Skip to content
Better HN
Top
New
Best
Ask
Show
Jobs
Search
⌘K
Is Chain-of-Thought Reasoning of LLMs a Mirage? A Data Distribution Lens (2025) | Better HN
Is Chain-of-Thought Reasoning of LLMs a Mirage? A Data Distribution Lens (2025)
(opens in new tab)
(arxiv.org)
1 points
SomaticPirate
16d ago
1 comments
Share
1 comments
default
newest
oldest
SomaticPirate
OP
16d ago
A useful reminder that just because an LLM "appears" to be thinking and reasoning that it likely isn't. If it hasn't seen something similar or "in distribution" before, then it typically doesn't perform well.
j
/
k
navigate · click thread line to collapse
