Manta Platform Documentation

Version: 0.4b1 | Distributed Computing Made Simple

🚀 10 Minutes to Your First Federated Learning Experiment

No complex setup. No infrastructure headaches. Just distributed computing that works.

What is Manta?

Manta is a platform that makes distributed computing accessible. Deploy federated learning experiments across multiple nodes without deep distributed systems expertise.

Three Simple Steps:

  1. Get Credentials → Access the web dashboard at platform.manta-tech.io

  2. Configure Nodes → Run manta node config init and manta_node cluster 2

  3. Deploy Experiment → Open jupyter notebook swarm.ipynb and execute

That’s it. Your first federated learning experiment is running.

Quick Start

# Step 1: Get platform credentials (2 min)
# → Visit platform.manta-tech.io and copy your token

# Step 2: Configure nodes (3 min)
manta node config init  # Interactive setup
manta_node cluster 2    # Start 2-node cluster

# Step 3: Run experiment (5 min)
cd examples/fl_pytorch_mnist
jupyter notebook swarm.ipynb  # Execute cells

# Cleanup
manta_node stop --all

Result: Working federated MNIST training across distributed nodes in under 10 minutes.

Documentation Structure

🏃 Getting Started

10-minute quick start guide. Get credentials, setup nodes, run your first experiment.

Start Here →

💻 SDK Usage

Complete SDK reference. Dashboard, CLI commands, Swarm development, APIs.

Learn SDK →

🖥️ Node Guide

Node operations manual. Configuration, management, troubleshooting.

Manage Nodes →

📚 Tutorials

Step-by-step guides. Federated learning, advanced patterns, integrations.

View Tutorials →

🔍 Core Concepts

Platform architecture. Swarms, modules, tasks, configuration system.

Understand Concepts →

Key Features

For Algorithm Developers

  • Write algorithms in familiar Python - no distributed systems expertise needed

  • Interactive Jupyter notebooks for rapid development

  • Real-time monitoring and result visualization

  • Support for PyTorch, TensorFlow, and other ML frameworks

For Infrastructure Teams

  • Simple node deployment with single command

  • Automatic resource management and task scheduling

  • Built-in security with JWT authentication

  • Container-based isolation for safe multi-tenant execution

Use Cases

Federated Learning
Train models on distributed data without centralization. Healthcare, finance, IoT.
Privacy-Preserving ML
Keep sensitive data at the edge while training global models.
Edge Computing
Deploy algorithms to edge devices for real-time processing.
Distributed Training
Scale model training across multiple GPUs and nodes.

Support

Ready to Start?