Podcast explores FinneGAN and FinneganLM, two models trained only on Finnegans Wake
begusgasper · x · 2026-07-28
The post points to a podcast discussion about FinneGAN and FinneganLM — two models trained exclusively on Finnegans Wake.
- The episode focuses on how to explore the models’ latent spaces.
- It was part of the Berkeley symposium on the Modes of Reading, organized by Yael Segalovitz.
- The quoted podcast lineup includes talks by Nicholas Baer, Pardis Dabashi, David Bates, Hannes Bajohr, and the speaker themselves.
- This is a niche but interesting example of literary-text-only model training and analysis.
More from Fun
- AI is exposing people's mediocre taste, but everyone blames the AI — AndyMasley · 2026-07-28
- Reddit turns “GPT 5.6” into a Cerebras-style hype meme — Independent-Wind4462 · 2026-07-28
- Bittensor SN95 post turns “Actual Kimi” into a running AI joke — markjeffrey · 2026-07-28
- EmblemVault pitches no-code onchain apps, and the screenshot turns into a meme — adamamcbride · 2026-07-28
- Same Claude, twice the tokens: summarize a call, then re-analyze its own summary — JenniferHli · 2026-07-28
- Agent Arcade: An Open-Source Tool to Play Retro Games While AI Codes — DanWahlin · 2026-07-28