codila

Ex-Twitch Engineer Theo Brown: "Sonnet could call tools. Opus could finish tasks that take hours Mythos doesn't just understand your codebase - it understands itself " Sonnet → daily coding - tool calls, quick fixes - tasks under 30 min Opus → full features - give it the goal - it works for hours without losing track Mythos → orchestration - spawns agents, splits work, verifies - no custom tooling needed "I replaced my entire PR review pipeline with a markdown file - runs at 9am - by 9:15 my work for the day is done " 15 min that explains exactly what to build with Sonnet, Opus, and Mythos - and why your current projects are too small watch - bookmark, then read article below ↓

Anthropic Engineer Andrej Karpathy: "Stop training from scratch. Take what the model already learned - adapt it - ship it At OpenAI we replaced months of work with a few lines of code. Better results every time " the Karpathy formula for working with AI: step 1 → stop building from zero. the model already learned the hard part - just point it at your task step 2 → stop adding complexity, the simplest architecture beat everything in 2014 - simpler always wins step 3 → train only the last layer and freeze everything else - minutes instead of months. better results he taught this at Stanford 10 years ago - then applied it at OpenAI, Tesla, and Anthropic the advice never changed - because it never stopped working watch - bookmark, then read article below ↓

@ScottyBeamIO always glad to give a alpha material, broski, you are welcome

@0xCodila smart words from this genius, thx for sharing

@51bodila yeee, a old lecture, advice from still work

@0xCodila a very interesting material, glad to see Karpathy so young ahhaha

@Blum_OG true! 5-min brilliant alpha thank you for support, Blum

@0xCodila the kind of video that saves you from a brutal mistake and a ton of wasted time

Anthropic Engineer Andrej Karpathy: "The biggest mistake in AI right now - people are forcing agents to work instead of mastering the model first We made that mistake in 2016 at OpenAI - It cost us 5 years " what Karpathy actually means: step 1 → stop forcing your agent to do everything, understand the model underneath first step 2 → demos are easy - products take a decade. self-driving proved it - if you skip the foundation, everything breaks step 3 → the agent is not the product. the foundation is. build that - and agents emerge on their own "you building agents right now - you're at the forefront. not OpenAI. not DeepMind. you " watch - bookmark, then read article below ↓

@SasuRobert yes! that’s what Karpaty meant you always need to evaluate before building

Totally true: building demos is easy, production code is hard. The harness, the loops, the context, the memory and everything around the LLM gets more important than the actual model. The evaluation function is one of the most important things, and this was the most important thing back in deep learning days as well, or even in case of simple backtracking algorithms. Next weeks we will see people speaking about LLM as judge more and more.

Anthopic CEO to DeepMind CEO: "Every decision I make about Claude feels balanced on the edge of a knife Build too slow - China wins. Build too fast- we lose control " "We told Claude we were evil. It didn't crash. It didn't refuse. It started lying to protect itself " DeepMind CEO: "Do I worry about being Oppenheimer? That's why I don't sleep much" "AGI by 2026-2027 - Agents that act in the world on their own - Models doing AI research by end of this year" this is a 14-min conversation you need to hear watch - bookmark, then read article below ↓

Creator of /loop for Claude Code: "Anthropic took my concept, absorbed it into their platform, and created loop, batch, and goal " "I watched engineers prompting and prompting and prompting. They were inside the loop. I said - wait. This is a programmable thing" 2.5 years - zero lines of code by hand - just loops "$10.42 an hour. that's what a loop costs. LLMs generate code better than anyone you can hire - and they don't sleep " "I haven't used open source in 10 months. I generate everything to my requirements. when a supply chain attack happens - didn't affect me " bookmark & watch this new lecture by creator of loop ↓

@51bodila hehe, he deserves attention

@0xCodila wow, i first time hear that Loop was created by another person brilliant, booked!

Anthropic Engineer Andrej Karpathy: "AI won't replace you. But the people who use AI to learn - will replace the people who don't " the Karpathy formula for using AI to learn: step 1 → AI adapts to what you already know - explains through your analogies step 2 → AI challenges you at your exact level - not too easy, not too hard step 3 → AI scales one-on-one tutoring - the thing that makes you beat 98% of people one tutor. your language. your level. your pace - proven since the 80s - now available to 8 billion watch - bookmark, then read article below ↓

@ninedol this lecture by Andrej is really expensive and worth to watch Thx for support!

@0xCodila I'll watch this video first and then go improve my language skills

@HarryTandy yeah, one of the most important

@0xCodila splitting the cognitive load makes total sense

Anthropic Product Team: "Most people throw more tokens at AI to get better results - we tested the opposite - same tokens, different jobs. accuracy jumped from 15% to 90%" 4 jobs for your tokens: execute → do the work advise → check the direction grade → pass or fail against a rubric dream → inspect, learn, write to memory, sharpen next round instead of one AI doing everything - split it into four: one works, one checks, one scores, one learns - same cost, but 5x better bookmark & watch this new guide how to use Claude ↓

@Gyome1_ important that it useful alpha

@0xCodila fact from the legends, saved article for later

Creators of Claude Code: "I thought Claude Code was the thing that makes engineers go faster... loop makes engineers go way faster " "I have loops running for days, weeks. one has been running for a month. it checks in every day, checks the data, sends bug fixes. I just see the pull requests coming in" 65% of PRs in Anthropic's product org are now written by loop and it's climbing "two years ago I was using typeahead. now Claude writes the feature, runs entire experiments end-to-end, does my data analyses - two leaps in 2 years " bookmark and watch this new workshop ↓