🔥The 9th Round of Easy Loan, Earn $40 Reward is in progress❗️ ⏰ Promotion Period: January 15th - Feburary 15th, 2025 👉 Register now and check more details at gate.io/campaigns/358

Wrote an intro to evals for long-context Q&A systems: • How it differs from basic Q&A • What dimensions & metrics to eval on • How to build llm-evaluators • How to build eval datasets • Benchmarks: narratives, technical docs, multi-docs eugeneyan.com/writing/qa-eva…

Overall site quality is super important. Google has explained this many times in the past and I have presented and written about this a thousand times (including one mammoth post where I used the "Gabeback Machine" to document every time Google has explained that site-level

Major career cheat code: Be the person who can figure it out. You’ll be handed tasks you have no idea how to complete. There's nothing more valuable than someone who can just figure it out. Do some work. Ask key questions. Get it done. If you do that, people will fight over you.

OpenAI COO Brad Lightcap sees o3 as a challenger to push back on the way he sees the world. To stress-test assumptions, surface blind spots, and question his priors. “o3… there's something deeply great.”

notes on context engineering with Walden of Cognition

Last year, I wrote a 100+ page thesis on China's AI industrial policy. It covers Beijing's spending across: (1) basic R&D, (2) chips, (3) B2G partnerships, and (4) traditional subsidies. Over the next few days, I'll break down the paper's findings, starting today with R&D.🧵

a complete guide to fine-tuning LLMs in 15 minutes. this covers how to use Unsloth to fine-tune models in notebooks, how to create your custom chat templates, datasets, and more. this guy deserves much more attention. here is also the full 1:20 hour video going in detail:

My take on AI talent. Basically, if you raise a fund, you can take 2% and 20% on the total funds. I think generally, machine learning research and software development have not been compensated well relative to the impact and leverage. That said, if you're going to be given

I open sourced Sniffly, a tool that analyzes Claude Code logs to help me understand my usage patterns and errors. Key learnings. 1. The biggest type of errors Claude Code made is Content Not Found (20 - 30%). It tries to find files or functions that don't exist. So I

Josh from Profound about the surge in Reddit citations in ChatGPT starting in May: "That just quietly happened because the folks over at OpenAI turned the dial." Sounds like another company we know. :) "At the same time, there's insane stuff happening in regards to the way that

"AI will eat SaaS" IMO, no, it will eat Services which is 25X larger. the prize isn’t the familiar $200B SaaS pool; it’s the $4.6T enterprises pour each year into salaries and outsourced services – the very labor-intelligent agents are now poised to absorb.

OpenAI says it will use Google's, $GOOGL, cloud for ChatGPT

High fidelity image editing is now in the OpenAI API and ChatGPT! Here's a quick demo I made that focuses on editing faces. It adds mustaches and facial expressions—but preserves all the other details of the face. Notice how my hair, glasses, nose, etc are perfectly maintained.

xmas in july ?? 🌶️

1/N Yesterday in Tokyo we OpenAI ran a 10‑hour live Humans vs AI exhibition at the AtCoder World Tour Finals Heuristic. We pointed an OpenAI reasoning model at the same brutal problem the finalists tackled—no human help, same rules, same clock. Buckle up. 👇

I know AI cat barber has mixed results, but I would still pay extra to see him

OpenAI's agent just competed in a 10hr AtCoder contest — no human help. Started simple, then self-learned wall placements, resource reallocations, and optimized paths. Recursive self-improvement loop? Looks like it. This might be the first real demo of long-term autonomous

OpenAI’s agent joined the AtCoder Finals — 10 hours, fully autonomous. It wrote code, built strategies, and improved itself — all without human help. Among the human contestants was Codeforces #76 ‘yosupo’… The agent quietly outperformed him. The day a top-tier human fell to AI

Cryptic!

Hey Elon Musk xAI Developers X Design where's the read thread aloud button? Can we have one please?