Bu TTS Modeli omersaidd/sesame_csm_nisan_kumru_turkish_4000_steps modelini fullfinetuning yöntemi ile omersaidd/tts_mazlum_kiper_tur veri seti ile eğitilmiştir.

Eğitim süresi 2 saattir. 2000 adım olarak eğitilmiş olup. GPU olarak RXT4090 kullanılmıştır.

Eğitim parametreleri:


from transformers import TrainingArguments, Trainer
from unsloth import is_bfloat16_supported

trainer = Trainer(
    model = model,
    train_dataset = processed_ds,
    args = TrainingArguments(
        per_device_train_batch_size = 2,
        gradient_accumulation_steps = 4,
        warmup_steps = 5,
        max_steps = 2000,
        learning_rate = 2e-4,
        fp16 = not is_bfloat16_supported(),
        bf16 = is_bfloat16_supported(),
        logging_steps = 1,
        optim = "adamw_8bit",
        weight_decay = 0.001, # Turn this on if overfitting
        lr_scheduler_type = "linear",
        seed = 3407,
        output_dir = "outputs",
        report_to = "none", # Use TrackIO/WandB etc
    ),
)

Örnek Çıktılar :

Downloads last month
3
Safetensors
Model size
2B params
Tensor type
F32
·
F16
·
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support

Model tree for omersaidd/csm-mazlum-kiper-tur_2000

Base model

sesame/csm-1b
Finetuned
unsloth/csm-1b
Finetuned
(1)
this model

Dataset used to train omersaidd/csm-mazlum-kiper-tur_2000