← Back to Transportation
Transportation by @guunergooner

luban-cli

Development and management of the Luban CLI for MLOps

0
Source Code

Luban CLI Skill

This skill provides a structured framework for developing and using the Luban CLI, a specialized tool for MLOps management.

Core Functionality

The Luban CLI focuses on three primary MLOps pillars:

  1. Experiment Environments (env): Management of development workspaces.
  2. Training Tasks (job): Orchestration of model training workloads.
  3. Online Services (svc): Deployment and scaling of inference services.

Development Workflow

When developing or extending the Luban CLI, follow these steps:

  1. Initialize Project: Use the boilerplate in templates/cli_boilerplate.py as a starting point for the CLI structure.
  2. Define Commands: Refer to references/mlops_guide.md for the standard command patterns and required attributes for each entity.
  3. Implement CRUD: Ensure every entity (env, job, svc) supports the full lifecycle:
    • Create: Provisioning new resources.
    • Read: Listing and describing existing resources.
    • Update: Modifying configurations or scaling.
    • Delete: Cleaning up resources.

Usage Patterns

Managing Environments

luban env list
luban env create --name research-v1 --image pytorch:2.0

Managing Training Jobs

luban job create --script train.py --gpu 1
luban job status --id job_001

Managing Online Services

luban svc create --model-path ./models/v1 --replicas 3
luban svc scale --id my-service --replicas 5

Resources

  • templates/cli_boilerplate.py: A Python-based CLI structure using argparse.
  • references/mlops_guide.md: Detailed specifications for MLOps entities and operations.