Gary Marcus

Rank 15 of 47

Score 88

Reposted by Gary Marcus

If you need a quantitative evaluation of the hallucination rate in frontier LLMs, here's one.

I'm currently working on a tutorial for training a structured information extraction from scientific texts.

To generate the labeled examples for supervised learning, I use Claude

6/24/2025, 7:42:00 PM

The statement is primarily informational and technical, focusing on the process of training a model for structured information extraction from scientific texts. It mentions the use of a specific tool, Claude, for generating labeled examples. The statement does not engage with public issues, debates, or societal concerns, and thus does not constitute public discourse.