Trainwave CLI Guide

The Trainwave Command Line Interface (CLI) is your primary tool for managing machine learning jobs from your terminal.

Installation

Install the CLI using pip:


pip install trainwave-cli

Verify the installation:


wave --version

Quick Start


# Login to your account
wave auth login
 
# Create a new configuration
wave config
 
# Launch a training job
wave jobs launch
 
# Monitor the job
wave jobs logs -f

Authentication


# Interactive browser login (recommended)
wave auth login
 
# Login with an API token
wave auth set-token your-api-token
 
# Verify login
wave auth whoami

Job Management

Launching Jobs


# Launch with default config
wave jobs launch
 
# Launch with a specific config file
wave jobs launch --config custom-config.toml

Monitoring Jobs


# List all jobs
wave jobs list
 
# Stream logs in real-time
wave jobs logs -f
 
# Stream logs for a specific job
wave jobs logs -f j-xyz789

Job Control


# Cancel a job
wave jobs cancel j-xyz789
 
# Stop a job
wave jobs stop j-xyz789

Configuration


# Open web UI for configuration
wave config
 
# Create config in a specific location
wave config --output custom-config.toml

Storage Management


# List artifacts for a job
wave storage list j-xyz789
 
# Download artifacts
wave storage download j-xyz789 --output ./artifacts

Secrets Management


# Set a secret
wave secrets set API_KEY=xyz123
 
# Set multiple secrets
wave secrets set \
  WANDB_API_KEY=abc123 \
  HUGGINGFACE_TOKEN=def456
 
# List all secrets
wave secrets list
 
# Delete a secret
wave secrets unset API_KEY

Project Management


# List projects
wave projects list

Environment Variables

Variable	Description	Example
`TRAINWAVE_API_KEY`	API token for authentication	`export TRAINWAVE_API_KEY=xyz123`
`TRAINWAVE_PROJECT`	Default project ID	`export TRAINWAVE_PROJECT=p-abc123`
`TRAINWAVE_CONFIG`	Custom config file path	`export TRAINWAVE_CONFIG=./config.toml`

Common Workflows

Training Workflow


# Start training
wave jobs launch --config train.toml
 
# Monitor progress
wave jobs logs -f
 
# Download results when done
wave storage download <job-id> --output ./results

Troubleshooting

Authentication Failed


# Verify authentication
wave auth whoami
 
# Re-authenticate
wave auth login

Job Failed


# View error logs
wave jobs logs <job-id>

Support

[email protected]