File size: 693 Bytes
1ae875b |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 |
from datasets import load_dataset
from transformers import Trainer, TrainingArguments, Tacotron2ForConditionalGeneration
# تحميل البيانات من Hugging Face Datasets
dataset = load_dataset("your_username/sada2022")
najdi_data = dataset.filter(lambda example: example['SpeakerDialect'] == 'Najdi')
# إعداد النموذج والمعالج
model = Tacotron2ForConditionalGeneration.from_pretrained("facebook/tacotron2")
# إعداد التدريب
training_args = TrainingArguments(output_dir="./results", per_device_train_batch_size=16, num_train_epochs=3)
trainer = Trainer(model=model, args=training_args, train_dataset=najdi_data)
# بدء التدريب
trainer.train()
|