps: 10000
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss
No log	1.92	250	0.6150
0.7483	3.85	500	0.5494
0.7483	5.77	750	0.5001
0.5482	7.69	1000	0.4861
0.5482	9.62	1250	0.4792
0.502	11.54	1500	0.4786
0.502	13.46	1750	0.4804
0.4794	15.38	2000	0.4803
0.4794	17.31	2250	0.4724
0.4685	19.23	2500	0.4801
0.4685	21.15	2750	0.4740
0.4553	23.08	3000	0.4840
0.4553	25.0	3250	0.4857
0.4567	26.92	3500	0.4792
0.4567	28.85	3750	0.4831
0.445	30.77	4000	0.4884
0.445	32.69	4250	0.4845
0.4412	34.62	4500	0.4944
0.4412	36.54	4750	0.4940
0.4373	38.46	5000	0.4863
0.4373	40.38	5250	0.4899
0.4353	42.31	5500	0.4954
0.4353	44.23	5750	0.5005
0.4265	46.15	6000	0.4994
0.4265	48.08	6250	0.4918
0.4285	50.0	6500	0.5022
0.4285	51.92	6750	0.4939
0.4209	53.85	7000	0.4989
0.4209	55.77	7250	0.4959
0.4206	57.69	7500	0.5013
0.4206	59.62	7750	0.5061
0.4189	61.54	8000	0.5092
0.4189	63.46	8250	0.5084
0.422	65.38	8500	0.5116
0.422	67.31	8750	0.5115
0.415	69.23	9000	0.5100
0.415	71.15	9250	0.5121
0.4179	73.08	9500	0.5112
0.4179	75.0	9750	0.5115
0.4139	76.92	10000	0.5139

Framework versions

• Transformers 4.36.0.dev0
• Pytorch 2.1.0+cu121
• Datasets 2.15.0
• Tokenizers 0.14.1