In science fiction movies, there are often scenes of dialogue between humans and computers, where you can simply give instructions to the computer and it will quickly execute the corresponding operations. For programmers, this seems like an absurd fantasy, and they scoff at these plotlines, believing that the screenwriters have no understanding of programming.
But! The emergence of Open-Interpreter has overturned this situation. Based on the most powerful artificial intelligence ChatGPT in the world today, it can understand human natural language and execute commands. Currently, it only supports text communication, but with the help of voice recognition technology, it is not far off to achieve the plotlines of science fiction movies.
Open-Interpreter: An open-source AI code interpreter that supports Windows, MacOS, and Linux platforms. It can run in a local environment, access the internet, execute Python scripts, and has no limitations on file size and runtime. As soon as the project was released, it skyrocketed to the top of GitHub's hot list with 31.2k stars. What is shocking is that the author of this project is just an undergraduate student in education at Western Washington University - Killian Lucas.
Open-Interpreter allows AI to execute code on your computer. It supports OpenAI and models hosted on Hugging Face. You can achieve your goals by giving instructions to Open-Interpreter in natural language. It's a bit like J.A.R.V.I.S. in Iron Man. For example, if you say to it, "Can you set my system to dark mode?", it will automatically write code to modify the computer's theme.
Installation and Usage:
Prerequisites:
- Python 3.10+
- Prepare an OpenAI API key
- Enable magic (you know what I mean~)
You can install Open-Interpreter with just one line of code:
pip install open-interpreter
Open the "cmd.exe" and enter the "interpreter" command to start Open-Interpreter.
interpreter
After executing the command, two language models will be provided:
- OpenAI's GPT-4 model (recommended)
- Mistral AI's open-source Mistral-7B language model
Here, we will use the GPT-4 model. Enter the OpenAI key and press enter.
Enter the command line interactive mode.
Open-Interpreter will write and execute code on its own to complete the tasks we give. To stop the execution of a task, press "ctrl + c" to terminate the task. If you want to close the program, simply close the terminal.
Important note: The code generated by Open-Interpreter runs in the local environment and interacts with files, system settings, etc., which may result in unexpected results such as data loss. Therefore, before executing any code, it will ask if you want to run it. Enter "y" to continue running or "n" to not execute. You can bypass this confirmation by specifying the command line parameter "-y" or setting "interpreter.auto_run=True" at startup.
Command Line Parameters:
-
--debug
Enable debug mode, which will output detailed running logs. -
--mode
Change the language model, as shown below:interpreter --model gpt-3.5-turbo interpreter --model claude-2 interpreter --model command-nightly -
--local
Run locally, using the Mistral-7B model by default. This model is 4.11GB in size, and the download speed depends on the actual internet speed! (Requires magic download)
To run Open Interpreter locally, the default language model is Mistral-7B, but you can also specify other language models, such as using the HuggingFace repo ID (tiiuae/falcon-180B).
interpreter --model tiiuae/falcon-180B
For example, using Win10 x64 system to use these models locally requires installing Visual Studio 2022 Community and NVIDIA CUDA Toolkit, and at least 30GB of disk space should be reserved!
Refer to the official documentation for detailed installation steps:
https://github.com/KillianLucas/open-interpreter/blob/main/docs/WINDOWS.md
https://github.com/KillianLucas/open-interpreter/blob/main/docs/GPU.md
https://github.com/KillianLucas/open-interpreter/blob/main/docs/MACOS.md
Compared to ChatGPT, Open-Interpreter is a groundbreaking open-source project. Its unique interactive and programming chat features allow users to have conversations with it using natural language in the terminal, providing more precise control over tasks and workflows.
Open-Interpreter not only solves many limitations of running OpenAI GPT-4 in a remote environment but also perfectly combines the powerful features of GPT-4 with the flexibility of a local development environment, providing developers with a wider range of functionality and flexibility. If Open-Interpreter is combined with powerful speech recognition capabilities, the future of artificial intelligence is full of even more infinite possibilities!