zlm_b32_le4_s8000 / README.md
mikhail-panzo's picture
End of training
1a88145 verified
metadata
license: mit
base_model: microsoft/speecht5_tts
tags:
  - generated_from_trainer
model-index:
  - name: zlm_b32_le4_s8000
    results: []

zlm_b32_le4_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: 0.3262

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: 16
  • eval_batch_size: 8
  • seed: 42
  • gradient_accumulation_steps: 2
  • 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: 2000
  • training_steps: 8000
  • mixed_precision_training: Native AMP

Training results

Training Loss Epoch Step Validation Loss
0.5552 0.2094 500 0.4883
0.4913 0.4188 1000 0.4266
0.446 0.6283 1500 0.3975
0.4222 0.8377 2000 0.3949
0.4273 1.0471 2500 0.3829
0.4028 1.2565 3000 0.3674
0.3941 1.4660 3500 0.3616
0.3871 1.6754 4000 0.3519
0.3828 1.8848 4500 0.3493
0.3954 2.0942 5000 0.3490
0.381 2.3037 5500 0.3398
0.372 2.5131 6000 0.3372
0.3642 2.7225 6500 0.3314
0.3692 2.9319 7000 0.3308
0.3555 3.1414 7500 0.3283
0.3536 3.3508 8000 0.3262

Framework versions

  • Transformers 4.41.0.dev0
  • Pytorch 2.2.1+cu121
  • Datasets 2.19.0
  • Tokenizers 0.19.1