-
Notifications
You must be signed in to change notification settings - Fork 0
/
run_whisper_train_single_task_1.sh
executable file
·81 lines (76 loc) · 4.84 KB
/
run_whisper_train_single_task_1.sh
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
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
## SINGLE-TASK TRAINING
# # indspeech_digit_cdsr
# OMP_NUM_THREADS=8 CUDA_VISIBLE_DEVICES=2 python src/train_seq2seq.py \
# --model_name_or_path=openai/whisper-base \
# --task_config_name indspeech_digit_cdsr_nusantara_sptext \
# --preprocessing_num_workers=16 --audio_column_name=audio --text_column_name=text \
# --per_device_train_batch_size=16 --per_device_eval_batch_size=16 \
# --dataloader_num_workers=16 --dataloader_pin_memory \
# --seed=14045 --num_train_epochs=30 --learning_rate=1e-5 --fp16 \
# --logging_strategy=steps --logging_steps=10 --report_to=tensorboard \
# --evaluation_strategy=epoch --eval_steps=1 --eval_accumulation_steps=1 \
# --save_strategy=epoch --save_steps=1 --save_total_limit=1 --load_best_model_at_end \
# --metric_for_best_model=wer --greater_is_better=False \
# --gradient_checkpointing=True \
# --cache_dir=./cache/indspeech_digit_cdsr/ \
# --output_dir=./save/indspeech_digit_cdsr/ 2>&1 | tee ./log/indspeech_digit_cdsr_exp.logs
# # indspeech_news_lvcsr
# OMP_NUM_THREADS=8 CUDA_VISIBLE_DEVICES=2 python src/train_seq2seq.py \
# --model_name_or_path=openai/whisper-base \
# --task_config_name indspeech_news_lvcsr_nusantara_sptext \
# --preprocessing_num_workers=16 --audio_column_name=audio --text_column_name=text \
# --per_device_train_batch_size=16 --per_device_eval_batch_size=16 \
# --dataloader_num_workers=16 --dataloader_pin_memory \
# --seed=14045 --num_train_epochs=30 --learning_rate=1e-5 --fp16 \
# --logging_strategy=steps --logging_steps=10 --report_to=tensorboard \
# --evaluation_strategy=epoch --eval_steps=1 --eval_accumulation_steps=1 \
# --save_strategy=epoch --save_steps=1 --save_total_limit=1 --load_best_model_at_end \
# --metric_for_best_model=wer --greater_is_better=False \
# --gradient_checkpointing=True \
# --cache_dir=./cache/indspeech_news_lvcsr/ \
# --output_dir=./save/indspeech_news_lvcsr/ 2>&1 | tee ./log/indspeech_news_lvcsr_exp.logs
# indspeech_teldialog_lvcsr
OMP_NUM_THREADS=8 CUDA_VISIBLE_DEVICES=2 python src/train_seq2seq.py \
--model_name_or_path=openai/whisper-base \
--task_config_name indspeech_teldialog_lvcsr_nusantara_sptext \
--preprocessing_num_workers=16 --audio_column_name=audio --text_column_name=text \
--per_device_train_batch_size=16 --per_device_eval_batch_size=16 \
--dataloader_num_workers=16 --dataloader_pin_memory \
--seed=14045 --num_train_epochs=30 --learning_rate=1e-5 --fp16 \
--logging_strategy=steps --logging_steps=10 --report_to=tensorboard \
--evaluation_strategy=epoch --eval_steps=1 --eval_accumulation_steps=1 \
--save_strategy=epoch --save_steps=1 --save_total_limit=1 --load_best_model_at_end \
--metric_for_best_model=wer --greater_is_better=False \
--gradient_checkpointing=True \
--cache_dir=./cache/indspeech_teldialog_lvcsr/ \
--output_dir=./save/indspeech_teldialog_lvcsr/ 2>&1 | tee ./log/indspeech_teldialog_lvcsr_exp.logs
# # librivox_indonesia_ind
# OMP_NUM_THREADS=8 CUDA_VISIBLE_DEVICES=2 python src/train_seq2seq.py \
# --model_name_or_path=openai/whisper-base \
# --task_config_name librivox_indonesia_ind_nusantara_sptext \
# --preprocessing_num_workers=16 --audio_column_name=audio --text_column_name=text \
# --per_device_train_batch_size=16 --per_device_eval_batch_size=16 \
# --dataloader_num_workers=16 --dataloader_pin_memory \
# --seed=14045 --num_train_epochs=30 --learning_rate=1e-5 --fp16 \
# --logging_strategy=steps --logging_steps=10 --report_to=tensorboard \
# --evaluation_strategy=epoch --eval_steps=1 --eval_accumulation_steps=1 \
# --save_strategy=epoch --save_steps=1 --save_total_limit=1 --load_best_model_at_end \
# --metric_for_best_model=wer --greater_is_better=False \
# --gradient_checkpointing=True \
# --cache_dir=./cache/librivox_indonesia_ind/ \
# --output_dir=./save/librivox_indonesia_ind/ 2>&1 | tee ./log/librivox_indonesia_ind_exp.logs
# # librivox_indonesia_ace
# OMP_NUM_THREADS=8 CUDA_VISIBLE_DEVICES=2 python src/train_seq2seq.py \
# --model_name_or_path=openai/whisper-base \
# --task_config_name librivox_indonesia_ace_nusantara_sptext \
# --preprocessing_num_workers=16 --audio_column_name=audio --text_column_name=text \
# --per_device_train_batch_size=16 --per_device_eval_batch_size=16 \
# --dataloader_num_workers=16 --dataloader_pin_memory \
# --seed=14045 --num_train_epochs=30 --learning_rate=1e-5 --fp16 \
# --logging_strategy=steps --logging_steps=10 --report_to=tensorboard \
# --evaluation_strategy=epoch --eval_steps=1 --eval_accumulation_steps=1 \
# --save_strategy=epoch --save_steps=1 --save_total_limit=1 --load_best_model_at_end \
# --metric_for_best_model=wer --greater_is_better=False \
# --gradient_checkpointing=True \
# --cache_dir=./cache/librivox_indonesia_ace/ \
# --output_dir=./save/librivox_indonesia_ace/ 2>&1 | tee ./log/librivox_indonesia_ace_exp.logs