Hugging Face's logo

Arbi-Houssem
/
speecht5_ar_tn_1.7

Text-to-Audio
Transformers
TensorBoard
Safetensors
Arabic
speecht5
Generated from Trainer
Inference Endpoints
Model card Files Metrics Community
speecht5_ar_tn_1.7 / README.md
Arbi-Houssem's picture
Arbi-Houssem
End of training
7f3e6d2 verified
preview code
|
raw
history blame contribute delete
1.75 kB
metadata
language:
  - ara
license: mit
base_model: MBZUAI/speecht5_tts_clartts_ar
tags:
  - generated_from_trainer
datasets:
  - Arbi-Houssem/Tunisian_dataset_STT-TTS15s_filtred1.0_Mixed
model-index:
  - name: SpeechT5 TTS Tunisien
    results: []

SpeechT5 TTS Tunisien

This model is a fine-tuned version of MBZUAI/speecht5_tts_clartts_ar on the Tunisian_dataset_STT-TTS15s_filtred1.0_Mixed dataset. It achieves the following results on the evaluation set:

  • Loss: 0.4833

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: 4
  • eval_batch_size: 2
  • seed: 42
  • gradient_accumulation_steps: 8
  • 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: 500
  • training_steps: 4000
  • mixed_precision_training: Native AMP

Training results

Training Loss Epoch Step Validation Loss
0.5487 53.69 1000 0.4952
0.4979 107.38 2000 0.4845
0.488 161.07 3000 0.4835
0.4856 214.77 4000 0.4833

Framework versions

  • Transformers 4.35.1
  • Pytorch 2.3.1+cu121
  • Datasets 2.14.7
  • Tokenizers 0.14.1