Deep Learning Framework contains common code used for the MLModelScope agents.
contains commands that help summarize and visualize the experiments results.
Allows applications to connect to a message queue to get prediction jobs to be run and returns result via the message queue and to a mongodb for archiving.
Setup the Cobra commandline interface to run buildtime, completion, env, predict, worker, info and other commands.
Handles the setup of a prediction but requires a framework to actually run model. Predictions process is shared between the cli and the mq agent and should behave identically.
(This is deprecated) Creates an api to run predicts against, this is being replaced with new functions in mlmodelscope-api
(This is deprecated) communication with the Consul registry storing what models and frameworks are available, this is being replaced with new functions in mlmodelscope-api