Posts

Tip Please check this out in colab to run the code easily! Introduction Goal I want jais-13B deployed with an API quickly and easily. Info In this blog you will learn: How to leverage TGI and Inference Endpoints with jais How to deploy a model on the HW of your choice using the Hub Client Library Fundamental concepts on how decoding works and why they matter Approach There are lots of options out there that are “1-click” which is really cool!...

Goal This is part 6 of 6 in our tutorial on Arabic RAG. We have created all of the components we need to make our RAG solution. All that is left is to stitch them together! Note In this blog you will learn how to: Quickly and efficiently deploy jais using Inference Endpoints Combine all the components of RAG into a functional system Create a beautiful Gradio App for RAG If you want to skip all this and actually try the app here it is: https://huggingface....

Goal This is part 5 of 6 in our tutorial on Arabic RAG. I’ll be using this blog as a guide, but to actually run this tutorial, its best that you run this notebook as described in part 1. Info In this blog you will learn: Important VectorDB considerations How to prepare your data for ingestion How to use LanceDB for RAG Approach VectorDB Why do we even need a VectorDB?...

Goals This is part 4 of 6 in our tutorial on Arabic RAG. I’ll be using this blog as a guide, but to actually run this tutorial, its best that you run this notebook as described in part 1. Before diving in, I want to you to think about how much money it costs to embed 2M articles. Make an estimate and see how accurate your guess is at the end of the blog....

Goal This is part 3 of 6 in our tutorial on Arabic RAG. I’ll be using this blog as a guide, but to actually run this tutorial, its best that you run this notebook as described in part 1. In this blog you will learn: Chunking Considerations How to leverage the very useful Haystack library from Deepset for preprocessing your data for RAG How to structure your data and code for parallel pre-processing Preprocessing import json from pathlib import Path import pickle from tqdm....

Goal This is part 2 of 6 in our tutorial on Arabic RAG. I’ll be using this blog as a guide, but to actually run this tutorial, its best that you run this notebook as described in part 1. In this blog you will learn: How to choose an Embedding Model Why you need to think about token analysis for Arabic RAG How to analyze a tokenizer to estimate words per token How to visualize this to justify your decisions Why Analyze Tokenization?...

Goal This is part 1 of 6 on a tutorial for Arabic RAG. RAG is short for Retrieval Augmented Generation. It took it’s name from Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks though it’s current usage is much more similar to the RALM Paper. In this tutorial you will learn: Why is RAG important How to download Wikipedia How to format Wikipedia for scalable processing Addressing Hallucinations Large Language Models (LLMs) get blamed (though unfairly IMHO) for “hallucinating”....

Introduction Welcome to my blog! My name is Derek Thomas and I am currently a Pre-Sales Machine Learning Success Engineer at Hugging Face 🤗! I’m really proud to work here as it was my dream when I first stumbled upon pytorch-pretrained-bert way back in November 2018 and was blown away. I live in Abu Dhabi, UAE, though I’m not from here. I’m an American from Kentucky. Un-related (possibly) I make superb fried chicken 🍗....