File size: 1,129 Bytes
ffb6ea4 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 |
#!/usr/bin/env bash
python -m torch.distributed.launch \
--nproc_per_node 8 run_speech_recognition_ctc.py \
--dataset_name="common_voice" \
--model_name_or_path="./" \
--tokenizer_name_or_path="./" \
--dataset_name="mozilla-foundation/common_voice_3_0" \
--path_to_eval_file="./valSeqs_1.0_uniform_new_version.text" \
--path_to_train_file="./trainSeqs_1.0_uniform_new_version.text" \
--dataset_config_name="sv-SE" \
--text_column_name="sentence" \
--output_dir="./" \
--overwrite_output_dir \
--num_train_epochs="150" \
--per_device_train_batch_size="4" \
--per_device_eval_batch_size="4" \
--gradient_accumulation_steps="1" \
--learning_rate="7.5e-5" \
--warmup_steps="1000" \
--evaluation_strategy="steps" \
--save_steps="500" \
--eval_steps="500" \
--logging_steps="100" \
--layerdrop="0.0" \
--activation_dropout="0.1" \
--save_total_limit="3" \
--freeze_feature_extractor \
--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 \
--fp16 \
--group_by_length \
--do_train --do_eval
|