python run_speech_recognition_ctc.py \ --dataset_name="common_voice" \ --model_name_or_path="./checkpoint-10500" \ --dataset_config_name="tr" \ --output_dir="./" \ --overwrite_output_dir \ --num_train_epochs="5" \ --per_device_train_batch_size="32" \ --gradient_accumulation_steps="1" \ --learning_rate="2e-4" \ --adam_beta1="0.999" \ --adam_beta2="0.9999" \ --warmup_steps="0" \ --evaluation_strategy="steps" \ --text_column_name="sentence" \ --save_steps="500" \ --eval_steps="500" \ --logging_steps="1" \ --layerdrop="0.0" \ --eval_metrics wer cer \ --save_total_limit="1" \ --mask_time_prob="0.3" \ --mask_time_length="10" \ --mask_feature_prob="0.1" \ --mask_feature_length="64" \ --chars_to_ignore , ? . ! - \; \: \" “ % ‘ ” � \ --fp16 \ --group_by_length \ --push_to_hub \ --do_train --do_eval \ --gradient_checkpointing \ --report_to tensorboard \ --use_auth_token