I have been using opus 4.6 (and sometimes codex) and u know I cant transition from that to gemini 3 flash 👀 it was just rate limit ... till it resets i also rest and learn other new staff 😁

I am currently inside a project pitch at the company I work at, and now the project being pitched is maths ai, it is an ai project for ATP( automated theorem proving) and it is quite interesting u know 😌😌

Before that music ai was pitched and it is for music generating ai models - what I learned there was ( jargon alert 😁) that diffusion based models can be seen as special schrodinger bridge problem where the end distribution is a normal gaussian ... I didn't know that till now ... (I will in the future post what exactly I jargoned 😁 now 😁)

If u don't know the jargons ... Don't worry the ones I mentioned here will also be included in the ai playground being built and u will get to know them when the time comes but if there is anyone who can't rest and is curios(which I am for example) I can write a simple post today evening or tomorrow ... 😁

Moltbook got acquired by Meta (along with the team - which aren't really engineers) It's good to start vibe-coding even if you're not in tech. You'd be surprised by how many creative ideas you can come up with :)

Afternoon Guys .... 🙌

https://youtu.be/R0uMcXsfo2o

this was an awesome video I watched back then ... but haven't even noticed the connection - poor me 😁

But really ai in the aggregate level is not seeming good ...

https://blog.mlreview.com/making-sense-of-the-bias-variance-trade-off-in-deep-reinforcement-learning-79cf1e83d565

This is a very nice blog about the bias-variance tradeoff in RL ... I think most of u would be familiar with bias-variance tradeoff in supervised ML ... but yeah in RL there is more to it ... I invite anyone interested to read it ...

Oh this is so nice, this is what community is for ... 👨‍❤️‍💋‍👨

https://youtu.be/byQmJ9x0RWA

These AI safety videos really are freaking me out ... 😱

Hey guys, I hope you didn't miss me too much! Today won't have too many exciting new things, but I think it can be considered a good take.
Today I was watching a Crash Course video about how to navigate the internet, and it mentioned two things: vertical reading and lateral reading. I want to explain these concepts and also ramble about what I was thinking at the time.
In short, vertical reading is when you read a piece of content directly from top to bottom. Lateral reading is when you first read about the source from other sources — to understand its intentions and agenda — and then read the original content with that context in mind.
Although this is a good habit to have, I don't think you or I, who are being bombarded with new information left and right, can realistically do this in-depth research for everything we hear. Just today I watched things about the central limit theorem, a method for building curves from moving straight lines, diffusion algorithms, CNNs, and graph neural networks — a lot of things just from morning to evening.
So what I think we should do, or at least the most efficient approach to lateral reading, is to only apply it when you encounter a contradictory claim. Until then, hold your current view as authoritative. But when you do see a contradiction, instead of trying to make the new claim fit your existing belief, first start by uncovering the hidden assumptions behind your own belief through lateral reading — before evaluating the new one. moreover i wander is an algorithm of this kind will bet the efficiency of A* and greedy algorithm, i asked chatgpt and it said it looks similar to lazy algorithms such as dynamic A** so if you guy know anything about this algorithm, or this topic feel free to comment🙂‍↔️🙂‍↔️🙂‍↔️🙂‍↔️

Morning fam 🙌🙌🙌

"የምትወዱት ሰዎች ፈጽሞ አልቅሱለት፡፡

ወየው ወየው ወየው አማኑኤል አምላካችን፡፡

ወየው ወየው ወየው መድኃኒታችን ኢየሱስ፡፡


ወየው ወየው ወየው ንጉሣችን ክርስቶስ፡፡

ወየው ወየው ወየው ጻድቃን ከዕንጨት አወረዱት፡፡

ሥጋውንም ለመገነዝ ከርቤ የሚባል የጣፈጠ ሽቱንና ንጹሕ በፍታን አመጡ፡፡

ቅዱስ እግዚአብሔር ቅዱስ ኃያል ቅዱስ ሕያው የማይሞት የማይለወጥ፡፡

ከድንግል ማርያም የተወለደ አቤቱ ይቅር በለን፡፡

ቅዱስ እግዚአብሔር ቅዱስ ኃያል ቅዱስ የማይሞት የማይለወጥ በዮርዳኖስ የተጠመቀ በመስቀልም የተሰቀለ አቤቱ ይቅር በለን፡፡"

ቅዱስ ዮሐንስ አፈወርቅ

https://github.com/danielgatis/rembg?tab=readme-ov-file

this is an awesome cli tool - u don't have to search for background remover ai free 😁 just use this using

rembg i path.png // or open the http server using rembg s --host 0.0.0.0 --port 7000 --log_level info

So today I was reading this paper: https://arxiv.org/html/2511.22570v1 -> It is the deepseekmath-v2 and u know it is insane, it scored Gold level on IMO 2025, while my brain here still glitches understanding the architecture 😁

It was a good read -> obviously used claude for studying deeply and I made the compressed version of the whole paper in the telegraph below ... anyone interested can read 🙂‍↕️🙂‍↕️🙂‍↕️

https://telegra.ph/DEEPSEEKMath-V2-03-13