-
Notifications
You must be signed in to change notification settings - Fork 5
/
abba_train_catboost.sh
58 lines (54 loc) · 7.95 KB
/
abba_train_catboost.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
# This script generates outputs for the hold out sets (val and test)
# Please adjust the --checkpoint files in the commands
# Next it uses the hold out outputs to train a second level stacking (catboost)
export DATA_ROOT_PATH=/media/alaska2/all_qfs/
export folders='Cover/ JUNIWARD/ JMiPOD/ UERD/'
export splits='val test'
cd abba/
for split in $splits
do
python3 predict/predict_folder_richmodels.py --folder 'Cover/' --subset $split --experiment JRM --quality-factor 75 --checkpoint weights/rich_models/QF75_JRM_Y_ensemble_v7.mat &
python3 predict/predict_folder_richmodels.py --folder 'Cover/' --subset $split --experiment JRM --quality-factor 90 --checkpoint weights/rich_models/QF90_JRM_Y_ensemble_v7.mat &
python3 predict/predict_folder_richmodels.py --folder 'Cover/' --subset $split --experiment JRM --quality-factor 95 --checkpoint weights/rich_models/QF95_JRM_Y_ensemble_v7.mat &
python3 predict/predict_folder_richmodels.py --folder 'Cover/' --subset $split --experiment DCTR --quality-factopr 75 --checkpoint weights/rich_models/QF75_DCTR_Y_ensemble_v7.mat &
python3 predict/predict_folder_richmodels.py --folder 'Cover/' --subset $split --experiment DCTR --quality-factor 90 --checkpoint weights/rich_models/QF90_DCTR_Y_ensemble_v7.mat &
python3 predict/predict_folder_richmodels.py --folder 'Cover/' --subset $split --experiment DCTR --quality-factor 95 --checkpoint weights/rich_models/QF95_DCTR_Y_ensemble_v7.mat &
python3 predict/predict_folder_richmodels.py --folder 'JUNIWARD/' --subset $split --experiment JRM --quality-factor 75 --checkpoint weights/rich_models/QF75_JRM_Y_ensemble_v7.mat &
python3 predict/predict_folder_richmodels.py --folder 'JUNIWARD/' --subset $split --experiment JRM --quality-factor 90 --checkpoint weights/rich_models/QF90_JRM_Y_ensemble_v7.mat &
python3 predict/predict_folder_richmodels.py --folder 'JUNIWARD/' --subset $split --experiment JRM --quality-factor 95 --checkpoint weights/rich_models/QF95_JRM_Y_ensemble_v7.mat &
python3 predict/predict_folder_richmodels.py --folder 'JUNIWARD/' --subset $split --experiment DCTR --quality-factor 75 --checkpoint weights/rich_models/QF75_DCTR_Y_ensemble_v7.mat &
python3 predict/predict_folder_richmodels.py --folder 'JUNIWARD/' --subset $split --experiment DCTR --quality-factor 90 --checkpoint weights/rich_models/QF90_DCTR_Y_ensemble_v7.mat &
python3 predict/predict_folder_richmodels.py --folder 'JUNIWARD/' --subset $split --experiment DCTR --quality-factor 95 --checkpoint weights/rich_models/QF95_DCTR_Y_ensemble_v7.mat &
python3 predict/predict_folder_richmodels.py --folder 'JMiPOD/' --subset $split --experiment JRM --quality-factor 75 --checkpoint weights/rich_models/QF75_JRM_Y_ensemble_v7.mat &
python3 predict/predict_folder_richmodels.py --folder 'JMiPOD/' --subset $split --experiment JRM --quality-factor 90 --checkpoint weights/rich_models/QF90_JRM_Y_ensemble_v7.mat &
python3 predict/predict_folder_richmodels.py --folder 'JMiPOD/' --subset $split --experiment JRM --quality-factor 95 --checkpoint weights/rich_models/QF95_JRM_Y_ensemble_v7.mat &
python3 predict/predict_folder_richmodels.py --folder 'JMiPOD/' --subset $split --experiment DCTR --quality-factor 75 --checkpoint weights/rich_models/QF75_DCTR_Y_ensemble_v7.mat &
python3 predict/predict_folder_richmodels.py --folder 'JMiPOD/' --subset $split --experiment DCTR --quality-factor 90 --checkpoint weights/rich_models/QF90_DCTR_Y_ensemble_v7.mat &
python3 predict/predict_folder_richmodels.py --folder 'JMiPOD/' --subset $split --experiment DCTR --quality-factor 95 --checkpoint weights/rich_models/QF95_DCTR_Y_ensemble_v7.mat &
python3 predict/predict_folder_richmodels.py --folder 'UERD/' --subset $split --experiment JRM --quality-factor 75 --checkpoint weights/rich_models/QF75_JRM_Y_ensemble_v7.mat &
python3 predict/predict_folder_richmodels.py --folder 'UERD/' --subset $split --experiment JRM --quality-factor 90 --checkpoint weights/rich_models/QF90_JRM_Y_ensemble_v7.mat &
python3 predict/predict_folder_richmodels.py --folder 'UERD/' --subset $split --experiment JRM --quality-factor 95 --checkpoint weights/rich_models/QF95_JRM_Y_ensemble_v7.mat &
python3 predict/predict_folder_richmodels.py --folder 'UERD/' --subset $split --experiment DCTR --quality-factor 75 --checkpoint weights/rich_models/QF75_DCTR_Y_ensemble_v7.mat &
python3 predict/predict_folder_richmodels.py --folder 'UERD/' --subset $split --experiment DCTR --quality-factor 90 --checkpoint weights/rich_models/QF90_DCTR_Y_ensemble_v7.mat &
python3 predict/predict_folder_richmodels.py --folder 'UERD/' --subset $split --experiment DCTR --quality-factor 95 --checkpoint weights/rich_models/QF95_DCTR_Y_ensemble_v7.mat
for folder in $folders
do
python3 predict/predict_folder_outofbounds.py --folder $folder --subset $split
python3 predict/predict_folder_tf.py --folder $folder --subset $split --quality-factor 75 --checkpoint weights/SRNet/QF75/model.ckpt-232000
python3 predict/predict_folder_tf.py --folder $folder --subset $split --quality-factor 90 --checkpoint weights/SRNet/QF90/model.ckpt-39000
python3 predict/predict_folder_tf.py --folder $folder --subset $split --quality-factor 95 --checkpoint weights/SRNet/QF95/model.ckpt-18000
python3 predict/predict_folder_pytorch.py --folder $folder --subset $split --model efficientnet-b4 --experiment efficientnet_b4_NR_mish --checkpoint weights/efficientnet_b4_NR_mish/best-checkpoint-017epoch.bin
python3 predict/predict_folder_pytorch.py --folder $folder --subset $split --model efficientnet-b5 --experiment efficientnet_b5_NR_mish --checkpoint weights/efficientnet_b5_NR_mish/best-checkpoint-018epoch.bin
python3 predict/predict_folder_pytorch.py --folder $folder --subset $split --model mixnet_xl --experiment mixnet_xL_NR_mish --surgery 1 --checkpoint weights/mixnet_xL_NR_mish/best-checkpoint-021epoch.bin
python3 predict/predict_folder_pytorch.py --folder $folder --subset $split --model efficientnet-b2 --experiment efficientnet_b2_NR --surgery 0 --fp16 0 --checkpoint weights/efficientnet_b2/NR/best-checkpoint-028epoch.bin
python3 predict/predict_folder_pytorch.py --folder $folder --subset $split --model efficientnet-b2 --experiment efficientnet_b2_R --decoder R --surgery 0 --fp16 0 --checkpoint weights/efficientnet_b2/R/best-checkpoint-028epoch.bin
python3 predict/predict_folder_pytorch.py --folder $folder --subset $split --model mixnet_s --test-time-augmentation 1 --experiment mixnet_S_R_seed0 --decoder R --surgery 0 --fp16 0 --checkpoint weights/mixnet_S/R_seed0/best-checkpoint-033epoch.bin
python3 predict/predict_folder_pytorch.py --folder $folder --subset $split --model mixnet_s --test-time-augmentation 1 --experiment mixnet_S_R_seed1 --decoder R --surgery 0 --fp16 0 --checkpoint weights/mixnet_S/R_seed1/best-checkpoint-035epoch.bin
python3 predict/predict_folder_pytorch.py --folder $folder --subset $split --model mixnet_s --test-time-augmentation 1 --experiment mixnet_S_R_seed2 --decoder R --surgery 0 --fp16 0 --checkpoint weights/mixnet_S/R_seed2/best-checkpoint-036epoch.bin
python3 predict/predict_folder_pytorch.py --folder $folder --subset $split --model mixnet_s --test-time-augmentation 1 --experiment mixnet_S_R_seed3 --decoder R --surgery 0 --fp16 0 --checkpoint weights/mixnet_S/R_seed3/best-checkpoint-038epoch.bin
python3 predict/predict_folder_pytorch.py --folder $folder --subset $split --model mixnet_s --test-time-augmentation 1 --experiment mixnet_S_R_seed4 --decoder R --surgery 0 --fp16 0 --checkpoint weights/mixnet_S/R_seed4/best-checkpoint-035epoch.bin
python3 predict/predict_folder_pytorch.py --folder $folder --subset $split --model mixnet_s --test-time-augmentation 1 --experiment mixnet_S_NR --surgery 0 --fp16 0 --checkpoint weights/mixnet_S/NR/best-checkpoint-058epoch.bin
python3 predict/group_experiments.py --folder $folder --subset $split --id 0000
done
done
python3 train/stracking/skopt_catboost.py --zoo-id 0000