xushengfeng/lime
An LLM-powered Pinyin Chinese input method using Qwen3-0.6B for intelligent typing suggestions.
Collecting fresh signals — velocity needs a few days of history.
collecting data…
star history
This project builds a Pinyin input method engine powered by a small language model (Qwen3-0.6B). It leverages the LLM’s next-token prediction capabilities, using user-provided Pinyin to guide the sampling of candidate characters and words. The server backend runs in Deno.js, receiving keystrokes and returning text suggestions, making typing with LLM inference comparable in speed to traditional input engines.