Want to create an interactive transcript for this episode?
Podcast: Arxiv Papers
Episode: OLMOTRACE: Tracing Language Model Outputs Back to Trillions of Training Tokens
Description: OLMOTRACE is a real-time system that traces language model outputs to their training data, enabling users to explore fact-checking, hallucination, and creativity in language models.https://arxiv.org/abs//2504.07096YouTube: https://www.youtube.com/@ArxivPapersTikTok: https://www.tiktok.com/@arxiv_papersApple Podcasts: https://podcasts.apple.com/us/podcast/arxiv-papers/id1692476016Spotify: https://podcasters.spotify.com/pod/show/arxiv-papers