<?xml version="1.0" encoding="utf-8" standalone="yes"?><rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom"><channel><title>Huggingface on [ MECANIK DEV ]</title><link>https://mecanik.dev/zh-tw/tags/huggingface/</link><description>Recent content in Huggingface on [ MECANIK DEV ]</description><generator>Hugo -- gohugo.io</generator><language>zh-Hant</language><copyright>版權所有 © 2020-{year} [ MECANIK DEV ]。 版權所有。</copyright><lastBuildDate>Thu, 07 Aug 2025 20:00:00 +0100</lastBuildDate><atom:link href="https://mecanik.dev/zh-tw/tags/huggingface/index.xml" rel="self" type="application/rss+xml"/><item><title>Tiny BPE Trainer – 一個快速且輕量的 C++ BPE 訓練器</title><link>https://mecanik.dev/zh-tw/posts/tiny-bpe-trainer-a-fast-and-lightweight-bpe-trainer-in-c++/</link><pubDate>Thu, 07 Aug 2025 20:00:00 +0100</pubDate><guid>https://mecanik.dev/zh-tw/posts/tiny-bpe-trainer-a-fast-and-lightweight-bpe-trainer-in-c++/</guid><description>Tiny BPE Trainer 介紹如今，從 GPT 到 RoBERTa，大多數現代 NLP 模型都依賴於使用位元組對編碼 (BPE) 的子詞標記化。但是，如果您想用純 C++ 訓練自己的詞彙表，該怎麼辦</description></item></channel></rss>