Offre de lancement : Accès à vie à 79€ (code MZMDKXNW, 30 premiers clients)·En profiter ➔
← Back to Blog

COMPLETE COURSE ON VECTOR DATABASES: EMBEDDINGS AND CHROMADB

Published on 2026-11-05

2 free analyses · No credit card · Results in 30s
🚀

COMPLETE COURSE ON VECTOR DATABASES: EMBEDDINGS AND CHROMADB

Source Title:
Vector Databases Full Course for Beginners – Embeddings, Similarity Search & ChromaDB
Source URL:
Scan Date:
05/11/2026 at 08:07
Generator:
VidScope Agent (Autonomous Mode)
📄

EXECUTIVE SUMMARY

This complete course offers an in-depth introduction to vector databases, explaining fundamental concepts such as vectors, tokenization, and embedding models. It details how semantic search works and high-performance indexing mechanisms like IVF and HNSW. Through practical demonstrations using the Google Gemini API, learners discover how to generate embeddings and efficiently manipulate them within ChromaDB. This training covers the entire pipeline, from raw data transformation to storage and semantic querying.

Tool Name Category Quick Description Official Link
Gemini
Used
AI
Used as an AI model to generate vector embeddings during a live demonstration.
« You will generate embeddings during a live demonstration with Google's Gemini model »
Visit
Chroma DB
Used
Database
Vector database used in the practical project to store embeddings and perform semantic searches.
« you will store embeddings in Chroma DB and perform semantic searches »
Visit
OpenAI Tokenizer
Used
Dev tool
Online platform allowing visualization of text tokenization.
« let's head now to OpenAI's tokenization platform. »
Visit
GloVe
Used
AI
Word embedding model used to illustrate vector relationships between words.
« the model I am using is Glove 300D. »
Visit
Word2Vec
Mentioned
AI
Classic word embedding model cited as a pioneer in the field.
« Among popular examples of embedding models, we have Word2Vec. »
Visit
CodeBERT
Mentioned
AI
Embedding model optimized for code understanding and similarity.
« you have other embedding models from Cohere and Microsoft CodeBERT, which is optimized for code understanding. »
Visit
Google Colab
Used
Dev tool
Notebook environment used to run the live demonstration.
« Alright, let's now switch to Google Colab. »
Visit
Pandas
Used
Library
Python library used to manipulate data and generate the similarity matrix.
« We need Google's genai. We also need pandas »
Visit
Scikit-learn
Used
Library
Library used to compute cosine similarity between vectors.
« we need to import cosine_similarity from sklearn. »
Visit
Plotly
Used
Library
Visualization library used to render interactive 3D charts.
« You can do this using plotly. »
Visit
ChromaDB
Used
Database
Vector database used to store and perform semantic searches on embeddings.
« In this demonstration, we will generate embeddings and store them in Chroma DB. »
Visit
Python
Used
Programming language
Language used to manipulate data, call APIs, and interact with ChromaDB.
« How to write a Python function? »
Visit
Java
Mentioned
Programming language
Language mentioned as an example of content stored in the vector database.
« And how can I create a class in Java? »
Visit
SQL
Mentioned
Database
Relational database system used to compare indexing concepts.
« You will encounter an indexing layer in any database, whether traditional, like SQL, or NoSQL. »
Visit
NoSQL
Mentioned
Database
Non-relational database type used to compare indexing concepts and collections.
« You will encounter an indexing layer in any database, whether traditional, like SQL, or NoSQL. »
Visit
🔍

UNRESOLVED MENTIONS

Tools or technologies mentioned without a specific name in the video:

embedding models ⏱️ 06:21
Unspecified machine learning models used to convert raw data into numerical vectors.
large language models ⏱️ 53:16
Reference to modern models that use subword tokenization.
tokenizer ⏱️ 48:09
System component that converts raw data (text, audio, video) into numerical tokens.
Cohere embedding models ⏱️ 10:31
Mentioned as popular alternatives in the industry.
⏱️

KEY MOMENTS

[02:03]
Explanation of the role of tokenization in the embedding pipeline.
[05:35]
Explanation of the concept of subword tokenization for token optimization.
[08:18]
Introduction to vector indexing techniques to speed up search.
[11:11]
Comparison between traditional relational databases and vector databases.
[26:30]
Explanation of the relationship between the number of features and vector dimensions.
[34:47]
Demonstration of the formula and usefulness of cosine similarity to compare directions.
[48:47]
Analysis of the three tokenization methods: character, word, and subword.
[01:01:23]
Mathematical demonstration of the relationship between vectors (King - Man + Woman = Queen).
[01:15:39]
Setting up the Google Gemini client and configuring vector dimensionality.
[01:23:22]
Presentation of the internal architecture of a vector database with its four main layers.
[01:26:04]
Explanation of IVF (Inverted File Index) and HNSW (Hierarchical Navigable Small Worlds) indexing techniques.
[01:33:04]
Creating a ChromaDB client and storing vectors in a collection.
⭐

TECHNOLOGIES COVERED

🔹 Embeddings
🔹 Vector Database
🔹 Semantic Search
🔹 RAG
🔹 Tokenization
🔹 Machine Learning
🔹 IVF
🔹 HNSW
🔹 Similarity Search
🔹 Vectors

Ready to save hours every week?

Let VidScope's AI analyze your YouTube videos and extract the best tech tools instantly.

2 free analyses · No credit card · Results in 30s