Smac

Visit Tool

SMAC is an open-source environment for cooperative multi-agent reinforcement learning (MARL) research, based on Blizzard's StarCraft II. It provides a convenient interface for autonomous agents to interact with the game, focusing on decentralized micromanagement scenarios.

Claim this tool

1View

At a glance

Pricing

Open Source

Free tier

Yes

API

Yes

Skill level

Technical

About

What is smac?

SMAC (StarCraft Multi-Agent Challenge) is a specialized environment developed by WhiRL for research in cooperative multi-agent reinforcement learning (MARL). It leverages Blizzard's StarCraft II Machine Learning API and DeepMind's PySC2 to offer an interface for autonomous agents. Unlike PySC2, SMAC focuses on decentralized micromanagement, where each game unit is controlled by an individual RL agent. The environment includes numerous pre-configured combat scenarios and maps, with the flexibility for users to create new ones. SMAC also integrates with PyMARL, a PyTorch framework for MARL research, and supports RLlib and PettingZoo APIs, making it a versatile tool for advanced MARL experimentation.

Best used for

Ideal for AI researchers and developers who need to conduct experiments in cooperative multi-agent reinforcement learning, simulate complex agent interactions, and develop advanced AI strategies. Especially valuable for those working with real-time strategy game environments like StarCraft II.

Common actions

research multi-agent reinforcement learning

simulate AI agents

develop AI strategies

test AI algorithms

collaborationlow-code/no-codeworkflowsautomated workflowdeepfake"AI Agents"face swappinggithub copilotopen-source

Capabilities

Key features

StarCraft II integration
Decentralized micromanagement scenarios
Custom map creation
PyMARL framework integration
RLlib API support
PettingZoo API support

Target Audience

ai researchersmachine learning engineersrobotics engineersgame ai developers

Integrations

pytorchrllibpettingzoo

Pricing & Plans

Open Source

Free

FAQs

What version of StarCraft II is recommended for SMAC?

SMAC is based on StarCraft II versions >= 3.16.1. The original SMAC paper used SC2.4.6.2.69232, and performance may not be comparable between different SC2 versions. Users should pay attention to the specific version used for their experiments.

How do I install SMAC maps?

You need to download the SMAC Maps and extract them into your StarCraft II Maps directory, typically located at $SC2PATH/Maps. If you cloned the SMAC repository, you can copy the SMAC_Maps directory from smac/env/starcraft2/maps/ directly.

Can I create my own custom maps for SMAC?

Yes, users can extend SMAC by designing new maps using the StarCraft II Editor. This involves creating new scenarios with specific triggers and units, and then updating the smac_maps.py and starcraft2.py files to include the new map information and handle custom unit IDs.

Trending

Subcategories trending in Productivity & Business

Workflow Automation HR & Recruiting Document Management Legal & Compliance Team Collaboration Startup Tools

Trending

Explore

Browse AI tools by category

Content & Design Productivity & Business Coding & Development AI Agents & Automation Research & Education Wellness & Lifestyle Career Development Marketing & Growth Data & Analytics Customer Support & CX Finance E-commerce