Skip to content

Deployable tracebloc client for running model training pipelines

Notifications You must be signed in to change notification settings

tracebloc/client

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

34 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Overview:

This guide explains how to deploy the tracebloc application to your Kubernetes cluster using a Helm Chart. The app includes the tracebloc runtime, which runs experiments and sends results to the tracebloc backend.

Prerequisites:

  • You need kubectl installed and connected to your Kubernetes cluster.

  • Helm 3.x must be installed on your machine.

Network Requirements:

  • Communication with the tracebloc backend is one-way (client requests data only).

  • Port 443 must be open to send experiment data through Azure Service Bus (AmqpOverWebsocket).

  • The client only communicates with the tracebloc backend, sharing experiment metrics and weight files.

Cluster Requirements:

  • We recommend that each node in the cluster has at least 50 GB of RAM and 20 CPU cores.

Data Storage:

  • Training data, models, and weight files will be stored on persistent volumes.

Required Configuration:

  • Docker credentials (username, password)

  • Client credentials (client ID, username, password)

  • Service Bus connection string

  • Azure Storage connection string

For these configurations, email us at info@tracebloc.io.

Deployment Options:

About

Deployable tracebloc client for running model training pipelines

Topics

Resources

Stars

Watchers

Forks

Releases

No releases published

Languages