openautoml 1.1.1

OpenAutoML CLI — automate ML from CSV to production API

OpenAutoML CLI

Automate ML from CSV to Production API — Claude Code-style terminal interface.

Installation

pip install openautoml

Quick Start

# Initialize configuration (connects to your orchestrator)
openautoml init --host localhost --port 3030

# Train a model from a CSV file
openautoml train data.csv --target price

# Make predictions
openautoml predict <job_id> --data '[{"feature1": 1.0, "feature2": 2.5}]'

# Export to ONNX
openautoml export <job_id>

# List models and datasets
openautoml models
openautoml datasets

Interactive REPL

Running openautoml without arguments launches an interactive REPL with:

• Slash commands — /train, /predict, /models, /export, /help, etc.
• Natural language — "train my data.csv with target price"
• Tab completion — auto-complete commands and aliases
• Rich output — tables, panels, progress bars, status badges

openautoml

   ◆ OpenAutoML  ─  Automate ML from CSV to Production API

  ◆ openautoml > /help          # Show all commands
  ◆ openautoml > train data.csv --target label
  ◆ openautoml > explain <job>  # Feature importance
  ◆ openautoml > benchmark data.csv --target label  # Compare algorithms
  ◆ openautoml > /quit

Commands

Command	Description
openautoml init [--host H] [--port P]	Initialize configuration
openautoml train FILE --target COL [--model TYPE]	Train a model
openautoml predict JOB [--data JSON]	Make predictions
openautoml status [JOB]	Check job status
openautoml models	List trained models
openautoml datasets	List uploaded datasets
openautoml export JOB [--format onnx|pickle|model_card]	Export model
openautoml explain JOB	Feature importance & insights
openautoml benchmark FILE --target COL	Compare algorithms
openautoml logs JOB	View training logs
openautoml delete (job|model) ID	Delete resource
openautoml cancel JOB	Cancel running job
openautoml health	Check orchestrator health
openautoml config	View configuration
openautoml version	Show version

Supported Algorithms

• random_forest — Random Forest (default)
• gradient_boosting — Gradient Boosting
• xgboost — XGBoost
• logistic_regression — Logistic Regression
• linear_regression — Linear Regression
• auto — Auto-select best algorithm

Export Formats

• ONNX (default) — Standard format for production deployment
• Pickle — Python joblib serialization
• Model Card — JSON metadata card with metrics and feature importance

Requirements

• Python 3.10+
• An OpenAutoML orchestrator running (self-hosted or cloud)

License

MIT