embedding
A vector representation of a concept, word, or entity in a high-dimensional space where distance and geometric relationships between vectors encode semantic or structural relationships between the entities
As defined in Harvard Presents NEW Knowledge-Graph AGENT (MedAI)
Other definitions
Placing a lower-dimensional surface inside a higher-dimensional space; a sphere can be embedded in 3D Euclidean space but the hyperbolic plane cannot.
211 | Solo: Secrets of Einstein's Equation
Where it comes up · 14
- Yann LeCun's $1B Bet Against LLMs [Part 1]11 claims
- Knowledge Graph or Vector Database… Which is Better?4 claims
- Ep.4 - Chunking Strategies Explained: How to Structure Data for LLMs #RAG #llm #ai #aiapplications4 claims
- Harvard Presents NEW Knowledge-Graph AGENT (MedAI)2 claims
- Ben Thompson on Big Tech, China, and the AI Boom Running Out of Money - [Invest Like the Best, EP.487]2 claims
- The 7 Levels of Claude Code & RAG2 claims
- Is Gemini File Search Actually a Game-Changer?2 claims
- This combo makes smart note systems magical.2 claims
and 6 more
Most important claims · 10 of 35Search all claims for “embedding” →
- Whereas a sphere can be constructed inside ordinary three-dimensional Euclidean space (which is why spherical geometry was not recognized as a distinct geometry first), the hyperbolic plane—a two-dimensional surface obeying the axioms of hyperbolic geometry—cannot be embedded in three-dimensional Euclidean space; the Pringle or saddle are only approximations, so the true hyperbolic plane exists only in the minds of mathematicians.211 | Solo: Secrets of Einstein's Equation
- DRIFT (Dynamic Reasoning in Inference with Flexible Traversal) search is a novel technique introduced by Microsoft that combines global and local search by using hypothetical document embedding to generate a document based on the query that is then embedded and used to retrieve similar community reports, followed by generation of an initial answer and follow-up questions, then local search on each follow-up question with relevance scoring to create a pseudo-refinement loop.Knowledge Graph or Vector Database… Which is Better?
- Representation collapse in joint embedding architectures—where networks trivially output identical embeddings for all inputs—was the primary unsolved problem preventing joint embeddings from learning useful representations until Barlow Twins and related approaches.Yann LeCun's $1B Bet Against LLMs [Part 1]
- Text is tokenized into discrete units, each token is assigned a numeric ID from a dictionary, and that ID is converted into a complex vector (embedding) that represents the token's meaning in mathematical form.Designing LLM Systems PART 1 Foundations Chapter 1 AI Systems Architecture The Real Currency of AI
- JEPA (Joint Embedding Predictive Architecture) avoids the pixel prediction problem by mapping observations to embeddings and training a predictor to forecast future embeddings rather than pixels, allowing the model to focus on salient scene features instead of random unpredictable details.Yann LeCun's $1B Bet Against LLMs [Part 1]
- VJEPA-2 conditions JEPA on action signals sent to a robot arm, allowing the predictor to learn how control signals change the robot's position in embedded space, enabling planning by exploring the model's predictions to find actions matching a goal state.Yann LeCun's $1B Bet Against LLMs [Part 1]
- VICReg is a simplified version of Barlow Twins that turned out to be quite effective, and DINO is a similar joint embedding technique developed by colleagues at Meta's Paris lab, demonstrating convergence on joint embedding approaches for vision.Yann LeCun's $1B Bet Against LLMs [Part 1]
- A semantic search system can be built for Obsidian by vibe-coding a TypeScript plugin that calls Claude Code, which uses an MCP server to query a Chroma vector database storing embeddings of every note, enabling similarity search on meaning rather than keywords.This combo makes smart note systems magical.
- Semantic chunking uses vector similarity to identify context boundaries by embedding chunks, measuring similarity between consecutive chunks, and merging chunks that exceed a similarity threshold while separating chunks that fall below it, thereby creating chunks with genuine semantic coherence rather than arbitrary size or structure boundaries.Ep.4 - Chunking Strategies Explained: How to Structure Data for LLMs #RAG #llm #ai #aiapplications
- Horace Barlow hypothesized in 1961 that neurons in animal and human vision systems operate by reducing redundant information between neurons, a principle later applied by Stefan Deny and colleagues to solve representation collapse in joint embedding architectures.Yann LeCun's $1B Bet Against LLMs [Part 1]