Testing Branch

Code: github.com/mpcsb/tb-llm-distillation An LLM can label a text-classification dataset without building a task-specific model first.

Code: github.com/mpcsb/reidentification (I had a draft of this experiment sitting around for a while.

Code: github.com/mpcsb/tb-embedding-quality Why this matters We often use cosine distance as if the resulting neighborhoods behaved like a clean metric space.

Code: github.com/mpcsb/tb_model_equivalence Most model replacement flows stop after validation accuracy.

(This post comes from a series of old notebook ideas I’m revisiting — notes written years ago, now turned into posts.)

This note walks through a simple but realistic case where Bayesian logistic regression helps simulate pricing scenarios — a model based way to explore sales decisions.

Why study model calibration under noisy data?

This post is about extending python code with Go.

This is the continuation to this post where we explore the changes in the output of machine learning models when they are trained on samples of varying sizes.

Modeling time series outliers with Gaussian Processes This is essentially a back of the envelope study for the identification of outliers in time series.