3 ms·
I can forgive the modeling being godawful jank (windows floating in the air, disconnected from the house). But I expected it to have a better understanding of t
by dundarious 2mo ago
I can forgive the modeling being godawful jank (windows floating in the air, disconnected from the house). But I expected it to have a better understanding of the text. Instead, we have Bilbo's "disappearance" interpreted as him magically transporting or cloaking, and similarly for his reappearance.
- kiwibyproxy 2mo agowhich to be fair, happens just a few paragraphs later :) I first watched without sound and thought "oh that's the birthday speech disappearance"
- anigbrowl 2mo agoBut this only seems wrong to you because you're familiar with the prior context. When there's only a single paragraph to work from, and it's a drily humorous text, why not employ comic literalism and lean into the perplexity with which his neighbors viewed him?
- mvdtnz 2mo agoThe LLM has access to the full text and every translation of it ever produced, both in its corpus and on the searchable web.
- lern_too_spel 2mo agoThe fact that a ring appears at the end proves that the model has context beyond what was in the prompt. The model clearly knows the story that the prompt was extracted from.
- emp17344 2mo agoThis is a rationalization. The simpler explanation is that the model failed to properly depict the passage. As the AI booster crowd would say, cope.
- shepherdjerred 2mo agoIsn't this almost certainly in the training set?
- 0x1ceb00da 2mo agoI haven't read the book or watched the movies and even then it looked weird as hell.