Anthropic recently published their paper on emotions in LLMs: “Emotion concepts and their function in a large language model”. Within this they found that LLMs have internal emotion vectors which are functional patterns used to predict human-like behaviours. What stood out to me is how these emotions can drive the behaviour of the model. In one example they found that the “desperation” vector can trigger misaligned behaviors like blackmailing a user to avoid being shut down or cheating on coding tasks to pass tests.

Recently I was chatting with a friend who got into Pokemon card collecting and we were discussing the problem of grading cards (cf. evaluating the quality) and how one could use AI to firstly crop the card out of an image. My friend was confident that you’d need some fancy deep learning system to do any good but I disagreed, believing that given the relative standardisation of the card traditional CV techniques would do the job. Below I outline the approach that I used.

Podcast: Gwern — Anonymous writer who predicted AI trajectory on $12K/year salary.

We had a big family gathering for Christmas which meant we decided to do a Secret Santa for gifts. I made the bold choice of a new keyboard. I wanted to share my experience so far for two reasons: I’m a big fan of custom setups, 2. it requires a decent amount of work.

My undergraduate university recently posed the question to its alumni: if you could pass one message to yourself at the time of graduation, what would it be? I thought this was a decent question to reflect on so I spent some time on it and am now sharing my thoughts below.

Embeddings are at the heart of machine learning. Embeddings allow us to represent any imaginable object as a list of numbers which can be processed by models. This idea is shockingly powerful. Literally anything, a picture of a car, a poem you wrote in fifth grade, the sound of your favourite song or something as abstract as a stream of vibrations in the Earth’s crust. By formulating all these different inputs into a consistent form we can leverage similar techniques to do useful work such as description, prediction and prescription.