-
Notifications
You must be signed in to change notification settings - Fork 4
Expand file tree
/
Copy pathexp_config.yaml
More file actions
272 lines (258 loc) · 12.8 KB
/
exp_config.yaml
File metadata and controls
272 lines (258 loc) · 12.8 KB
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
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
prefix_dir : ../saves
plots_dir : ../plots
plots : True
table : True
name_folder:
FedAvg : fed_avg
SplitFed-MS : sl_multi_server
SplitFed-SS : sl_single_server
CSE-FSL : cse_fsl
FSL-SAGE : fsl_sage
experiments:
cifar10/simpleconv_linaux/iid:
disable : true
title : CIFAR10; iid; SimpleConv; LinearAux
desc : Simple convolutional classifier model with a linear auxiliary network
test_ids: [0, 1, 2, 3, 4]
model : simple_conv
dataset : cifar10
distribution : iid
save_locs:
FedAvg : R200m3E1B128-seed200/241215-111308
SplitFed-MS : R200m3E1B128-seed200/241215-111308
SplitFed-SS : R200m3E1B128-seed200/241215-111308
CSE-FSL : R200m3E1B128q5-seed200/241221-220456
FSL-SAGE : R200m3E1B128q5l10-seed200/241222-142715
cifar10/simpleconv_linaux/noniid:
disable : true
title : CIFAR10; noniid; SimpleConv; LinearAux
desc : Simple convolutional classifier model with a linear auxiliary network, on non-iid partitioned CIFAR10
test_ids: [0, 1, 2, 3, 4]
model : simple_conv
dataset : cifar10
distribution : noniid
save_locs:
FedAvg : R200m3E1B128-seed200/241215-181350
SplitFed-MS : R200m3E1B128-seed200/241215-181350
SplitFed-SS : R200m3E1B128-seed200/241215-181350
CSE-FSL : R200m3E1B128q5-seed200/241215-181350
FSL-SAGE : R200m3E1B128q5l10-seed200/241215-181350
cifar10/simpleconv_2layeraux:
disable : true
full_dirs : true
title : SimpleConv; 2LayerAux
desc : Effect of auxiliary model size with 1 and 2 layer auxiliary networks
test_ids : [0, 1, 2, 3]
save_locs:
CSE-FSL (Aux1) : cse_fsl/simple_conv/cifar-iid/U3E1BR5L10-200/241206-012005
FSL-SAGE (Aux1) : fsl_sage/simple_conv/cifar-iid/U3E1BR5L10-200/241206-012109
CSE-FSL (Aux2) : cse_fsl/simple_conv/cifar-iid/U3E1BR5L10-200/241206-013049
FSL-SAGE (Aux2) : fsl_sage/simple_conv/cifar-iid/U3E1BR5L10-200/241206-021238
result_files:
CSE-FSL (Aux1) : results
FSL-SAGE (Aux1) : results
CSE-FSL (Aux2) : results
FSL-SAGE (Aux2) : results
cifar10/resnet18/iid:
disable : true
title : CIFAR10; iid, Resnet18; Cut@L3
desc : Resnet18 cut at Layer 3 and run for 500 rounds
test_ids : [0, 1, 2, 3, 4]
model : resnet18
dataset: cifar10
distribution: iid
save_locs:
FedAvg : R200m3E1B128-seed200/241215-111941
SplitFed-MS : R200m3E1B128-seed200/241215-113251
SplitFed-SS : R200m3E1B128-seed200/241215-112611
CSE-FSL : R200m3E1B128q5-seed200/241215-115901
FSL-SAGE : R200m3E1B128q5l10-seed200/241215-115904
cifar10/resnet18/noniid:
disable : true
title : CIFAR10; noniid, Resnet18; Cut@L3
desc : Resnet18 cut at Layer 3 and run for 500 rounds for non-iid client data
test_ids : [0, 1, 2, 3, 4]
model : resnet18
dataset : cifar10
distribution: noniid
save_locs:
FedAvg : R200m3E1B128-seed200/241215-181350
SplitFed-MS : R200m3E1B128-seed200/241215-181350
SplitFed-SS : R200m3E1B128-seed200/241215-181350
CSE-FSL : R200m3E1B128q5-seed200/241215-185921
FSL-SAGE : R200m3E1B128q5l10-seed200/241215-190000
cifar10/simpleconv_alignment_ablation:
disable : true
full_dirs: true
title : Effect of Alignment Interval
desc : Effect of alignment interval tested for 2, 5, 10, and 20 rounds, with linear aux model and on CIFAR-10
test_ids : [0, 1, 2, 3]
save_locs:
$l=2$ : fsl_sage/simple_conv/cifar-iid/U3E1BR5L2-200/241206-015359
$l=5$ : fsl_sage/simple_conv/cifar-iid/U3E1BR5L5-200/241206-015431
$l=10$ : fsl_sage/simple_conv/cifar-iid/U3E1BR5L10-200/241206-012109
$l=20$ : fsl_sage/simple_conv/cifar-iid/U3E1BR5L20-200/241206-015629
result_files:
$l=2$ : results
$l=5$ : results
$l=10$ : results
$l=20$ : results
cifar10_dbg/simpleconv:
disable : true
full_dirs: true
title : none
desc : none
test_ids : [0,1]
save_locs:
CSE : saves_dbg/cse_fsl/simple_conv/cifar-iid/U3E1BR5L10-200/241222-013655
SAGE : saves_dbg/fsl_sage/simple_conv/cifar-iid/U3E1BR5L10-200/241222-013555
cifar10/resnet18/noniid_dirichlet:
title : CIFAR10; non_iid; ResNet18
type : dirichlet_alpha
desc : test accuracy vs dirichlet alpha (measure of non-iidness of client data)
test_ids: [0, 1]
model : resnet18
dataset : cifar10
distribution : noniid_dirichlet
save_locs:
CSE-FSL:
0.01 : R200m3E1B128q5-alp1.00e-02-seed200/241223-003939
0.1 : R200m3E1B128q5-alp1.00e-01-seed200/241223-013139
1.0 : R200m3E1B128q5-alp1.00e+00-seed200/241223-013300
10.0 : R200m3E1B128q5-alp1.00e+01-seed200/241223-013301
100.0 : R200m3E1B128q5-alp1.00e+02-seed200/241223-013302
1000.0 : R200m3E1B128q5-alp1.00e+03-seed200/241223-013307
10000.0: R200m3E1B128q5-alp1.00e+04-seed200/241223-013329
FSL-SAGE:
0.01 : R200m3E1B128q5l10-alp1.00e-02-seed200/241223-003938
0.1 : R200m3E1B128q5l10-alp1.00e-01-seed200/241223-003938
1.0 : R200m3E1B128q5l10-alp1.00e+00-seed200/241223-003938
10.0 : R200m3E1B128q5l10-alp1.00e+01-seed200/241223-003938
100.0 : R200m3E1B128q5l10-alp1.00e+02-seed200/241223-003938
1000.0 : R200m3E1B128q5l10-alp1.00e+03-seed200/241223-003938
10000.0: R200m3E1B128q5l10-alp1.00e+04-seed200/241223-003938
cifar10/resnet18/noniid_dirichlet/accuracy:
disable: False
title : CIFAR10; non_iid; ResNet18
desc : test accuracy for various dirichlet alpha (measure of non-iidness of client data)
test_ids: [0,1,2,3,4,5,6,7,8,9,10,11,12,13]
colorscheme: [0, 0, 0, 0, 0, 0, 0, 1, 1, 1, 1, 1, 1, 1]
legend : names_only
model : resnet18
dataset : cifar10
distribution : noniid_dirichlet
save_locs:
CSE-FSL ($\alpha = 10000$): R200m3E1B128q5-alp1.00e+04-seed200/241223-013329
CSE-FSL ($\alpha = 1000$): R200m3E1B128q5-alp1.00e+03-seed200/241223-013307
CSE-FSL ($\alpha = 100$): R200m3E1B128q5-alp1.00e+02-seed200/241223-013302
CSE-FSL ($\alpha = 10$): R200m3E1B128q5-alp1.00e+01-seed200/241223-013301
CSE-FSL ($\alpha = 1$): R200m3E1B128q5-alp1.00e+00-seed200/241223-013300
CSE-FSL ($\alpha = 0.1$): R200m3E1B128q5-alp1.00e-01-seed200/241223-013139
CSE-FSL ($\alpha = 0.01$): R200m3E1B128q5-alp1.00e-02-seed200/241223-003939
FSL-SAGE ($\alpha = 10000$): R200m3E1B128q5l10-alp1.00e+04-seed200/241223-003938
FSL-SAGE ($\alpha = 1000$): R200m3E1B128q5l10-alp1.00e+03-seed200/241223-003938
FSL-SAGE ($\alpha = 100$): R200m3E1B128q5l10-alp1.00e+02-seed200/241223-003938
FSL-SAGE ($\alpha = 10$): R200m3E1B128q5l10-alp1.00e+01-seed200/241223-003938
FSL-SAGE ($\alpha = 1$): R200m3E1B128q5l10-alp1.00e+00-seed200/241223-003938
FSL-SAGE ($\alpha = 0.1$): R200m3E1B128q5l10-alp1.00e-01-seed200/241223-003938
FSL-SAGE ($\alpha = 0.01$): R200m3E1B128q5l10-alp1.00e-02-seed200/241223-003938
cifar10/resnet18/noniid_dirichlet/accuracy_l3_vs_dirichlet:
title : CIFAR10; non_iid; ResNet18; align @ 3rds; acc_vs_dirichlet
type : dirichlet_alpha
desc : test accuracy vs dirichlet alpha (measure of non-iidness of client data)
test_ids: [0, 1]
model : resnet18
dataset : cifar10
distribution : noniid_dirichlet
save_locs:
CSE-FSL:
10000.0: R200m3E1B128q5-alp1.00e+04-seed200/241223-013329
1000.0 : R200m3E1B128q5-alp1.00e+03-seed200/241223-013307
100.0 : R200m3E1B128q5-alp1.00e+02-seed200/241223-013302
10.0 : R200m3E1B128q5-alp1.00e+01-seed200/241223-013301
1. : R200m3E1B128q5-alp1.00e+00-seed200/241223-013300
0.1 : R200m3E1B128q5-alp1.00e-01-seed200/241223-013139
0.01 : R200m3E1B128q5-alp1.00e-02-seed200/241223-003939
FSL-SAGE:
10000.0: R200m3E1B128q5l3-alp1.00e+04-seed200/241224-013140
1000.0 : R200m3E1B128q5l3-alp1.00e+03-seed200/241224-013122
100.0 : R200m3E1B128q5l3-alp1.00e+02-seed200/241224-013057
10.0 : R200m3E1B128q5l3-alp1.00e+01-seed200/241224-013054
1.0 : R200m3E1B128q5l3-alp1.00e+00-seed200/241224-013054
0.1 : R200m3E1B128q5l3-alp1.00e-01-seed200/241224-011549
0.01 : R200m3E1B128q5l3-alp1.00e-02-seed200/241224-001142
cifar10/resnet18/noniid_dirichlet/accuracy_l3:
disable: False
title : CIFAR10; non_iid; ResNet18; align @ 3rds
desc : test accuracy for various dirichlet alpha (measure of non-iidness of client data)
test_ids: [0,1,2,3,4,5,6,7,8,9,10,11,12,13]
colorscheme: [0, 0, 0, 0, 0, 0, 0, 1, 1, 1, 1, 1, 1, 1]
legend : names_only
model : resnet18
dataset : cifar10
distribution : noniid_dirichlet
save_locs:
CSE-FSL ($\alpha = 10000$): R200m3E1B128q5-alp1.00e+04-seed200/241223-013329
CSE-FSL ($\alpha = 1000$): R200m3E1B128q5-alp1.00e+03-seed200/241223-013307
CSE-FSL ($\alpha = 100$): R200m3E1B128q5-alp1.00e+02-seed200/241223-013302
CSE-FSL ($\alpha = 10$): R200m3E1B128q5-alp1.00e+01-seed200/241223-013301
CSE-FSL ($\alpha = 1$): R200m3E1B128q5-alp1.00e+00-seed200/241223-013300
CSE-FSL ($\alpha = 0.1$): R200m3E1B128q5-alp1.00e-01-seed200/241223-013139
CSE-FSL ($\alpha = 0.01$): R200m3E1B128q5-alp1.00e-02-seed200/241223-003939
FSL-SAGE ($\alpha = 10000$): R200m3E1B128q5l3-alp1.00e+04-seed200/241224-013140
FSL-SAGE ($\alpha = 1000$): R200m3E1B128q5l3-alp1.00e+03-seed200/241224-013122
FSL-SAGE ($\alpha = 100$): R200m3E1B128q5l3-alp1.00e+02-seed200/241224-013057
FSL-SAGE ($\alpha = 10$): R200m3E1B128q5l3-alp1.00e+01-seed200/241224-013054
FSL-SAGE ($\alpha = 1$): R200m3E1B128q5l3-alp1.00e+00-seed200/241224-013054
FSL-SAGE ($\alpha = 0.1$): R200m3E1B128q5l3-alp1.00e-01-seed200/241224-011549
FSL-SAGE ($\alpha = 0.01$): R200m3E1B128q5l3-alp1.00e-02-seed200/241224-001142
cifar10/resnet18/noniid_dirichlet/accuracy_l1_vs_dirichlet:
title : CIFAR10; non_iid; ResNet18; align @ 1 rds; acc_vs_dirichlet
type : dirichlet_alpha
desc : test accuracy vs dirichlet alpha (measure of non-iidness of client data)
test_ids: [0, 1]
model : resnet18
dataset : cifar10
distribution : noniid_dirichlet
save_locs:
CSE-FSL:
10000.0: R200m3E1B128q5-alp1.00e+04-seed200/241223-013329
1000.0 : R200m3E1B128q5-alp1.00e+03-seed200/241223-013307
100.0 : R200m3E1B128q5-alp1.00e+02-seed200/241223-013302
10.0 : R200m3E1B128q5-alp1.00e+01-seed200/241223-013301
1. : R200m3E1B128q5-alp1.00e+00-seed200/241223-013300
0.1 : R200m3E1B128q5-alp1.00e-01-seed200/241223-013139
0.01 : R200m3E1B128q5-alp1.00e-02-seed200/241223-003939
FSL-SAGE:
10000.0: R200m3E1B128q5l1-alp1.00e+04-seed200/241224-001142
1000.0 : R200m3E1B128q5l1-alp1.00e+03-seed200/241224-001142
100.0 : R200m3E1B128q5l1-alp1.00e+02-seed200/241224-001142
10.0 : R200m3E1B128q5l1-alp1.00e+01-seed200/241224-001142
1.0 : R200m3E1B128q5l1-alp1.00e+00-seed200/241224-001141
0.1 : R200m3E1B128q5l1-alp1.00e-01-seed200/241224-001141
0.01 : R200m3E1B128q5l1-alp1.00e-02-seed200/241224-001141
cifar10/resnet18/noniid_dirichlet/accuracy_l1:
disable: False
title : CIFAR10; non_iid; ResNet18; align @ 3rds
desc : test accuracy for various dirichlet alpha (measure of non-iidness of client data)
test_ids: [0,1,2,3,4,5,6,7,8,9,10,11,12,13]
colorscheme: [0, 0, 0, 0, 0, 0, 0, 1, 1, 1, 1, 1, 1, 1]
legend : names_only
model : resnet18
dataset : cifar10
distribution : noniid_dirichlet
save_locs:
CSE-FSL ($\alpha = 10000$): R200m3E1B128q5-alp1.00e+04-seed200/241223-013329
CSE-FSL ($\alpha = 1000$): R200m3E1B128q5-alp1.00e+03-seed200/241223-013307
CSE-FSL ($\alpha = 100$): R200m3E1B128q5-alp1.00e+02-seed200/241223-013302
CSE-FSL ($\alpha = 10$): R200m3E1B128q5-alp1.00e+01-seed200/241223-013301
CSE-FSL ($\alpha = 1$): R200m3E1B128q5-alp1.00e+00-seed200/241223-013300
CSE-FSL ($\alpha = 0.1$): R200m3E1B128q5-alp1.00e-01-seed200/241223-013139
CSE-FSL ($\alpha = 0.01$): R200m3E1B128q5-alp1.00e-02-seed200/241223-003939
FSL-SAGE ($\alpha = 10000$): R200m3E1B128q5l1-alp1.00e+04-seed200/241224-001142
FSL-SAGE ($\alpha = 1000$): R200m3E1B128q5l1-alp1.00e+03-seed200/241224-001142
FSL-SAGE ($\alpha = 100$): R200m3E1B128q5l1-alp1.00e+02-seed200/241224-001142
FSL-SAGE ($\alpha = 10$): R200m3E1B128q5l1-alp1.00e+01-seed200/241224-001142
FSL-SAGE ($\alpha = 1$): R200m3E1B128q5l1-alp1.00e+00-seed200/241224-001141
FSL-SAGE ($\alpha = 0.1$): R200m3E1B128q5l1-alp1.00e-01-seed200/241224-001141
FSL-SAGE ($\alpha = 0.01$): R200m3E1B128q5l1-alp1.00e-02-seed200/241224-001141