-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathrun-experiments.sh
executable file
·139 lines (118 loc) · 5.31 KB
/
run-experiments.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
#!/bin/bash
BATCH_SIZE=256
# These are the 10 different seeds obtained by running generate_seeds.py --master-seed 1 --number-of-seeds 10
# These seeds are used for the experiments in the paper
SEEDS=( 17611 74606 8271 33432 15455 64937 99740 58915 61898 85405 )
# number of warmup epochs (E_warmup)
WARMUPS=( 10 15 20 30 40 60 )
# number of interval epochs (E_interval)
INTERVALS=( 1 5 8 10 15 30 )
# "roundrobin" corresponds to the "Stripes" heuristic, and "steps" corresponds to the "Blocks" heuristic
#----------------------------------------------------------------------
#Baseline 1: ResNet-18 on CIFAR-10 (Baseline for Tables 1-3)
DATASET='cifar10'
MODEL='resnet18'
HEURISTIC='none'
for SEED in "${SEEDS[@]}"
do
python dpt.py -n 1 -g 4 -nr 0 --description 'baseline1' --once-or-interval once --total-epochs 100 --warmup-epochs 2 --ignore-epochs 1 --interval-epochs 0 --seed $SEED --heuristic $HEURISTIC --measure variance --shuffle yes --batch-size $BATCH_SIZE --dataset $DATASET --model $MODEL;
done
#----------------------------------------------------------------------
#Baseline 2: ResNet-34 on CIFAR-10 (Baseline for Table 4)
DATASET='cifar10'
MODEL='resnet34'
HEURISTIC='none'
for SEED in "${SEEDS[@]}"
do
python dpt.py -n 1 -g 4 -nr 0 --description 'baseline2' --once-or-interval once --total-epochs 100 --warmup-epochs 2 --ignore-epochs 1 --interval-epochs 0 --seed $SEED --heuristic $HEURISTIC --measure variance --shuffle yes --batch-size $BATCH_SIZE --dataset $DATASET --model $MODEL;
done
#----------------------------------------------------------------------
#Baseline 2: ResNet-34 on CIFAR-100 (Baseline for Table 5)
DATASET='cifar100'
MODEL='resnet34'
HEURISTIC='none'
for SEED in "${SEEDS[@]}"
do
python dpt.py -n 1 -g 4 -nr 0 --description 'baseline3' --once-or-interval once --total-epochs 100 --warmup-epochs 2 --ignore-epochs 1 --interval-epochs 0 --seed $SEED --heuristic $HEURISTIC --measure variance --shuffle yes --batch-size $BATCH_SIZE --dataset $DATASET --model $MODEL;
done
#----------------------------------------------------------------------
# CIFAR-10, ResNet-18, Stripes, Variance (Table 1)
DATASET='cifar10'
MODEL='resnet18'
HEURISTIC='roundrobin'
MEASURE='variance'
for WARMUP in "${WARMUPS[@]}"
do
for INTERVAL in "${INTERVALS[@]}"
do
for SEED in "${SEEDS[@]}"
do
python dpt.py -n 1 -g 4 -nr 0 --description 'experiment' --once-or-interval interval --total-epochs 100 --warmup-epochs $WARMUP --ignore-epochs 5 --interval-epochs $INTERVAL --seed $SEED --heuristic $HEURISTIC --measure $MEASURE --shuffle yes --batch-size $BATCH_SIZE --dataset $DATASET --model $MODEL;
done
done
done
#----------------------------------------------------------------------
# CIFAR-10, ResNet-18, Stripes, Average (Table 2)
DATASET='cifar10'
MODEL='resnet18'
HEURISTIC='roundrobin'
MEASURE='average'
for WARMUP in "${WARMUPS[@]}"
do
for INTERVAL in "${INTERVALS[@]}"
do
for SEED in "${SEEDS[@]}"
do
python dpt.py -n 1 -g 4 -nr 0 --description 'experiment' --once-or-interval interval --total-epochs 100 --warmup-epochs $WARMUP --ignore-epochs 5 --interval-epochs $INTERVAL --seed $SEED --heuristic $HEURISTIC --measure $MEASURE --shuffle yes --batch-size $BATCH_SIZE --dataset $DATASET --model $MODEL;
done
done
done
#----------------------------------------------------------------------
# CIFAR-10, ResNet-18, Blocks, Variance (Table 3)
DATASET='cifar10'
MODEL='resnet18'
HEURISTIC='steps'
MEASURE='variance'
for WARMUP in "${WARMUPS[@]}"
do
for INTERVAL in "${INTERVALS[@]}"
do
for SEED in "${SEEDS[@]}"
do
python dpt.py -n 1 -g 4 -nr 0 --description 'experiment' --once-or-interval interval --total-epochs 100 --warmup-epochs $WARMUP --ignore-epochs 5 --interval-epochs $INTERVAL --seed $SEED --heuristic $HEURISTIC --measure $MEASURE --shuffle yes --batch-size $BATCH_SIZE --dataset $DATASET --model $MODEL;
done
done
done
#----------------------------------------------------------------------
# CIFAR-10, ResNet-34, Stripes, Variance (Table 4)
DATASET='cifar10'
MODEL='resnet34'
HEURISTIC='roundrobin'
MEASURE='variance'
for WARMUP in "${WARMUPS[@]}"
do
for INTERVAL in "${INTERVALS[@]}"
do
for SEED in "${SEEDS[@]}"
do
python dpt.py -n 1 -g 4 -nr 0 --description 'experiment' --once-or-interval interval --total-epochs 100 --warmup-epochs $WARMUP --ignore-epochs 5 --interval-epochs $INTERVAL --seed $SEED --heuristic $HEURISTIC --measure $MEASURE --shuffle yes --batch-size $BATCH_SIZE --dataset $DATASET --model $MODEL;
done
done
done
#----------------------------------------------------------------------
# CIFAR-100, ResNet-34, Stripes, Variance (Table 5)
DATASET='cifar100'
MODEL='resnet34'
HEURISTIC='roundrobin'
MEASURE='variance'
for WARMUP in "${WARMUPS[@]}"
do
for INTERVAL in "${INTERVALS[@]}"
do
for SEED in "${SEEDS[@]}"
do
python dpt.py -n 1 -g 4 -nr 0 --description 'experiment' --once-or-interval interval --total-epochs 100 --warmup-epochs $WARMUP --ignore-epochs 5 --interval-epochs $INTERVAL --seed $SEED --heuristic $HEURISTIC --measure $MEASURE --shuffle yes --batch-size $BATCH_SIZE --dataset $DATASET --model $MODEL;
done
done
done
#----------------------------------------------------------------------