blank

Between 2017 and 2025, transformer architectures for LLMs underwent rapid exploration followed by striking convergence.

A Quick Background on Tokenization Large language models don’t see text - they see sequences of integers.

Related to my earlier post on designing AI for human augmentation, localized to the field of software engineering.

The narrative around AI has long been dominated by automation — the idea that AI will progressively take over human tasks, making certain jobs obsolete while increasing efficiency in others.

As Large Language Models (LLMs) gain mass adoption and excitement, there is no shortage of benchmarks within the LLM community; benchmarks like HellaSwag tests for commonsense inference via sentence completion, while TruthfulQA seeks to measure a model’s…

Note (2024-05-01): This post is due for an update!

The Case for Model Monitoring Model monitoring is crucial because the effectiveness of ML models in production degrades over time.

The problem we’re discussing in this post appears in Bayesian Data Analysis, 3rd edition (Gelman et al., 2013).

In statistics, copulas are functions that allow us to define a multivariate distribution by specifying their univariate marginals and interdependencies separately.