Gpt Vs Gemini For Structured Information Extraction

A systematic approach for evaluating the Structured Output accuracy of Large Language Models

What you'll learn

How to use the Structured Output feature in GPT

How to use the Structured Output feature in Gemini

How to extract different data types like numerical values, booleans etc

How to measure the accuracy of the structured information you extracted

Requirements

Fairly proficient in Python

You should already know how to use Jupyter

Preferable: basic knowledge of the spaCy NLP library

Description

Natural Language Processing (NLP) is often* considered to be the combination of two branches of study - Natural Language Understanding (NLU) and Natural Language Generation (NLG). Large Language Models can do both NLU and NLG. In this course we are primarily interested in the NLU aspect - more specifically we are interested in how to extract structured information from free form text. (There is also an NLG aspect to the course which you will notice as you watch the video lessons).Recently both GPT and Gemini introduced the ability to extract structured output from the prompt text. As of this writing (November 2024), they are the only LLMs which provide native support for this feature via their API itself - in other words, you can simply specify the response schema as a Python class, and the LLMs will give you a "best effort" response which is guaranteed to follow the schema. It is best effort because while the response is guaranteed to follow the schema, sometimes the fields are empty.  How can we assess the accuracy of this structured information extraction?This course provides a practical and systematic approach for assessing the accuracy of LLM Structured Output responses. So which one is better - GPT or Gemini? Watch the course to find out :-)*For example, that is how Ines Montani, co-founder of spaCy recently described the fields in a podcast interview.

Overview

Section 1: Introduction

Lecture 1 Is this meme still true?

Lecture 2 About this course

Lecture 3 Why not use client libraries

Section 2: Getting started

Lecture 4 Install libraries

Lecture 5 Set environment variables

Lecture 6 Download the Jupyter notebook

Section 3: Numerical values

Lecture 7 Exploring numerical values in the dataset

Lecture 8 Extracting numerical values using Gemini

Lecture 9 Measuring Gemini accuracy for numerical values

Lecture 10 Extracting numerical values using GPT

Lecture 11 Measuring GPT accuracy for numerical values

Lecture 12 Comparing Gemini and GPT accuracy for numerical values

Section 4: Date values

Lecture 13 Exploring date values in the dataset

Lecture 14 Extracting date values using Gemini

Lecture 15 Measuring Gemini accuracy for date values

Lecture 16 Extracting date values using GPT

Lecture 17 Measuring GPT accuracy for date values

Lecture 18 Comparing GPT and Gemini accuracy for date values

Section 5: Boolean values

Lecture 19 Exploring boolean values in the dataset

Lecture 20 Extracting boolean values using Gemini

Lecture 21 Measuring Gemini accuracy for boolean values

Lecture 22 Extracting boolean values using GPT

Lecture 23 Measuring GPT accuracy for boolean values

Lecture 24 Comparing GPT and Gemini accuracy for boolean values

Section 6: Why use an Explanation

Lecture 25 Downsides of using the Explanation class

Lecture 26 Explanation provides a future reference

Lecture 27 Explanation can speed up annotation for spaCy Prodigy

Lecture 28 Explanation can provide more accurate responses

Lecture 29 Better responses: an example

Lecture 30 What we can infer from the quality of GPT and Gemini explanations

Intermediate Python developers who want to learn how to use GPT and Gemini to extract structured information from any dataset