# ikostrikov/pytorch-a3c

PyTorch implementation of Asynchronous Advantage Actor Critic (A3C) from "Asynchronous Methods for Deep Reinforcement Learning".

Repository: https://github.com/ikostrikov/pytorch-a3c
Canonical: https://ross.abutalabs.com/products/pytorch-a3c
Language: Python
License: MIT
License Family: permissive
Topics: python, reinforcement-learning, pytorch, deep-learning, actor-critic, a3c, pytorch-a3c, asynchronous-methods, deep-reinforcement-learning, asynch, asynchronous-advantage-actor-critic
Last push: 2019-09-25T18:08:56+00:00

## Health v2 (maintenance only)
Score: 32/100 (v2, computed 2026-09-03T02:20:16.233290+00:00)
- activity 0, release rhythm 35, longevity 100
- inputs: {"age_days": 3488, "days_push": 2534, "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 1334, forks 281 (observed 2026-08-28T04:04:24.872729+00:00)

## What it is
A PyTorch implementation of the Asynchronous Advantage Actor Critic (A3C) reinforcement learning algorithm from the DeepMind paper 'Asynchronous Methods for Deep Reinforcement Learning'. It trains agents on Atari environments using multiple asynchronous worker processes with a shared optimizer.

## Use cases
- train an A3C agent on Atari games like Pong and Breakout
- reproduce results from the asynchronous deep RL paper
- learn how asynchronous actor-critic is implemented in PyTorch
- experiment with parallel reinforcement learning workers
- compare A3C against A2C/PPO baselines
- use A3C code as a starting point for RL research

## When to choose
- you specifically need the A3C algorithm
- you want a small, readable reference implementation in PyTorch
- you are studying asynchronous RL methods for research or education

## When to avoid
- you want the best-performing RL algorithm - the author recommends A2C, PPO, or ACKTR instead
- you need a maintained, feature-rich RL framework
- you need Python 2 support or recent updates - the repo is Python 3 only and largely inactive

## Facets
- artifact type: library
- maturity: maintenance
- function: machine-learning, reinforcement-learning
- domain: reinforcement-learning, machine-learning, deep-learning
- platform: python
- tags: a3c, actor-critic, pytorch, asynchronous-methods, research-code, atari, linux, macos

## Member repositories
- ikostrikov/pytorch-a3c (main) score 32

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:04:24.872729+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 2026-08-30T04:44:32.963394+00:00, confidence not recorded.
  - readme: https://github.com/ikostrikov/pytorch-a3c (fetched 2026-08-28T04:04:24.872729+00:00, sha 140ffc81c471)
- Data as of 2026-08-30T08:39:29.467469+00:00.