metadata

library_name: transformers
license: mit
base_model: microsoft/speecht5_tts
tags:
  - generated_from_trainer
model-index:
  - name: speecht5_finetuned_Mar
    results: []

speecht5_finetuned_Mar

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: 0.4785

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: 1
eval_batch_size: 1
seed: 42
gradient_accumulation_steps: 8
total_train_batch_size: 8
optimizer: Use 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: 50
training_steps: 3000
mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss
0.6818	0.7976	100	0.6565
0.6107	1.5952	200	0.5797
0.6033	2.3928	300	0.5716
0.5893	3.1904	400	0.5686
0.5919	3.9880	500	0.5624
0.5732	4.7856	600	0.5932
0.5677	5.5833	700	0.5485
0.5521	6.3809	800	0.5407
0.5513	7.1785	900	0.5348
0.5488	7.9761	1000	0.5366
0.5344	8.7737	1100	0.5267
0.5329	9.5713	1200	0.5163
0.5118	10.3689	1300	0.5172
0.5126	11.1665	1400	0.5089
0.5256	11.9641	1500	0.5104
0.5126	12.7617	1600	0.5054
0.5034	13.5593	1700	0.5023
0.4986	14.3569	1800	0.4974
0.4926	15.1545	1900	0.5081
0.4961	15.9521	2000	0.4929
0.4886	16.7498	2100	0.4863
0.491	17.5474	2200	0.4925
0.4845	18.3450	2300	0.4859
0.4726	19.1426	2400	0.4835
0.4786	19.9402	2500	0.4829
0.4732	20.7378	2600	0.4800
0.4713	21.5354	2700	0.4737
0.4571	22.3330	2800	0.4789
0.4559	23.1306	2900	0.4771
0.4552	23.9282	3000	0.4785

Framework versions

Transformers 4.46.3
Pytorch 2.5.1+cu121
Datasets 3.1.0
Tokenizers 0.20.3