A unified interface for computing surprisal (log probabilities) from language models! Supports neural, symbolic, and black-box API models.

A playful experiment turning LLM surprisal values into musical melodies (and vice versa)

Calculate surprisal for words in text.

Jupyter notebooks on Google Colab to calculate surprisal for experimental stimuli, using the minicons package.

Reverse-Engineering the Reader - code for aligning language models to human psychometric data (EMNLP 2024).

Quantifying surprise in clinical care

On the Proper Treatment of Units in Surprisal Theory - code for computing surprisal over arbitrary unit inventories from language models (ACL 2026)

Quantifying surprise in clinical care

Reproducible real-EEG workbench for exact language-model surprisal, human N400 effects, held-out alignment, and causal auditing.

Using NLP to generate an index for my blog.

Multilingual toolkit for evaluating LLMs using embeddings

Открыл секрет стихотворения "Жди меня" https://dzen.ru/video/watch/69426bca41a09c78095e98d2

The C-SALT-mix corpus and analysis for the Open Mind paper 'Signal Smoothing and Syntactic Choices: A Critical Reflection on the UID Hypothesis' (MIT Press).

Interactive GPT-2 surprisal calculator - AI-assisted development demo

This repository is for the paper Word Surprisal Correlates with Sentential Contradiction in LLMs. In Proceedings of the 19th Conference of the European Chapter of the Association for Computational Linguistics (Volume 1: Long Papers), pages 4549–4564, Rabat, Morocco. Association for Computational Linguistics.

Evaluating syntactic comprehension of large language models on garden-path sentences by comparing computational parsing behaviors against human cognitive patterns.

ALSI/ILSA is a lexical and syntactic analyzer