Install
On Medium, anyone can share insightful perspectives, useful knowledge, and life wisdom with the world.
- 15,833articles · 30d
- 56+ min agolatest article
- Aug 14, 2026earliest in window
- 68%with images
- 78avg words
- software 4,967
- ai 4,620
- community 4,391
- technology 4,173
- coding 3,912
- development 3,879
- engineering 3,839
- lifestyle 3,839
- opinion 3,783
- inclusive 3,678
- artificial intelligence 3,426
- business 2,955
- health 2,422
- productivity 1,961
- politics 1,906
- education 1,852
- psychology 1,826
- culture 1,780
- webdev 1,771
- relationships 1,611
- science and technology 3,661
- arts culture entertainment and media 3,642
- PS 3,028
- economy business and finance 1,953
- FIB 1,950
- CE 1,914
- BI 1,864
- SOL 1,672
Please confirm you are human
This browser or connection looks automated. Press and continuously hold the control for 3 seconds to enable Google-hosted web results and, when separately allowed, AI-assisted answers.
A successful check enables 100 search requests. Interactive access does not authorize scraping, systematic collection, or reuse of search output.
News
I Gave a Toy Neural Network More Concepts Than Neurons. It Did Something Beautiful.
15+ hour, 30+ min ago (32+ words) A hands-on tour of superposition, sparse autoencoders, and the young science of reading AI’s mind- with code you …...
N-Grams in NLP: How Machines Learn Language One Word at a Time
3+ week, 2+ hour ago (565+ words) Hey folks, welcome back to another blog! If you have been following my NLP series, you already know that teaching a machine to …...
Diffusion Beat GANs in 2021. GAN Loss Is in Stable Diffusion
54+ min ago (717+ words) Diffusion models passed GANs on FID in 2021, yet GAN loss still runs inside Stable Diffusion's autoencoder and three distillation methods....
Be Careful What You Train (2.3)
4+ week, 1+ day ago (731+ words) A projection is only as honest as the priors it is built from — and the better trained it is, the more convincingly it can lie. The last two pieces built a hopeful machine. The present is a projection; error is…...
The Attention War Chapter 5
2+ week, 4+ day ago (31+ words) The Federation Torn Apart — Yugoslavia’s Economic Collapse and Manufactured Ethnic War We now move from the slow ideological erosion of the late Soviet Union to a …...
I Cut My Reasoning Tokens 92% by Making My Model “Think” in 5 Words
3+ week, 17+ hour ago (39+ words) The reasoning was never the reasoning. It was 1,800 tokens of a model clearing its throat. Here is a real fragment out of my …...
The Attention War Chapter 4
2+ week, 6+ day ago (32+ words) The Empire That Ate Itself — Late Soviet Stagnation and the Hollowing of Utopian Promise We turn now from post-bubble democratic Japan to a superpower built on a …...
I Upgraded My Embedding Model. Top-1 Recall Dropped From 89% to 61% Overnight.
4+ week, 22+ hour ago (289+ words) The Upgrade That Looked Perfect on Paper On a Tuesday night we swapped the embedding model behind our semantic search …...
The Attention War Chapter 12
1+ day, 6+ hour ago (32+ words) The Post‑Cold‑War Vacuum — Searching for Enemies in a Unipolar World The collapse of the Soviet Union in 1991 removed the central adversarial figure that had …...