python run_speech_recognition_ctc.py \ --dataset_name="timit_asr" \ --model_name_or_path="facebook/wav2vec2-xls-r-300m" \ --dataset_config_name="clean" \ --train_split_name="train" \ --output_dir="./" \ --overwrite_output_dir \ --num_train_epochs="50" \ --max_train_samples="50000" \ --max_duration_in_seconds="8" \ --per_device_train_batch_size="8" \ --per_device_eval_batch_size="8" \ --gradient_accumulation_steps="4" \ --logging_steps="100" \ --save_steps="1000" \ --eval_steps="1000" \ --learning_rate="5e-5" \ --warmup_steps="2000" \ --eval_metrics="wer" \ --length_column_name="input_length" \ --evaluation_strategy="steps" \ --text_column_name="sentence" \ --chars_to_ignore , ? . ! \- \; \: \" “ % ‘ ” � — ’ … – \ --layerdrop="0.0" \ --activation_dropout="0.1" \ --save_total_limit="3" \ --freeze_feature_encoder \ --feat_proj_dropout="0.0" \ --mask_time_prob="0.75" \ --mask_time_length="10" \ --mask_feature_prob="0.25" \ --mask_feature_length="64" \ --gradient_checkpointing \ --use_auth_token \ --fp16 \ --group_by_length \ --do_train --do_eval \ --push_to_hub