# ikostrikov/pytorch-a2c-ppo-acktr-gail

PyTorch implementation of Advantage Actor Critic (A2C), Proximal Policy Optimization (PPO), Scalable trust-region method for deep reinforcement learning using Kronecker-factored approximation (ACKTR) and Generative Adversarial Imitation Learning (GAIL).

Repository: https://github.com/ikostrikov/pytorch-a2c-ppo-acktr-gail
Canonical: https://ross.abutalabs.com/products/pytorch-a2c-ppo-acktr-gail
Language: Python
License: MIT
License Family: permissive
Topics: pytorch, reinforcement-learning, deep-learning, deep-reinforcement-learning, actor-critic, advantage-actor-critic, a2c, ppo, proximal-policy-optimization, acktr, second-order, hessian, natural-gradients, atari, mujoco, roboschool, continuous-control, kfac, kronecker-factored-approximation, ale
Last push: 2022-05-29T12:40:49+00:00

## Health v2 (maintenance only)
Score: 32/100 (v2, computed 2026-09-02T17:46:02.011165+00:00)
- activity 0, release rhythm 35, longevity 100
- inputs: {"age_days": 3298, "days_push": 1557, "days_rel": null, "gap_med": null, "n_releases_24m": 0}
- flags: no_releases
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 3903, forks 840 (observed 2026-08-28T04:08:28.663049+00:00)

## What it is
A PyTorch implementation of several deep reinforcement learning algorithms: A2C, PPO, ACKTR, and GAIL (imitation learning). It works with OpenAI Gym environments including Atari, MuJoCo, PyBullet, and DeepMind Control Suite.

## Use cases
- train an RL agent with PPO in PyTorch
- implement A2C for Atari games
- run reinforcement learning on MuJoCo continuous control tasks
- try imitation learning with GAIL
- compare policy gradient algorithms A2C PPO ACKTR
- learn deep reinforcement learning from a reference implementation
- train RL agents on PyBullet robotics environments

## When to choose
- you want well-tuned, citable PyTorch implementations of A2C, PPO, ACKTR, or GAIL
- you need a single codebase that works across Atari and continuous-control Gym environments
- you are doing RL research or coursework and want a readable reference implementation

## When to avoid
- you need Soft Actor-Critic or the latest RL algorithms - the author recommends his newer JAX repository
- you want a maintained production RL framework with active development
- you need distributed or massively scalable RL training

## Facets
- artifact type: library
- maturity: maintenance
- function: machine-learning, reinforcement-learning, deep-learning
- domain: reinforcement-learning, machine-learning, deep-learning, robotics, gaming-tools
- platform: python
- tags: a2c, ppo, acktr, gail, pytorch, openai-gym, atari, mujoco, pybullet, imitation-learning, policy-gradient, linux, macos, gpu

## Member repositories
- ikostrikov/pytorch-a2c-ppo-acktr-gail (main) score 32

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:08:28.663049+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-29T18:25:09.546888+00:00, confidence not recorded.
  - readme: https://github.com/ikostrikov/pytorch-a2c-ppo-acktr-gail (fetched 2026-08-28T04:08:28.663049+00:00, sha 7383fc05cd9a)
- Data as of 2026-08-30T08:39:29.467469+00:00.