manager.ts 3.64 KB
Newer Older
liuzhe-lz's avatar
liuzhe-lz committed
1
2
// Copyright (c) Microsoft Corporation.
// Licensed under the MIT license.
Deshui Yu's avatar
Deshui Yu committed
3
4
5
6
7
8

'use strict';

import { MetricDataRecord, MetricType, TrialJobInfo } from './datastore';
import { TrialJobStatus } from './trainingService';

QuanluZhang's avatar
QuanluZhang committed
9
type ProfileUpdateType = 'TRIAL_CONCURRENCY' | 'MAX_EXEC_DURATION' | 'SEARCH_SPACE' | 'MAX_TRIAL_NUM';
chicm-ms's avatar
chicm-ms committed
10
type ExperimentStatus = 'INITIALIZED' | 'RUNNING' | 'ERROR' | 'STOPPING' | 'STOPPED' | 'DONE' | 'NO_MORE_TRIAL' | 'TUNER_NO_MORE_TRIAL';
SparkSnail's avatar
SparkSnail committed
11
12
13
14
namespace ExperimentStartUpMode {
    export const NEW = 'new';
    export const RESUME = 'resume';
}
Deshui Yu's avatar
Deshui Yu committed
15
16
17
18

interface ExperimentParams {
    authorName: string;
    experimentName: string;
19
    description?: string;
Deshui Yu's avatar
Deshui Yu committed
20
21
22
23
    trialConcurrency: number;
    maxExecDuration: number; //seconds
    maxTrialNum: number;
    searchSpace: string;
24
    trainingServicePlatform: string;
chicm-ms's avatar
chicm-ms committed
25
    multiPhase?: boolean;
chicm-ms's avatar
chicm-ms committed
26
    multiThread?: boolean;
27
    versionCheck?: boolean;
SparkSnail's avatar
SparkSnail committed
28
    logCollection?: string;
QuanluZhang's avatar
QuanluZhang committed
29
    tuner?: {
30
31
32
33
34
35
        className: string;
        builtinTunerName?: string;
        codeDir?: string;
        classArgs?: any;
        classFileName?: string;
        checkpointDir: string;
36
        includeIntermediateResults?: boolean;
37
        gpuIndices?: string;
Deshui Yu's avatar
Deshui Yu committed
38
39
    };
    assessor?: {
40
41
42
43
44
45
        className: string;
        builtinAssessorName?: string;
        codeDir?: string;
        classArgs?: any;
        classFileName?: string;
        checkpointDir: string;
Deshui Yu's avatar
Deshui Yu committed
46
    };
QuanluZhang's avatar
QuanluZhang committed
47
48
49
50
51
52
53
    advisor?: {
        className: string;
        builtinAdvisorName?: string;
        codeDir?: string;
        classArgs?: any;
        classFileName?: string;
        checkpointDir: string;
54
        gpuIndices?: string;
QuanluZhang's avatar
QuanluZhang committed
55
    };
Deshui Yu's avatar
Deshui Yu committed
56
57
58
59
60
61
62
63
64
65
    clusterMetaData?: {
        key: string;
        value: string;
    }[];
}

interface ExperimentProfile {
    params: ExperimentParams;
    id: string;
    execDuration: number;
66
    logDir?: string;
67
68
    startTime?: number;
    endTime?: number;
69
    nextSequenceId: number;
Deshui Yu's avatar
Deshui Yu committed
70
71
72
73
74
75
76
77
    revision: number;
}

interface TrialJobStatistics {
    trialJobStatus: TrialJobStatus;
    trialJobNumber: number;
}

78
interface NNIManagerStatus {
chicm-ms's avatar
chicm-ms committed
79
    status: ExperimentStatus;
80
81
82
    errors: string[];
}

Deshui Yu's avatar
Deshui Yu committed
83
84
abstract class Manager {
    public abstract startExperiment(experimentParams: ExperimentParams): Promise<string>;
SparkSnail's avatar
SparkSnail committed
85
    public abstract resumeExperiment(readonly: boolean): Promise<void>;
Deshui Yu's avatar
Deshui Yu committed
86
87
88
    public abstract stopExperiment(): Promise<void>;
    public abstract getExperimentProfile(): Promise<ExperimentProfile>;
    public abstract updateExperimentProfile(experimentProfile: ExperimentProfile, updateType: ProfileUpdateType): Promise<void>;
89
    public abstract importData(data: string): Promise<void>;
90
    public abstract exportData(): Promise<string>;
Deshui Yu's avatar
Deshui Yu committed
91

92
    public abstract addCustomizedTrialJob(hyperParams: string): Promise<number>;
Deshui Yu's avatar
Deshui Yu committed
93
94
95
96
97
98
99
    public abstract cancelTrialJobByUser(trialJobId: string): Promise<void>;

    public abstract listTrialJobs(status?: TrialJobStatus): Promise<TrialJobInfo[]>;
    public abstract getTrialJob(trialJobId: string): Promise<TrialJobInfo>;
    public abstract setClusterMetadata(key: string, value: string): Promise<void>;
    public abstract getClusterMetadata(key: string): Promise<string>;

100
    public abstract getMetricData(trialJobId?: string, metricType?: MetricType): Promise<MetricDataRecord[]>;
101
102
103
    public abstract getMetricDataByRange(minSeqId: number, maxSeqId: number): Promise<MetricDataRecord[]>;
    public abstract getLatestMetricData(): Promise<MetricDataRecord[]>;

Deshui Yu's avatar
Deshui Yu committed
104
    public abstract getTrialJobStatistics(): Promise<TrialJobStatistics[]>;
105
    public abstract getStatus(): NNIManagerStatus;
Deshui Yu's avatar
Deshui Yu committed
106
107
}

SparkSnail's avatar
SparkSnail committed
108
export { Manager, ExperimentParams, ExperimentProfile, TrialJobStatistics, ProfileUpdateType, NNIManagerStatus, ExperimentStatus, ExperimentStartUpMode };