File size: 897 Bytes
3308cff 8dfd77e 3308cff e73fa51 3308cff 7c3634e 8dfd77e 3308cff 8dfd77e |
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 |
python run_speech_recognition_seq2seq.py \
--dataset_name="NbAiLab/NPSC" \
--dataset_config_name="16K_mp3" \
--data_cache_dir="/mnt/lv_ai_1_ficino/rolvb/cache" \
--model_name_or_path="./" \
--output_dir="./" \
--preprocessing_num_workers="16" \
--length_column_name="input_length" \
--overwrite_output_dir \
--num_train_epochs="50" \
--per_device_train_batch_size="8" \
--per_device_eval_batch_size="8" \
--gradient_accumulation_steps="8" \
--learning_rate="1e-4" \
--warmup_steps="1000" \
--evaluation_strategy="steps" \
--text_column_name="text" \
--save_steps="400" \
--eval_steps="400" \
--logging_steps="10" \
--save_total_limit="1" \
--freeze_feature_encoder \
--gradient_checkpointing \
--fp16 \
--group_by_length \
--predict_with_generate \
--generation_max_length="40" \
--generation_num_beams="1" \
--do_train --do_eval \
--do_lower_case \
--push_to_hub
|