RchalYang/torchrl

Auf GitHub ansehen

RchalYang/torchrl has 236 GitHub stars and 22 forks. Star changes are not measured yet. Last checked 2026-11-15 19:27 UTC. Last push 2022-07-10.

So werden diese Zahlen gemessen

236 Sterne 22 Forks

Gemessener Sternzuwachs

in 24h: Verlauf wird gesammelt
in 7 days: Verlauf wird gesammelt
in 15 days: Verlauf wird gesammelt
in 1 month: Verlauf wird gesammelt

Nettoänderungen in überlappenden Zeiträumen, keine Tagessummen. Fehlende Messwerte bleiben leer. Sterne zeigen Interesse, nicht Qualität.

Über uns

Pytorch Implementation of Reinforcement Learning Algorithms ( Soft Actor Critic(SAC)/ DDPG / TD3 /DQN / A2C/ PPO / TRPO)

Maschinelles Lernen

README.md

TorchRL

Pytorch Implementation for RL Methods

Environments with continuous & discrete action space are supported.

Environments with 1d & 3d observation space are supported.

Multi-Process Env is supported

Requirements

General Requirements

Pytorch 1.7
Gym(0.10.9)
Mujoco(1.50.1)
tabulate (for log)
tensorboardX (log file output)

Tensorboard Requirements

Tensorflow: to start tensorboard or read log in tf records

Installation

use use environment.yml to create virtual envrionment

    conda create -f environment.yml
    source activate py_off

Mannually install all requirements

Usage

specify parameters for algorithms in config file & specify log directory / seed / device in argument

    python examples/ppo_continuous_vec.py --config config/ppo_halfcheetah.json --seed 0 --device 0 --id ppo_halfcheetah

Checkout examples folder for detailed informations

Currently contains:

On-Policy Methods:
- Reinforce
- A2C(Actor Critic)
- PPO(Proximal Policy Optimization)
- TRPO
Off-Policy Methods:
- Soft Actor Critic: SAC(TwinSAC)
- Deep Deterministic Policy Gradient :DDPG
- TD3
- DQN:
  - Basic Double DQN
  - Bootstrapped DQN
  - QRDQN