AI Research Posts

🌁#75: What is Metacognitive AI

Metacognitive AI — LLMs that monitor their own limits — is the next research frontier. Covers AI welfare, anthropomorphism debates, and top papers of the week.

10 xLSTM Models

A curated list of 10+ xLSTM variants — Bio-xLSTM, AxLSTM, Vision-LSTM, xLSTM-UNet — with use cases across vision, audio, genomics, and time series.

Inside Les Ministraux: Mistral's Small Model Strategy

Trace Mistral AI's roadmap from Mistral 7B to Mixtral 8×7B and Ministral. Architecture, benchmarks, and edge computing use cases.

FOD#74: Sparks of AGI – OpenAI’s plans to get there

we discuss Sébastien Bubeck at OpenAI, what is going on with SLMs, and offer you a collection of interesting articles, relevant news and research papers. Dive in!

10 Open Multimodal Models

🦸🏻#4: From Butler to J.A.R.V.I.S to Butler Again?

Google Project Jarvis is a Gemini-powered browser AI agent. Covers J.A.R.V.I.S. origins, Zuckerberg's Jarvis, and the π0 robot foundation model.

Concepts: RLHF, RLAIF, RLEF, RLCF

4 RL+F approaches that guide a model with targeted feedback

FOD#73: Apple Intelligence is undercooked

We discuss Apple’s unfortunate underdelivery and provide a collection of immensely interesting articles, interviews, news, and research papers. Dive in!

6+ Free Diffusion Model Tutorials and Courses

6 free resources to learn diffusion models: tutorials from DeepLearning.AI, Hugging Face course, Berkeley lecture, and research papers. No cost.

Guest Post: The Elusive Definition of Artificial General Intelligence (AGI)

What is AGI? Compare Google DeepMind vs OpenAI 5-level frameworks, the Turing Test debate, and why general intelligence remains so hard to define.

What is Whiteboard-of-Thought?

Whiteboard-of-Thought (WoT) lets multimodal LLMs reason visually by drawing intermediate steps. How it works, benchmarks vs CoT, and current limitations