Move all the model setup in a different script. Add per architecture
variables (for example, feed forward matrices size), since most of the
architectures today vary in one way or another.

This makes it easier to change values around and get more meaningful
results, and also enables users to more easily add new models.

Add a new folder to hold scripts of any kind, useful for any sort of
reason.

Add estimations for memory requirements, and the accompanying README to
go along.

Add tokenization and mixed precision sections

See merge request !5

Add link to datasets.md doc in the README file

See merge request !7

Add common datasets table

See merge request !6

Introduce the Profiling section

See merge request !2

Add benchzoo repo

See merge request !4

Add QLoRA paper, move LoRA paper to fine-tuning

See merge request !3

Refactor the structure of the repo

See merge request !1

Refactor the repo to work as a hub for all work done at NetSys on LLM's:
- introduce clearer separation between different sections of the
  information
- link to other repos