This is exciting; I expect we are going to see a lot more things like this and it will be one of the most important impacts of AI. Congrats to the Future House team. https://t.co/Cxeh8UlWdk

most people haven’t updated on this add greg’s “any task that frontier ai can sort of do today, it’ll likely be able to do reliably one year from now” and your ability to trust the work is about to go up a lot as ilya said, reliability is the biggest bottleneck to usefulness

We’ve developed a new way to train small AI models with internal mechanisms that are easier for humans to understand. Language models like the ones behind ChatGPT have complex, sometimes surprising structures, and we don’t yet fully understand how they work. This approach

6 months ago @Sentry acquired our company & since then I've been experimenting a lot with AI I wrote up some of my thoughts on vibe coding & dealing with "AiDHD" https://t.co/PvIVKp5esD

A common mistake that AI companies make nowadays is to not give their engineers enough time and mental calm to do their best work. Constant deadlines, pressure and distractions from daily AI news are poison for writing good code and systems that scale well. That’s why most AI

@josh1yan I joined @OpenAI as a resident. First, get the fundamentals down. If there's one subject you need to know inside and out, it's linear algebra. Read and understand a classic textbook like Bishop's Pattern Recognition and Machine Learning. Then, take on an ambitious project. I

The first fantastic paper on scaling RL with LLMs just dropped. I strongly recommend taking a look and will be sharing more thoughts on the blog soon. The Art of Scaling Reinforcement Learning Compute for LLMs Khatri & Madaan et al.

Here's my talk on the year in open models from The Curve. This felt like one of my better talks -- uncovering a ton of information that way more people should be familiar with. I feel like I'm just scratching the surface and I'm seen as the "leading expert" here. We need more

This memory lane tour reminded me of a few anecdotes. Here is a quirkier, somewhat nostalgic, alternative path through the history of AI. We held a CIFAR workshop at @UofT to learn about how to use GPUs for Deep Learning. We learned a lot from @npinto. Soon after we had the

Karpathy’s thinking is sublime it goes without saying Yes, ghosts in the machine and from the machine. They are missing billions of years of evolution. Will robots help ground them? Maybe I am very bullish on what LLMs are and will become by 2030 even if they “need several

Tinker is cool. If you're a researcher/developer, tinker dramatically simplifies LLM post-training. You retain 90% of algorithmic creative control (usually related to data, loss function, the algorithm) while tinker handles the hard parts that you usually want to touch much less

Finally had a chance to listen through this pod with Sutton, which was interesting and amusing. As background, Sutton's "The Bitter Lesson" has become a bit of biblical text in frontier LLM circles. Researchers routinely talk about and ask whether this or that approach or idea

LoRA makes fine-tuning more accessible, but it's unclear how it compares to full fine-tuning. We find that the performance often matches closely---more often than you might expect. In our latest Connectionism post, we share our experimental results and recommendations for LoRA.

required reading for anyone who works with computers

Another impressive paper by Google DeepMind. It takes a closer look at the limits of embedding-based retrieval. If you work with vector embeddings, bookmark this one. Let's break down the technical details:

I was using Claude Code wrong... Here’s what I learnt and how I maximise Claude Code performance + Best tips that ACTUALLY useful 👇 Thread below

few underrated yt channels (or rather super niche rn) i have been watching on/off to understand gpu programming basics cuda mode - various lectures including inference optimisation 0mean1sigma - nice animation tunadorable - triton kernel focused simon oz - nice explanations

Here you go — 1. Transformers - Visual intro to Transformers (3b1b) [YouTube] -nanoGPT & tokenization (Karpathy) [YouTube] -Decoding strategies in LLMs (Maxime Labonne) [GitHub] 2. Pre training -Distributed training techniques (Duan et al.) [Paper] -nanotron: Minimal

This guy literally explains MCP in under 20 minutes https://t.co/WRUNiqOVY1

Noticing myself adopting a certain rhythm in AI-assisted coding (i.e. code I actually and professionally care about, contrast to vibe code). 1. Stuff everything relevant into context (this can take a while in big projects. If the project is small enough just stuff everything