---
name: "slmthread"
section: "1"
platform: "linux"
package: "sunpinyin"
version: "3.0.0rc2-3"
release: "Arch Linux 2026-11-01"
platforms: ["linux"]
url: "https://new.carta.tech/slmthread"
---

# slmthread(1)

threads the language model

## NAME

slmthread - threads the language model

## SYNOPSIS

**slmthread** *primitive\_slm* *threaded\_slm*

## DESCRIPTION

**slmthread** add back-off-state for each slm node in the *primitive\_slm*.  Also it compresses 32-bit float into 16 bit representation. These processing  speeds up the looking up.

The *primitive\_slm* is always generated by **slmprune**. And  the *threaded\_slm* can be used to feed **slmseg** as a reference to segment  Chinese text.

## AUTHOR

Originally written by Phill.Zhang \<phill.zhang@sun.com\>. Currently maintained by Kov.Chai \<tchaikov@gmail.com\>.

## SEE ALSO

[**slmprune**(1)](/slmprune/1.md), [**slmseg**(1)](/slmseg/1.md).