Language Models

AI Large Language Models Natural Language Processing

Beginner Friendly Step-by-Step Example of RAG

Imagine your large language model (LLM) is like a vast library brimming with books containing general knowledge about almost everything. Now, suppose you run a specialized store—maybe a boutique with exclusive fashion items or a tech shop with unique gadgets. Customers pop in with questions specific to what you offer. While your LLM is a […]

Read More
AI Large Language Models

What is Multimodal RAG? Unlocking LLMs with Vector Databases

Imagine you’re a wizard tinkering away in a grand library, surrounded by documents, diagrams, and artifacts, each telling a unique story with different sensory languages. Text stitches the narrative, images give it color, and sounds add rhythm. Your challenge? To swiftly pull the right information together in response to a query. The magic spell you […]

Read More
AI Large Language Models

How to Fine-Tuning Local LLMs with Axolotl in Python

Today, let’s dive into the intricacies of fine-tuning large language models (LLMs) locally using Axelottle, a Python framework designed for such tasks. Imagine you’re a mechanic, tweaking a well-designed car to perform specific stunts – that’s pretty much what fine-tuning an LLM entails. You start with a solid base—a pre-trained model—and then customize it to […]

Read More
AI Large Language Models

LLMs Aren’t Just Predicting Words. Here’s the Truth About AI Reasoning.

Have you ever pondered whether algorithms can actually “think”? It’s a question that seems straight out of a sci-fi novel, yet here we are, discussing it in the context of modern artificial intelligence. The essence of this question reminds me of the misconceptions that cloud our understanding of intricate machines. To say that a Ferrari […]

Read More
AI Large Language Models

AI Coherence vs Correctness

Today, I want to explore an intriguing and somewhat alarming phenomenon in the world of large language models (LLMs) that’s been highlighted in a recent study by Trent Slade. This issue isn’t about simple errors like a misplaced date or a typographical oversight. It’s about a new and profoundly insidious type of error—semantic stabilization—where the […]

Read More
AI Red Hat Research

Pre-Training vs Fine-Tuning Explained: OSFT, LoRA, RL & Red Hat Training Hub Breakdown

Understanding the nuts and bolts of how AI learns can sometimes feel like deciphering a secret code. In a recent enlightening discussion I came across about language models—specifically regarding pre-training and various fine-tuning techniques—I realized just how intricate and fascinating the world of AI training really is. Let’s unpack that together, shall we? Starting with […]

Read More
AI Red Hat Research

Training Hub: continuous learning for LLMs

As we delve deeper into the complexities and capabilities of AI, it’s crucial to understand not just how we build these models, but how we refine and perfect them over time to better meet our needs and solve real-world problems. Today, let’s explore the fascinating world of post-training language models. It’s a bit like tuning […]

Read More
AI Large Language Models Red Hat

How to Connect data to LLMs with model customization

In today’s digital age, the art of adapting technology to meet specific business needs is not just useful; it’s essential. And nowhere is this more evident than in the realm of artificial intelligence—specifically in the tuning of language models for tailored business functions. As I delved into a fascinating session from Red Hat AI, led […]

Read More
AI Large Language Models

How to Train a Reasoning-Capable LLM in One Weekend

As a passionate advocate for democratizing technology, nothing excites me more than the idea that complex tools and capabilities are becoming more accessible to a wider audience. Today, let’s dive deep into an incredible example of this: using NVIDIA NeMo to train a large language model with reasoning capabilities. This process, which might have once […]

Read More
AI Large Language Models Red Hat

Cedric’s Got 99 Problems, but LLMs Ain’t One! With Red Hat

Whether you’re entrenched in the tech trenches of Brooklyn, like our protagonist from Red Hat, or elsewhere in the vibrant AI landscape, self-hosting your own language models (LLMs) is becoming not just a viable option, but increasingly, a strategic necessity. This video is from Ai4. The motivation behind running one’s own LLMs stems largely from […]

Read More