| export WANDB_ENTITY="cahya" | |
| export WANDB_PROJECT="xlsr-luganda" | |
| export WANDB_LOG_MODEL=true | |
| python run_speech_recognition_ctc.py \ | |
| --dataset_name="mozilla-foundation/common_voice_7_0" \ | |
| --model_name_or_path="indonesian-nlp/wav2vec2-luganda" \ | |
| --dataset_config_name="lg" \ | |
| --output_dir="./" \ | |
| --overwrite_output_dir \ | |
| --num_train_epochs="10" \ | |
| --per_device_train_batch_size="64" \ | |
| --per_device_eval_batch_size="2" \ | |
| --gradient_accumulation_steps="4" \ | |
| --learning_rate="1e-8" \ | |
| --warmup_steps="10" \ | |
| --length_column_name="input_length" \ | |
| --evaluation_strategy="steps" \ | |
| --text_column_name="sentence" \ | |
| --save_steps="10" \ | |
| --eval_steps="10" \ | |
| --logging_steps="10" \ | |
| --layerdrop="0.0" \ | |
| --activation_dropout="0.1" \ | |
| --save_total_limit="3" \ | |
| --freeze_feature_encoder \ | |
| --feat_proj_dropout="0.0" \ | |
| --mask_time_prob="0.4" \ | |
| --mask_time_length="10" \ | |
| --mask_feature_prob="0.25" \ | |
| --mask_feature_length="64" \ | |
| --gradient_checkpointing \ | |
| --use_auth_token \ | |
| --fp16=true \ | |
| --group_by_length \ | |
| --do_train=true \ | |
| --do_eval=true \ | |
| --push_to_hub=true \ | |
| --chars_to_ignore , ? . ! \; \: \"\" \% \' \" \' \' \` … \’ » « \‘ '“' '”' � é û \‘ \’ \( \) | |