To estimate $P(\mathbf{w})$, the LD requires a probabilistic model of the text generator, which we refer to as the language model. For most artificial tasks, the language modeling problem is quite simple. Often the language is specified by …etc., from which word sequences can be derived. Fig. 7 is the model of the artificial Raleigh language which has been used in some of our experiments. The output alphabet is the 250-word vocabulary of the language. For diagrammatic convenie…
Raleigh
other · 2 mentions across 1 reading
In this course
Raleigh appears in these excerpts only as a reference to a constructed artificial language used in machine learning experiments, serving as a controlled test case for language modeling. The authors deploy "the artificial Raleigh language" as a simplified system with a 250-word vocabulary to evaluate probabilistic models of text generation, which is foundational to how language models estimate word sequences. This appears to be a passing technical reference rather than a figure of conceptual weight in the course's broader arguments about AI and society.
Mentioned in 1 reading