-
Notifications
You must be signed in to change notification settings - Fork 0
/
run_segan+_train.sh
executable file
·31 lines (27 loc) · 1.03 KB
/
run_segan+_train.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
#!/bin/bash
#
#train on time field
#
# CUDA_VISIBLE_DEVICES=0 CUDA_CACHE_PATH='~/.cudacache' \
# python -u train.py --save_path ppgan_time/model0307/epoch50 \
# --clean_trainset data_veu4/expanded_segan1_additive/clean_trainset \
# --noisy_trainset data_veu4/expanded_segan1_additive/noisy_trainset \
# --cache_dir data_tmp --batch_size 100 --no_bias --epoch 50 --slice_size 256 \
# --l1_weight 0 --g_lr 4e-5 --d_lr 5e-5 --z_dim 512 --dpool_slen 8 --ppgantype ppgan_time
#
# train on frequency field
#
#change dpool_slen ,srate,generator lablefc
export CUDA_VISIBLE_DEVICES=0
CUDA_CACHE_PATH='~/.cudacache' \
python train.py --save_path ppg_time/model0715/train1 \
--clean_trainset NewData/clean_trainset \
--noisy_trainset NewData/noisy_trainset \
--val_spilt 0.9 \
--cache_dir data_tmp --batch_size 50 --no_bias \
--epoch 500 \
--l1_weight 100 --g_lr 5e-5 --d_lr 1e-5 --z_dim 256 --dpool_slen 31 \
--opt adam\
--ppgantype ppgan_freq
#--clean_valset NewData/clean_valset \
#--noisy_valset NewData/noisy_valset \