| python join_datasets_asr_ctc.py \ | |
| --dataset_name="mozilla-foundation/common_voice_9_0,marinone94/nst_sv" \ | |
| --model_name_or_path="KBLab/wav2vec2-large-voxrex" \ | |
| --dataset_config_name="sv-SE,distant_channel" \ | |
| --train_split_name="train+validation,train" \ | |
| --eval_split_name="test,None" \ | |
| --output_dir="./" \ | |
| --overwrite_output_dir \ | |
| --num_train_epochs="10" \ | |
| --per_device_train_batch_size="32" \ | |
| --per_device_eval_batch_size="32" \ | |
| --gradient_accumulation_steps="4" \ | |
| --learning_rate="7.5e-4" \ | |
| --max_train_samples="10000" \ | |
| --max_eval_samples="100" \ | |
| --warmup_ratio="0.0" \ | |
| --length_column_name="input_length" \ | |
| --evaluation_strategy="steps" \ | |
| --save_strategy="steps" \ | |
| --eval_steps="50" \ | |
| --save_steps="50" \ | |
| --text_column_name="sentence" \ | |
| --logging_steps="20" \ | |
| --layerdrop="0.0" \ | |
| --activation_dropout="0.1" \ | |
| --save_total_limit="2" \ | |
| --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" \ | |
| --dataset_seed="42" \ | |
| --gradient_checkpointing \ | |
| --use_auth_token \ | |
| --preprocessing_only \ | |
| --group_by_length \ | |
| --do_train --do_eval \ | |
| --push_to_hub |