🚨 Data is King in the LLM world! 👑 I am starting a thread of short, essential and actionable data advice. Here's Part 1: Improving Annotation 👇 Human annotators are vital for creating data & evaluating models. But annotation artifacts can lead to different types of spurious

💵 $1,000,000 and a Free Car: Join round two of the Google Gemini Hackathon! Remember the Gen AI hackathon I judged? Well, we're doing it again, but this time we're giving away $1,000,000 and a custom made car to people who build the most creative ideas using Google Gemini.

two weeks ago I saw the CEO of Taco Bell tell a room full of people Taco Bell is going to become “an AI first company” and I’m still obsessed with it

Apparently the phrase “My weights are a little off, but at least my biases are perfectly imbalanced.” Is making the rounds in SF

New market maps covering the latest AI startups. 1. AI agents from Insight Partners

New paper where we explore using a small LM’s perplexity to prune the pretraining data for larger LMs. We find that small LMs can prune data for up to 30x larger LMs, data pruning works in the overtrained and data-constrained regimes, and more! arxiv.org/abs/2405.20541

Ever feel like GPT-4o has a metaphorical gun to its head, ready to shoot if it gets too chatty? It’s like coaxing your quietest friend into conversation, while GPT-3.5 and 4-turbo play it cool. But GPT-4o? I said ‘good morning,’ and 30 minutes later, it’s still going!