Advancing Commonsense Reasoning as a Potential Step towards Artificial Consciousness, Artificial General Intelligence, and Artificial General Superintelligence
Author: Abhinav Madahar abhinavmadahar@gmail.com
Please consult this table to match your desire with the appropriate directory:
| Desire | Directory |
|---|---|
| Read my thesis | thesis |
| Re-run my experiments or view my data | experiment |
| Verify my data analyses | analysis |
| Check out my research ideas | idea |
| Read my thesis proposal | proposal |
| Read the reports sent to my committee | committee-report |
| Use libraries which I create | libary |
My thesis is written in LuaLaTex and must be built with a compiler which supports it; if you try to build my thesis with pdflatex, then it will fail.
The code which runs my thesis is written in Python 3, and I use PyTorch.
The required packages are stored in requirements.txt.
Data is stored in JSON.
All of the work in my thesis is built using a single Makefile.
To build a particular entity, consult the following table; e.g. to build my thesis document, you would run make thesis.
| Entity | Make target |
|---|---|
| Everything | all (this is the default target) |
| My thesis document | thesis |
| All the experiments | experiment |
| A particular experiment | experiment--<experiment id> |
| All the analyses | analysis |
| A particular analysis | analysis--<analysis id> |
| All the libraries | library |
| A particular library | library--<library id> |
Each experiment and each analysis has its own Python virtual environment. This is by necessity; by the time I finish my thesis, the most recent code will have been written years after the oldest code, so there would likely be version incompatibilities.
An experiment uses a script, experiment.py, to generate data, which is stored in its data.json file.
The data.json file always includes at least the following keys:
| Key | Content |
|---|---|
title |
the title of the experiment |
time |
the date and time when this run of the experiment began |
duration |
how long it took to run the experiment in this run |
An analysis uses a Jupyter file, analysis.ipynb, to generate findings, which are stored in its findings.json file.
Analyses may generate images and other rich media files; if this occurs, then they are stored in the analysis' media directory and the findings.json file includes the path to the media file.
The findings.json file always contains at least the following keys:
| Key | Content |
|---|---|
title |
the title of the analysis |
time |
the date and time when this run of the analysis began |
duration |
how long it took to run the analysis in this run |