Embeddings

What Is an Embedding?
Meaning Becomes Geometry
Tokens, Embeddings and Probability
Learning an Embedding Space
Similarity Is a Decision
Zero-Shot Decisions
Dimensions Do Not Mean What You Think
The same cosine means two different things in two spaces
Neighborhoods and Hubs
Dense Retrieval without Re-teaching Embeddings
How Many Dimensions Does Meaning Need?
Feature Space: What Does a Linear Model Actually See?
The nearest neighbour can be wrong, and a readout can be right
Hard Negatives
Retrieval Is a Policy
Calibration
A threshold calibrated on one domain does not transfer to another
Is Similarity One-Dimensional?
Change the Model, Change the Universe
Can One Embedding Space Be Translated Into Another?
Retrieval Is Not Geometry
What Should a Translation Preserve?
Can a Smaller Representation Preserve a Larger One?
From Deltas to Operators
Turn a delta into an operator, and measure what it costs
Building an Embedding Runtime
Walk the whole embedding runtime in one command
Preference Rankers — Learning Which Answer Is Better
MR.Q — Building a Neural Quality Model From Two Embeddings
RELATE: Searching Embeddings by Relation, Not Just Similarity