A practical guide to building, testing, and documenting SQL transformations
Understanding permutation symmetry in deep learning, and what it means for weight averaging and model merging
Latest
How we split a tightly coupled Python pipeline into independently deployable services
Statistical thinking beyond formulas
My first impressions of OpenAI's new frontier model
How model validation standards are changing for LLM-based systems: what breaks, what carries over, and how to test output quality
Learn how to simulate reality with Python
Towards Data Science launches a video showcase for real-world AI work
Context Windows Don’t Know What’s Still True — I Built a Validity Layer That Does
Large Language ModelsA context window can be technically complete and still describe a world that no longer exists. I built a deterministic benchmark to measure the cost of acting on stale context.
I saved my day with a mere $52 bill. But you don't have to repeat my mistake.
How to catch a payload that looks correct but isn't, using a watchdog pattern with working Python.
Editor's Picks
Understanding permutation symmetry in deep learning, and what it means for weight averaging and model merging
How we split a tightly coupled Python pipeline into independently deployable services
How model validation standards are changing for LLM-based systems: what breaks, what carries over, and how to test output quality
I saved my day with a mere $52 bill. But you don't have to repeat my mistake.
From scalar observations to self-attention, and how positional information restores sequence order
Why the default 50/50 split is the wrong move when your treatment is more expensive than your control, and how cost-based sampling weights fix it
The Variable Newsletter
Sorting through the good, bad, and ambiguous aspects of vibe coding
Don't miss our most-read and -shared stories of the past month
Deep Dives
Statistical thinking beyond formulas
Context Windows Don’t Know What’s Still True — I Built a Validity Layer That Does
Large Language ModelsA context window can be technically complete and still describe a world that no longer exists. I built a deterministic benchmark to measure the cost of acting on stale context.
Linear Discriminant Analysis (LDA) in Real-Life: Dimensionality Reduction in a Real-Estate Dataset
Machine LearningUsing LDA for dimensionality reduction in classification problems
Improving reinforcement learning for complex physics
Learn how to run a lot of parallel coding agents without expensive, powerful hardware at home
Enterprise Document Intelligence [Vol.1 #B4] - A diagnostic and five composable operations, not a decision tree
