This is a way to run open source models locally. You need the right hardware but it is a very efficient way to experiment with the newest models, fine tuning etc. ChatGPT uses massive model which are not practical to run on your own hardware. Privacy is also an issue for many people, particularly enterprise.
That said though, there's currently no practical way to reach ChatGPT's performance locally, but I have hope that will change eventually. Of course, even today you can reach quite good performance locally, and for a lot of people it's more than good enough for their needs.
Comments
This is a way to run open source models locally. You need the right hardware but it is a very efficient way to experiment with the newest models, fine tuning etc. ChatGPT uses massive model which are not practical to run on your own hardware. Privacy is also an issue for many people, particularly enterprise.
That said though, there's currently no practical way to reach ChatGPT's performance locally, but I have hope that will change eventually. Of course, even today you can reach quite good performance locally, and for a lot of people it's more than good enough for their needs.