0x_arjunghosh_ai

Me <==> loyla.ai

𝐕𝐢𝐬𝐮𝐚𝐥 𝐛𝐥𝐨𝐠 on Vision Transformers is live. vizuaranewsletter.com/p/vision-trans… Learn how ViT works from the ground up, and fine-tune one on a real classification dataset. CNNs process images through small sliding filters. Each filter only sees a tiny local region, and the model has to stack many layers before distant parts of an image can even talk to each other. Vision Transformers threw that whole approach out. ViT chops an image into patches, treats each patch like a token, and runs self-attention across the full sequence. Every patch can attend to every other patch from the very first layer. No stacking required. That global view from layer one is what made ViT surpass CNNs on large-scale benchmarks. 𝐖𝐡𝐚𝐭 𝐭𝐡𝐞 𝐛𝐥𝐨𝐠 𝐜𝐨𝐯𝐞𝐫𝐬: - Introduction to Vision Transformers and comparison with CNNs - Adapting transformers to images: patch embeddings and flattening - Positional encodings in Vision Transformers - Encoder-only structure for classification - Benefits and drawbacks of ViT - Real-world applications of Vision Transformers - Hands-on: fine-tuning ViT for image classification The Image below shows Self-attention connects every pixel to every other pixel at once. Convolution only sees a small local window. That's why ViT captures things CNNs miss, like the optical illusion painting where distant patches form a hidden face. The architecture is simple. Split image into patches, flatten them into embeddings (like words in a sentence), run them through a Transformer encoder, and the class token collects info from all patches for the final prediction. Patch in, class out. Inside attention: each patch (query) compares itself to all other patches (keys), softmax gives attention weights, and the weighted sum of values produces a new representation aware of the full image, visualizes what the CLS token actually attends to through attention heatmaps. The second half of the blog is hands-on code. I fine-tuned ViT-Base from google (86M params) on the Oxford-IIIT Pet dataset, 37 breeds, ~7,400 images. 𝐁𝐥𝐨𝐠 𝐋𝐢𝐧𝐤 vizuaranewsletter.com/p/vision-trans… 𝐒𝐨𝐦𝐞 𝐑𝐞𝐬𝐨𝐮𝐫𝐜𝐞𝐬 Dr @sreedathpanat Videos on ViT ViT paper dissection youtube.com/watch?v=U_sdod… Build ViT from Scratch youtube.com/watch?v=ZRo74x… Original Paper arxiv.org/abs/2010.11929 Next up: demystifying Low-Rank Adaptation (LoRA) in PEFT! Follow me @Mayank_022 along for more deep learning insights, cool fine-tuning projects, and updates from the upcoming blog posts.

@harjtaggar @guilhemherail After @deepseek_ai, I think @openclaw release was the next defining moment in the #AI timeline. What do you think ?

Does anyone know what’s going on with the lobster on Wall Street lol?

@harjtaggar @guilhemherail Lolz....

At @ycombinator Alumni Demo Day! W26 is an unfair concentration of talent 🤯

Hey @instagram @Meta WTF? When I see the reel in normal view mode the audio is in english, but when I try to repost, it changes to a southern regional language (no english!). Is the AI Audio translation failing on repost share? instagram.com/reel/DVYp0XMj-…

Wow. @GarryTan (@ycombinator's CEO) just dropped the ultimate cheat code for software engineers. 🔥 He just open-sourced gstack, his personal toolkit that transforms Claude Code from a basic chatbot into an entire virtual engineering department. Instead of asking Claude to "build a feature" and hoping for the best, gstack lets you summon specific "brains" on demand: → The Visionary: /plan-ceo-review acts like Brian Chesky. It stops you from building boring features and pushes for magic. → The Architect: /plan-eng-review draws sequence diagrams and state machines before coding. → The Paranoid Reviewer: /review looks for N+1 queries and stale reads. → The QA Lead: /qa literally logs into your staging environment, clicks around, takes screenshots, and gives your app a health score in 60 seconds. The QA tool alone is built on Bun and Playwright, running 20x faster than Claude’s native Chrome MCP with zero context bloat. He used this exact setup to ship 100 PRs a week for the last 50 days. Get the repo in 🧵 ↓

"I realized tech is this thing that can bring people out of whatever situation they're in and often into prosperity. And that's what I want for everyone." @ycombinator’s @garrytan tells @emilychangtv how tech changed his family's life. Watch here: trib.al/sxg1VGR

@kevinrose Can you share the link from @garrytan's timeline?

and the ref=gstack link to track incoming from people actually coding... so well executed.

possibly the sharpest VC marketing move I've seen... @garrytan ships 15 claude code skills, the repo hits 37k stars and 4.6k forks, then -- only after delivering real value - drops the pitch, bravo 👏:

I am loving @kevinrose that @digg is coming back! Yes I am that old and from the previous Dot Com bubble era to remember and was a power user of same. Maybe you still have the DB details and just switch us back on 😎

Hey @AnthropicAI @claudeai , How do you support #startups in #India ? Like #AWS @AWSCloudIndia & @GoogleIndia is doing via Token & subscription support? I am a solopreneur (loyla.ai) and how can I avail ?

We are already building#AI software with @claudeai #sdk inside it! @AnthropicAI

@t_blom @noahzweben We are already doing that 😎

Pretty soon, I think we’ll see software shipping with Claude Code SDK embedded inside. Users will use it to configure and modify the software to meet their exact needs. The best changes will get passed back to the software developer and reincorporated in the master release.

@elonmusk But was that with consent?

“Anthropic”

Weird realization: The best AI coding is in the morning when you are fresh from a night full of dreaming about latent space. Sleep early. Wake up early. The best ideas are in the morning. It's not just about raw token maxxing. It is about teaching the machines the right abstraction that comes out of your own personal experience and the synthesis that comes from a good night's sleep.

WTF @facebook, why can't u still fix the simple asynchronous JS loader "see more" for ur Birthday wishes page on FB wall ur app?After 4-4 batches it started crashing & Why not even an infinite scroll? In this day & age of #vibe coding, should I send u guys the code @MetaforDevs ?

@CodeByPoonam It was always so and so was any android phone app! Text scam mode was always there 🤦😎

Google Drive just made every document scanner app on your phone irrelevant.

Ofocuse why would LLM model not?Let me ask u as Human Intelligence to talk,read & solve a math problems written 4 you in Nagamese language,u will also brainfreeze & zero shot #epicfail 😎 I mean do u really not listen to @ylecun & his rant that LLMs r just hyper text predictors?

Ofocuse why would LLM model not?Let me ask u as Human Intelligence to talk,read & solve a math problems written 4 you in Nagamese language,u will also brainfreeze & zero shot #epicfail 😎 I mean do u really not listen to @ylecun & his rant that LLMs r just hyper text predictors?

@paraschopra Ofocuse why would LLM model not?Let me ask u as Human Intelligence to talk,read & solve a math problems written 4 you in Nagamese language,u will also brainfreeze & zero shot #epicfail 😎 I mean do u really not listen to @ylecun & his rant that LLMs r just hyper text predictors?

We found a task where LLMs struggle massively! Give them a coding problem in Python and they'd work great. Give the same problem in brainfuck and zero-shot their performance is ~0% +[--------->+<]>+.++[--->++<]>+.