My point is that the LLM is the result of the training. It is not what is doing the training. Like a program is the result of coding. A program doesn’t itself perform the coding. An LLM being created by a training process is a different mode of operation than using the resulting LLM to generate content.
So, you are trying to manipulate the words here, and you are trying to say that training is not learning. You are talking out of your ass, dude. Just read about how neural networks work, especially on unsupervised training.
