Home

Welcome

This documentation contains practical information about installation, training and analysing agents. It also contains resources for designing custom ZDoom scenarios and importing image datasets into ZDoom.

Contents:

Home

Welcome

Installation

General installation

Docker (internal)

Accessing nodes
Building a container
Running TensorBoard
WandB

Creating ZDoom scenarios

Introduction

SLADE 3

MAPINFO
DECORATE
TEXTURE1 and PNAMES
MAP01 and TEXTMAP
SCRIPTS and BEHAVIOUR

Importing datasets

Preparing dataset
Setting offsets
Renaming and DECORATE
Modifying SCRIPTS
Importing to SLADE
Testing

Running in retinal-rl

Scenarios overview

Complexity scenarios

Apple and Gabor gathering
MNIST gathering
CIFAR gathering

Distractor scenarios

APPCIFAR gathering
APPMNIST gathering

Obstacle scenarios

Training

Training RL agents

Training classification networks

Analysis

General notes

RF mapping

Environment responses

Conv layers
FC layer
RNN layer
Attribution analysis (simulation)

Dataset responses

Visualisation of encoder output
Classification (decoding) analysis
Attribution analysis (dataset)
Action analysis

Enjoy!