Kosti's Projects
Experimental code: adaptive importance sampling for bayesian networks.
Robust recipes to align language models with human and AI preferences
Quick web interface for browsing recent arxiv submissions
Search arXiv from the command line. Easily open as URL, PDF or output a generated BibTex entry. Integration with BibDesk to automatically insert references. Naive keyword scoring system to filter results to those likely to be relevant.
Auriga is a minimalist LaTeX beamer presentation theme 📽
Official code of our work, AVATAR: A Parallel Corpus for Java-Python Program Translation.
A collection of papers of neural-symbolic AI (mainly focus on NLP applications)
A collection of resources and papers on Diffusion Models
My bashrc config file.
The simplest implementation of Bayesian optimisation.
A minimalist and modern Beamer theme
Beamer color theme using the Solarized palette
Biostatistics in Practice course notes
Files that I use on my blog, mainly code.
Jupyter notebook companion for "a tolerance for the relative entropy?" blogpost.
Jupyter Notebook to create your own Brownian mountains.
This is the repo for a project. The relevant class is PUBHLTH 590F, UMass Amherst.
Experiments with Dirichlet Laplacian eigenvalues and MNIST (just for fun).
ChatArena (or Chat Arena) is a Multi-Agent Language Game Environments for LLMs. The goal is to develop communication and collaboration capabilities of AIs.
Evaluating an LLM trained on chess PGN strings using techniques from the Othello World Models paper.
Example of calculations in IPython involving exponential of operators and Lie brackets.
Showcasing the computation of the concentration-information bounds for the bias of an observable.
Official content for Harvard CS109
Implementation of debiasing algorithm in "Debiasing Representations by Removing Unwanted Variation Due to Protected Attributes" on ProPublica's COMPAS data set
[arXiv preprint 2024] Official code release accompanying the paper "diff History for Neural Language Agents" (Piterbarg, Pinto, Fergus)
Forcing Diffuse Distributions out of Language Models