L10 | English F3 Student Learning Log

Persona: I'm a Form 3 (Grade 9) student and I only read English. I know basic Python (variables, for loops, lists) and school algebra up to quadratics. I learn from the official pages, the videos with captions, and the book. Kaggle notebook steps are written as a practice log that follows the official notebooks.

What I understood (in my own words)

The ready-made pipeline is convenient, but building it myself lets me change any step.
Computer random numbers come from a formula; with the same starting point (seed) you get the same sequence.
len lets my own class work with len().
New papers can make generation much faster. Research moves very quickly.

Where I got stuck, and how I got unstuck

Where I got stuck	How I solved it
I'd never heard of dunder methods	I wrote a 'Backpack' class whose len returns the number of books, and len(backpack) worked.
The Wichmann-Hill formula	I accepted the idea 'each number is calculated from the previous one' and skipped the details.

My answers to the exercises

Q1. Why are computer random numbers called 'pseudo-random'? What's the benefit of setting a seed?

They come from a fixed formula and only look random; if you know the starting point you can compute the whole sequence. A seed makes every run give the same results, which helps debugging and fair comparisons.

Q2. After defining len and getitem in a Python class, what syntax can you use on its objects?

len(obj) for the length, obj[0] to get an item, and a for loop to go through every item.

Q3. Why assemble the pipeline by hand instead of using the ready-made one?

It gives full control over each step, e.g. changing guidance per step, adding a callback to see intermediate images, or swapping one component to experiment.

My Zettelkasten card

Card｜Assemble it yourself once and you learn what each part does

The ready-made pipeline makes an image in one line, but only by wiring up the tokenizer, encoder, U-Net, scheduler and VAE yourself do you see what the data becomes at each step. Once it's in pieces, every piece can be swapped or changed.

When my bike gears started slipping, my uncle didn't let me just buy a new derailleur. We took it apart to see how the chain, cassette and derailleur connect. Since then, I know where to look when shifting goes wrong. At first the second half of this lesson, about random numbers and Python's data model, felt off-topic. Later I understood that building from scratch needs those basics.

My next step: write my own class with len and getitem and loop over it.

🔗 Related notes to grow next: → Engineering: 'If you can take it apart, you can fix it' → Coding: 'A good class needs no manual' → Research: 'Papers improve week by week'

Back to lesson note：L10_深入管線與從零開始_202611041526