Skip to content
Better HN
Top
Best
Ask
Show
New
Jobs
Search
⌘K
Prompt Repetition Improves Non-Reasoning LLMs
(opens in new tab)
(arxiv.org)
1 points
ajoy
4mo ago
1 comments
Save
Share
1 comments
1 comments · 1 top-level
top
newest
oldest
ajoy
OP
4mo ago
"When not using reasoning, repeating the input prompt improves performance for popular models (Gemini, GPT, Claude, and Deepseek) without increasing the number of generated tokens or latency."
j
/
k
navigate · click thread line to collapse
