ChatGPT is an excellent AI capable of generating all kinds of texts, but it has a major drawback. Being a web application, you need to be connected to the internet, and also, the servers can become saturated or hang at the most inopportune moment.
That’s where it comes into play GPT4Alla chatbot similar to ChatGPT, but with several important differences: it works locally on your PC, it’s completely free and does not require an internet connection. Let’s see how it works.
What is GPT4All Exactly?
GPT4All is an open source project created by Nomic AI, a US-based developer group that seeks to democratize access to Artificial Intelligence. GPT4All has several language models based on GPT-J and LLaMA.
LLaMA is the language model developed by Meta, and GPT-J, for its part, is based on GPT-2. Therefore, the foundations of Artificial Intelligence could be considered inferior to the current model of ChatGPT based on GPT-4.
However, everything we lose in raw power we gain in privacy, since all our data and queries are done locally on our PC, without going through another company’s external servers.
How to Install GPT4All on your PC
Unlike other chatbots that can be run from a local PC (such as the famous AutoGPTanother open source AI based on GPT-4), installing GPT4All is surprisingly easy.
- Go into gpt4all.iothe official website of the project.
- In the “Download Desktop Chat Client” section, click on “Windows Installer -> Windows Installer” to download the program installation file. Note: A version for macOS and another for Ubuntu are also available.
- Install the application following the instructions of the wizard (basically it consists of pressing “Next” until the installation is complete).
- Once the installation is complete, open the GPT4All application from the shortcut that you will see on the desktop.
- The first time you start the application you will have to download some of the language models in order to start working. The developer recommends downloading the model “gpt4all-j-v1.3-groovy”, which is the best commercial model they currently have, so we will click on the “Download” button that we will see right next to this model.
- If you want to download any other language model, you can also do so by clicking on the corresponding “Download” button. In this sense, it may be interesting to test the LLaMA 13b model trained by Nomic AI, although all this is completely optional.
- Once the download of the language model is complete, we will go to the main window of GPT4All. From the drop-down menu in the upper area we will select the language model we want to use. In our case, we will select gpt4all-j-v1.3-groovy.
- Now everything is ready. Now, you just have to place the cursor on “Send a message” (located in the lower area) to start chatting with the AI.
As it happens with ChatGPT, if we do not like any answer we can ask it to generate a new one by clicking on the button “Regenerate response” that we will see at the end of each conversation.
If we display the side menu on the left we will find our chat history, being able to create a new conversation from the button “new chat”. From here you can also update the application and download new language models.
From what we have been able to verify during the tests, the chatbot can speak in Spanish, although it is better handled in English. When we ask it questions that require elaborate answers it also spends a lot of time thinking, and at least for now, it seems to be behind ChatGPT in processing power.
However, GPT4All has the great advantage of being able to download different language models, so it is a matter of time before new, more powerful and efficient versions appear. In any case, a most interesting tool.