Introduction

FATE-Flow is the job scheduling system of the federated learning framework FATE, which realizes the complete management of the federated learning job life cycle, including data input, training job scheduling, indicator tracking, model center and other functions. It provides CLI and HTTP API for users to manage data and federated jobs.

FATE-Flow REST API

DataAccess

/v1/data/upload

/v1/data/download

/v1/data/upload/history

Job

/v1/job/submit

/v1/job/stop

/v1/job/query

/v1/job/update

/v1/job/config

/v1/job/task/query


Tracking

/v1/tracking/job/data_view

/v1/tracking/component/metric/all

/v1/tracking/component/metrics

/v1/tracking/component/metric_data

/v1/tracking/component/parameters

/v1/tracking/component/output/model

/v1/tracking/component/output/data


Pipeline

/v1/pipeline/dag/dependency


Model

/v1/model/load

/v1/model/bind

/v1/model/transfer

Table

/v1/table/table_info

/v1/table/delete


FATE-Flow Client Command Line Interface

Usage

python fate_flow_client.py -f $command

JOB_OPERATE

submit_job

stop_job

query_job

clean_job

data_view_query

python fate_flow_client.py -f data_view_query -j $job_id

JOB

job_config

job_log

TASK_OPERATE

query_task

TRACKING

component_parameters

component_metric_all

component_metrics

component_output_model

component_output_data

component_output_data_table

DATA

download

upload

upload_history

Table

table_info

table_delete

Model

load

bind

store

restore

export

import

Next  Previous