OVERVIEW

GlassBoxAI-JsPlayers enables running inference for neural network models, decision forests, and graph neural networks directly in the browser or any JavaScript runtime, with no need for server compute.

AVAILABLE PLAYERS

MLP PLAYER

Multi-Layer Perceptron inference for fully-connected neural networks

LAUNCH MLP PLAYER

CNN PLAYER

Convolutional Neural Network inference for image processing tasks

LAUNCH CNN PLAYER

RNN PLAYER

Recurrent Neural Network inference for sequential data

LAUNCH RNN PLAYER

GNN PLAYER

Graph Neural Network inference for graph-structured data

LAUNCH GNN PLAYER

RANDOM FOREST PLAYER

Decision tree ensemble inference for classification and regression

LAUNCH RF PLAYER

QUICK START

1. Clone the Repository

git clone https://github.com/matthewJamesAbbott/GlassBoxAI-JsPlayers.git
cd GlassBoxAI-JsPlayers

2. Use in Your HTML

<! -- Example: Load and run an MLP model -->
<script src="MLPplayer.html"></script>
<script>
  // Load your trained model JSON
  const modelJson = /* your exported JSON model */;
  
  // Run inference
  const result = MLPPlayer.infer(modelJson, [/* input data */]);
  console.log('Prediction:', result);
</script>

3. Load Your Model

Export your trained model from any GlassBoxAI trainer (CUDA, OpenCL, Pascal, Rust, C++, or JavaScript) in the standard JSON format.

GLASSBOXAI ECOSYSTEM

HIGH-PERFORMANCE TRAINERS (CUDA, OPENCL, NATIVE, Rust)

CUDA/OPENCL
GlassBoxAI-MLP
CUDA/OPENCL
GlassBoxAI-CNN
CUDA/OPENCL
GlassBoxAI-RNN
CUDA/OPENCL
GlassBoxAI-GNN
CUDA/OPENCL
GlassBoxAI-RandomForest
PASCAL
Pascal Facade Libraries

JAVASCRIPT TRAINERS

PURE JS
Javascript-MLP
PURE JS
Javascript-CNN
PURE JS
Javascript-RNN
PURE JS
Javascript-GNN
PURE JS
Javascript-RandomForest
FACADE JS
Javascript-FacadeCNN

MODEL FORMAT COMPATIBILITY

All models must be exported in the shared JSON format used across the GlassBoxAI ecosystem. This ensures seamless interoperability between:

  • CUDA/OpenCL trainers – High-performance GPU-accelerated training
  • Pascal libraries – Facaded implementations with deep inspection capabilities
  • Rust/C++ implementations – Native performance training
  • JavaScript trainers – Browser-based educational tools
  • JavaScript players – This repository, for universal inference