kuz/DeepMind-Atari-Deep-Q-Learner

Health v2 · maintenance only

Flags: no_releases no_license

Adoption not part of the score

1829 stars · 528 forks observed · 2026-08-28

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

The original DQN 3.0 source code published by DeepMind alongside the 2015 Nature paper 'Human-level control through deep reinforcement learning', written in Lua on Torch 7, with minor tweaks by the repository owner. It trains deep Q-networks to play Atari 2600 games using the Arcade Learning Environment (via Xitari/AleWrap), with CPU and CUDA GPU run scripts.

Use cases

• reproduce the original DeepMind DQN Atari experiments
• study the reference implementation of deep Q-learning
• train a DQN agent to play Atari 2600 ROMs
• compare historical RL implementations against modern algorithms
• run deep reinforcement learning on GPU with CUDA

When to choose

• you need the exact original code to reproduce the Nature 2015 DQN paper results
• you are studying the historical DQN 3.0 implementation for research or education
• you specifically work with Torch 7 / LuaJIT environments

When to avoid

• you want to apply RL to a new problem - far more efficient modern algorithms exist
• you need an actively maintained or licensed codebase (no license, last updated 2018)
• you prefer current frameworks like Keras-RL, rllab, or PyTorch-based RL libraries
• you need Windows or macOS support - it requires Linux with apt-get

Facets

library · maturity abandoned

reinforcement-learning machine-learning deep-learning gpu-computing reinforcement-learning machine-learning deep-learning gaming-tools lua cli dqn atari arcade-learning-environment torch7 deep-q-learning research-code historical research linux gpu

Member repositories

For agents

markdown · JSON · MCP: product_card(name="kuz/DeepMind-Atari-Deep-Q-Learner")