python main.py `
  --model_name_or_path="openai/whisper-medium" `
  --dataset_name="mozilla-foundation/common_voice_17_0" `
  --dataset_config_name="kk" `
  --language="kazakh" `
  --train_split_name="train+validation" `
  --eval_split_name="test" `
  --model_index_name="Whisper Medium Kazakh" `
  --max_steps="5000" `
  --output_dir="./" `
  --per_device_train_batch_size="16" `
  --per_device_eval_batch_size="8" `
  --logging_steps="25" `
  --learning_rate="1e-5" `
  --warmup_steps="500" `
  --evaluation_strategy="steps" `
  --eval_steps="1000" `
  --save_strategy="steps" `
  --save_steps="1000" `
  --generation_max_length="225" `
  --length_column_name="input_length" `
  --max_duration_in_seconds="30" `
  --text_column_name="sentence" `
  --freeze_feature_encoder="False" `
  --report_to="tensorboard" `
  --metric_for_best_model="wer" `
  --greater_is_better="False" `
  --load_best_model_at_end `
  --gradient_checkpointing `
  --fp16 `
  --overwrite_output_dir `
  --do_train `
  --do_eval `
  --predict_with_generate `
  --do_normalize_eval `
  --streaming `
  --push_to_hub