Yh
My Telegram is 3 years old. Older than 52% of all accounts.
My Age pays $AGE and $GRAM for old accounts. Check yours: myageapp.xyz/r/5955295642-ia…
17
bare metal retweeted
X premium free for a month, thank me later!
Expires on: Monday, November 30, 2026 at 10:30:00 PM
Redeem At: grok.com/plans/referral/?sku…
Readers added context they thought people might want to know
The post's claim of "X premium free for a month" via referral is inaccurate. Official program offers eligible new users a 14-day trial requiring a credit card at checkout with auto-renewal unless cancelled; referrers must meet US eligibility rules and receive usage resets.
x.ai/legal/referral…
docs.x.ai/x-subscription…
bare metal retweeted
“Islamophobia” If you're not phobic from this, there's something genuinly dark and twisted inside you and you're an extremely evil person lacking of sympathy.
Incredible
Fun Fact - I know my drawing below does not look that crazy, but it was by far the most technically difficult drawing I have completed.
This is one non-intersecting line. I drew it with a pen. I wanted to share it.
#singlelineart #loop
12
bare metal retweeted
I improved NanoGPT Speedrun from 39.9s → 6.7s (-83.2%), more than every record since 2025 combined (-81.4%)!
The web is full of rare, repeated content that the LM can’t memorize in time. I solve this with retrieval-augmented n-gram hints🧵
Interesting
The drug class we are developing, Claromers, kill nearly all known germs (bacteria, viruses, fungi, and biofilms) with one simple devastating mechanism that has avoided microbial resistance for over 200 million years.
Claromers mimic a natural defense peptide in the human body called LL-37. Each Claromer molecule is very small, only dozens of atoms big - about 1/3 the size of the natural peptide. It is like a spear - a destructive end and a holding shaft or tail end. The shaft side is hydrophobic, which means it is water-avoiding and does not like to touch water. The other side - the pathogen destroyer - is highly positively charged (called “cationic” in science literature), like a powerful magnet that pulls toward anything negatively charged (called “anionic” in science).
Stanford bioengineering Assoc Prof Barron and colleagues wrote in a recent study of Claromers: “These amphipathic peptoids self-assembled in different ways, as determined by small-angle X-ray scattering; some adopt helical bundles, while others form core−shell ellipsoidal or worm-like micelles.” That means the groups form either short twisted bundles, like worms, or small oval shapes like a football. 🏈
The self-assembly into these nano structures I. the watery fluids of the body because the water-avoiding shaft-tails point inward toward each other , away from the water. The positive pathogen-killer ends stick outward like sharpened spears. The molecules stick together on their own into tight groups.
These groups attack like a three-dimensional Greek phalanx, a tight formation of soldiers. Like a football made of spears. The spears face out. Because each molecule is so tiny its mass is almost zero. When the positive killer ends feel the pull of the germ’s outer covering they accelerate from standing still to nearly the speed of light with almost no friction, like lightning bolt ⚡️ leaping towards the pathogen. The killer-football formation hits all at once the with huge kinetic energy and blows apart the outer membrane of the pathogen.
The germ cannot survive. It is destroyed in a microsecond.
Barron and colleagues also wrote: “While self-assembly correlated overall with antibacterial efficacy... Certain self-assembled morphologies seem better-suited for antibacterial activity.” The football 🏈 hits harder.
We choose the super effective oval shapes that mass the spears for the strongest hit.
16
Interesting
1/9 Current optimizers are static during training. What if the optimizer could think and react while the network trains?
We let an agent watch the training dynamics and rewrite the optimizer's source code as training goes: a reactive optimizer.
In every language and image setting we tried that has no tuned recipe, it matches or beats the best recipe one can one-shot, and it guards against loss spikes and overfitting by itself. In the near future, optimization decisions may belong to agents, not humans.
📝 xingjianbai.com/reactive-opt…
💻 Try it on your task: github.com/xingjian-bai/opal
🧵
1
78






















