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