What's the best way to monitor LLMs that use the Gemini API?

I used to use langfuse but it doesn't seem work as nicely as it does with openai.

Using Few Shot examples to boost LLM data extraction by over 50%?

If you spent countless hours fine-tuning prompts, testing different parsing libraries, and trying to craft perfect solutions only to get mediocre results, this is for you.

My 3 mantras to stay sane as an entrepreneur.

Always visible in my desk.

I should probably have a nicer version framed or something, but hey, who has time for that? 😂

🚀 Preventing Hallucinations in Vision-Based Data Extraction

Vision models are coming up as the best way to deal with documents with complex layouts. On the flip side, they are more likely to hallucinate results.

How can we address that? With OCR based data validations. 👇

It feels like chess engines are so powerful now that they become a bit useless in chess commentary. Even GMs can't make sense of the eval bar sometimes. It would be better maybe to have a more "human" eval bar that actually helps the audience and commentators.

I love how chess player assign so much meaning, personality, and purpose to chess pieces throughout games. So much passion and emotions on a board game.

Super excited about PydanticAI. Looking forward to taking it out for a spin.

🚀 [NEW POST] Show, Don’t Tell: How Dynamic Examples Boosted Accuracy from 61% to 97%

Ever spent hours fine-tuning prompts or testing document parsing libraries, only to end up with meh results? What if I told you that one simple change could drastically improve your results?

Interesting paper on Entity Matching using LLMs. I think I'll work on a demo of this soon.

Adding and example increased our LLM data extraction accuracy from 61% to 97%.

No complex prompt engineering. No fine-tuning. Just showing the model what good output looks like made all the difference.

Sometimes the simplest solutions are the most powerful.

It's a funny that LLMs never suggest LLMs as a solution to a problem. Like they are not aware of their own capabilities. I guess it's lack of representation of those solutions in the training data, but still, clear sign that we are not at AGI levels yet 🤣.

Working on Mechanistic Interpretability at @anthropic.com must be super exciting.

🚀 [NEW POST] The Minimum Viable Model (MVM) Framework

The most common mistake I saw in more than 10 years of building AI products? overcomplicating things. You can avoid that with these 5 steps and build AI products that actually work.

Here's how 🧵👇

🧠 Keeping up with research doesn’t have to be painful: LLMs to the rescue!

If you’ve ever tried staying up to date with platforms like arXiv, you know how overwhelming it can be. We built a demo notebook to show how LLMs can simplify this process for you.

👇 More details: