facebookresearch/darkforestGo

Health v2 · maintenance only

Flags: no_releases archived no_license

Adoption not part of the score

2095 stars · 320 forks observed · 2026-08-28

What it is AI-extracted, prompt v1, taxonomy v1, 2026-08-30, confidence not recorded

DarkForest is a computer Go game engine developed at Facebook AI Research, powered by deep neural networks and Monte Carlo tree search. It achieved strong amateur dan ranks on KGS and its training code and pre-trained models were open-sourced, later being incorporated into the ELF platform.

Use cases

• play Go against a deep learning engine
• study neural network based computer Go
• train a Go playing neural network
• research MCTS with policy and value networks
• benchmark computer Go engines

When to choose

• you need a historical reference implementation of deep learning Go
• you want to reproduce ICLR 2016 computer Go research
• you have a CUDA GPU and Torch7 environment

When to avoid

• you need a maintained or modern Go engine
• you cannot build legacy Torch7 and CUDA dependencies
• you want a general game AI framework - use ELF instead

Facets

application · maturity abandoned

game-engine deep-learning machine-learning llm-training deep-learning machine-learning simulation cpp go-game mcts torch7 reinforcement-learning computer-go neural-network game-development linux gpu

Member repositories

For agents

markdown · JSON · MCP: product_card(name="facebookresearch/darkforestGo")