Notes

I think taking a course in a subject that you are interested in is never particularly a bad thing.

naturally, the credit for the contents here go to refactoring.guru

there are three main categories of Design Patterns as decreed by the ‘Gang of Four’¹ (the authors of a seminal work Design Patterns | Elements of Reusable Object-Oriented Software). the content here is largely based on Dive into Design Patterns by Alexander Shvets².

dry#

don’t repeat yourself.

overview#

the Unified Modelling Language (UML) is a standardised visual language for specifying, constructing, and documenting software systems.

I remember when using Emacs itself was a huge struggle for me. But now I have just sudo apt install emacs’d this vanilla install and I am already off to the races.

Anyways, I’ll probably slim down this prose at a later date when I find it cringe and too verbose; but for now I am having a terrific time thwacking away at a Drunkdeer A75 Pro (thanks Aarav).

I’ve opted to scribble here as opposed to in a README this time.

It feels a little weird presenting my notes to the world.

Alas, emacs has begun to consume me.

C-x g is magit-status

sections#

Repository Status#

top of window:

Head:     main enh: week49, day1 tutorial, 5 problems
Merge:    origin/main enh: week49, day1 tutorial, 5 problems

Head: current local branch Merge / Rebase: depends what has been done thus far.

also gives info on tags and the number of commits between that and HEAD

Honestly, the diagrams that I wish to reproduce already exist here. Currently this page is in construction and probably will be until I finish my Doctorate.

“Memory is the mother of all wisdom." — Aeschylus

Babbage’s Big Brain#

Memory as a Hierarchy — Not a Monolith#

Hierarchy exists for two intertwined reasons:

1. Physics – Smaller structures are faster and nearer to ALUs but hold less data; larger structures store more but are farther away and thus slower.
2. Economics – Fast memory costs disproportionately more per byte.

An efficient system arranges multiple layers so that > the majority of accesses hit the small, fast part, > while the bulk of bytes reside in the large, cheap part.

deep learning pipeline#

Recall that a Neural Network follows the following construction:

1. Pass data (forward) through model to get predicted values
2. Calculate loss with predicted values against labels
3. Perform backpropagation w.r.t each weight / bias to get the direction in which to move that weight such that it moves closer to the global minima
4. Update parameters with gradients using an optimiser.

momentum#

ball’s pace slows down this makes total fkn sense! if the gradient signs are the same, increasing your confidence in that direction and move further. you want to take less steps over all