How would I compile an entirely custom-made GPT model and run it locally on an Intel-based Mac? Most of the sources that I've consulted have yielded unsatisfactory methods, so I'm hoping that someone here knows.
comments
you could probably use llama or something
by building, do you mean training?
I just need an entirely local GPT client and a way to make a model that runs entirely on provided data. I'm not familiar with the technical side of this.
you might want to look into fine-tuning? https://www.datacamp.com/tutorial/fine-tuning-large-language-models
I’m not looking to fine-tune an existing GPT; I'm looking to make an entirely new model that runs on entirely new data.
that might be a bit hard, you'd probably need a lot of money and data.
People don’t do it like that
Plus, if you insist, your GPU (or worse your CPU) would be absolutely tortured on the training process
People don’t do it like that
Plus, if you insist, your GPU (or worse your CPU) would be absolutely tortured on the training process