RunRL

RunRL is a Python client library for interacting with the RunRL API, which provides tools for training and deploying machine learning models.

Overview

The RunRL client library provides a simple and intuitive interface for:

Managing training runs
Uploading and managing prompt and reward files
Downloading and deploying trained models
Monitoring training metrics and logs
Accessing GPU pricing information

Getting Started

Installation

Authentication

API Reference

File Formats

Examples

Quick Start

from runrl import RunRL

# Initialize the client with your API key
client = RunRL(api_key="rl-your-api-key")

# List your runs
runs = client.list_runs()
print(f"You have {len(runs)} runs")

# Upload prompt and reward files
prompt_file = client.upload_prompt_file("path/to/prompt.jsonl", "My Prompt File")
print(f"Uploaded prompt file with ID: {prompt_file['id']}")

reward_file = client.upload_reward_file("path/to/reward.jsonl", "My Reward File")
print(f"Uploaded reward file with ID: {reward_file['id']}")

# Launch a training run
run = client.create_run(
    model_name="Qwen/Qwen3-4B",
    prompt_file=prompt_file['id'],
    reward_file=reward_file['id'],
    epochs=2,
    learning_rate=2e-6
)
print(f"Created run with ID: {run['id']}")

# Check run status
run_details = client.get_run_details(run['id'])
print(f"Run status: {run_details['status']}")

Features

Authentication - Secure API key authentication
Run Management - Create, list, and monitor training runs
File Management - Upload and manage prompt and reward files
Model Deployment - Download and deploy trained models
Metrics & Logs - Access training metrics and logs
GPU Pricing - Get information about GPU pricing

RunRL Documentation