To a layman, no it’s not and should not be viewed as the same, either. One is a MASSIVE stack of precomputed matrices (NOT a neural network), the other a stack of predetermined algorithms with some random() thrown in.
They are all actually and literally completely different in form, function, and computing requirement.
The precomputed matrices are definitely a neural network. GPT literally stands for generative pre-trained transformer, a transformer being a well known neural network architecture.
No, a neural network is explicitly a type of architecture that mimics neurons. A transformer is just that: a methematical operation.
You’re making the mistake of calling a backhoe a “tractor” just because it’s also used for moving dirt around… Yes, both are heavy machinery, but that doesn’t magically make the terms interchangeable.
To a layman, no it’s not and should not be viewed as the same, either. One is a MASSIVE stack of precomputed matrices (NOT a neural network), the other a stack of predetermined algorithms with some random() thrown in.
They are all actually and literally completely different in form, function, and computing requirement.
The precomputed matrices are definitely a neural network. GPT literally stands for generative pre-trained transformer, a transformer being a well known neural network architecture.
No, a neural network is explicitly a type of architecture that mimics neurons. A transformer is just that: a methematical operation.
You’re making the mistake of calling a backhoe a “tractor” just because it’s also used for moving dirt around… Yes, both are heavy machinery, but that doesn’t magically make the terms interchangeable.
https://arxiv.org/pdf/1706.03762
Like I said. You’re mixing your terms. Linking a paper does not prove you understand what you’re talking about.
Then explain to me how the neural networks behind LLMs are not mimicking neurons. Because that’s the basic idea behind all of deep learning.