One of the best when it comes to writing and explaining about llm is Sebastian Raschka magazine.sebastianraschka.com/p/the-big-llm-… Must read for anyone trying to understand the differences

Took Shoffr twice in Delhi and I don’t think I can choose anything else now. One of the best experiences, clean cars and great drivers. Always on time and very polite 🙏 I really wish they start in Mumbai soon

#ChatGPT5 is here. I guessed it would focus on tools and bigger context, and it did. 1. Smarter reasoning 2. 1M-token context 3. Text, image, audio, video 4. Personalities: Cynic, Robot, Listener, Nerd 5. Google/tool integration 6. Safer, more accurate replies #openai

Nano Banana + N8N = UGC Content Factory This system generates scroll stopping video ads on autopilot using Google's newest image generation model. No models needed No video crews No $10K agencies Just endless UGC-style content at the speed of thought. Here's how it works: →

Today I downloaded sublime text back to avoid the ai code completions. Learning something new and don’t want ai to complete all the code 😅

🚫 No peeking! LLMs don’t get to cheat by looking ahead when predicting the next word. In my new post, I explain causal attention — the masking trick that keeps models honest, one token at a time. open.substack.com/pub/viveknayya…

RoPE is such a well thought way of positional information. I have understood like 90% of it and it seems so simple now 🙌

Finally wrapped up coding gpt 2 from scratch. Thanks to Sebastian Raschka and his book that I was able to do it. Coded everything myself and feel more confident in now writing new architectures from scratch too 🙌

There’s always that one person who explains things so clearly that you suddenly believe, ‘I can understand this too.’ This year, for me, that person is Sebastian Raschka Grateful. 🙏

No Claude, I don't want the 5th README in this project, thanks.