Japanese brothers have you tried a patty melt?
It’s easy to make
1. Cook a burger patty
2. Cook onions on a stove until slightly brown (this is called caramelizing and makes the onions sweet)
3. Put it in of two slices of thick bread
4. Toast on your skillet with butter
Enjoy
Something I’ve noticed about Grok 4.3 is that it’s suffering from something I call Branch Context Bleeding which leads to poisoned contexts.
The best way I can describe it is by example. If I try to ask Grok “Can you tell me more about marauders” but my phone lovingly auto corrupts to “Can you tell me more about murders” and hit send. It’ll start processing it. After noticing this I hit edit > change murders to the correct word “marauders” and then re-send.
You’d think this would force grok to reconsider the context from scratch, however Grok feeds its previous considerations into the change meaning Grok will be improperly biased towards talking about murders.
This has lead me to scrapping many an attempted conversation due to the poisoned context.
Playing the steam version of Tomba 2. Anyone else run into a glitch where fighting an evil pig (ghost in my case) with taboo on then losing it in the fight will give you taboo back after the fight ends but make it impossible to lose on taking damage?
Fruit still rids taboo.
Counterpoint. If you can’t support 3 themes then delete light. Dim is the best theme by far.
But until they fix their mistakes check out this CSS by @izutorishima.
xitter.cf/izutorishima/status/20…
Also this extension
soitis.dev/control-panel-for…
If you’re a user script kind of guy then here’s one I used Grok to whip up based off of Torishima‘s CSS
```
// ==UserScript==
// @name X (Twitter) Restore Dim Theme
// @namespace grok.x.ai
// @version 1.0
// @description Restores a dim dark mode appearance on X.com (since the official dim option was removed) based off the CSS from Torishima. Thanks!
// @author Ebsolas/Grok/izutorishima
// @match xitter.cf/*
// @match xitter.cf/*
// @grant none
// @run-at document-start
// ==/UserScript==
(function () {
'use strict';
const css = `
body, .r-kemksi, [style*="background-color: rgb(0, 0, 0);"] {
background-color: rgb(21, 32, 43) !important;
}
.r-5zmot {
background-color: rgb(21, 32, 43, 0.65) !important;
}
.r-1kqtdi0, .r-1roi411, .border-gray-100 {
border-color: rgb(56 68 77) !important;
}
.r-2sztyj {
border-top-color: rgb(56 68 77) !important;
}
.r-1igl3o0 {
border-bottom-color: rgb(56 68 77) !important;
}
[style*="color: rgb(113, 118, 123);"], .text-gray-700 {
color: rgb(139 152 165) !important;
}
.from-background {
--tw-gradient-from: none !important;
}
`;
const style = document.createElement('style');
style.type = 'text/css';
style.appendChild(document.createTextNode(css));
document.head.appendChild(style);
})();
```
Replying to @izutorishima
【修正版】微妙に従来の色が残ってる箇所があったので修正版です
```
body, .r-kemksi, [style*="background-color: rgb(0, 0, 0);"] {
background-color: rgb(21, 32, 43) !important;
}
.r-5zmot {
background-color: rgb(21, 32, 43, 0.65) !important;
}
.r-1kqtdi0, .r-1roi411, .border-gray-100 {
border-color: rgb(56 68 77) !important;
}
.r-2sztyj {
border-top-color: rgb(56 68 77) !important;
}
.r-1igl3o0 {
border-bottom-color: rgb(56 68 77) !important;
}
[style*="color: rgb(113, 118, 123);"], .text-gray-700 {
color: rgb(139 152 165) !important;
}
.from-background {
--tw-gradient-from: none !important;
}
```
Replying to @AntonHand
A Few things.
1 simulation and render distance are separate. Simulation distance determines how many chunks things tick. Render is for visuals.
2 voxy is being used here. It’s a mod that caches visual data about further chunks. It’s not having the server or game actually render
You ever have an intrusive thought that’s the dumbest idea ever.
Mines worse.
What would happen if I made an AI in Perl
Grok:
In regards to deepseek. I even tried to raise this as a concept on X a while ago too. I called it, knew that it’d be a path forward. I ended up furthering my concept and still think that using an LLM to create a dataset of symbols/concepts and training a new AI on that would lead to an even more mentally flexible and efficient AI.
Thank you @HerobrinesArmy for both making ANGRY and letting us know he’s back.
xitter.cf/i/grok/share/Bw02hvpni…











