<?xml version="1.0" encoding="utf-8" ?><rss version="2.0"><channel><title><![CDATA[2401_82889064的博客]]></title><description><![CDATA[]]></description><link>https://blog.youkuaiyun.com/2401_82889064</link><language>zh-cn</language><generator>https://blog.youkuaiyun.com/</generator><copyright><![CDATA[Copyright &copy; 2401_82889064]]></copyright><item><title><![CDATA[通用唤醒词识别模型 - Wav2Vec2]]></title><link>https://blog.youkuaiyun.com/2401_82889064/article/details/158235098</link><guid>https://blog.youkuaiyun.com/2401_82889064/article/details/158235098</guid><author>2401_82889064</author><pubDate>Fri, 20 Feb 2026 12:04:13 +0800</pubDate><description><![CDATA[基于 Wav2Vec2 实现的通用唤醒词识别模型，最终测试效果基本满足需求，但特征样本对模型效果影响较大。]]></description><category></category></item><item><title><![CDATA[ 正/余弦位置编码 Sinusoidal Encoding]]></title><link>https://blog.youkuaiyun.com/2401_82889064/article/details/142696435</link><guid>https://blog.youkuaiyun.com/2401_82889064/article/details/142696435</guid><author>2401_82889064</author><pubDate>Mon, 16 Feb 2026 12:43:03 +0800</pubDate><description><![CDATA[其中，pos 为词汇在句子中的位置索引；i 为特征维度索引，2i 代表偶数位置，2i + 1 代表奇数位置；d 为特征维度，在传统 Transformer 模型中默认为 512。]]></description><category></category></item><item><title><![CDATA[实现扩散模型 Stable Diffusion - MNIST 数据集]]></title><link>https://blog.youkuaiyun.com/2401_82889064/article/details/158126092</link><guid>https://blog.youkuaiyun.com/2401_82889064/article/details/158126092</guid><author>2401_82889064</author><pubDate>Mon, 16 Feb 2026 12:01:23 +0800</pubDate><description><![CDATA[基于 CNN 和 Attention 实现的扩散模型 Stable Diffusion，此模型在手写数字数据集 MNIST 上训练了 40+ 轮，最终测试效果良好，且能够胜任简单的手写数字图片生成任务。]]></description><category></category></item><item><title><![CDATA[实现时间步调度器]]></title><link>https://blog.youkuaiyun.com/2401_82889064/article/details/158101966</link><guid>https://blog.youkuaiyun.com/2401_82889064/article/details/158101966</guid><author>2401_82889064</author><pubDate>Sun, 15 Feb 2026 20:33:37 +0800</pubDate><description><![CDATA[基于扩散模型 Diffusion 原理实现的时间步调度器，以此支撑后续的 Stable Diffusion 训练任务中的噪声前向扩散和反向采样。]]></description><category></category></item><item><title><![CDATA[实现变分自编码器 VAE- MNIST 数据集]]></title><link>https://blog.youkuaiyun.com/2401_82889064/article/details/158101550</link><guid>https://blog.youkuaiyun.com/2401_82889064/article/details/158101550</guid><author>2401_82889064</author><pubDate>Sun, 15 Feb 2026 20:01:56 +0800</pubDate><description><![CDATA[基于 CNN 和 Attention 实现的变分自编码器 VAE，此模型在手写数字数据集 MNIST 上训练了 60+ 轮，最终测试效果良好，且能够胜任后续的 Stable Diffusion 训练任务。]]></description><category></category></item><item><title><![CDATA[使用 BERT 实现意图理解和实体识别]]></title><link>https://blog.youkuaiyun.com/2401_82889064/article/details/151365647</link><guid>https://blog.youkuaiyun.com/2401_82889064/article/details/151365647</guid><author>2401_82889064</author><pubDate>Tue, 09 Sep 2025 16:50:59 +0800</pubDate><description><![CDATA[本文使用 Base BERT 模型结构，训练 200 余条关于 ”开关灯“ 语料，以实现基本的意图理解和实体识别。]]></description><category></category></item><item><title><![CDATA[手动实现 RNN 和 LSTM]]></title><link>https://blog.youkuaiyun.com/2401_82889064/article/details/149335693</link><guid>https://blog.youkuaiyun.com/2401_82889064/article/details/149335693</guid><author>2401_82889064</author><pubDate>Mon, 14 Jul 2025 17:39:53 +0800</pubDate><description><![CDATA[手动实现 RNN 和 LSTM]]></description><category></category></item><item><title><![CDATA[使用 Conformer 模型实现自动语音识别，ASR]]></title><link>https://blog.youkuaiyun.com/2401_82889064/article/details/148261847</link><guid>https://blog.youkuaiyun.com/2401_82889064/article/details/148261847</guid><author>2401_82889064</author><pubDate>Tue, 27 May 2025 17:41:51 +0800</pubDate><description><![CDATA[使用 Conformer 模型实现自动语音识别，ASR]]></description><category></category></item><item><title><![CDATA[手动实现 Transformer 模型]]></title><link>https://blog.youkuaiyun.com/2401_82889064/article/details/148030733</link><guid>https://blog.youkuaiyun.com/2401_82889064/article/details/148030733</guid><author>2401_82889064</author><pubDate>Sat, 17 May 2025 17:08:34 +0800</pubDate><description><![CDATA[本文使用 Pytorch 库手动实现了传统 Transformer 模型中的多头自注意力机制、残差连接和层归一化、前馈层、编码器、解码器等子模块，进而实现了对 Transformer 模型的构建。]]></description><category></category></item><item><title><![CDATA[中文 Tokenization&Tokenizer]]></title><link>https://blog.youkuaiyun.com/2401_82889064/article/details/146309259</link><guid>https://blog.youkuaiyun.com/2401_82889064/article/details/146309259</guid><author>2401_82889064</author><pubDate>Mon, 17 Mar 2025 10:50:32 +0800</pubDate><description><![CDATA[本文收集封装了几种常用的中文分词器。]]></description><category></category></item><item><title><![CDATA[中文句子分词]]></title><link>https://blog.youkuaiyun.com/2401_82889064/article/details/145298088</link><guid>https://blog.youkuaiyun.com/2401_82889064/article/details/145298088</guid><author>2401_82889064</author><pubDate>Wed, 22 Jan 2025 10:18:34 +0800</pubDate><description><![CDATA[基于 jieba 库的中文句子分词工具，侧重识别句子中的数字词汇。]]></description><category></category></item><item><title><![CDATA[基于 Transformer 的英译中翻译项目实战]]></title><link>https://blog.youkuaiyun.com/2401_82889064/article/details/144327859</link><guid>https://blog.youkuaiyun.com/2401_82889064/article/details/144327859</guid><author>2401_82889064</author><pubDate>Sun, 08 Dec 2024 18:18:44 +0800</pubDate><description><![CDATA[基于 Transformer 的英译中翻译项目实战（附完整代码）。]]></description><category></category></item><item><title><![CDATA[基于 Transformer 的新闻标题文本分类项目实战]]></title><link>https://blog.youkuaiyun.com/2401_82889064/article/details/144167487</link><guid>https://blog.youkuaiyun.com/2401_82889064/article/details/144167487</guid><author>2401_82889064</author><pubDate>Sun, 01 Dec 2024 19:34:54 +0800</pubDate><description><![CDATA[基于 Transformer 的新闻标题文本分类项目实战（附完整代码）。]]></description><category></category></item><item><title><![CDATA[填充句子到相同长度]]></title><link>https://blog.youkuaiyun.com/2401_82889064/article/details/143677291</link><guid>https://blog.youkuaiyun.com/2401_82889064/article/details/143677291</guid><author>2401_82889064</author><pubDate>Mon, 11 Nov 2024 14:58:29 +0800</pubDate><description><![CDATA[在进行词向量转换时（词嵌入），由于各个句子长短不一，无法直接转换成张量，这时就需要对较短的句子进行填充（padding），以满足张量输入对于形状的要求。]]></description><category></category></item><item><title><![CDATA[One-Hot 编码]]></title><link>https://blog.youkuaiyun.com/2401_82889064/article/details/142700199</link><guid>https://blog.youkuaiyun.com/2401_82889064/article/details/142700199</guid><author>2401_82889064</author><pubDate>Fri, 04 Oct 2024 10:38:39 +0800</pubDate><description><![CDATA[One-Hot 编码就是用于将离散的分类（文本）标签转换为二进制向量。]]></description><category></category></item><item><title><![CDATA[OpenCV-图像基础处理]]></title><link>https://blog.youkuaiyun.com/2401_82889064/article/details/136672852</link><guid>https://blog.youkuaiyun.com/2401_82889064/article/details/136672852</guid><author>2401_82889064</author><pubDate>Wed, 13 Mar 2024 10:37:05 +0800</pubDate><description><![CDATA[/]]></description><category></category></item><item><title><![CDATA[Python-sklearn-diabetes项目实战]]></title><link>https://blog.youkuaiyun.com/2401_82889064/article/details/136628591</link><guid>https://blog.youkuaiyun.com/2401_82889064/article/details/136628591</guid><author>2401_82889064</author><pubDate>Mon, 11 Mar 2024 20:16:10 +0800</pubDate><description><![CDATA[/]]></description><category></category></item><item><title><![CDATA[Python-sklearn-Optimizer]]></title><link>https://blog.youkuaiyun.com/2401_82889064/article/details/136601687</link><guid>https://blog.youkuaiyun.com/2401_82889064/article/details/136601687</guid><author>2401_82889064</author><pubDate>Sun, 10 Mar 2024 14:30:14 +0800</pubDate><description><![CDATA[/]]></description><category></category></item><item><title><![CDATA[Python-sklearn-datasets]]></title><link>https://blog.youkuaiyun.com/2401_82889064/article/details/136558073</link><guid>https://blog.youkuaiyun.com/2401_82889064/article/details/136558073</guid><author>2401_82889064</author><pubDate>Fri, 08 Mar 2024 11:58:21 +0800</pubDate><description><![CDATA[/]]></description><category></category></item><item><title><![CDATA[Python-sklearn-LossFunction]]></title><link>https://blog.youkuaiyun.com/2401_82889064/article/details/136536422</link><guid>https://blog.youkuaiyun.com/2401_82889064/article/details/136536422</guid><author>2401_82889064</author><pubDate>Thu, 07 Mar 2024 15:53:33 +0800</pubDate><description><![CDATA[/]]></description><category></category></item></channel></rss>