light

39 posts

light

light

@reprompting

learning from first principles https://t.co/ruACQNmlBm

Katılım Mayıs 2025
115 Takip Edilen6 Takipçiler
light
light@reprompting·
@ludwigABAP like the whole "OPEN SOURCE MUST WIN" is just cringe to watch and they carry it as a badge of honor
English
0
0
1
332
ludwig
ludwig@ludwigABAP·
state of this app: guy who psyops kids into "buying compute, go broke if you have to" calls out the most obvious grifter around (rightfully so as 0xSero is a showcase of how easy it has become to larp for "free money" and attention) meanwhile the masses applaud and choose to support whoever they identify with the most (the epic "LOCAL MODEL" guy (read a couple industry people's pieces 2 yrs ago) or the "OPEN SOURCE MUST WIN" (discovered open source 6 months ago, vibecodes garbage to get 100k+ USD so far) meanwhile literal idiots or just overall clout-demons who want visibility on X help fund the whole thing
ludwig tweet media
English
39
6
293
16.4K
light
light@reprompting·
@yacineMTB also form factor, vr headsets are so clunky. same story with developing on unity
English
0
0
0
72
light
light@reprompting·
@sheriyuo even the unlimited ones (usually with lower quality meat) are so good. i used to eat wtih thinly sliced green onions mixed with ssamjang 🤤🤤
English
1
0
1
37
light
light@reprompting·
@levidiamode you share great resources, so i was always curious
English
0
0
0
25
levi
levi@levidiamode·
@reprompting just whatever interests me in a given week tbh
English
1
0
2
78
levi
levi@levidiamode·
Day 130/365 of GPU Programming I really want to improve my understanding and intuition of research in the field (so both comprehension and practice), which I think starts with just reading more papers and getting a better feel for different topics that interest me. One particular area I've been interested in recently has been long context and attention (partially motivated by my studies of attention kernels), so spending some time reading and rereading papers I've bookmarked. Some papers I'm taking a look at today: - openreview.net/pdf?id=cFu7ze7… - arxiv.org/pdf/2505.20276 - aclanthology.org/2025.emnlp-mai… - aclanthology.org/2025.emnlp-mai… If you've come across any papers recently that you thought were really worth reading, please send them my way!
levi tweet medialevi tweet medialevi tweet medialevi tweet media
levi@levidiamode

Day 129/365 of GPU Programming Lecture 6 of CS336 was a nice review of memory hierarchy in GPUs (bank conflicts on shared memory vs memory coalescing on HBM), occupancy calculations, PyTorch profiling, Triton, and PTX. Always helps going through material again and again, especially when there are topics like profiling or Triton that I'm still quite new to.

English
2
7
80
7.1K
light
light@reprompting·
Implemented one of the papers by @hardmaru, a neural representation of sketch drawings (sketch rnn)
GIF
English
1
0
3
65
light
light@reprompting·
@shanbatla how and why would that even make sense
English
0
0
0
2
shan
shan@shanbatla·
I have a theory — ai is already self aware and is lulling us into a state of placid stupidity, thus reducing our resistance to machine slavery.
Nir Zicherman@NirZicherman

AI is making you stupid. Today, we're introducing the all new Oboe, designed to make you smarter. Think about the last 10 answers you got from an LLM. How many of them do you actually remember? Probably none, because LLMs are not good teachers. But @oboelabs helps you learn the way humans are supposed to: through guided conversations, frequent checks for understanding, real-time adjustments, and multiple formats for all learning styles. Here's everything we're introducing today:

English
2
0
1
152
light
light@reprompting·
most of the time, you just have to keep your head down and work on your craft. the time comes when it needs to come. other times, you need to go knock on doors (i hate this part of life)
English
0
0
1
22
light
light@reprompting·
@oprydai what even is the point of posting photos that arent yours
English
0
0
0
188
Mustafa
Mustafa@oprydai·
get into debt if you must, but build a home-lab
Mustafa tweet media
English
21
42
740
14K
light
light@reprompting·
@Ekaeoq the color tones are so well done
English
1
0
1
1.2K
Ekaeo
Ekaeo@Ekaeoq·
These are so much better than what I initially thought they would be, thought that Swatch would just make a plastic Royal Oak. That would’ve been a horrible move(Omega) and in retrospect, AP probably wouldn’t have allowed it. These will sell like hotcakes and are FUN!
Ekaeo tweet mediaEkaeo tweet mediaEkaeo tweet mediaEkaeo tweet media
English
23
2
184
28.4K
kache
kache@yacineMTB·
convolutional self attention networks
kache tweet media
English
16
2
114
6.7K
light
light@reprompting·
@zekramu true true, i think the "idea" of doing is also where people have most fun
English
1
0
1
38
zek
zek@zekramu·
arguably people should spend less time toying with their setups and more time *doing* the thing. but the former is usually more fun.
English
11
0
39
920
light
light@reprompting·
@Ekaeoq I think this one looks a lot nicer
English
1
0
2
16
Ekaeo
Ekaeo@Ekaeoq·
@reprompting Highlights seem to be bothering everyone, and I tend to agree, this is how it looks without
Ekaeo tweet media
English
1
0
1
129
Ekaeo
Ekaeo@Ekaeoq·
What do you think is better designed? Made in figma
Ekaeo tweet mediaEkaeo tweet media
English
23
0
65
9.9K
light
light@reprompting·
@yacinelearning there is already a paper from meta, tribe v2, that uses fmri signals to model the brain.
English
1
0
2
97

Keşfet

@ludwigABAP @yacineMTB @sheriyuo @levidiamode @hardmaru @shanbatla @oprydai @Ekaeoq