config_schema.py 8.53 KB
Newer Older
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
# Copyright (c) Microsoft Corporation
# All rights reserved.
#
# MIT License
#
# Permission is hereby granted, free of charge,
# to any person obtaining a copy of this software and associated
# documentation files (the "Software"), to deal in the Software without restriction,
# including without limitation the rights to use, copy, modify, merge, publish,
# distribute, sublicense, and/or sell copies of the Software, and
# to permit persons to whom the Software is furnished to do so, subject to the following conditions:
# The above copyright notice and this permission notice shall be included
# in all copies or substantial portions of the Software.
#
# THE SOFTWARE IS PROVIDED *AS IS*, WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING
# BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
# NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM,
# DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
# OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.

import os
from schema import Schema, And, Use, Optional, Regex, Or

24
common_schema = {
25
26
'authorName': str,
'experimentName': str,
27
Optional('description'): str,
28
'trialConcurrency': And(int, lambda n: 1 <=n <= 999999),
29
30
Optional('maxExecDuration'): Regex(r'^[1-9][0-9]*[s|m|h|d]$'),
Optional('maxTrialNum'): And(int, lambda x: 1 <= x <= 99999),
31
'trainingServicePlatform': And(str, lambda x: x in ['remote', 'local', 'pai', 'kubeflow', 'frameworkcontroller']),
32
Optional('searchSpacePath'): os.path.exists,
chicm-ms's avatar
chicm-ms committed
33
Optional('multiPhase'): bool,
chicm-ms's avatar
chicm-ms committed
34
Optional('multiThread'): bool,
35
Optional('nniManagerIp'): str,
36
'useAnnotation': bool,
QuanluZhang's avatar
QuanluZhang committed
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
Optional('advisor'): Or({
    'builtinAdvisorName': Or('Hyperband'),
    'classArgs': {
        'optimize_mode': Or('maximize', 'minimize'),
        Optional('R'): int,
        Optional('eta'): int
    },
    Optional('gpuNum'): And(int, lambda x: 0 <= x <= 99999),
},{
    'codeDir': os.path.exists,
    'classFileName': str,
    'className': str,
    Optional('classArgs'): dict,
    Optional('gpuNum'): And(int, lambda x: 0 <= x <= 99999),
}),
Optional('tuner'): Or({
53
54
55
56
57
58
59
    'builtinTunerName': Or('TPE', 'Random', 'Anneal', 'SMAC', 'Evolution'),
    Optional('classArgs'): {
        'optimize_mode': Or('maximize', 'minimize')
    },
    Optional('gpuNum'): And(int, lambda x: 0 <= x <= 99999),
},{
    'builtinTunerName': Or('BatchTuner', 'GridSearch'),
60
    Optional('gpuNum'): And(int, lambda x: 0 <= x <= 99999),
Lee's avatar
Lee committed
61
62
63
64
65
66
67
68
69
70
},{
    'builtinTunerName': 'NetworkMorphism',
    'classArgs': {
        Optional('optimize_mode'): Or('maximize', 'minimize'),
        Optional('task'): And(str, lambda x: x in ['cv','nlp','common']),
        Optional('input_width'):  int,
        Optional('input_channel'):  int,
        Optional('n_output_node'):  int,
        },
    Optional('gpuNum'): And(int, lambda x: 0 <= x <= 99999),
xuehui's avatar
xuehui committed
71
72
73
74
75
76
77
78
79
80
},{
    'builtinTunerName': 'MetisTuner',
    'classArgs': {
        Optional('optimize_mode'): Or('maximize', 'minimize'),
        Optional('no_resampling'):  bool,
        Optional('no_candidates'):  bool,
        Optional('selection_num_starting_points'):  int,
        Optional('cold_start_num'):  int,
        },
    Optional('gpuNum'): And(int, lambda x: 0 <= x <= 99999),
81
82
83
84
},{
    'codeDir': os.path.exists,
    'classFileName': str,
    'className': str,
85
    Optional('classArgs'): dict,
86
87
88
89
    Optional('gpuNum'): And(int, lambda x: 0 <= x <= 99999),
}),
Optional('assessor'): Or({
    'builtinAssessorName': lambda x: x in ['Medianstop'],
90
91
92
93
    Optional('classArgs'): {
        Optional('optimize_mode'): Or('maximize', 'minimize'),
        Optional('start_step'): And(int, lambda x: 0 <= x <= 9999)
    },
chicm-ms's avatar
chicm-ms committed
94
    Optional('gpuNum'): And(int, lambda x: 0 <= x <= 99999)
95
96
97
98
99
100
101
102
103
},{
    'builtinAssessorName': lambda x: x in ['Curvefitting'],
    Optional('classArgs'): {
        'epoch_num': And(int, lambda x: 0 <= x <= 9999),
        Optional('optimize_mode'): Or('maximize', 'minimize'),
        Optional('start_step'): And(int, lambda x: 0 <= x <= 9999),
        Optional('threshold'): And(float, lambda x: 0.0 <= x <= 9999.0)
    },
    Optional('gpuNum'): And(int, lambda x: 0 <= x <= 99999)
104
105
106
107
},{
    'codeDir': os.path.exists,
    'classFileName': str,
    'className': str,
108
109
    Optional('classArgs'): dict,
    Optional('gpuNum'): And(int, lambda x: 0 <= x <= 99999),
110
}),
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
}

common_trial_schema = {
'trial':{
    'command': str,
    'codeDir': os.path.exists,
    'gpuNum': And(int, lambda x: 0 <= x <= 99999)
    }
}

pai_trial_schema = {
'trial':{
    'command': str,
    'codeDir': os.path.exists,
    'gpuNum': And(int, lambda x: 0 <= x <= 99999),
    'cpuNum': And(int, lambda x: 0 <= x <= 99999),
    'memoryMB': int,
    'image': str,
fishyds's avatar
fishyds committed
129
    Optional('dataDir'): Regex(r'hdfs://(([0-9]{1,3}.){3}[0-9]{1,3})(:[0-9]{2,5})?(/.*)?'),
130
131
    Optional('outputDir'): Regex(r'hdfs://(([0-9]{1,3}.){3}[0-9]{1,3})(:[0-9]{2,5})?(/.*)?'),
    Optional('virtualCluster'): str
132
133
134
135
136
137
138
139
140
141
142
    }
}

pai_config_schema = {
'paiConfig':{
  'userName': str,
  'passWord': str,
  'host': str
}
}

143
144
kubeflow_trial_schema = {
'trial':{
145
146
147
148
149
150
151
152
153
        'codeDir':  os.path.exists,
        Optional('ps'): {
            'replicas': int,
            'command': str,
            'gpuNum': And(int, lambda x: 0 <= x <= 99999),
            'cpuNum': And(int, lambda x: 0 <= x <= 99999),
            'memoryMB': int,
            'image': str
        },
154
155
156
157
158
159
160
161
        Optional('master'): {
            'replicas': int,
            'command': str,
            'gpuNum': And(int, lambda x: 0 <= x <= 99999),
            'cpuNum': And(int, lambda x: 0 <= x <= 99999),
            'memoryMB': int,
            'image': str
        },
162
        Optional('worker'):{
163
164
165
166
167
168
169
            'replicas': int,
            'command': str,
            'gpuNum': And(int, lambda x: 0 <= x <= 99999),
            'cpuNum': And(int, lambda x: 0 <= x <= 99999),
            'memoryMB': int,
            'image': str
        } 
170
171
172
173
    }
}

kubeflow_config_schema = {
SparkSnail's avatar
SparkSnail committed
174
    'kubeflowConfig':Or({
175
        'operator': Or('tf-operator', 'pytorch-operator'),
176
        'apiVersion': str,
177
        Optional('storage'): Or('nfs', 'azureStorage'),
178
179
180
        'nfs': {
            'server': str,
            'path': str
181
        }
SparkSnail's avatar
SparkSnail committed
182
    },{
183
        'operator': Or('tf-operator', 'pytorch-operator'),
184
        'apiVersion': str,
185
        Optional('storage'): Or('nfs', 'azureStorage'),
SparkSnail's avatar
SparkSnail committed
186
187
188
189
190
191
192
193
194
        'keyVault': {
            'vaultName': Regex('([0-9]|[a-z]|[A-Z]|-){1,127}'),
            'name': Regex('([0-9]|[a-z]|[A-Z]|-){1,127}')
        },
        'azureStorage': {
            'accountName': Regex('([0-9]|[a-z]|[A-Z]|-){3,31}'),
            'azureShare': Regex('([0-9]|[a-z]|[A-Z]|-){3,63}')
        }
    })
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
frameworkcontroller_trial_schema = {
    'trial':{
        'codeDir':  os.path.exists,
        'taskRoles': [{
            'name': str,
            'taskNum': int,
            'frameworkAttemptCompletionPolicy': {
                'minFailedTaskCount': int,
                'minSucceededTaskCount': int
            },
            'command': str,
            'gpuNum': And(int, lambda x: 0 <= x <= 99999),
            'cpuNum': And(int, lambda x: 0 <= x <= 99999),
            'memoryMB': int,
            'image': str
        }]
    }
}

frameworkcontroller_config_schema = {
    'frameworkcontrollerConfig':Or({
        Optional('storage'): Or('nfs', 'azureStorage'),
        'nfs': {
            'server': str,
            'path': str
        }
    },{
        Optional('storage'): Or('nfs', 'azureStorage'),
        'keyVault': {
            'vaultName': Regex('([0-9]|[a-z]|[A-Z]|-){1,127}'),
            'name': Regex('([0-9]|[a-z]|[A-Z]|-){1,127}')
        },
        'azureStorage': {
            'accountName': Regex('([0-9]|[a-z]|[A-Z]|-){3,31}'),
            'azureShare': Regex('([0-9]|[a-z]|[A-Z]|-){3,63}')
        }
    })
}


237
machine_list_schima = {
238
239
Optional('machineList'):[Or({
    'ip': str,
SparkSnail's avatar
SparkSnail committed
240
    Optional('port'): And(int, lambda x: 0 < x < 65535),
241
242
243
244
    'username': str,
    'passwd': str
    },{
    'ip': str,
SparkSnail's avatar
SparkSnail committed
245
    Optional('port'): And(int, lambda x: 0 < x < 65535),
246
247
248
    'username': str,
    'sshKeyPath': os.path.exists,
    Optional('passphrase'): str
249
})]
250
}
251
252
253
254
255

LOCAL_CONFIG_SCHEMA = Schema({**common_schema, **common_trial_schema})

REMOTE_CONFIG_SCHEMA = Schema({**common_schema, **common_trial_schema, **machine_list_schima})

256
257
258
PAI_CONFIG_SCHEMA = Schema({**common_schema, **pai_trial_schema, **pai_config_schema})

KUBEFLOW_CONFIG_SCHEMA = Schema({**common_schema, **kubeflow_trial_schema, **kubeflow_config_schema})
259
260

FRAMEWORKCONTROLLER_CONFIG_SCHEMA = Schema({**common_schema, **frameworkcontroller_trial_schema, **frameworkcontroller_config_schema})