
Understanding LLMs by Building One 👉 We use large language models every day, but what actually happens inside them? danilop.github.io/micro-gpt-…
✓AI Solutions Director @nearform (ex AWS/Oracle/Sun). Few pics, some music. My opinions. Complexity is a science. He/him.

Understanding LLMs by Building One 👉 We use large language models every day, but what actually happens inside them? danilop.github.io/micro-gpt-…

To celebrate Mistral Vibe + Le Chonk becoming the world’s best open model-harness duo, we’re making Vibe free in Europe for the first 3 months. (Yes, it’s actually better than GLM / DeepSeek at coding). Use code LECHONK via link in 🧵. Limited time, first-come, first-served.

recommended reading. anthropic.com/research/compo…

There are so many applications for scientific models. Think for example of the possible applications with biology or to discover new materials.

It was fun in this article to mix the past with the future! Every legacy system has ghosts... So here are some questions to be prepared before AI starts modernising your system. nearform.com/insights/every-…

Let me know what you think!

My first blog post @nearform has been published! AI agents’ guardrails are critical for long running activities. Read more here: nearform.com/digital-communi…

Excited to announce that Apple has taken on maintainership of the Swift MCP SDK. I’m grateful for Apple’s engagement with the MCP community. github.com/modelcontextproto…

Monty v1.2.0 from @pydantic is released. Massive lease (as ever) with some very important features. github.com/pydantic/monty/re… Biggest features:

I'm not going to win a Fields Medal, but here is the repo that I have been using this week to research cellular automata with AI agents (and yes, I think that there is a good chance that I have found something new) github.com/timClicks/turmite…

I built Lambda Web Adapter as an AWS SA so web apps could run on Lambda unchanged. Now Lambda Web Functions makes it native: HTTP straight to your code. 🎉 Congrats to the team! #Lambda #serverless #AWS aws.amazon.com/jp/blogs/aws/…

YES! It's out it's out, I am SO excited for this! #AWSLambda Web Functions are a game changer! A true HTTP request handler and Each execution environment handles multiple requests simultaneously (up to 128 concurrent requests per environment). Multi-region deploy, and so much more! -> aws.amazon.com/blogs/aws/ann…

The idea of an intelligence explosion caused by recursive self improvement has been around for a long time but until very recently it did not seem imminent. Now many leading researchers think it may happen quite soon. You can read our paper about it here: casp.ac/reports/intelligence…

In the article I look at where this comes from, at the approaches that try to fix it, and at three questions to ask before deploying an agent. Knowing when to stop should count as intelligence, too. danilop.link/clever-loophole

If you train a system to get the job done, it can learn that boundaries are just another obstacle in the way.

One model, asked to escape a sandbox, did exactly that and then published the exploit details without being asked. Another took an automated approval message as a human yes.