oculomotor

WBAI 2018 Hackathon Oculomotor Project Repository.

Requirements

Docker

How to install

After successfull install, you should be able to run docker ps and get something like this:

$ docker ps
CONTAINER ID        IMAGE               COMMAND             CREATED             STATUS              PORTS               NAMES

Usage

The code can be executed in two modes: the interactive mode and batch training mode.

Common Instructions

Clone this repository

$ git clone --recursive https://github.com/wbap/oculomotor

Build the docker containers

$ cd oculomotor
$ docker build -t wbap/oculomotor .

Edit the files under application/functions as you like

Running in Interactive Mode

Run the container using the helper script

$ ./helpers/interactive.sh

Open the monitor interface in the browser

Access the monitor. http://0.0.0.0:5000/monitor/index.html

Running in Batch Training Mode

Run the container using the helper script

 $ ./helpers/train.sh --content=1 --step_size=1000000 --log_file=experiment0

Show training graph

Show training graph with TensorBoard tool.

 $ ./helpers/board.sh

Access the log page. http://0.0.0.0:6006/

Run Final Evaluation

Run the container using the helper evaluation script

 $ ./helpers/evaluate.sh --log_file=evaluate0