-
Notifications
You must be signed in to change notification settings - Fork 0
/
run_whisper_train_monolingual_multi_task.sh
executable file
·49 lines (46 loc) · 2.61 KB
/
run_whisper_train_monolingual_multi_task.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
## MONOLINGUAL MULTI-TASK TRAINING
# # ind_lang
# OMP_NUM_THREADS=8 CUDA_VISIBLE_DEVICES=6 python src/train_seq2seq.py \
# --model_name_or_path=openai/whisper-small \
# --lang ind \
# --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-4 --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/ind_lang/ \
# --output_dir=./save/ind_lang/ 2>&1 | tee ./log/ind_lang_exp.logs
# sun_lang
OMP_NUM_THREADS=8 CUDA_VISIBLE_DEVICES=1 python src/train_seq2seq.py \
--model_name_or_path=openai/whisper-small \
--lang sun \
--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-4 --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/sun_lang/ \
--output_dir=./save/sun_lang/ 2>&1 | tee ./log/sun_lang_exp.logs
# # jav_lang
# OMP_NUM_THREADS=8 CUDA_VISIBLE_DEVICES=6 python src/train_seq2seq.py \
# --model_name_or_path=openai/whisper-small \
# --lang jav \
# --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-4 --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/jav_lang/ \
# --output_dir=./save/jav_lang/ 2>&1 | tee ./log/jav_lang_exp.logs