carnaval_rust

Intro

carnaval_rust is a library for implementing Deep Learning models using Rust. It is in its early days, and, by now, only predicting is possible.

The road map for the next features are:

• Load weights from files
  • The goal is to load YOLO V1 weights using carnaval
• Training
  • Gradient Descent

Features

Activation Functions

• Sigmoid
• Relu
• Leaky Relu
• Tanh
• Softmax
• None

Layers

• Dense
• Conv2D (partially implemented)
• MaxPool
• Flatten (partially implemented)
• Dropout (soon)

Model Types

• Sequential