-
Notifications
You must be signed in to change notification settings - Fork 4
/
awa2_vit.sh
154 lines (99 loc) · 33.7 KB
/
awa2_vit.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
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
#!/bin/sh
#SBATCH --output=path/awa2_vit_%j.out
pwd; hostname; date
CURRENT=`date +"%Y-%m-%d_%T"`
echo $CURRENT
slurm_output_bb_train=awa2_vit_bb_train_$CURRENT.out
slurm_output_bb_test=awa2_vit_bb_test_$CURRENT.out
slurm_output_t_train=awa2_vit_t_train_$CURRENT.out
slurm_output_t_test=awa2_vit_t_test_$CURRENT.out
slurm_output_iter1_g_train=awa2_vit_iter1_g_train_$CURRENT.out
slurm_output_iter1_g_test=awa2_vit_iter1_g_test_$CURRENT.out
slurm_output_iter1_residual_train=awa2_vit_iter1_residual_train_$CURRENT.out
slurm_output_iter2_g_train=awa2_vit_iter2_g_train_$CURRENT.out
slurm_output_iter2_g_test=awa2_vit_iter2_g_test_$CURRENT.out
slurm_output_iter2_residual_train=awa2_vit_iter2_residual_train_$CURRENT.out
slurm_output_iter3_g_train=awa2_vit_iter3_g_train_$CURRENT.out
slurm_output_iter3_g_test=awa2_vit_iter3_g_test_$CURRENT.out
slurm_output_iter3_residual_train=awa2_vit_iter3_residual_train_$CURRENT.out
slurm_output_iter4_g_train=awa2_vit_iter4_g_train_$CURRENT.out
slurm_output_iter4_g_test=awa2_vit_iter4_g_test_$CURRENT.out
slurm_output_iter4_residual_train=awa2_vit_iter4_residual_train_$CURRENT.out
slurm_output_iter5_g_train=awa2_vit_iter5_g_train_$CURRENT.out
slurm_output_iter5_g_test=awa2_vit_iter5_g_test_$CURRENT.out
slurm_output_iter5_residual_train=awa2_vit_iter5_residual_train_$CURRENT.out
slurm_output_iter6_g_train=awa2_vit_iter6_g_train_$CURRENT.out
slurm_output_iter6_g_test=awa2_vit_iter6_g_test_$CURRENT.out
slurm_output_iter6_residual_train=awa2_vit_iter6_residual_train_$CURRENT.out
slurm_output_iter6_residual_test=awa2_vit_iter6_residual_test_$CURRENT.out
slurm_explanations=awa2_vit_explanations_$CURRENT.out
echo "awa2 VIT-B_16"
source path-of-conda/anaconda3/etc/profile.d/conda.sh
conda activate python_3_7_rtx_6000
# BB model
# BB Training scripts
python ../codebase/BB/VIT/TransFG-master/train_awa2.py --seed 42 --bs 16 --learning_rate 0.03 --arch "ViT-B_16" --eval_every 10 --lr 0.03 --weight-decay 0 --num_steps 10000 --decay_type "cosine" --warmup_steps 500 --max_grad_norm 1.0 --seed 42 --gradient_accumulation_steps 1 --split non-overlap > $slurm_output_bb_train
# T model
# train
python ../codebase/train_t_awa2.py --checkpoint-file "VIT_CUBS_7000_checkpoint.bin" --bs 32 --layer "VIT" --flattening-type "VIT" --arch "VIT-B_16" > $slurm_output_t_train
# Test
python ../codebase/test_t_awa2.py --checkpoint-file "VIT_CUBS_7000_checkpoint.bin" --checkpoint-file-t "g_best_model_epoch_39.pth.tar" --save-concepts True --bs 16 --solver-LR "sgd" --loss-LR "BCE" --layer "VIT" --flattening-type "VIT" --arch "VIT-B_16"> $slurm_output_t_test
# MoIE Training scripts
#---------------------------------
# # iter 1
#---------------------------------
python ../codebase/train_explainer_awa2.py --root-bb "lr_0.03_epochs_95" --checkpoint-bb "VIT_CUBS_7000_checkpoint.bin" --iter 1 --expert-to-train "explainer" --dataset "awa2" --cov 0.2 --bs 16 --dataset-folder-concepts "lr_0.03_epochs_95_ViT-B_16_layer4_VIT_sgd_BCE" --lr 0.01 --input-size-pi 2048 --temperature-lens 0.7 --lambda-lens 0.0001 --alpha-KD 0.9 --temperature-KD 10 --hidden-nodes 10 --layer "VIT" --arch "VIT-B_16" > $slurm_output_iter1_g_train
python ../codebase/test_explainer_awa2.py --checkpoint-model "model_g_best_model.pth.tar" --root-bb "lr_0.03_epochs_95" --checkpoint-bb "VIT_CUBS_7000_checkpoint.bin" --iter 1 --expert-to-train "explainer" --dataset "awa2" --cov 0.2 --bs 16 --dataset-folder-concepts "lr_0.03_epochs_95_ViT-B_16_layer4_VIT_sgd_BCE" --lr 0.01 --input-size-pi 2048 --temperature-lens 0.7 --lambda-lens 0.0001 --alpha-KD 0.9 --temperature-KD 10 --hidden-nodes 10 --layer "VIT" --arch "VIT-B_16"> $slurm_output_iter1_g_test
python ../codebase/train_explainer_awa2.py --checkpoint-model "model_g_best_model.pth.tar" --root-bb "lr_0.03_epochs_95" --checkpoint-bb "VIT_CUBS_7000_checkpoint.bin" --iter 1 --expert-to-train "residual" --dataset "awa2" --cov 0.2 --bs 16 --dataset-folder-concepts "lr_0.03_epochs_95_ViT-B_16_layer4_VIT_sgd_BCE" --lr 0.01 --input-size-pi 2048 --temperature-lens 0.7 --lambda-lens 0.0001 --alpha-KD 0.9 --temperature-KD 10 --hidden-nodes 10 --layer "VIT" --arch "VIT-B_16"> $slurm_output_iter1_residual_train
#---------------------------------
# # iter 2
#---------------------------------
# Train explainer
python ../codebase/train_explainer_awa2.py --checkpoint-model "model_g_best_model.pth.tar" --checkpoint-residual "model_residual_best_model.pth.tar" --prev_explainer_chk_pt_folder "/ocean/projects/asc170022p/shg121/PhD/ICLR-2022/checkpoints/awa2/explainer/ViT-B_16/lr_0.01_epochs_500_temperature-lens_6.0_use-concepts-as-pi-input_True_input-size-pi_2048_cov_0.2_alpha_0.5_selection-threshold_0.5_lambda-lens_0.0001_alpha-KD_0.99_temperature-KD_10.0_hidden-layers_1_layer_VIT_explainer_init_none/iter1" --root-bb "lr_0.03_epochs_95" --checkpoint-bb "VIT_CUBS_7000_checkpoint.bin" --iter 2 --expert-to-train "explainer" --dataset "awa2" --cov 0.2 0.2 --bs 16 --dataset-folder-concepts "lr_0.03_epochs_95_ViT-B_16_layer4_VIT_sgd_BCE" --lr 0.01 0.01 --input-size-pi 2048 --temperature-lens 0.7 --lambda-lens 0.0001 --alpha-KD 0.9 --temperature-KD 10 --hidden-nodes 10 --layer "VIT" --arch "VIT-B_16" > $slurm_output_iter2_g_train
# # Test explainer
python ../codebase/test_explainer_awa2.py --checkpoint-model "model_g_best_model.pth.tar" "model_g_best_model.pth.tar" --checkpoint-residual "model_residual_best_model.pth.tar" --prev_explainer_chk_pt_folder "/ocean/projects/asc170022p/shg121/PhD/ICLR-2022/checkpoints/awa2/explainer/ViT-B_16/lr_0.01_epochs_500_temperature-lens_6.0_use-concepts-as-pi-input_True_input-size-pi_2048_cov_0.2_alpha_0.5_selection-threshold_0.5_lambda-lens_0.0001_alpha-KD_0.99_temperature-KD_10.0_hidden-layers_1_layer_VIT_explainer_init_none/iter1" --root-bb "lr_0.03_epochs_95" --checkpoint-bb "VIT_CUBS_7000_checkpoint.bin" --iter 2 --expert-to-train "explainer" --dataset "awa2" --cov 0.2 0.2 --bs 16 --dataset-folder-concepts "lr_0.03_epochs_95_ViT-B_16_layer4_VIT_sgd_BCE" --lr 0.01 0.01 --input-size-pi 2048 --temperature-lens 0.7 --lambda-lens 0.0001 --alpha-KD 0.9 --temperature-KD 10 --hidden-nodes 10 --layer "VIT" --arch "VIT-B_16" > $slurm_output_iter2_g_test
# # # Train residual
python ../codebase/train_explainer_awa2.py --checkpoint-model "model_g_best_model.pth.tar" "model_g_best_model.pth.tar" --checkpoint-residual "model_residual_best_model.pth.tar" --prev_explainer_chk_pt_folder "/ocean/projects/asc170022p/shg121/PhD/ICLR-2022/checkpoints/awa2/explainer/ViT-B_16/lr_0.01_epochs_500_temperature-lens_6.0_use-concepts-as-pi-input_True_input-size-pi_2048_cov_0.2_alpha_0.5_selection-threshold_0.5_lambda-lens_0.0001_alpha-KD_0.99_temperature-KD_10.0_hidden-layers_1_layer_VIT_explainer_init_none/iter1" --root-bb "lr_0.03_epochs_95" --checkpoint-bb "VIT_CUBS_7000_checkpoint.bin" --iter 2 --expert-to-train "residual" --dataset "awa2" --cov 0.2 0.2 --bs 16 --dataset-folder-concepts "lr_0.03_epochs_95_ViT-B_16_layer4_VIT_sgd_BCE" --lr 0.01 0.01 --input-size-pi 2048 --temperature-lens 0.7 --lambda-lens 0.0001 --alpha-KD 0.9 --temperature-KD 10 --hidden-nodes 10 --layer "VIT" --arch "VIT-B_16" > $slurm_output_iter2_residual_train
#---------------------------------
# # iter 3
#---------------------------------
# Train explainer
python ../codebase/train_explainer_awa2.py --checkpoint-model "model_g_best_model.pth.tar" "model_g_best_model.pth.tar" --checkpoint-residual "model_residual_best_model.pth.tar" "model_residual_best_model.pth.tar" --prev_explainer_chk_pt_folder "/ocean/projects/asc170022p/shg121/PhD/ICLR-2022/checkpoints/awa2/explainer/ViT-B_16/lr_0.01_epochs_500_temperature-lens_6.0_use-concepts-as-pi-input_True_input-size-pi_2048_cov_0.2_alpha_0.5_selection-threshold_0.5_lambda-lens_0.0001_alpha-KD_0.99_temperature-KD_10.0_hidden-layers_1_layer_VIT_explainer_init_none/iter1" "/ocean/projects/asc170022p/shg121/PhD/ICLR-2022/checkpoints/awa2/explainer/ViT-B_16/lr_0.01_epochs_500_temperature-lens_6.0_use-concepts-as-pi-input_True_input-size-pi_2048_cov_0.2_alpha_0.5_selection-threshold_0.5_lambda-lens_0.0001_alpha-KD_0.99_temperature-KD_10.0_hidden-layers_1_layer_VIT_explainer_init_none/cov_0.2_lr_0.01/iter2" --root-bb "lr_0.03_epochs_95" --checkpoint-bb "VIT_CUBS_7000_checkpoint.bin" --iter 3 --expert-to-train "explainer" --dataset "awa2" --cov 0.2 0.2 0.2 --bs 16 --dataset-folder-concepts "lr_0.03_epochs_95_ViT-B_16_layer4_VIT_sgd_BCE" --lr 0.01 0.01 0.01 --input-size-pi 2048 --temperature-lens 0.7 --lambda-lens 0.0001 --alpha-KD 0.9 --temperature-KD 10 --hidden-nodes 10 --layer "VIT" --arch "VIT-B_16" > $slurm_output_iter3_g_train
# Test explainer
python ../codebase/test_explainer_awa2.py --checkpoint-model "model_g_best_model.pth.tar" "model_g_best_model.pth.tar" "model_g_best_model.pth.tar" --checkpoint-residual "model_residual_best_model.pth.tar" "model_residual_best_model.pth.tar" --prev_explainer_chk_pt_folder "/ocean/projects/asc170022p/shg121/PhD/ICLR-2022/checkpoints/awa2/explainer/ViT-B_16/lr_0.01_epochs_500_temperature-lens_6.0_use-concepts-as-pi-input_True_input-size-pi_2048_cov_0.2_alpha_0.5_selection-threshold_0.5_lambda-lens_0.0001_alpha-KD_0.99_temperature-KD_10.0_hidden-layers_1_layer_VIT_explainer_init_none/iter1" "/ocean/projects/asc170022p/shg121/PhD/ICLR-2022/checkpoints/awa2/explainer/ViT-B_16/lr_0.01_epochs_500_temperature-lens_6.0_use-concepts-as-pi-input_True_input-size-pi_2048_cov_0.2_alpha_0.5_selection-threshold_0.5_lambda-lens_0.0001_alpha-KD_0.99_temperature-KD_10.0_hidden-layers_1_layer_VIT_explainer_init_none/cov_0.2_lr_0.01/iter2" --root-bb "lr_0.03_epochs_95" --checkpoint-bb "VIT_CUBS_7000_checkpoint.bin" --iter 3 --expert-to-train "explainer" --dataset "awa2" --cov 0.2 0.2 0.2 --bs 16 --dataset-folder-concepts "lr_0.03_epochs_95_ViT-B_16_layer4_VIT_sgd_BCE" --lr 0.01 0.01 0.01 --input-size-pi 2048 --temperature-lens 0.7 --lambda-lens 0.0001 --alpha-KD 0.9 --temperature-KD 10 --hidden-nodes 10 --layer "VIT" --arch "VIT-B_16" > $slurm_output_iter3_g_test
# # # # Train residual
python ../codebase/train_explainer_awa2.py --checkpoint-model "model_g_best_model.pth.tar" "model_g_best_model.pth.tar" "model_g_best_model.pth.tar" --checkpoint-residual "model_residual_best_model.pth.tar" "model_residual_best_model.pth.tar" --prev_explainer_chk_pt_folder "/ocean/projects/asc170022p/shg121/PhD/ICLR-2022/checkpoints/awa2/explainer/ViT-B_16/lr_0.01_epochs_500_temperature-lens_6.0_use-concepts-as-pi-input_True_input-size-pi_2048_cov_0.2_alpha_0.5_selection-threshold_0.5_lambda-lens_0.0001_alpha-KD_0.99_temperature-KD_10.0_hidden-layers_1_layer_VIT_explainer_init_none/iter1" "/ocean/projects/asc170022p/shg121/PhD/ICLR-2022/checkpoints/awa2/explainer/ViT-B_16/lr_0.01_epochs_500_temperature-lens_6.0_use-concepts-as-pi-input_True_input-size-pi_2048_cov_0.2_alpha_0.5_selection-threshold_0.5_lambda-lens_0.0001_alpha-KD_0.99_temperature-KD_10.0_hidden-layers_1_layer_VIT_explainer_init_none/cov_0.2_lr_0.01/iter2" --root-bb "lr_0.03_epochs_95" --checkpoint-bb "VIT_CUBS_7000_checkpoint.bin" --iter 3 --expert-to-train "residual" --dataset "awa2" --cov 0.2 0.2 0.2 --bs 16 --dataset-folder-concepts "lr_0.03_epochs_95_ViT-B_16_layer4_VIT_sgd_BCE" --lr 0.01 0.01 0.01 --input-size-pi 2048 --temperature-lens 0.7 --lambda-lens 0.0001 --alpha-KD 0.9 --temperature-KD 10 --hidden-nodes 10 --layer "VIT" --arch "VIT-B_16" > $slurm_output_iter3_residual_train
#---------------------------------
# # iter 4
#---------------------------------
# Train explainer
python ../codebase/train_explainer_awa2.py --checkpoint-model "model_g_best_model.pth.tar" "model_g_best_model.pth.tar" "model_g_best_model.pth.tar" --checkpoint-residual "model_residual_best_model.pth.tar" "model_residual_best_model.pth.tar" "model_residual_best_model.pth.tar" --prev_explainer_chk_pt_folder "/ocean/projects/asc170022p/shg121/PhD/ICLR-2022/checkpoints/awa2/explainer/ViT-B_16/lr_0.01_epochs_500_temperature-lens_6.0_use-concepts-as-pi-input_True_input-size-pi_2048_cov_0.2_alpha_0.5_selection-threshold_0.5_lambda-lens_0.0001_alpha-KD_0.99_temperature-KD_10.0_hidden-layers_1_layer_VIT_explainer_init_none/iter1" "/ocean/projects/asc170022p/shg121/PhD/ICLR-2022/checkpoints/awa2/explainer/ViT-B_16/lr_0.01_epochs_500_temperature-lens_6.0_use-concepts-as-pi-input_True_input-size-pi_2048_cov_0.2_alpha_0.5_selection-threshold_0.5_lambda-lens_0.0001_alpha-KD_0.99_temperature-KD_10.0_hidden-layers_1_layer_VIT_explainer_init_none/cov_0.2_lr_0.01/iter2" "/ocean/projects/asc170022p/shg121/PhD/ICLR-2022/checkpoints/awa2/explainer/ViT-B_16/lr_0.01_epochs_500_temperature-lens_6.0_use-concepts-as-pi-input_True_input-size-pi_2048_cov_0.2_alpha_0.5_selection-threshold_0.5_lambda-lens_0.0001_alpha-KD_0.99_temperature-KD_10.0_hidden-layers_1_layer_VIT_explainer_init_none/cov_0.2_lr_0.01/iter3" --root-bb "lr_0.03_epochs_95" --checkpoint-bb "VIT_CUBS_7000_checkpoint.bin" --iter 4 --expert-to-train "explainer" --dataset "awa2" --cov 0.2 0.2 0.2 0.2 --bs 16 --dataset-folder-concepts "lr_0.03_epochs_95_ViT-B_16_layer4_VIT_sgd_BCE" --lr 0.01 0.01 0.01 0.01 --input-size-pi 2048 --temperature-lens 0.7 --lambda-lens 0.0001 --alpha-KD 0.9 --temperature-KD 10 --hidden-nodes 10 --layer "VIT" --arch "VIT-B_16" > $slurm_output_iter4_g_train
# # # Test explainer
python ../codebase/test_explainer_awa2.py --checkpoint-model "model_g_best_model.pth.tar" "model_g_best_model.pth.tar" "model_g_best_model.pth.tar" "model_g_best_model.pth.tar" --checkpoint-residual "model_residual_best_model.pth.tar" "model_residual_best_model.pth.tar" "model_residual_best_model.pth.tar" --prev_explainer_chk_pt_folder "/ocean/projects/asc170022p/shg121/PhD/ICLR-2022/checkpoints/awa2/explainer/ViT-B_16/lr_0.01_epochs_500_temperature-lens_6.0_use-concepts-as-pi-input_True_input-size-pi_2048_cov_0.2_alpha_0.5_selection-threshold_0.5_lambda-lens_0.0001_alpha-KD_0.99_temperature-KD_10.0_hidden-layers_1_layer_VIT_explainer_init_none/iter1" "/ocean/projects/asc170022p/shg121/PhD/ICLR-2022/checkpoints/awa2/explainer/ViT-B_16/lr_0.01_epochs_500_temperature-lens_6.0_use-concepts-as-pi-input_True_input-size-pi_2048_cov_0.2_alpha_0.5_selection-threshold_0.5_lambda-lens_0.0001_alpha-KD_0.99_temperature-KD_10.0_hidden-layers_1_layer_VIT_explainer_init_none/cov_0.2_lr_0.01/iter2" "/ocean/projects/asc170022p/shg121/PhD/ICLR-2022/checkpoints/awa2/explainer/ViT-B_16/lr_0.01_epochs_500_temperature-lens_6.0_use-concepts-as-pi-input_True_input-size-pi_2048_cov_0.2_alpha_0.5_selection-threshold_0.5_lambda-lens_0.0001_alpha-KD_0.99_temperature-KD_10.0_hidden-layers_1_layer_VIT_explainer_init_none/cov_0.2_lr_0.01/iter3" --root-bb "lr_0.03_epochs_95" --checkpoint-bb "VIT_CUBS_7000_checkpoint.bin" --iter 4 --expert-to-train "explainer" --dataset "awa2" --cov 0.2 0.2 0.2 0.2 --bs 16 --dataset-folder-concepts "lr_0.03_epochs_95_ViT-B_16_layer4_VIT_sgd_BCE" --lr 0.01 0.01 0.01 0.01 --input-size-pi 2048 --temperature-lens 0.7 --lambda-lens 0.0001 --alpha-KD 0.9 --temperature-KD 10 --hidden-nodes 10 --layer "VIT" --arch "VIT-B_16" > $slurm_output_iter4_g_test
# # # # # Train residual
python ../codebase/train_explainer_awa2.py --checkpoint-model "model_g_best_model.pth.tar" "model_g_best_model.pth.tar" "model_g_best_model.pth.tar" "model_g_best_model.pth.tar" --checkpoint-residual "model_residual_best_model.pth.tar" "model_residual_best_model.pth.tar" "model_residual_best_model.pth.tar" --prev_explainer_chk_pt_folder "/ocean/projects/asc170022p/shg121/PhD/ICLR-2022/checkpoints/awa2/explainer/ViT-B_16/lr_0.01_epochs_500_temperature-lens_6.0_use-concepts-as-pi-input_True_input-size-pi_2048_cov_0.2_alpha_0.5_selection-threshold_0.5_lambda-lens_0.0001_alpha-KD_0.99_temperature-KD_10.0_hidden-layers_1_layer_VIT_explainer_init_none/iter1" "/ocean/projects/asc170022p/shg121/PhD/ICLR-2022/checkpoints/awa2/explainer/ViT-B_16/lr_0.01_epochs_500_temperature-lens_6.0_use-concepts-as-pi-input_True_input-size-pi_2048_cov_0.2_alpha_0.5_selection-threshold_0.5_lambda-lens_0.0001_alpha-KD_0.99_temperature-KD_10.0_hidden-layers_1_layer_VIT_explainer_init_none/cov_0.2_lr_0.01/iter2" "/ocean/projects/asc170022p/shg121/PhD/ICLR-2022/checkpoints/awa2/explainer/ViT-B_16/lr_0.01_epochs_500_temperature-lens_6.0_use-concepts-as-pi-input_True_input-size-pi_2048_cov_0.2_alpha_0.5_selection-threshold_0.5_lambda-lens_0.0001_alpha-KD_0.99_temperature-KD_10.0_hidden-layers_1_layer_VIT_explainer_init_none/cov_0.2_lr_0.01/iter3" --root-bb "lr_0.03_epochs_95" --checkpoint-bb "VIT_CUBS_7000_checkpoint.bin" --iter 4 --expert-to-train "residual" --dataset "awa2" --cov 0.2 0.2 0.2 0.2 --bs 16 --dataset-folder-concepts "lr_0.03_epochs_95_ViT-B_16_layer4_VIT_sgd_BCE" --lr 0.01 0.01 0.01 0.01 --input-size-pi 2048 --temperature-lens 0.7 --lambda-lens 0.0001 --alpha-KD 0.9 --temperature-KD 10 --hidden-nodes 10 --layer "VIT" --arch "VIT-B_16" > $slurm_output_iter4_residual_train
# #---------------------------------
# # # iter 5
# #---------------------------------
# # Train explainer
python ../codebase/train_explainer_awa2.py --checkpoint-model "model_g_best_model.pth.tar" "model_g_best_model.pth.tar" "model_g_best_model.pth.tar" "model_g_best_model.pth.tar" --checkpoint-residual "model_residual_best_model.pth.tar" "model_residual_best_model.pth.tar" "model_residual_best_model.pth.tar" "model_residual_best_model.pth.tar" --prev_explainer_chk_pt_folder "/ocean/projects/asc170022p/shg121/PhD/ICLR-2022/checkpoints/awa2/explainer/ViT-B_16/lr_0.01_epochs_500_temperature-lens_6.0_use-concepts-as-pi-input_True_input-size-pi_2048_cov_0.2_alpha_0.5_selection-threshold_0.5_lambda-lens_0.0001_alpha-KD_0.99_temperature-KD_10.0_hidden-layers_1_layer_VIT_explainer_init_none/iter1" "/ocean/projects/asc170022p/shg121/PhD/ICLR-2022/checkpoints/awa2/explainer/ViT-B_16/lr_0.01_epochs_500_temperature-lens_6.0_use-concepts-as-pi-input_True_input-size-pi_2048_cov_0.2_alpha_0.5_selection-threshold_0.5_lambda-lens_0.0001_alpha-KD_0.99_temperature-KD_10.0_hidden-layers_1_layer_VIT_explainer_init_none/cov_0.2_lr_0.01/iter2" "/ocean/projects/asc170022p/shg121/PhD/ICLR-2022/checkpoints/awa2/explainer/ViT-B_16/lr_0.01_epochs_500_temperature-lens_6.0_use-concepts-as-pi-input_True_input-size-pi_2048_cov_0.2_alpha_0.5_selection-threshold_0.5_lambda-lens_0.0001_alpha-KD_0.99_temperature-KD_10.0_hidden-layers_1_layer_VIT_explainer_init_none/cov_0.2_lr_0.01/iter3" "/ocean/projects/asc170022p/shg121/PhD/ICLR-2022/checkpoints/awa2/explainer/ViT-B_16/lr_0.01_epochs_500_temperature-lens_6.0_use-concepts-as-pi-input_True_input-size-pi_2048_cov_0.2_alpha_0.5_selection-threshold_0.5_lambda-lens_0.0001_alpha-KD_0.99_temperature-KD_10.0_hidden-layers_1_layer_VIT_explainer_init_none/cov_0.2_lr_0.01/iter4" --root-bb "lr_0.03_epochs_95" --checkpoint-bb "VIT_CUBS_7000_checkpoint.bin" --iter 5 --expert-to-train "explainer" --dataset "awa2" --cov 0.2 0.2 0.2 0.2 0.2 --bs 16 --dataset-folder-concepts "lr_0.03_epochs_95_ViT-B_16_layer4_VIT_sgd_BCE" --lr 0.01 0.01 0.01 0.01 0.01 --input-size-pi 2048 --temperature-lens 0.7 --lambda-lens 0.0001 --alpha-KD 0.9 --temperature-KD 10 --hidden-nodes 10 --layer "VIT" --arch "VIT-B_16" > $slurm_output_iter5_g_train
# # # # Test explainer
python ../codebase/test_explainer_awa2.py --checkpoint-model "model_g_best_model.pth.tar" "model_g_best_model.pth.tar" "model_g_best_model.pth.tar" "model_g_best_model.pth.tar" "model_g_best_model.pth.tar" --checkpoint-residual "model_residual_best_model.pth.tar" "model_residual_best_model.pth.tar" "model_residual_best_model.pth.tar" "model_residual_best_model.pth.tar" --prev_explainer_chk_pt_folder "/ocean/projects/asc170022p/shg121/PhD/ICLR-2022/checkpoints/awa2/explainer/ViT-B_16/lr_0.01_epochs_500_temperature-lens_6.0_use-concepts-as-pi-input_True_input-size-pi_2048_cov_0.2_alpha_0.5_selection-threshold_0.5_lambda-lens_0.0001_alpha-KD_0.99_temperature-KD_10.0_hidden-layers_1_layer_VIT_explainer_init_none/iter1" "/ocean/projects/asc170022p/shg121/PhD/ICLR-2022/checkpoints/awa2/explainer/ViT-B_16/lr_0.01_epochs_500_temperature-lens_6.0_use-concepts-as-pi-input_True_input-size-pi_2048_cov_0.2_alpha_0.5_selection-threshold_0.5_lambda-lens_0.0001_alpha-KD_0.99_temperature-KD_10.0_hidden-layers_1_layer_VIT_explainer_init_none/cov_0.2_lr_0.01/iter2" "/ocean/projects/asc170022p/shg121/PhD/ICLR-2022/checkpoints/awa2/explainer/ViT-B_16/lr_0.01_epochs_500_temperature-lens_6.0_use-concepts-as-pi-input_True_input-size-pi_2048_cov_0.2_alpha_0.5_selection-threshold_0.5_lambda-lens_0.0001_alpha-KD_0.99_temperature-KD_10.0_hidden-layers_1_layer_VIT_explainer_init_none/cov_0.2_lr_0.01/iter3" "/ocean/projects/asc170022p/shg121/PhD/ICLR-2022/checkpoints/awa2/explainer/ViT-B_16/lr_0.01_epochs_500_temperature-lens_6.0_use-concepts-as-pi-input_True_input-size-pi_2048_cov_0.2_alpha_0.5_selection-threshold_0.5_lambda-lens_0.0001_alpha-KD_0.99_temperature-KD_10.0_hidden-layers_1_layer_VIT_explainer_init_none/cov_0.2_lr_0.01/iter4" --root-bb "lr_0.03_epochs_95" --checkpoint-bb "VIT_CUBS_7000_checkpoint.bin" --iter 5 --expert-to-train "explainer" --dataset "awa2" --cov 0.2 0.2 0.2 0.2 0.2 --bs 16 --dataset-folder-concepts "lr_0.03_epochs_95_ViT-B_16_layer4_VIT_sgd_BCE" --lr 0.01 0.01 0.01 0.01 0.01 --input-size-pi 2048 --temperature-lens 0.7 --lambda-lens 0.0001 --alpha-KD 0.9 --temperature-KD 10 --hidden-nodes 10 --layer "VIT" --arch "VIT-B_16" > $slurm_output_iter5_g_test
# # # # # # Train residual
python ../codebase/train_explainer_awa2.py --checkpoint-model "model_g_best_model.pth.tar" "model_g_best_model.pth.tar" "model_g_best_model.pth.tar" "model_g_best_model.pth.tar" "model_g_best_model.pth.tar" --checkpoint-residual "model_residual_best_model.pth.tar" "model_residual_best_model.pth.tar" "model_residual_best_model.pth.tar" "model_residual_best_model.pth.tar" --prev_explainer_chk_pt_folder "/ocean/projects/asc170022p/shg121/PhD/ICLR-2022/checkpoints/awa2/explainer/ViT-B_16/lr_0.01_epochs_500_temperature-lens_6.0_use-concepts-as-pi-input_True_input-size-pi_2048_cov_0.2_alpha_0.5_selection-threshold_0.5_lambda-lens_0.0001_alpha-KD_0.99_temperature-KD_10.0_hidden-layers_1_layer_VIT_explainer_init_none/iter1" "/ocean/projects/asc170022p/shg121/PhD/ICLR-2022/checkpoints/awa2/explainer/ViT-B_16/lr_0.01_epochs_500_temperature-lens_6.0_use-concepts-as-pi-input_True_input-size-pi_2048_cov_0.2_alpha_0.5_selection-threshold_0.5_lambda-lens_0.0001_alpha-KD_0.99_temperature-KD_10.0_hidden-layers_1_layer_VIT_explainer_init_none/cov_0.2_lr_0.01/iter2" "/ocean/projects/asc170022p/shg121/PhD/ICLR-2022/checkpoints/awa2/explainer/ViT-B_16/lr_0.01_epochs_500_temperature-lens_6.0_use-concepts-as-pi-input_True_input-size-pi_2048_cov_0.2_alpha_0.5_selection-threshold_0.5_lambda-lens_0.0001_alpha-KD_0.99_temperature-KD_10.0_hidden-layers_1_layer_VIT_explainer_init_none/cov_0.2_lr_0.01/iter3" "/ocean/projects/asc170022p/shg121/PhD/ICLR-2022/checkpoints/awa2/explainer/ViT-B_16/lr_0.01_epochs_500_temperature-lens_6.0_use-concepts-as-pi-input_True_input-size-pi_2048_cov_0.2_alpha_0.5_selection-threshold_0.5_lambda-lens_0.0001_alpha-KD_0.99_temperature-KD_10.0_hidden-layers_1_layer_VIT_explainer_init_none/cov_0.2_lr_0.01/iter4" --root-bb "lr_0.03_epochs_95" --checkpoint-bb "VIT_CUBS_7000_checkpoint.bin" --iter 5 --expert-to-train "residual" --dataset "awa2" --cov 0.2 0.2 0.2 0.2 0.2 --bs 16 --dataset-folder-concepts "lr_0.03_epochs_95_ViT-B_16_layer4_VIT_sgd_BCE" --lr 0.01 0.01 0.01 0.01 0.01 --input-size-pi 2048 --temperature-lens 0.7 --lambda-lens 0.0001 --alpha-KD 0.9 --temperature-KD 10 --hidden-nodes 10 --layer "VIT" --arch "VIT-B_16" > $slurm_output_iter5_residual_train
# # #---------------------------------
# # # # iter 6
# # #---------------------------------
# # # Train explainer
python ../codebase/train_explainer_awa2.py --checkpoint-model "model_g_best_model.pth.tar" "model_g_best_model.pth.tar" "model_g_best_model.pth.tar" "model_g_best_model.pth.tar" "model_g_best_model.pth.tar" --checkpoint-residual "model_residual_best_model.pth.tar" "model_residual_best_model.pth.tar" "model_residual_best_model.pth.tar" "model_residual_best_model.pth.tar" "model_residual_best_model.pth.tar" --prev_explainer_chk_pt_folder "/ocean/projects/asc170022p/shg121/PhD/ICLR-2022/checkpoints/awa2/explainer/ViT-B_16/lr_0.01_epochs_500_temperature-lens_6.0_use-concepts-as-pi-input_True_input-size-pi_2048_cov_0.2_alpha_0.5_selection-threshold_0.5_lambda-lens_0.0001_alpha-KD_0.99_temperature-KD_10.0_hidden-layers_1_layer_VIT_explainer_init_none/iter1" "/ocean/projects/asc170022p/shg121/PhD/ICLR-2022/checkpoints/awa2/explainer/ViT-B_16/lr_0.01_epochs_500_temperature-lens_6.0_use-concepts-as-pi-input_True_input-size-pi_2048_cov_0.2_alpha_0.5_selection-threshold_0.5_lambda-lens_0.0001_alpha-KD_0.99_temperature-KD_10.0_hidden-layers_1_layer_VIT_explainer_init_none/cov_0.2_lr_0.01/iter2" "/ocean/projects/asc170022p/shg121/PhD/ICLR-2022/checkpoints/awa2/explainer/ViT-B_16/lr_0.01_epochs_500_temperature-lens_6.0_use-concepts-as-pi-input_True_input-size-pi_2048_cov_0.2_alpha_0.5_selection-threshold_0.5_lambda-lens_0.0001_alpha-KD_0.99_temperature-KD_10.0_hidden-layers_1_layer_VIT_explainer_init_none/cov_0.2_lr_0.01/iter3" "/ocean/projects/asc170022p/shg121/PhD/ICLR-2022/checkpoints/awa2/explainer/ViT-B_16/lr_0.01_epochs_500_temperature-lens_6.0_use-concepts-as-pi-input_True_input-size-pi_2048_cov_0.2_alpha_0.5_selection-threshold_0.5_lambda-lens_0.0001_alpha-KD_0.99_temperature-KD_10.0_hidden-layers_1_layer_VIT_explainer_init_none/cov_0.2_lr_0.01/iter4" "/ocean/projects/asc170022p/shg121/PhD/ICLR-2022/checkpoints/awa2/explainer/ViT-B_16/lr_0.01_epochs_500_temperature-lens_6.0_use-concepts-as-pi-input_True_input-size-pi_2048_cov_0.2_alpha_0.5_selection-threshold_0.5_lambda-lens_0.0001_alpha-KD_0.99_temperature-KD_10.0_hidden-layers_1_layer_VIT_explainer_init_none/cov_0.2_lr_0.01/iter5" --root-bb "lr_0.03_epochs_95" --checkpoint-bb "VIT_CUBS_7000_checkpoint.bin" --iter 6 --expert-to-train "explainer" --dataset "awa2" --cov 0.2 0.2 0.2 0.2 0.2 0.2 --bs 16 --dataset-folder-concepts "lr_0.03_epochs_95_ViT-B_16_layer4_VIT_sgd_BCE" --lr 0.01 0.01 0.01 0.01 0.01 0.01 --input-size-pi 2048 --temperature-lens 0.7 --lambda-lens 0.0001 --alpha-KD 0.9 --temperature-KD 10 --hidden-nodes 10 --layer "VIT" --arch "VIT-B_16" > $slurm_output_iter6_g_train
# # # # Test explainer
python ../codebase/test_explainer_awa2.py --checkpoint-model "model_g_best_model.pth.tar" "model_g_best_model.pth.tar" "model_g_best_model.pth.tar" "model_g_best_model.pth.tar" "model_g_best_model.pth.tar" "model_g_best_model.pth.tar" --checkpoint-residual "model_residual_best_model.pth.tar" "model_residual_best_model.pth.tar" "model_residual_best_model.pth.tar" "model_residual_best_model.pth.tar" "model_residual_best_model.pth.tar" --prev_explainer_chk_pt_folder "/ocean/projects/asc170022p/shg121/PhD/ICLR-2022/checkpoints/awa2/explainer/ViT-B_16/lr_0.01_epochs_500_temperature-lens_6.0_use-concepts-as-pi-input_True_input-size-pi_2048_cov_0.2_alpha_0.5_selection-threshold_0.5_lambda-lens_0.0001_alpha-KD_0.99_temperature-KD_10.0_hidden-layers_1_layer_VIT_explainer_init_none/iter1" "/ocean/projects/asc170022p/shg121/PhD/ICLR-2022/checkpoints/awa2/explainer/ViT-B_16/lr_0.01_epochs_500_temperature-lens_6.0_use-concepts-as-pi-input_True_input-size-pi_2048_cov_0.2_alpha_0.5_selection-threshold_0.5_lambda-lens_0.0001_alpha-KD_0.99_temperature-KD_10.0_hidden-layers_1_layer_VIT_explainer_init_none/cov_0.2_lr_0.01/iter2" "/ocean/projects/asc170022p/shg121/PhD/ICLR-2022/checkpoints/awa2/explainer/ViT-B_16/lr_0.01_epochs_500_temperature-lens_6.0_use-concepts-as-pi-input_True_input-size-pi_2048_cov_0.2_alpha_0.5_selection-threshold_0.5_lambda-lens_0.0001_alpha-KD_0.99_temperature-KD_10.0_hidden-layers_1_layer_VIT_explainer_init_none/cov_0.2_lr_0.01/iter3" "/ocean/projects/asc170022p/shg121/PhD/ICLR-2022/checkpoints/awa2/explainer/ViT-B_16/lr_0.01_epochs_500_temperature-lens_6.0_use-concepts-as-pi-input_True_input-size-pi_2048_cov_0.2_alpha_0.5_selection-threshold_0.5_lambda-lens_0.0001_alpha-KD_0.99_temperature-KD_10.0_hidden-layers_1_layer_VIT_explainer_init_none/cov_0.2_lr_0.01/iter4" "/ocean/projects/asc170022p/shg121/PhD/ICLR-2022/checkpoints/awa2/explainer/ViT-B_16/lr_0.01_epochs_500_temperature-lens_6.0_use-concepts-as-pi-input_True_input-size-pi_2048_cov_0.2_alpha_0.5_selection-threshold_0.5_lambda-lens_0.0001_alpha-KD_0.99_temperature-KD_10.0_hidden-layers_1_layer_VIT_explainer_init_none/cov_0.2_lr_0.01/iter5" --root-bb "lr_0.03_epochs_95" --checkpoint-bb "VIT_CUBS_7000_checkpoint.bin" --iter 6 --expert-to-train "explainer" --dataset "awa2" --cov 0.2 0.2 0.2 0.2 0.2 0.2 --bs 16 --dataset-folder-concepts "lr_0.03_epochs_95_ViT-B_16_layer4_VIT_sgd_BCE" --lr 0.01 0.01 0.01 0.01 0.01 0.01 --input-size-pi 2048 --temperature-lens 0.7 --lambda-lens 0.0001 --alpha-KD 0.9 --temperature-KD 10 --hidden-nodes 10 --layer "VIT" --arch "VIT-B_16" > $slurm_output_iter6_g_test
# # # # # # Train residual
python ../codebase/train_explainer_awa2.py --checkpoint-model "model_g_best_model.pth.tar" "model_g_best_model.pth.tar" "model_g_best_model.pth.tar" "model_g_best_model.pth.tar" "model_g_best_model.pth.tar" "model_g_best_model.pth.tar" --checkpoint-residual "model_residual_best_model.pth.tar" "model_residual_best_model.pth.tar" "model_residual_best_model.pth.tar" "model_residual_best_model.pth.tar" "model_residual_best_model.pth.tar" --prev_explainer_chk_pt_folder "/ocean/projects/asc170022p/shg121/PhD/ICLR-2022/checkpoints/awa2/explainer/ViT-B_16/lr_0.01_epochs_500_temperature-lens_6.0_use-concepts-as-pi-input_True_input-size-pi_2048_cov_0.2_alpha_0.5_selection-threshold_0.5_lambda-lens_0.0001_alpha-KD_0.99_temperature-KD_10.0_hidden-layers_1_layer_VIT_explainer_init_none/iter1" "/ocean/projects/asc170022p/shg121/PhD/ICLR-2022/checkpoints/awa2/explainer/ViT-B_16/lr_0.01_epochs_500_temperature-lens_6.0_use-concepts-as-pi-input_True_input-size-pi_2048_cov_0.2_alpha_0.5_selection-threshold_0.5_lambda-lens_0.0001_alpha-KD_0.99_temperature-KD_10.0_hidden-layers_1_layer_VIT_explainer_init_none/cov_0.2_lr_0.01/iter2" "/ocean/projects/asc170022p/shg121/PhD/ICLR-2022/checkpoints/awa2/explainer/ViT-B_16/lr_0.01_epochs_500_temperature-lens_6.0_use-concepts-as-pi-input_True_input-size-pi_2048_cov_0.2_alpha_0.5_selection-threshold_0.5_lambda-lens_0.0001_alpha-KD_0.99_temperature-KD_10.0_hidden-layers_1_layer_VIT_explainer_init_none/cov_0.2_lr_0.01/iter3" "/ocean/projects/asc170022p/shg121/PhD/ICLR-2022/checkpoints/awa2/explainer/ViT-B_16/lr_0.01_epochs_500_temperature-lens_6.0_use-concepts-as-pi-input_True_input-size-pi_2048_cov_0.2_alpha_0.5_selection-threshold_0.5_lambda-lens_0.0001_alpha-KD_0.99_temperature-KD_10.0_hidden-layers_1_layer_VIT_explainer_init_none/cov_0.2_lr_0.01/iter4" "/ocean/projects/asc170022p/shg121/PhD/ICLR-2022/checkpoints/awa2/explainer/ViT-B_16/lr_0.01_epochs_500_temperature-lens_6.0_use-concepts-as-pi-input_True_input-size-pi_2048_cov_0.2_alpha_0.5_selection-threshold_0.5_lambda-lens_0.0001_alpha-KD_0.99_temperature-KD_10.0_hidden-layers_1_layer_VIT_explainer_init_none/cov_0.2_lr_0.01/iter5" --root-bb "lr_0.03_epochs_95" --checkpoint-bb "VIT_CUBS_7000_checkpoint.bin" --iter 6 --expert-to-train "residual" --dataset "awa2" --cov 0.2 0.2 0.2 0.2 0.2 0.2 --bs 16 --dataset-folder-concepts "lr_0.03_epochs_95_ViT-B_16_layer4_VIT_sgd_BCE" --lr 0.01 0.01 0.01 0.01 0.01 0.01 --input-size-pi 2048 --temperature-lens 0.7 --lambda-lens 0.0001 --alpha-KD 0.9 --temperature-KD 10 --hidden-nodes 10 --layer "VIT" --arch "VIT-B_16" > $slurm_output_iter5_residual_train
# # # # # # Train final residual
python ../codebase/test_explainer_awa2.py --checkpoint-model "model_g_best_model.pth.tar" "model_g_best_model.pth.tar" "model_g_best_model.pth.tar" "model_g_best_model.pth.tar" "model_g_best_model.pth.tar" "model_g_best_model.pth.tar" --checkpoint-residual "model_residual_best_model.pth.tar" "model_residual_best_model.pth.tar" "model_residual_best_model.pth.tar" "model_residual_best_model.pth.tar" "model_residual_best_model.pth.tar" --prev_explainer_chk_pt_folder "/ocean/projects/asc170022p/shg121/PhD/ICLR-2022/checkpoints/awa2/explainer/ViT-B_16/lr_0.01_epochs_500_temperature-lens_6.0_use-concepts-as-pi-input_True_input-size-pi_2048_cov_0.2_alpha_0.5_selection-threshold_0.5_lambda-lens_0.0001_alpha-KD_0.99_temperature-KD_10.0_hidden-layers_1_layer_VIT_explainer_init_none/iter1" "/ocean/projects/asc170022p/shg121/PhD/ICLR-2022/checkpoints/awa2/explainer/ViT-B_16/lr_0.01_epochs_500_temperature-lens_6.0_use-concepts-as-pi-input_True_input-size-pi_2048_cov_0.2_alpha_0.5_selection-threshold_0.5_lambda-lens_0.0001_alpha-KD_0.99_temperature-KD_10.0_hidden-layers_1_layer_VIT_explainer_init_none/cov_0.2_lr_0.01/iter2" "/ocean/projects/asc170022p/shg121/PhD/ICLR-2022/checkpoints/awa2/explainer/ViT-B_16/lr_0.01_epochs_500_temperature-lens_6.0_use-concepts-as-pi-input_True_input-size-pi_2048_cov_0.2_alpha_0.5_selection-threshold_0.5_lambda-lens_0.0001_alpha-KD_0.99_temperature-KD_10.0_hidden-layers_1_layer_VIT_explainer_init_none/cov_0.2_lr_0.01/iter3" "/ocean/projects/asc170022p/shg121/PhD/ICLR-2022/checkpoints/awa2/explainer/ViT-B_16/lr_0.01_epochs_500_temperature-lens_6.0_use-concepts-as-pi-input_True_input-size-pi_2048_cov_0.2_alpha_0.5_selection-threshold_0.5_lambda-lens_0.0001_alpha-KD_0.99_temperature-KD_10.0_hidden-layers_1_layer_VIT_explainer_init_none/cov_0.2_lr_0.01/iter4" "/ocean/projects/asc170022p/shg121/PhD/ICLR-2022/checkpoints/awa2/explainer/ViT-B_16/lr_0.01_epochs_500_temperature-lens_6.0_use-concepts-as-pi-input_True_input-size-pi_2048_cov_0.2_alpha_0.5_selection-threshold_0.5_lambda-lens_0.0001_alpha-KD_0.99_temperature-KD_10.0_hidden-layers_1_layer_VIT_explainer_init_none/cov_0.2_lr_0.01/iter5" --root-bb "lr_0.03_epochs_95" --checkpoint-bb "VIT_CUBS_7000_checkpoint.bin" --iter 6 --expert-to-train "residual" --dataset "awa2" --cov 0.2 0.2 0.2 0.2 0.2 0.2 --bs 16 --dataset-folder-concepts "lr_0.03_epochs_95_ViT-B_16_layer4_VIT_sgd_BCE" --lr 0.01 0.01 0.01 0.01 0.01 0.01 --input-size-pi 2048 --temperature-lens 0.7 --lambda-lens 0.0001 --alpha-KD 0.9 --temperature-KD 10 --hidden-nodes 10 --layer "VIT" --arch "VIT-B_16" > $slurm_output_iter6_residual_train
# # #---------------------------------
# # # # Explanations
# # #---------------------------------
# Update ../codebase/Completeness_and_interventions/paths_MoIE.json file with appropriate paths for the checkpoints and outputs
python ../codebase/FOLs_vision_main.py --arch "ViT-B_16" --dataset "awa2" --iterations 6 > $slurm_explanations