It's not an unreasonable mental model to form to be honest – enormous weird blobs of vector floating point matrices and token embeddings are hardly an obvious way that this technology might work
Understanding How Large Language Models Actually Work
By
–
Leave a Reply