Reactions from ranked influencers
2 posts
If you ask a frontier LLM a multi-hop reasoning question, e.g., "Who won the Nobel Prize for Chemistry in (1900 + Mozart's age when he died)?", it usually can't answer correctly immediately (no thinking) BUT if you ask the same question & append 300 dots, suddenly it can answer?
Think tokens (still) exist in 2026! Nice study. Doesn't even need to be LLM... https://www.isca-archive.org/interspeech_2024/yang24g_interspeech.html "Empty" tokens are not always what they appear to be http://www.catb.org/jargon/html/koans.html#id3141241
If you ask a frontier LLM a multi-hop reasoning question, e.g., "Who won the Nobel Prize for Chemistry in (1900 + Mozart's age when he died)?", it usually can't answer correctly immediately (no thinking) BUT if you ask the same question & append 300 dots, suddenly it can answer?