Phi-3+ContinueDev+Ollama: STOP PAYING for Github's Copilot with this NEW, LOCAL & FREE Alternative

AICodeKing
25 Apr 202403:54

TLDRIn this episode of AI Code King, the host discusses the challenges of coding and the popularity of GitHub's Copilot tool. However, for those who prefer not to pay for such services, the host introduces a new, local, and free alternative using Microsoft's 53 model. This model is praised for its coding capabilities and is comparable to Llama and GPT 3.5. The video demonstrates how to set up a local co-pilot alternative using Continue Dev as a VS Code extension and Ollama to serve the API. The host guides viewers through the installation process and showcases the model's ability to generate code, answer questions, and provide auto-completion. The video concludes with a Python program example for a snake game, emphasizing the alternative's accessibility and performance on most computers.

Takeaways

  • 💻 Coding can be challenging, involving tasks like naming files and variables, but tools like GitHub's Copilot can help automate these tasks.
  • 🚀 GitHub's Copilot gained popularity for assisting with coding, but it requires a subscription fee, which not all developers are willing to pay.
  • 🌟 Open-source alternatives to Copilot have emerged, offering similar functionality without the cost, though they often require third-party APIs or self-hosting.
  • 🔍 The video discusses the use of Microsoft's 53 model, which is a new, smaller model with a reported performance comparable to Llama and GPT-3.5.
  • 📈 The 53 model is notable for its smaller size and the ability to run on less powerful hardware, making it more accessible to developers.
  • 🔧 The video demonstrates how to set up a local alternative to Copilot using the 53 model, Continue Dev as a VS Code extension, and Ollama for the API.
  • 📝 The setup process involves installing Ollama, setting up the 53 model, and configuring the Continue Dev extension to work with it.
  • 💬 The 53 model can be interacted with through a chat interface, where users can ask questions and receive instant answers or generate code snippets.
  • 🐍 An example given in the video is generating a Python program for a snake game, showcasing the model's ability to produce code based on user prompts.
  • ✅ The alternative is praised for its ease of use, compatibility with most computers, and the fact that it doesn't require a high-end graphics card.
  • 📈 The video encourages viewers to share their thoughts on the alternative in the comments and to subscribe to the channel for more content.

Q & A

  • What is the main topic of the video?

    -The video is about an alternative to Github's Copilot that is local, free, and can be used without paying for the service.

  • Why did developers start using Github's Copilot?

    -Developers started using Github's Copilot to avoid having to do basic coding tasks, as it assists in writing code.

  • What are some issues developers face with open-source alternatives to Github's Copilot?

    -Open-source alternatives often require using a third-party API or self-hosting a large model that demands significant computational resources, including a powerful graphics card.

  • What is the significance of Microsoft's 53 model in the context of this video?

    -Microsoft's 53 model is significant because it is a smaller model with 12 million parameters, which is efficient for coding and can run on less powerful hardware compared to larger models.

  • How does the 53 model compare to other models like Llama and GPT-3.5?

    -The 53 model is said to be on par with Llama and GPT-3.5 in terms of its coding capabilities.

  • What is the process of setting up the local co-pilot alternative using Microsoft's 53 model?

    -The process involves installing the Ollama model via its website, setting up the AMA chat interface, installing the Continue Dev extension in VS Code, and configuring it to use the 53 model for code generation and assistance.

  • What is Continue Dev and how does it relate to the local co-pilot alternative?

    -Continue Dev is a VS Code extension that, when combined with the Ollama model, serves as a local co-pilot alternative for code generation and assistance.

  • How can users generate code with the local co-pilot alternative?

    -Users can generate code by pressing Control+I and typing in their prompt in the Continue Dev extension interface.

  • What is the advantage of using the 53 model for larger code bases?

    -The 53 model has a 12 million parameter version that is optimized for handling larger code bases, providing a more efficient and effective code generation experience.

  • Can the local co-pilot alternative be used for auto-completion and code analysis?

    -Yes, the local co-pilot alternative can be used for auto-completion, and it can also analyze selected code to provide insights or explanations.

  • What is the hardware requirement for running the local co-pilot alternative?

    -The local co-pilot alternative can run on most computers, as it utilizes the 53 model which has lower hardware requirements compared to models with billions of parameters.

  • How can viewers provide feedback on the alternative presented in the video?

    -Viewers can leave comments on the video to express their opinion on the alternative, and they can also show their support by liking the video and subscribing to the channel.

Outlines

00:00

👨‍💻 Introduction to AI Code King and Co-Pilot Alternatives

The video begins with a candid acknowledgment of the difficulty of coding, from the initial setup to the naming of files and variables. The host discusses the popularity of GitHub's Co-Pilot tool, which automates basic coding tasks, but also highlights the financial barriers that deter some developers from using it. The host then introduces the audience to open-source alternatives that have emerged since Co-Pilot's release, mentioning that they have covered some on their channel. These alternatives often require a powerful computer or third-party API usage. However, the game-changing Microsoft 53 model is introduced as a more accessible alternative that performs well for coding tasks without the need for extensive computational resources. The host outlines the plan to create a local Co-Pilot alternative using the Microsoft 53 model and the Continue Dev extension for VS Code.

Mindmap

Keywords

💡Coding

Coding is the process of writing computer programs in a specific programming language. It is a fundamental skill for software development and is essential for creating applications, websites, and other digital products. In the video, coding is presented as a challenging task that requires focus and creativity, which is why the use of AI tools like GitHub's Copilot is attractive to many developers.

💡GitHub's Copilot

GitHub's Copilot is an AI-powered code generation tool that assists developers by suggesting code as they type. It uses machine learning to predict what the developer might want to write next, making the coding process more efficient. In the video, it is mentioned as a popular tool that many developers have adopted, but it also has a subscription cost which some developers prefer to avoid.

💡Open-source alternatives

Open-source alternatives refer to software options that are available with source code that anyone can inspect, modify, and enhance. These alternatives are typically developed collaboratively by communities and are free to use. In the context of the video, the presenter discusses open-source alternatives to GitHub's Copilot, emphasizing the availability of free options for developers who do not wish to pay for code assistance.

💡Microsoft's 53 model

The Microsoft 53 model is an AI model developed by Microsoft, which is designed to be proficient in coding tasks. It is mentioned in the video as being on par with other advanced AI models like Llama and GPT-3.5, suggesting that it can serve as an effective alternative to GitHub's Copilot for code generation and assistance.

💡VS Code Extension

A VS Code Extension is a software component that adds functionality to Visual Studio Code, a popular source-code editor. In the video, the presenter uses the term to refer to the Continue Dev extension, which integrates the Microsoft 53 model into VS Code, allowing developers to use it directly within their coding environment.

💡Continue Dev

Continue Dev is mentioned as a VS Code extension in the video. It serves as a platform for integrating AI models like Microsoft's 53 model into the development workflow, enabling developers to generate code, get instant answers, and enhance their productivity without leaving the VS Code environment.

💡Llama

Llama, in the context of the video, refers to a specific AI model or possibly a service that hosts AI models for use in various applications. It is used in conjunction with Continue Dev to serve the API that powers the local co-pilot alternative discussed in the video.

💡Autocompletion

Autocompletion is a feature in many coding environments and tools that suggests or automatically completes code for the developer as they type. This feature saves time and reduces the potential for errors. In the video, autocompletion is one of the functionalities provided by the local co-pilot alternative using the Microsoft 53 model.

💡API

API stands for Application Programming Interface, which is a set of protocols and tools that allows different software applications to communicate with each other. In the video, the presenter discusses using a Llama API to serve the Microsoft 53 model, which is then utilized by the Continue Dev extension in VS Code.

💡Python Program

A Python program is a set of instructions written in the Python programming language, which is known for its simplicity and readability. In the video, the presenter demonstrates the capabilities of the local co-pilot alternative by generating a Python program for a snake game, showcasing the AI's ability to produce functional code snippets.

💡Terminal

A terminal is a text-based interface for interacting with a computer system. It allows users to execute commands and receive output in text form. In the video, the terminal is used to install and interact with the Llama model, which is a part of setting up the local co-pilot alternative.

Highlights

Coding can be a challenging and time-consuming process.

GitHub's Copilot has become popular for assisting with coding tasks.

Many developers prefer not to pay for services like Copilot and seek alternatives.

Open-source alternatives to Copilot have emerged, requiring less computational resources.

Microsoft's 53 model is a new, efficient alternative for coding assistance, comparable to Llama and GPT-3.5.

The 53 model has a 12 128k context version, suitable for larger codebases.

The video demonstrates creating a local, free alternative to Copilot using Microsoft's 53 model.

Continue Dev is used as a VS Code extension for the alternative.

Ollama serves as the API for the local coding assistance tool.

Installation instructions for Ollama and the 53 model are provided.

Continue Dev can be installed from the VS Code extensions marketplace.

The 53 model can be selected within Continue Dev for instant coding assistance.

The tool can generate code based on user prompts, demonstrated with a Python program for the snake game.

Auto-completion and code analysis are additional features of the alternative tool.

The alternative is accessible on most computers and does not require a high-end graphics card.

The presenter encourages viewers to share their thoughts on the alternative in the comments.

A call to action for viewers to like, comment, and subscribe to the channel is made.