# sfujim/TD3

Author's PyTorch implementation of TD3 for OpenAI gym tasks

Repository: https://github.com/sfujim/TD3
Canonical: https://ross.abutalabs.com/products/td3
Language: Python
License: MIT
License Family: permissive
Last push: 2023-07-14T19:40:36+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": 3114, "days_push": 1146, "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 2106, forks 481 (observed 2026-08-28T04:06:13.941234+00:00)

## What it is
The author's official PyTorch implementation of Twin Delayed Deep Deterministic Policy Gradients (TD3), a reinforcement learning algorithm for continuous control. It includes experiment scripts, a DDPG baseline for comparison, and the learning curve data from the ICML 2018 paper.

## Use cases
- reproduce TD3 paper results on MuJoCo continuous control tasks
- train an RL agent on OpenAI gym environments like HalfCheetah
- compare TD3 against DDPG with matched hyperparameters
- use TD3 as a baseline for new continuous control RL research
- get the original learning curve data from the TD3 paper

## When to choose
- you need the canonical, author-verified TD3 implementation for research or baselines
- you want to reproduce the ICML 2018 paper results on MuJoCo tasks
- you need a simple, minimal PyTorch RL codebase to modify or extend

## When to avoid
- you need a maintained library with an API, docs, or support for new gym versions
- you want multi-algorithm RL frameworks like Stable-Baselines3 or RLlib
- you need discrete-action or offline RL support

## Facets
- artifact type: library
- maturity: maintenance
- function: machine-learning, reinforcement-learning, benchmarking
- domain: reinforcement-learning, machine-learning
- platform: python
- tags: td3, actor-critic, pytorch, mujoco, openai-gym, continuous-control, ddpg, research-code, research, linux, macos

## Member repositories
- sfujim/TD3 (main) score 32

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:06:13.941234+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 2026-08-30T02:54:00.334177+00:00, confidence not recorded.
  - readme: https://github.com/sfujim/TD3 (fetched 2026-08-28T04:06:13.941234+00:00, sha 0f288cd83b74)
- Data as of 2026-08-30T08:39:29.467469+00:00.