Most of my contributions are unfortunately private, but I do have some "extra-curricular" projects for which I created public repositories. These repos essentially revolve around either finance or the "Abstraction & Reasoning Corpus" challenge.
- GridCoder 2024: my code submission to ARC Prize 2024 and the companion code to my paper "Towards Efficient Neurally-Guided Program Induction for ARC-AGI". It is a series of experiments on the generalization capabilities of neurally-guided program induction in the context of the ARC-AGI problem domain.
- CountingWithTransformers: these scripts are the companion code to my paper "Counting and Algorithmic Generalization with Transformers".
- ARC_gym: a framework in development to help generate training data for the Abstraction & Reasoning Corpus challenge.
- MLC-ARC_gym: an experiment that applies the MLC method (Lake & Baroni (2023)), a meta-learning Transformer, on the ARC gym.
- Microstructure: a framework in development to help generate simulated financial data, to evaluate algorithms & approaches to ML trading.
- RLNonStationary: companion code for a talk I gave at Quantopian about using simulated data to train a reinforcement learning algorithm when non-stationarity is a problem.