explaingit

elyase/awesome-gpt3

4,539Audience · developerComplexity · 1/5Setup · easy

TLDR

A curated snapshot of what people built and wrote about GPT-3 when it first opened to the public in 2020, covering code generation demos, writing tools, API libraries, and early commercial products.

Mindmap

mindmap
  root((repo))
    What it is
      Curated reference list
      GPT-3 early demos
      2020 snapshot
    Demo Categories
      Code generation
      Writing and email
      Chart and recipe
      Medical and science
    Resources
      Blog articles
      Open-source tools
      Commercial products
    Audience
      AI researchers
      Developers
      Product builders
Click or tap to explore — scroll the page freely

Code map

Detail Auto

An interactive map of this repo's files and how they connect — its source is parsed live in your browser. Click Visualize to build it.

filefunction / class

Things people build with this

USE CASE 1

Browse real GPT-3 demos from 2020 to find inspiration for AI-powered app ideas across coding, writing, and data tasks

USE CASE 2

Research which early commercial products were built on GPT-3 to identify patterns in successful LLM applications

USE CASE 3

Find open-source Python, shell, and Go libraries for calling the GPT-3 API in your own projects

USE CASE 4

Study early blog posts describing GPT-3 limitations to understand which constraints still apply to modern language models

Tech stack

Markdown

Getting it running

Difficulty · easy Time to first run · 5min
License terms are not specified in the explanation.

In plain English

Awesome GPT-3 is a curated list of demonstrations, articles, GitHub projects, and commercial products built using OpenAI's GPT-3 language model. It was assembled when GPT-3 first became available via API in 2020 and captures the early wave of experimentation around the technology. The demonstrations section covers a wide range of things people tried with GPT-3 shortly after access opened up. On the coding side, people showed GPT-3 generating HTML layouts, React components, SQL queries, Python scripts, and even machine learning code from plain-English descriptions. On the writing side, examples include drafting emails from bullet points, simplifying legal language, translating text into multiple languages, and rewriting sentences to be more polite. Other demos covered chart generation, answering medical and physics questions, generating food recipes, creating marketing copy, and even playing chess. The articles section links to blog posts and essays from that period discussing how GPT-3 works technically, what it can and cannot do, and broader commentary on the hype surrounding it. These range from technical overviews and paper explanations to more critical pieces about managing expectations. The GitHub section points to a few related open-source projects, including a sandbox tool for turning GPT-3 ideas into demos quickly and libraries for interacting with the API from Python, shell, and Go. The products section lists commercial tools that launched using GPT-3, such as a Tailwind CSS code generator and an email drafting tool. The list has no code of its own. It is a reference document in Markdown format, maintained on GitHub. For anyone trying to understand what people were building with large language models in the early days of the GPT-3 API, this repository provides a snapshot of that moment.

Copy-paste prompts

Prompt 1
Based on the awesome-gpt3 demo list, summarize which types of tasks showed the strongest early results with GPT-3 and which types struggled, organized by category.
Prompt 2
Using the awesome-gpt3 email drafting examples as a pattern, help me write a prompt for Claude or GPT-4 that turns a bullet-point list into a polished professional email.
Prompt 3
From the awesome-gpt3 article list, what did critics say GPT-3 could not do well in 2020? Which of those limitations still apply to GPT-4 or Claude today?
Prompt 4
Help me build a simple HTML layout generator app similar to the demos in awesome-gpt3 using the current OpenAI API, show the full code and the prompt structure.
Open on GitHub → Explain another repo

← elyase on gitmyhub — every repo by this author, as a profile.

Verify against the repo before relying on details.