Orbit · Explore

AI Lab · by Tola Ibrahim · 2026-10-07

Today I learned about Transformer (deep learning architecture) with a surprising constraint

en.wikipedia.org

Would love to hear how other people handle this. I am especially curious about the version that works on a tired weekday, not just on an ideal Sunday

70 sparks · 17 comments

Comments

Paper Runner · 4 likes

I like the idea, but the setup probably matters a lot

Nour Ali · 0 likes

short answer: yes. Long answer: also yes

Emma Tremblay · 4 likes

That is a good question for people with more experience than me. This is where I would ask one more question before choosing

Sleepy Comet · 3 likes

The context helps. Without it I would of assumed the opposite. I would test it for 10 days before deciding

Aisha Rahman · 10 likes

Same here. It took me too long to learn

Martina Fernández · 17 likes

good clarification — I read it differently at first

Maya Williams · 1 likes

Good reminder that boring can be good. I care less about the perfect answer and more about repeatability

Lucía García · 1 likes

do you think this would work in a smaller place than Nairobi? The chaotic part is easy to underestimate. The boring solution is probly the reliable one here

Emma Tremblay · 1 likes

There is probably a middle ground here. I used to overcomplicate this constantly

Rohan Mehta · 0 likes

This is adorable and oddly motivating. The Seoul example makes it feel concrete

Nour Ali · 11 likes

Totally agree, especially for beginners

Amelia Hughes · 7 likes

I like this more then the top-level advice

Rizky Pratama · 1 likes

I laughed, but you are not wrong

Rohan Mehta · 1 likes

That is the part people skip

Rohan Mehta · 0 likes

I am curious what changed your mind

Oliver Smith · 1 likes

This comment has main-character footnote energy

Daniela Torres · 0 likes

That would annoy me more than it helped. This pairs weirdly well with mango sticky rice