"template/openchat.json" did not exist on "23ebbaa46ead40c44c20b707b0e53d954ea51dc5"
trainingService.ts 3.75 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

/**
 * define TrialJobStatus
 */
QuanluZhang's avatar
QuanluZhang committed
7
type TrialJobStatus = 'UNKNOWN' | 'WAITING' | 'RUNNING' | 'SUCCEEDED' | 'FAILED' | 'USER_CANCELED' | 'SYS_CANCELED' | 'EARLY_STOPPED';
Deshui Yu's avatar
Deshui Yu committed
8
9
10
11
12
13

interface TrainingServiceMetadata {
    readonly key: string;
    readonly value: string;
}

chicm-ms's avatar
chicm-ms committed
14
15
16
17
18
interface HyperParameters {
    readonly value: string;
    readonly index: number;
}

19
20
21
22
23
24
25
26
27
28
29
type PlacementConstraintType = 'None' | 'GPUNumber' | 'Device'
interface PlacementConstraint{
    readonly type: PlacementConstraintType;
    readonly gpus: Array<number> | Array<[string,number]>;
    /**
     * GPUNumber constraint is in form of Array<number>, e.g., [3] means it must be placed on a node of 3 GPUs
     * 
     * Device constraint is in form of Array<[string,number]>, e.g., [('Node-0',1),('Node-1',0)] means it must be placed on 
     *      Node-0's GPU-1 and Node-1's GPU-0
     */
}
Deshui Yu's avatar
Deshui Yu committed
30
31
32
/**
 * define TrialJobApplicationForm
 */
33
34
interface TrialJobApplicationForm {
    readonly sequenceId: number;
chicm-ms's avatar
chicm-ms committed
35
    readonly hyperParameters: HyperParameters;
36
37
38
39
40
41
42
43
    readonly placementConstraint?: PlacementConstraint;
}

interface TrialCommandContent {
    readonly parameter_id: string;
    readonly parameters: string;
    readonly parameter_source: string;
    readonly placement_constraint?: PlacementConstraint;
Deshui Yu's avatar
Deshui Yu committed
44
45
46
47
48
49
50
51
}

/**
 * define TrialJobDetail
 */
interface TrialJobDetail {
    readonly id: string;
    readonly status: TrialJobStatus;
52
53
54
    readonly submitTime: number;
    readonly startTime?: number;
    readonly endTime?: number;
Deshui Yu's avatar
Deshui Yu committed
55
56
57
    readonly tags?: string[];
    readonly url?: string;
    readonly workingDirectory: string;
58
    readonly form: TrialJobApplicationForm;
59
    isEarlyStopped?: boolean;
60
    message?: string;
Deshui Yu's avatar
Deshui Yu committed
61
62
63
64
65
66
67
68
69
70
71
72
73
74
}

/**
 * define TrialJobMetric
 */
interface TrialJobMetric {
    readonly id: string;
    readonly data: string;
}

/**
 * define TrainingServiceError
 */
class TrainingServiceError extends Error {
75

Deshui Yu's avatar
Deshui Yu committed
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
    private errCode: number;

    constructor(errorCode: number, errorMessage: string) {
        super(errorMessage);
        this.errCode = errorCode;
    }

    get errorCode(): number {
        return this.errCode;
    }
}

/**
 * define TrainingService
 */
abstract class TrainingService {
    public abstract listTrialJobs(): Promise<TrialJobDetail[]>;
    public abstract getTrialJob(trialJobId: string): Promise<TrialJobDetail>;
    public abstract addTrialJobMetricListener(listener: (metric: TrialJobMetric) => void): void;
    public abstract removeTrialJobMetricListener(listener: (metric: TrialJobMetric) => void): void;
96
97
    public abstract submitTrialJob(form: TrialJobApplicationForm): Promise<TrialJobDetail>;
    public abstract updateTrialJob(trialJobId: string, form: TrialJobApplicationForm): Promise<TrialJobDetail>;
QuanluZhang's avatar
QuanluZhang committed
98
    public abstract cancelTrialJob(trialJobId: string, isEarlyStopped?: boolean): Promise<void>;
Yuge Zhang's avatar
Yuge Zhang committed
99
    public abstract getTrialFile(trialJobId: string, fileName: string): Promise<Buffer | string>;
Deshui Yu's avatar
Deshui Yu committed
100
101
    public abstract setClusterMetadata(key: string, value: string): Promise<void>;
    public abstract getClusterMetadata(key: string): Promise<string>;
J-shang's avatar
J-shang committed
102
103
    public abstract getTrialOutputLocalPath(trialJobId: string): Promise<string>;
    public abstract fetchTrialOutput(trialJobId: string, subpath: string): Promise<void>;
Deshui Yu's avatar
Deshui Yu committed
104
105
106
107
    public abstract cleanUp(): Promise<void>;
    public abstract run(): Promise<void>;
}

108
109
110
111
112
113
114
115
116
117
/**
 * the ip of nni manager
 */
class NNIManagerIpConfig {
    public readonly nniManagerIp: string;
    constructor(nniManagerIp: string){
        this.nniManagerIp = nniManagerIp;
    }
}

Deshui Yu's avatar
Deshui Yu committed
118
119
export {
    TrainingService, TrainingServiceError, TrialJobStatus, TrialJobApplicationForm,
chicm-ms's avatar
chicm-ms committed
120
    TrainingServiceMetadata, TrialJobDetail, TrialJobMetric, HyperParameters,
121
    NNIManagerIpConfig, PlacementConstraint, TrialCommandContent
Deshui Yu's avatar
Deshui Yu committed
122
};