File size: 1,192 Bytes
264a841 |
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 39 40 |
#!/usr/bin/env bash
accelerate launch --mixed_precision=bf16 run_distillation.py \
--model_name_or_path "./distil-large-v3-init" \
--teacher_model_name_or_path "openai/whisper-large-v3" \
--train_dataset_name "mozilla-foundation/common_voice_15_0" \
--train_dataset_config_name "de" \
--train_split_name "train" \
--text_column_name "sentence" \
--eval_dataset_name "mozilla-foundation/common_voice_15_0" \
--eval_dataset_config_name "de" \
--eval_split_name "validation" \
--eval_text_column_name "sentence" \
--eval_steps 5000 \
--save_steps 5000 \
--warmup_steps 500 \
--learning_rate 1e-4 \
--lr_scheduler_type "linear" \
--logging_steps 25 \
--save_total_limit 1 \
--max_steps 50000 \
--per_device_train_batch_size 64 \
--per_device_eval_batch_size 64 \
--dataloader_num_workers 16 \
--preprocessing_num_workers 16 \
--ddp_timeout 7200 \
--dtype "bfloat16" \
--output_dir "./" \
--use_pseudo_labels "false" \
--wandb_project "huggingface" \
--condition_on_prev_probability "0.0" \
--do_train \
--do_eval \
--gradient_checkpointing \
--overwrite_output_dir \
--predict_with_generate \
--freeze_encoder \
--streaming
|