Alternatives to Github Copilot and ChatGPT

Curating the ocean of AI news keeping you informed on topics key to the craft of software development

🧰 AI tools that help you build software

Huge improvements in the newest release of gpt 3 and 4

All this available immediately from OpenAI. Some highlights:
- Tokens 25% cheaper for input tokens on 3.5
- There is now a 16k token context length option (former max was 4k)
- There is now a function calling mechanism. In an API call, you can describe functions and have the model intelligently choose to output a JSON object containing arguments to call those functions. The API does not call the function; instead, the model generates JSON that you can use to call the function in your code.

Alternatives to Github Copilot and ChatGPT

There are a growing number of AI coding tools that are alternatives to Copilot and ChatGPT. The Pragmatic Engineer blog offers a nicely put together summary. There is a tool for every developer, no matter what their needs are.

What's your Big O?

Timecomplexity.ai is a nicely designed tool for a specific task; determining the efficiency of your algorithms using Big O notation. It offers up to 50 free daily responses for logged-in users, and unlimited for paid users. Uploaded code is public, but private uploads may be an option in the future.

Show Me ChatGPT Plugin

This plugin allows you to create and edit diagrams in ChatGPT. It uses the Kroki API to generate SVG diagrams from text descriptions. The plugin is still under development, but it is already capable of creating a wide variety of diagrams, including flowcharts, UML diagrams, and entity-relationship diagrams.

Generate React Components and Webpages from Figma Designs

Quest AI is a no-code platform that helps developers build React components and webpages from Figma designs. It uses artificial intelligence to generate code that is both accurate and well-formatted. Quest helps you build the full stack applications faster than ever.

CodeSquire: An AI Code Assistant for Data Scientists

CodeSquire is an AI code assistant that helps data scientists write better code faster. It uses AI to understand the code you are writing and provide suggestions for improvements. CodeSquire can also generate code for you, so you can focus on the creative aspects of data science. Supported platforms include Jupyter, Colab, Databricks, Google BigQuery.

🏗️ Utilizing AI in your software

Go from prototype to production

Building a prototype with GPT-4 is relatively straightforward. However, scaling it for production can be challenging. Chariot can help with this process by providing tools for configuring your language model, uploading your data, generating embeddings, streaming messages, and managing conversations.

Postgres as a vector Db

Supabase Vector is a new tool that allows developers to store and search embedding vectors in Postgres. “The best vector database is the database you already have.”

Python package for easily interfacing with chat apps

simpleaichat provides a number of features that make it easier to develop chat applications, including:

• A simple API for sending and receiving messages

• Support for multiple chat apps

• Automatic handling of authentication and authorization

• A number of helper functions for common chat tasks

Analyze multiple LLMs in parallel

Aviary is a platform that allows users to interact with multiple large language models (LLMs) simultaneously. Users can compare the outputs of different models, rank them by quality, and get cost and latency estimates. Access via an intuitive UI or CLI.

Keep Track of AI Progress with Open LLM Leaderboard

I came across the 🤗 Open LLM Leaderboard, a tool that tracks, ranks, and evaluates open LLMs and chatbots as they're released. An automation evaluates submitted models on four key benchmarks from the Eleuther AI Language Model Evaluation Harness, a unified framework that tests generative language models across various tasks.

GPT Best Practices

OpenAI’s GPT best practices guide covers methods to improve model reasoning, reduce the likelihood of model hallucinations, and more. The guide provides six strategies for getting better results including writing clear instructions and providing reference text.

📰 AI news for devs

Generative AI support added to Google's Vertex AI

Google Cloud's Vertex AI now offers generative AI support, providing access to over 60 generative AI models hosted on Google Cloud. The platform provides an intuitive interface, including the Generative AI Studio, which allows users to harness the power of generative AI without extensive technical knowledge.

DeepMind’s Game-Playing AIs Optimize Code and Infrastructure

DeepMind’s Alpha series of AIs have been repurposed to optimize code and infrastructure. AlphaZero, exposed to Borg data, reduced underused hardware by up to 19%. MuZero reduced the bitrate of YouTube videos by 4%. It is amazing that an AI developed for winning games could learn and generalize its approach in unrelated fields like compression.

Eight Things to Know about Large Language Models

Samuel R. Bowman’s paper surveys evidence for eight potentially surprising points about LLMs. Much of the points made harken back to the mysterious nature of neural nets at scale, specifically stated in 5. Experts are not yet able to interpret the inner workings of LLMs.

ChatGPT’s Impact on Open Source Software

OpenAI’s ChatGPT has created a buzz among developers, especially those in the open source community. Concerns for open source contributors and developers are sparking debate as questions surrounding the origin of generated source code surface, and the possible ethical and legal implications are discussed. However, developers should not fear or stray away from using ChatGPT but rather shift their focus to understanding how to embrace it for positive outcomes.

Review of a long history of developer assisting tools

AI-assisted software development tools automatically suggest or generate software source code to programmers. This may feel like cutting-edge technology -- in many respects, it is. However, AI-assisted development isn't entirely new. In fact, it has a longer history that stretches back to other tools that used AI or algorithmic processing to help coders do their work.

Asana enters the AI ring

The popular team collaboration tool Asana has entered the ring with a suite of AI assistance features. Sign up for the live demo

A better search for ChatGPT plugins

We appreciate that OpenAI released plugin support, as it greatly enhances the product. Most would agree the UX for the feature is a “work in progress” 🙂. WhatPlugin to the rescue. This is much more of what you would expect in a tool to search and browse the suite of plugins available.

🏫 Read/Listen/Watch: Learn

Learn while you commute, do chores, exercise…

A Very Gentle Introduction to Large Language Models without the Hype [read: 30min]

Mark Riedl’s article provides a gentle introduction to large language models without the hype. It’s designed for people with no computer science background and uses metaphors to explain concepts. I found it interesting how he breaks down the terminology associated with large language models and ChatGPT without any jargon

Introducing LLM University from Cohere [course]

Cohere has launched LLM University, a comprehensive learning resource for anyone interested in natural language processing (NLP) and LLMs. The curriculum covers everything from the basics of LLMs to advanced topics like generative AI. I found it interesting that they cater to learners from all backgrounds and offer hands-on exercises.

Build an LLM-powered app in 18 lines of code [tutorial]

Chanin Nantasenamat’s LangChain tutorial shows how to build an LLM-powered Streamlit app in four steps using OpenAI, LangChain, and Streamlit. This really shows how easy it is to rapidly prototype robust applications with LangChain. Well written and easy to follow.

LangChain for LLM Application Development [course]

DeepLearning.AI offers a short course on LangChain for LLM Application Development, instructed by Harrison Chase and Andrew Ng. The course teaches essential skills in expanding the use cases and capabilities of language models in application development using the LangChain framework. I really appreciate in these courses that the concepts are applied and you are even given a pre-populated notebook so you can follow along.

 Thanks for reading and we will see you next time

Follow me on twitter, DM me links you would like included in a future newsletters.