How much of NVIDIA stock price surge is due to Ring Attention?

If you're interested in learning about best practices for building and evaluating language models for agents, I'll be hosting Shishir Patil to talk about the Gorilla project. Should be informative! continual.ai/events/gorilla…

A hugely important commitment to the openness of Meta's AI ecosystem by Mark: "Open Source AI Is the Path Forward " Llama 3.1 is free, open, and on par with the best proprietary systems. To maximize performance, safety, customizability, and efficiency, AI platforms must be open,

Pretty sure agentic workflow frameworks are going to end up exactly where workflow frameworks ended up: durable execution not graphs, nodes, or events.

In a multi-step agent loop, if each step improves the probability of success by some positive (bounded) amount then task performance will approach 100%.

Conversational UI is here to stay. v0 switching from bespoke versioning + instruction workflow to chat is one example.

AI is going multimodal and realtime. The infra opportunities – from inference to observability – seem bigger than for LLMs.

RL is back

This project is extremely well done. Definitely worth looking at if you have a TypeScript stack.

This is definitely worth checking out if you have been dreaming about git like workflows for data and machine learning pipelines. It's a first principles rethink of what a data platforms should look like.

It's nice to be able to sand down the little frustrations so easily these days

All the best AI products put you in flow state.

Fix your underlying systems + add agents + do an iota of error analysis. Skip step 1 or 3 and you'll join the "agents don't work" crowd.

Linear (Linear) is really pulling ahead in terms of enabling external agents on their platform. Their Agent Interaction Guidelines is the latest example linear.app/developers/age…. Every SaaS app needs to make agent integrations first class or risk becoming irrelevant.