Class Business
Mid-course assessments of practicums (in your Canvas inbox)
- Readings for Next Classes
- For Thursday
- Chris Woodford, “How Neural Networks Work: A Simple Introduction” (2023)
- Cornellius Yudha Wijaya, “Large Language Models Explained in 3 Levels of Difficulty” (2024)
- Optional: For a slightly deeper, more detailed explanation of neural networks as part of an explanation of large language models, see Andreas Stöffelbauer, “How Large Language Models Work: From Zero to ChatGPT” (2023).
- Minh Hua and Rita Raley, “Playing With Unicorns: AI Dungeon and Citizen NLP” (2020) [If you can’t finish this article by this class, try to read at least the first two sections; and then finish the rest of the article for the next class.]
- For next Tuesday (May 21)
- Emily M. Bender et al., “On the Dangers of Stochastic Parrots: Can Language Models Be Too Big?” (2021)
- [Optional: If you are interested in the controversy and background behind this article, see Tim Simonite, “What Really Happened When Google Ousted Timnit Gebru” (2021)]
- Alan Liu, “What Is Good Writing in the Age of ChatGPT?” (Commencement Address for the UCSB English Dept., 2023).
- Emily M. Bender et al., “On the Dangers of Stochastic Parrots: Can Language Models Be Too Big?” (2021)
Due Tuesday, May 21: Large Language Models & Text-to-Image Large Models Exercise
- For Thursday
The 20th-C. “Linguistic Turn”: Structural Linguistics & Distributional Semantics (continued from last class)
Word Embedding
Aurelie Herbelot, “Distributional Semantics: A Light Introduction”
Once we have such measures, we can build a simulation of how the words in a particular language relate to each other. This is done in a so-called ‘semantic space’ which, mathematically, corresponds to a vector space…. Here is a very simplified example, where I define the words ‘dragon’, ‘lion’ and ‘democracy’ with respect to only two dimensions: ‘dangerous’ and ‘political’.
Idea of embedding words in semantic space: Analogy of force-directed graph layout
Luis Serrano, “What Are Word and Sentence Embeddings?” (2023)
“Infinite Craft” by Neal Agarwal
Tensorflow Embedding Projector (Google A)
Saptarashmi Bandyopadhyay et al., “Word Embedding Demo: Tutorial” (2022) — Interactive demo
Practicum 5: Word Embedding Exercise
Nika Mavrody, Laura B. McGrath, Nichole Nomura, and Alexander Sherman, “Voice” (2021) — “Abstract” and pp. 155-164.
Ultimately, we show that voice, style, and genre operate in a unified vernacular critical system, that voice (along with genre) is a subcategory of style, and that voice consists of the parts of style not otherwise captured by genre.
“Voice”
“Style”
“Style – Voice = Genre”
“Style – Genre = Voice”
Ryan Heuser, “Word Vectors in the Eighteenth Century” (conference proceedings abstract) (2017)
- James Lee, James, Blaine Greteman, Jason Lee, and David Eichmann, “Linked Reading: Digital Historicism and Early Modern Discourses of Race around Shakespeare’s Othello” (2018)
“Word2Vec Model of “Complexion” (EEBO Corpus from 1500 to 1623)” — Fig. 14 in James Lee, Blaine Greteman, Jason Lee, and David Eichmann. “Linked Reading: Digital Historicism and Early Modern Discourses of Race around Shakespeare’s Othello.” Journal of Cultural Analytics, 2018.
- Fabian Offert, “Intuition and Epistemology of High-Dimensional Vector Space” (2021)
- Jacques Derrida, Margins of Philosophy (1982)
- Catachresis “concerns first the violent, forced, abusive inscription of a sign, the imposition of a sign upon a meaning which did not yet have its own proper sign in language” (255)
- J. Hillis Miller, “The Figure in the Carpet” (1980):
- “Catachresis is the name for that procedure whereby [Henry] James uses all the realistic detail of his procedure as a novelist to name in figure, abusive transfer, something else for which there is no literal name and therefore, within the convention of referentiality which the story as a realistic novel accepts, no existence” (111)
