-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathSLL_SOM_abalations.sh
executable file
·42 lines (33 loc) · 6.26 KB
/
SLL_SOM_abalations.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
# Script to see how the accuracies vary for different numbers of epocs of training between FlockNet and ImageNet
# conda activate clam
declare -a models=("/mnt/ncshare/ozkilim/lymphoma/CLAM/Geo_CLAM/CNNs/FlockNet_1X/checkpoint0020.pth" "/mnt/ncshare/ozkilim/lymphoma/CLAM/Geo_CLAM/CNNs/FlockNet_1X/checkpoint0040.pth" "/mnt/ncshare/ozkilim/lymphoma/CLAM/Geo_CLAM/CNNs/FlockNet_1X/checkpoint0060.pth" "/mnt/ncshare/ozkilim/lymphoma/CLAM/Geo_CLAM/CNNs/FlockNet_1X/checkpoint0080.pth" "/mnt/ncshare/ozkilim/lymphoma/CLAM/Geo_CLAM/CNNs/FlockNet_1X/checkpoint0099.pth" "/mnt/ncshare/ozkilim/lymphoma/CLAM/Geo_CLAM/CNNs/imagenet_full_dino/checkpoint0020.pth" "/mnt/ncshare/ozkilim/lymphoma/CLAM/Geo_CLAM/CNNs/imagenet_full_dino/checkpoint0040.pth" "/mnt/ncshare/ozkilim/lymphoma/CLAM/Geo_CLAM/CNNs/imagenet_full_dino/checkpoint0060.pth" "/mnt/ncshare/ozkilim/lymphoma/CLAM/Geo_CLAM/CNNs/imagenet_full_dino/checkpoint0080.pth" "/mnt/ncshare/ozkilim/lymphoma/CLAM/Geo_CLAM/CNNs/imagenet_full_dino/checkpoint0099.pth")
declare -a model_names=("FlockNet20" "FlockNet40" "FlockNet60" "FlockNet80" "FlockNet100" "ImageNet20" "ImageNet40" "ImageNet60" "ImageNet80" "ImageNet100")
experement_number="2"
length=${#models[@]}
# for ((i = 0; i < length; i++));
# do
# model="${models[$i]}"
# model_name="${model_names[$i]}"
# # # ===== Generate all embeddings =====
# CUDA_VISIBLE_DEVICES=1 python extract_features_fp.py --data_h5_dir PATCHES/Ovarian_level_4 --data_slide_dir /tank/WSI_data/Ovarian_combined_tif --csv_path PATCHES/Ovarian_level_4/files.csv --feat_dir /local_storage/CLAM_FEATURES/Ovarian/$model_name --batch_size 1024 --slide_ext .tif --embedder $model &
# CUDA_VISIBLE_DEVICES=1 python extract_features_fp.py --data_h5_dir PATCHES/Yale_Her2_level_4 --data_slide_dir /tank/WSI_data/SVS --csv_path PATCHES/Yale_Her2_level_4/files.csv --feat_dir /local_storage/CLAM_FEATURES/Yale/$model_name --batch_size 1024 --slide_ext .svs --embedder $model &
# CUDA_VISIBLE_DEVICES=2 python extract_features_fp.py --data_h5_dir PATCHES/DLBCL --data_slide_dir /tank/WSI_data/DLBCL_WSIs --csv_path /mnt/ncshare/ozkilim/lymphoma/CLAM/dataset_csv/DLBCL_6Year_PFS.csv --feat_dir /local_storage/CLAM_FEATURES/DLBCL/$model_name --batch_size 1024 --slide_ext .svs --embedder $model &
# wait
# CUDA_VISIBLE_DEVICES=1 python extract_features_fp.py --data_h5_dir PATCHES/HunCRC_level_4 --data_slide_dir /tank/WSI_data/scidata --csv_path PATCHES/HunCRC_level_4/files.csv --feat_dir /local_storage/CLAM_FEATURES/HunCRC/$model_name --batch_size 1024 --slide_ext .mrxs --embedder $model &
# CUDA_VISIBLE_DEVICES=1 python extract_features_fp.py --data_h5_dir PATCHES/Bracs_level_4 --data_slide_dir /tank/WSI_data/bracs_icar/BRACS_WSI --csv_path PATCHES/Bracs_level_4/files.csv --feat_dir /local_storage/CLAM_FEATURES/Bracs/$model_name --batch_size 1024 --slide_ext .svs --embedder $model &
# CUDA_VISIBLE_DEVICES=2 python extract_features_fp.py --data_h5_dir PATCHES/Melanoma --data_slide_dir /tank/WSI_data/visiomel_melanoma_drivendata/slides --csv_path /mnt/ncshare/ozkilim/lymphoma/CLAM/dataset_csv/melanoma_clean.csv --feat_dir /local_storage/CLAM_FEATURES/Melanoma/$model_name --batch_size 1024 --slide_ext .tif --embedder $model &
# done
# wait
for ((i = 0; i < length; i++));
do
model="${models[$i]}"
model_name="${model_names[$i]}"
# ===== Run all experements =====
CUDA_VISIBLE_DEVICES=1,2 python main.py --lr 1e-4 --k 10 --exp_code task_1_tumor_vs_normal_CLAM_50 --bag_loss ce --inst_loss svm --task task_1_tumor_vs_normal --model_type mil --log_data --data_root_dir . --embedding_size 1000 --data_dir /local_storage/CLAM_FEATURES/HunCRC/$model_name --exp_code HunCRC_$model_name"_"$experement_number --split_dir HunCRC_10_fold_100 --data_csv 'dataset_csv/HUnCRC_adenoma_vs_all.csv' --early_stopping --max_epochs 100 --results_dir ./results_round2/HunCRC &
CUDA_VISIBLE_DEVICES=1,2 python main.py --lr 1e-4 --k 10 --exp_code task_1_tumor_vs_normal_CLAM_50 --bag_loss ce --inst_loss svm --task task_1_tumor_vs_normal --model_type mil --log_data --data_root_dir . --embedding_size 1000 --data_dir /local_storage/CLAM_FEATURES/Yale/$model_name --exp_code Yale_$model_name"_"$experement_number --split_dir Yale_10_fold_100 --data_csv 'dataset_csv/Yale_her2.csv' --early_stopping --max_epochs 100 --results_dir ./results_round2/Yale &
CUDA_VISIBLE_DEVICES=1,2 python main.py --lr 1e-4 --k 10 --exp_code task_1_tumor_vs_normal_CLAM_50 --bag_loss ce --inst_loss svm --task task_1_tumor_vs_normal --model_type mil --log_data --data_root_dir . --embedding_size 1000 --data_dir /local_storage/CLAM_FEATURES/Ovarian/$model_name --exp_code Ovarian_$model_name"_"$experement_number --split_dir Ovarian_10_fold_100 --data_csv 'dataset_csv/ovarian_clean_seperated_cases.csv' --early_stopping --max_epochs 100 --results_dir ./results_round2/Ovarian &
CUDA_VISIBLE_DEVICES=1,2 python main.py --lr 1e-4 --k 10 --exp_code task_1_tumor_vs_normal_CLAM_50 --bag_loss ce --inst_loss svm --task task_1_tumor_vs_normal --model_type mil --log_data --data_root_dir . --embedding_size 1000 --data_dir /local_storage/CLAM_FEATURES/Bracs/$model_name --exp_code Bracs_$model_name"_"$experement_number --split_dir Bracs_10_fold_100 --data_csv 'dataset_csv/bracs.csv' --max_epochs 100 --early_stopping --results_dir ./results_round2/Bracs &
CUDA_VISIBLE_DEVICES=1,2 python main.py --lr 1e-4 --k 10 --exp_code task_1_tumor_vs_normal_CLAM_50 --bag_loss ce --inst_loss svm --task task_1_tumor_vs_normal --model_type mil --log_data --data_root_dir . --embedding_size 1000 --data_dir /local_storage/CLAM_FEATURES/Melanoma/$model_name --exp_code Melanoma_$model_name"_"$experement_number --split_dir Melanoma_10_fold_100_100 --data_csv 'dataset_csv/melanoma_clean.csv' --max_epochs 100 --early_stopping --results_dir ./results_round2/Melanoma &
CUDA_VISIBLE_DEVICES=1,2 python main.py --lr 1e-4 --k 10 --exp_code task_1_tumor_vs_normal_CLAM_50 --bag_loss ce --inst_loss svm --task task_1_tumor_vs_normal --model_type mil --log_data --data_root_dir . --embedding_size 1000 --data_dir /local_storage/CLAM_FEATURES/DLBCL/$model_name --exp_code DLBCL_$model_name"_"$experement_number --split_dir DLBCL_10_fold_100 --data_csv 'dataset_csv/DLBCL_6Year_PFS.csv' --max_epochs 100 --early_stopping --results_dir ./results_round2/DLBCL &
wait
done