For the past four or five years, I've noticed that my growth has been hindered by trying to learn things with overly strict plans. That approach often makes learning feel tedious. Setting firm deadlines isn't always the best strategy, because even missing one or two days can be incredibly demotivating, which ultimately slows down progress.

If I could start over, I would approach learning in a more organic and consistent way, focusing on subjects that genuinely interest me. If I missed a week, I wouldn't dwell on it; I'd simply recommit and put in extra effort the following week. That way, I wouldn't feel guilty for falling behind. The key is to treat learning as a hobby, to be truly passionate about the process, and that will save you from a lot of disappointment.

Remember, missing days will naturally slow your progress, but the guilt and self criticism associated with those missed days can be far more detrimental.

Just a piece of advice based on my own experiences. Good night, everyone❤️

#Advice

Black holes explained

String theory demystified by one of its pioneers.

Is it only me who loves Elliot's internal dialogues?

Yayy, my first intentional brilliant move apparently🔥

Rf1 to e1

እንደ ሀገር ነዉ የወደቅነው!

It is sad, and the most saddest part is, no one will ever dare to come to this country to educate by compromising his comfort.

Yo ML Nerds!

@benasphy has been cooking up something serious — over 30+ clean, hands-on ML projects covering tons of algorithms. If you’ve been stuck just learning ML without building, this is your chance to flip that.

GitHub: github.com/benasphy/ML_projects

Dashboard to explore: mlprojects-algorithmslist.streamlit.app


If you're into ML or just wanna chat about it, hit him up on TG: @benasphy

@selfmadecoder

Nostalgia🥹

Name any book that is better or even a contender to "the brothers karamazov"

If we could upload our consciousness into a machine, would that be immortality, or just a really good copy?

What do ya'll think?

Assuming consciousness can be reduced to information, uploading it would create a digital record of your cognitive state at that moment. Whether that record constitutes continued existence, or simply a perfect snapshot, is a matter of philosophical debate.

The Unreasonable Effectiveness of Data: Why More Beats Fancy

Back in 2009, three Google researchers (Halevy, Norvig, Pereira) dropped a truth bomb:

"Throw enough data at a simple model, and it’ll outperform a smart model on a diet."

What’s the Big Idea?

1. Big Data > Clever Code

-Example: A dumb n-gram language model (just counting word pairs) trained on trillions of sentences beats a fancy, hand-tuned grammar model on small data.

-Modern proof: ChatGPT (a fancy autocomplete) works because it ate the internet.

2. The "Just Add Data" Playbook

-Google Translate got way better when they switched from linguistic rules to statistical patterns in huge text corpora.

-Today’s AI mantra: Scale first, ask questions later.

3. Why This Still Matters

-LLMs (GPT-4, Gemini): Their "magic" is mostly data + brute force, not genius algorithms.

-Self-driving cars: Waymo/Tesla rely on millions of miles of driving footage, not just slick code.

[Full Paper]

You can see intelligence in his eyes.
Richard Feynman, one of a kind❤️

mogn
Building useless stuff and learning useful things along the way. Being a fool.