Visual Gallery

Different approaches to breaking down text into manageable chunks for processing

Breaking text into tokens for processing by language models

How text is transformed into vector embeddings using neural networks

Methods for comparing and measuring similarity between vector embeddings

How similarity search works with vector embeddings

Traditional keyword-based search mechanisms and algorithms

Compressing vectors to reduce memory usage and improve performance

Product quantization technique for efficient vector compression

Comparison of different search methodologies and their use cases

How traditional keyword-based search works

Semantic search using vector embeddings

Combining keyword and vector search for optimal results

The motivation and use cases for vector databases

High-level architecture of a complete vector database system

Components and structure of a vector database instance

How data is organized into collections within a vector database

How objects are stored and structured in vector databases

How vectors are stored and managed in the database

How indexes are structured within vector databases

Distributing data across multiple shards for scalability

How collections are distributed across shards

Comparison of single-tenant and multi-tenant architectures

Simple flat index structure for exhaustive search

How search is performed in flat indexes

High-level view of the Hierarchical Navigable Small World algorithm

Visual representation of HNSW graph structure

How new vectors are inserted into HNSW indexes

How search works in HNSW indexes layer by layer

Key parameters and their impact on HNSW performance

Performance trade-offs between HNSW and flat indexes

Managing indexes that change over time

Introduction to the differences between standalone LLMs and RAG systems

How RAG enhances LLM responses with external knowledge

Benefits and use cases of RAG over standalone LLMs

Implementing RAG architectures in practice

Building a RAG system step by step

RAG with multimodal data using Weaviate

Understanding parameters and weights in neural networks

Deep dive into neural network parameter configurations

How neural networks learn through training

Using trained models for predictions and inference

Understanding and interpreting model outputs

Neural networks designed specifically for creating embeddings

How generative models produce text autoregressively

Overview and comparison of different embedding models

Details about the Arctic embedding model

Details about the FastText embedding model

The iterative process of evaluating and improving models

Core concepts in evaluating model performance

High-level architecture of a personalization agent

How personalization agents ingest and process user data

Detailed workflow of a personalization agent system

Agent workflows for data transformation tasks

Interactive coding guide and tutorial