LLMs Can’t Do Probability
I’ve seen a couple of recent posts where the writer mentioned asking LLMs to do something with a certain probability or a certain percentage of the time. There is a particular example that stuck in my...
View ArticleLLMs: To Fine-Tune or Not to Fine-Tune?
Knowing when to fine-tune LLMs and when to use an off-the-shelf model is a tricky question. New research can help shed a light on when each approach makes more sense and eke more performance out of...
View ArticleDitch that ChatGPT Subscription: Moving to Pay-as-you-Go AI usage with Open...
Introduction In the world of AI assistants, subscription services like ChatGPT Plus, Claude Pro and Google One AI have become increasingly popular amongst knowledge workers. However these subscription...
View ArticleA Personal Experiment with Coffee, Walking and Anxiety
I noticed that in the last few weeks I’ve been pretty anxious. More so than my normal background level of anxiety. I realised that I’ve been drinking a lot more heavily caffeinated iced coffees and...
View ArticleGenAI and the Trough of Disillusionment
There have been a bunch of recent stories about the GenAI hype not panning out so well. Likes The Game Theory of AI CapEx by David Cahn. A story from Sequoia Capital, who have a lot of money riding on...
View ArticleWatched: Long Legs
Just returned from seeing Long Legs in the cinema. I went into this movie pretty much blind, not knowing much except that Nicolas Cage was involved. It was a really interesting thriller and Cage was...
View ArticleWatched: Trap
We went to see Trap, M. Night Shyamalan‘s latest starring Josh Hartnett as a somewhat creepy antagonist trying to escape a concert without being arrested (because he’s actually a serial killer and the...
View ArticleVisiting Bletchley Park
Yesterday Mrs R and I visited Bletchley Park as part of our third wedding anniversary celebration. As a computer scientist, BP has been on my bucket list since forever. It’s the cradle of many...
View ArticleData Export in Bulk inside your CI Pipeline With Sling
In this post I will discuss my migration journey from Airbyte to Sling. Sling is a new lightweight data sync product that is small enough that it can run in a CI pipeline. What’s the Problem? I’m...
View ArticleRunning Phi MoE 3.5 on Macbook Pro
The relatively recently released Phi 3.5 model series includes a mixture-of-experts model featuring 16 x 3.3 Billion parameter expert models. It activates these experts two at a time resulting in...
View Article