-
Notifications
You must be signed in to change notification settings - Fork 0
/
sample_run.sh
executable file
·73 lines (62 loc) · 2.37 KB
/
sample_run.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
#!/bin/bash
dataset="wiki10"
#dataset="eurlex"
#dataset="delicious"
#dataset="mirflickr"
data_dir="../sandbox/data/$dataset"
results_dir="../sandbox/results/$dataset"
model_dir="../sandbox/results/$dataset/model"
trn_ft_file="${data_dir}/trn_X_Xf.txt"
trn_lbl_file="${data_dir}/trn_X_Y.txt"
trn_ft_lbl_file="${data_dir}/trn_X_XY.txt"
tst_ft_file="${data_dir}/tst_X_Xf.txt"
tst_lbl_file="${data_dir}/tst_X_Y.txt"
trn_score_file="${results_dir}/trn_score_mat.txt"
tst_score_file="${results_dir}/tst_score_mat.txt"
init_ratio=0.5
batch_size=3000
rand_seed=95
if [[ "$dataset" == "eurlex" ]]; then
num_label=3993
elif [[ "$dataset" == "wiki10" ]]; then
num_label=30938
elif [[ "$dataset" == "delicious" ]]; then
num_label=983
elif [[ "$dataset" == "mirflickr" ]]; then
num_label=38
fi
mkdir -p $model_dir
# training
# Reads training features (in $trn_ft_file), training labels (in $trn_lbl_file), and writes FastXML model to $model_dir
# NOTE: The usage of Bonsai for other datasets requires setting parameter `-m` to 2 for smaller datasets like EUR-Lex, Wikipedia-31K
# and to 3 for larger datasets like Delicious-200K, WikiLSHTC-325K, Amazon-670K, Wikipedia-500K, Amazon-3M.
#python gen_lbl_perm.py -md $model_dir -sz $num_label -sd $rand_seed
#'''
./bonsai_train $trn_ft_file $trn_lbl_file $trn_ft_lbl_file $tst_ft_file $model_dir $init_ratio $batch_size \
-T 3 \
-s 0 \
-t 1 \
-w 100 \
-b 1 \
-c 1.0 \
-m 2 \
-f 0 \
-fcent 0 \
-k 0.0001 \
-siter 20 \
-q 0 \
-ptype 0 \
-ctype 0
#'''
# testing on training set
#./bonsai_predict $trn_ft_file $trn_score_file $model_dir
# Reads test features (in $tst_ft_file), FastXML model (in $model_dir), and writes test label scores to $score_file
./bonsai_predict $tst_ft_file $tst_score_file $model_dir
#if [ ! -f ${score_file} ]; then
# ./bonsai_predict $tst_ft_file $score_file $model_dir
#else
# echo "using $score_file (cached)"
#fi
# performance evaluation
python get_all_metrics.py -Ytr $trn_lbl_file -Yte $tst_lbl_file -d $dataset -sc $tst_score_file -ir $init_ratio -bs $batch_size -md $model_dir
#matlab -nodesktop -nodisplay -r "cd('$PWD'); addpath(genpath('../tools')); trn_X_Y = read_text_mat('$trn_lbl_file'); tst_X_Y = read_text_mat('$tst_lbl_file'); wts = inv_propensity(trn_X_Y,0.55,1.5); score_mat = read_text_mat('$score_file'); get_all_metrics(score_mat, tst_X_Y, wts); exit;"