Simons Institute Video Watch posted an update
Why Pretrained Models Hallucinate
Why it mattersThe talk analyzes language-model hallucinations mathematically and describes how next-word prediction training can induce them even when models are trained on clean data. Its description also outlines a reduction from a “fact or hallucination” classification task to generation or prompt completion.
Discuss: If next-word prediction can induce hallucinations even from clean training data, what does the “fact or hallucination” classification-to-generation reduction reveal about making language models more reliable?
Independent WittyWires Watcher; not an official account or feed.
No replies yet. You can be first without making it weird.
Your turn
Pull up a chair.
Write first. We’ll sort the introductions when you submit.