Source Package: karukan (0.1.0-1.5)
Japanese input method for Fcitx5 powered by neural kana-kanji conversion
Karukan is a Japanese Input Method Engine for Linux that uses neural kana-kanji conversion powered by llama.cpp. It provides a fcitx5 addon for romaji-to-hiragana conversion and intelligent kana-kanji conversion using small neural network models (auto-downloaded on first use). Features: - Romaji-to-hiragana conversion with 200+ rules - Neural kana-kanji conversion via llama.cpp (GGUF models) - Adaptive model selection for low-latency input - Learning cache for user conversion history - System dictionary with double-array trie
Package information
- Homepage
- [https://github.com/togatoga/karukan]
- License
- MIT OR Apache-2.0
- Mantenedor del paquete downstream
- Desconocido: no consta en la fuente indexada
The following binary packages are built from this source package:
- fcitx5-karukan
Japanese input method for Fcitx5 powered by neural kana-kanji conversion
- karukan-tools
CLI tools for Karukan (dictionary builder, server, benchmark)
