Llama 2 has arrived! The highly anticipated update to Meta’s language model is now available for local installation. We know many of you have been eager to get your hands on this powerful AI assistant. In this post, we’ll walk you through the steps for setting up Llama 2 locally on your own machine.
- Preparing for Local Use
Whether you’re an AI enthusiast, developer, or business leader, having access to Llama 2 locally unlocks a world of possibilities. You’ll be able to utilize Llama’s advanced natural language capabilities for a wide range of applications, while keeping your data private and secure.
We’re thrilled to help guide you through the local setup process. With some simple configuration, you’ll have this remarkable AI assistant running smoothly in no time. The team at Meta has put in long hours to deliver this major update, and we think you’re going to love exploring everything Llama 2 has to offer.
Preparing for Local Use
Running Llama 2 locally provides a lot of flexibility since it doesn’t require an Internet connection. We’ve seen fascinating examples of its use, such as creating websites to showcase the cool factors of llamas. And with the release of Llama 2, we now have access to open-source tools that allow running it locally. Here are the main ones:
- Llama.cpp (Mac/Windows/Linux)
- Ollama (Mac)
- MLC LLM (iOS/Android)
Let’s dive into each one of them.
Llama.cpp: A Versatile Port of Llama
Llama.cpp is a C/C++ port of the Llama, enabling the local running of Llama 2 using 4-bit integer quantization on Macs. However, it extends its support to Linux and Windows as well.
To install it on your M1/M2 Mac, here is a line you can use:
“`bash curl -L “https://replicate.fyi/install-llama-cpp” | bash “` This installation command will also run fine on an Intel Mac or Linux machine, but without the `LLAMA_METAL=1` flag:
“`bash curl -L “https://replicate.fyi/install-llama-cpp-cpu” | bash “`
For Windows on WSL, use:
“`bash curl -L “https://replicate.fyi/windows-install-llama-cpp” | bash “`
Ollama: A macOS App
Ollama is a macOS open-source app that lets you run, create, and share large language models with a command-line interface, and it already supports Llama 2.
To use the Ollama CLI, download the macOS app at ollama.ai/download. Once installed, you can freely download Lllama 2 and start chatting with the model.
Here are the lines you can use to download the model:
```bash download the 7B model (3.8 GB) ollama pull llama2 or the 13B model (7.3 GB) ollama pull llama2:13b ```
And then run the model:
“`bash ollama run llama2 “`
Windows: A Detailed Guide
To install Llama on Windows, you need to follow these steps:
- Clone and download the Llama repository.
- Visit the Meta website and register to download the model/s. Once registered, you will get an email with a URL to download the models. You will need this URL when you run the download.sh script.
- Once you get the email, navigate to your downloaded llama repository and run the download.sh script. Make sure to grant execution permissions to the download.sh script.
- During this process, you will be prompted to enter the URL from the email. Do not use the “Copy Link” option but rather make sure to manually copy the link from the email.
- Once the model/s you want have been downloaded, you can run the model locally using the command provided in the Quick Start section of the Llama repository.
Windows users have a step-by-step guide for downloading and running the Llama model using Nvidia GPU’s CUDA Toolkit and cloning the relevant GitHub repository.
After following these steps, you can create a powershell function that can quickly run prompts with `llama “prompt goes here”`.
Running Llama 2 locally is becoming easier with the release of Llama 2 and the development of open-source tools designed to support its deployment across various platforms. Whether you are on a Mac, Windows, Linux, or even a mobile device, you can now harness the power of Llama 2 without the need for an Internet connection. As Llama 2 continues to evolve, we can expect even more exciting developments in the near future.