Aidan Cooper

I recently repeated an experiment that I've run a few times over the past of couple years: pitting GPT-3.5 Turbo Instruct against frontier LLMs at chess.

The senior-most engineer I worked with at Past Company had exacting standards when it came to the low-level craft of writing code.

Today, I’m releasing PGN2FEN — a new benchmark that tests the ability of language models to understand and simulate chess games by converting PGN move sequences into FEN board positions.

A few months ago, I wrote about building LLM systems with self-hosted, open source models that beat proprietary services like GPT-4 and Gemini.

Introduction When designing systems that use text generation models, many people first turn to proprietary services such as OpenAI's GPT-4 or Google's Gemini.

Introduction We often want specific outputs when interacting with generative language models.

It wasn't long ago that the constraints imposed by finite storage and compute meant that data modelling was a necessity.

The world is captivated by the unrelenting progress of generative artificial intelligence (AI).

In a previous post, I explained the theory behind Shapley values.

In the 1960s, Thomas Schelling developed a computational model that demonstrated how even a mild preference for neighbours of the same race or ethnicity can lead to high levels of segregation in a community.

Shapley values - and their popular extension, SHAP - are machine learning explainability techniques that are easy to use and interpret.

For tabular data, gradient boosted trees (GBTs) perform better than neural networks (NNs).

pandas is a powerful data analysis library with a rich API that offers multiple ways to perform any given data manipulation task.

Cluster analysis is a popular method for identifying subgroups within a population, but the results are often challenging to interpret and action.

The empirical utility of some fields of machine learning has rapidly outpaced our understanding of the underlying theory: the models are unreasonably effective, but we're not entirely sure why.

With interpretability becoming an increasingly important requirement for machine learning projects, there's a growing need to communicate the complex outputs of model interpretation techniques to non-technical stakeholders.