Medium

Dr. Leon Eversberg

Featured Book

See all (2)

Understanding Large Language Models and Generative AI: Inside the Technology Behind the Hype

Understanding Large Language Models and Generative AI: Inside the Technology Behind the Hype

An in-depth explanation of how LLMs work. The book covers the models, algorithms, and data that power the fast-moving world of generative AI.

2026

Stories

Pinned

In

Towards AI

by

Dr. Leon Eversberg

·

Aug 17

How Does Anthropic Watermark AI-Generated Content?

Learn how AI watermarks are hidden in text and files, how they can be detected, and what happens when you try to remove them

How Does Anthropic Watermark AI-Generated Content?

Pinned

In

AI Advances

by

Dr. Leon Eversberg

·

Dec 1, 2025

Why Do LLMs Hallucinate?

Why AI models produce plausible but false information and what you can do to avoid hallucinations

Why Do LLMs Hallucinate?

Pinned

In

TDS Archive

by

Dr. Leon Eversberg

·

Aug 11, 2024

How to Use Hybrid Search for Better LLM RAG Retrieval

Building an advanced local LLM RAG pipeline by combining dense embeddings with BM25

How to Use Hybrid Search for Better LLM RAG Retrieval

Pinned

In

TDS Archive

by

Dr. Leon Eversberg

·

May 2, 2024

How to Use Re-Ranking for Better LLM RAG Retrieval

Building an advanced local LLM RAG pipeline with two-step retrieval using open-source bi-encoders and cross-encoders

An image of an advanced RAG pipeline with two-step retrieval. First, a bi-encoder is used to find similar embedding vectors. Then, a cross-encoder model is used to narrow these candidates down to the top k most relevant documents.

In

Towards AI

by

Dr. Leon Eversberg

·

1d ago

How to Build a RAG Evaluation Dataset From Documents

Generate synthetic question-answer pairs from PDFs, validate them with an LLM judge, and create a dataset to evaluate your RAG pipeline

The Hugging Face dataset, shown as a table with the following columns: context, question, answer, source_doc, page, rationale, groundedness, stand_alone, faithfulness, answer_relevance, and score.

In

AI Advances

by

Dr. Leon Eversberg

·

Oct 4

Beyond Vector Search: Structure-Aware RAG With STAIR And Jev

How table-of-contents-based retrieval can guide LLMs to the right section

Workflow diagram divided into two steps: Input Preparation and Inference. Input Preparation shows a Book (PDF) leading to ToC Extraction. An arrow leads to the Inference stage, where Query Input and ToC feed into Jev, followed by Constrained Generation, and outputting to a Predicted Leaf Node.

In

AI Advances

by

Dr. Leon Eversberg

·

Sep 22

I Got Early Access to Jev, Typesafe AI’s New System One Model

How TypeSafe AI uses parallel sampling and typed outputs to deliver faster, cheaper decisions than standard LLMs

I Got Early Access to Jev, Typesafe AI’s New System One Model

In

Towards AI

by

Dr. Leon Eversberg

·

Sep 7

Shieldstral: Build an AI Content Moderation Classifier with Python

Learn how Mistral.AI’s 3B safety model uses natural-language policies for flexible, adaptive content moderation

Shieldstral: Build an AI Content Moderation Classifier with Python

In

AI Advances

by

Dr. Leon Eversberg

·

Aug 9

Why Human-Made Content Feels More Valuable Than AI-Generated Content

What research on effort, authenticity, and authorship can teach us about how we perceive generative AI

Why Human-Made Content Feels More Valuable Than AI-Generated Content

In

Level Up Coding

by

Dr. Leon Eversberg

·

Jul 28

The Practical Python Guide to Concurrency

Learn when to use asyncio, threads, and processes with a real PDF-to-Markdown pipeline

Blocking I/O sync code versus bon-blocking async code in Python

Dr. Leon Eversberg

Dr. Leon Eversberg

Book Author

Deep Learning PhD | AI Engineer | Author on LLMs & GenAI 📖 | Boosted & Featured Writer 🏆

Following

See all (1,302)

Help

Status

About

Careers

Press

Blog

Store

Privacy

Rules

Terms

Text to speech