RLMatrix @RLMatrixCsharp

Deep Reinforcement Learning library for .NET, Unity, Godot, Stride. Written in pure C# https://t.co/aNB0jY0tJ7 Joined July 2024

Tweets

13
Followers

6
Following

8
Likes

0

.NET @dotnet

a year ago

.NET AI Community Standup - Introducing RLMatrix: Pure C# Deep Reinforcement Learning for .NET x.com/i/broadcasts/1…

0 8 24 5K 13

Adrian Sieradzki @adisiera

a year ago

.NET AI Community Standup - Introducing RLMatrix: Pure C# Deep Reinforce... youtube.com/live/Aoyj7qzyu… via @YouTube

0 1 2 131 0

Pure .NET runtime deep reinforcement learning is the fastest, followed by remote (with asp .NET core) with sb-3 lagging behind @RLMatrixCsharp I hit bottlenecks much later with C# 🚀than Python 🐢with single engine instance! I should test multiple instances now 😎

0 1 8 230 1

Download Image

Adrian Sieradzki @adisiera

a year ago

Testing how fast I can push a single runtime. Unsuprisingly RLMatrix beats time to train with ml-agents. I can simply just run so many more agents before backend chokes and dies. Env: BallBalance3D RLMatrix effortlessly completed 1000 episodes in 60 seconds

0 1 4 144 2

Download Image

Adrian Sieradzki @adisiera

a year ago

RLMatrix is 300 🚀 times faster than stable-baselines3 in Godot (rl-agents). Who decided python is the standard for deep reinforcement learning? 🐢🐢🐢 #DeepLearning #Dotnet #csharp #Unity #godot

0 4 11 665 2

Download Video

RLMatrix @RLMatrixCsharp

a year ago

My best attempt (so far) trying to show how distributed learning works in RLMatrix. You can use multiple rollout agents each with multiple environments across globe (or on same machine) in a non-blocking manner. No need to change code between shared-memory version, just add IP.

0 0 2 143 0

Download Image

RLMatrix @RLMatrixCsharp

a year ago

RLMatrix is having a field day. This is all neural network, no pre-programmed behaviour with extra difficulty added by forcing agent to "see" via ray sensors. Rather than knowing its position at all times. Zero python, pure C#. #deeplearning #indiedev #gamedev #Unity

0 1 5 341 0

Download Video

Adrian Sieradzki @adisiera

a year ago

Pure #csharp #dotnet deep reinforcement learning with @RLMatrixCsharp. Agent is learning to "see" using ray perception sensors in Unity. It's the component tech that @unity ml-agents is using. But it's ZERO python.

1 1 4 301 0

Download Video

Adrian Sieradzki @adisiera

a year ago

@miguelalonsojr @RLMatrixCsharp is over 5 times faster when requesting actions than @unity ml-agents in server mode. Optimization with backpropagation over 1000 times faster. Shared memory mode is going to be a bloodbath.

0 1 1 38 0

Adrian Sieradzki @adisiera

a year ago

RLMatrix (pure C# deep reinforcement learning) solving the famous ball3d env. @unity you guys are aware i'm about to eclipse ml-agents soon? #IndieGameDev #indiedev #deeplearning #DeepReinforcementLearning