zmb-chapter-audio-dataset-force-aligned-speecht5

This model is a fine-tuned version of microsoft/speecht5_tts on the None dataset. It achieves the following results on the evaluation set:

Model description

More information needed

More information needed

More information needed

The following hyperparameters were used during training:

learning_rate: 0.0001
train_batch_size: 8
eval_batch_size: 8
seed: 3407
gradient_accumulation_steps: 4
total_train_batch_size: 32
optimizer: Use OptimizerNames.ADAMW_TORCH_FUSED with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
lr_scheduler_type: cosine
lr_scheduler_warmup_steps: 4000
training_steps: 40000
mixed_precision_training: Native AMP

Training Loss	Epoch	Step	Validation Loss
0.0625	40.0	1000	0.0545
0.055	80.0	2000	0.0515
0.0476	120.0	3000	0.0528
0.0464	160.0	4000	0.0543
0.0448	200.0	5000	0.0574
0.0417	240.0	6000	0.0585
0.0367	280.0	7000	0.0590
0.0417	320.0	8000	0.0628
0.0344	360.0	9000	0.0627
0.0334	400.0	10000	0.0657
0.0311	440.0	11000	0.0672
0.031	480.0	12000	0.0685
0.0309	520.0	13000	0.0691
0.0285	560.0	14000	0.0699
0.0313	600.0	15000	0.0720
0.0275	640.0	16000	0.0720
0.0283	680.0	17000	0.0724
0.0281	720.0	18000	0.0727
0.0276	760.0	19000	0.0758
0.0256	800.0	20000	0.0741
0.0267	840.0	21000	0.0741
0.0287	880.0	22000	0.0763
0.0261	920.0	23000	0.0761
0.0246	960.0	24000	0.0776
0.0229	1000.0	25000	0.0778
0.0241	1040.0	26000	0.0784
0.0233	1080.0	27000	0.0794
0.0233	1120.0	28000	0.0785
0.0237	1160.0	29000	0.0787
0.0235	1200.0	30000	0.0791
0.0227	1240.0	31000	0.0799
0.0217	1280.0	32000	0.0805
0.0228	1320.0	33000	0.0804
0.0242	1360.0	34000	0.0798
0.0238	1400.0	35000	0.0803
0.0229	1440.0	36000	0.0804
0.0229	1480.0	37000	0.0803
0.0235	1520.0	38000	0.0802
0.0232	1560.0	39000	0.0798
0.0224	1600.0	40000	0.0800

Safetensors

Model size

0.1B params

Tensor type

F32

Base model

Finetuned

this model