File size: 965 Bytes
084bb18 |
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 |
#!/usr/bin/env bash
CUDA_VISIBLE_DEVICES="1" python run_speech_recognition_ctc.py \
--dataset_name="timit_asr" \
--model_name_or_path="microsoft/unispeech-large-1500h-cv" \
--overwrite_output_dir \
--output_dir="./unispeech-large-1500h-cv-timit" \
--train_split_name="train" \
--num_train_epochs="20" \
--per_device_train_batch_size="32" \
--per_device_eval_batch_size="1" \
--weight_decay="0.005" \
--learning_rate="1e-4" \
--warmup_steps="1000" \
--evaluation_strategy="steps" \
--text_column_name="text" \
--save_steps="400" \
--eval_steps="100" \
--logging_steps="10" \
--layerdrop="0.0" \
--save_total_limit="3" \
--freeze_feature_extractor \
--chars_to_ignore , ? . ! - \; \: \" “ % ‘ ” � \
--fp16 \
--group_by_length \
--push_to_hub \
--do_train --do_eval \
|