Edit model card

fil_b128_le5_s8000

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: 0.4053

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: 32
  • eval_batch_size: 8
  • 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: linear
  • lr_scheduler_warmup_steps: 2000
  • training_steps: 8000
  • mixed_precision_training: Native AMP

Training results

Training Loss Epoch Step Validation Loss
0.5575 43.4783 500 0.4795
0.4877 86.9565 1000 0.4371
0.4582 130.4348 1500 0.4196
0.4486 173.9130 2000 0.4145
0.4353 217.3913 2500 0.4096
0.4261 260.8696 3000 0.4057
0.4206 304.3478 3500 0.4049
0.4117 347.8261 4000 0.4065
0.4123 391.3043 4500 0.4029
0.4117 434.7826 5000 0.4046
0.4053 478.2609 5500 0.4045
0.4064 521.7391 6000 0.4047
0.4058 565.2174 6500 0.4042
0.3995 608.6957 7000 0.4058
0.4076 652.1739 7500 0.4054
0.4042 695.6522 8000 0.4053

Framework versions

  • Transformers 4.41.0.dev0
  • Pytorch 2.2.1+cu121
  • Datasets 2.19.0
  • Tokenizers 0.19.1
Downloads last month
0
Safetensors
Model size
144M params
Tensor type
F32
·

Finetuned from