patrickvonplaten commited on
Commit
0301ea3
1 Parent(s): 451c409
Files changed (1) hide show
  1. run.sh +31 -0
run.sh ADDED
@@ -0,0 +1,31 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ #!/usr/bin/env bash
2
+ python -m torch.distributed.launch \
3
+ --nproc_per_node 8 run_speech_recognition_ctc.py \
4
+ --dataset_name="librispeech_asr" \
5
+ --model_name_or_path="asapp/sew-mid-100k" \
6
+ --dataset_config_name="clean" \
7
+ --train_split_name="train.100" \
8
+ --eval_split_name="validation" \
9
+ --output_dir="./sew-mid-100k-librispeech-clean-100h-ft" \
10
+ --preprocessing_num_workers="16" \
11
+ --overwrite_output_dir \
12
+ --num_train_epochs="3" \
13
+ --per_device_train_batch_size="4" \
14
+ --gradient_accumulation_steps="1" \
15
+ --learning_rate="3e-5" \
16
+ --warmup_steps="500" \
17
+ --evaluation_strategy="steps" \
18
+ --audio_column_name="file" \
19
+ --text_column_name="text" \
20
+ --save_steps="400" \
21
+ --eval_steps="100" \
22
+ --logging_steps="1" \
23
+ --layerdrop="0.0" \
24
+ --save_total_limit="3" \
25
+ --freeze_feature_extractor \
26
+ --gradient_checkpointing \
27
+ --chars_to_ignore , ? . ! - \; \: \" “ % ‘ ” \
28
+ --fp16 \
29
+ --group_by_length \
30
+ --push_to_hub \
31
+ --do_train --do_eval \