av/klmbr
View on GitHub ↗ ★ 89 TeX
klmbr - a prompt pre-processing technique to break through the barrier of entropy while generating text with LLMs
3 mentions on Hacker News by 1 person: 1 story, 2 comments · first 2024-09-19 · latest 2025-11-24
20082026
-
There was another technique "klmbr" a year or so ago: https://github.com/av/klmbr At a highest setting, It was unparseable by the LLMs at the time. Now, however, it looks like all major foundational models handle it easily, so some similar…
comment by everlier · 2025-11-24 · in Show HN: Stun LLMs with thousands of invisible Unicode characters -
Something like https://github.com/av/klmbr could also work
comment by everlier · 2025-07-07 · in Mercury: Ultra-fast language models based on diffusion