yes, but on this n-gram vs transformers; if you consider more general paradigm, self attention mechanism is basically a special form of a graph neural networks

yes, but on this n-gram vs transformers; if you consider more general paradigm, self attention mechanism is basically a special form of a graph neural networks [1]. [1] Bridging Graph Neural Networks and Large Language Models: A Survey and Unified Perspective https://infoscience.epfl.ch/server/api/core/bitstreams/7e6f86c2-04cf-4866-8d2e-5ad08272ca0b/content https://infoscience.epfl.ch/server/api/core/bitstreams/7e6f8...