<?xml version="1.0" encoding="utf-8" ?><rss version="2.0"><channel><title><![CDATA[FortuneLegend的博客]]></title><description><![CDATA[]]></description><link>https://blog.csdn.net/FortuneLegend</link><language>zh-cn</language><generator>https://blog.csdn.net/</generator><copyright><![CDATA[Copyright &copy; FortuneLegend]]></copyright><item><title><![CDATA[LLM大模型基础认知]]></title><link>https://blog.csdn.net/FortuneLegend/article/details/145019706</link><guid>https://blog.csdn.net/FortuneLegend/article/details/145019706</guid><author>FortuneLegend</author><pubDate>Wed, 08 Jan 2025 22:59:13 +0800</pubDate><description><![CDATA[关于大模型LLM的总结与梳理。]]></description><category></category></item><item><title><![CDATA[Python时间序列分析库Kats的安装]]></title><link>https://blog.csdn.net/FortuneLegend/article/details/132694576</link><guid>https://blog.csdn.net/FortuneLegend/article/details/132694576</guid><author>FortuneLegend</author><pubDate>Tue, 05 Sep 2023 16:35:42 +0800</pubDate><description><![CDATA[Kats (Kits to Analyze Time Series) 是一个轻量级、易于使用、可扩展和通用的框架，用于在 Python 中执行时间序列分析。时间序列分析是数据科学和工程工作的重要组成部分，当然我们做运维的也主要是和时序数据打交道，尤其是做异常检测、时序预测的项目。Kats 旨在为单变量和多变量时间序列的技术提供一站式服务，包括：预测（Forecasting）异常和变化点检测（Anomaly and Change Point Detection）]]></description><category></category></item><item><title><![CDATA[解决python xlrd.biffh.XLRDError: Excel xlsx file； not supported]]></title><link>https://blog.csdn.net/FortuneLegend/article/details/130285618</link><guid>https://blog.csdn.net/FortuneLegend/article/details/130285618</guid><author>FortuneLegend</author><pubDate>Fri, 21 Apr 2023 12:31:23 +0800</pubDate><description><![CDATA[raise XLRDError(FILE_FORMAT_DESCRIPTIONS[file_format]+'; not supported')
xlrd.biffh.XLRDError: Excel xlsx file; not supported]]></description><category></category></item><item><title><![CDATA[Pytorch nn.LSTM 使用注意事项]]></title><link>https://blog.csdn.net/FortuneLegend/article/details/126306636</link><guid>https://blog.csdn.net/FortuneLegend/article/details/126306636</guid><author>FortuneLegend</author><pubDate>Fri, 12 Aug 2022 17:11:01 +0800</pubDate><description><![CDATA[1、单向lstm、batch_size在第一维
2、单向lstm、batch_size在第二维
3、双向lstm、batch_size在第一维
4、双向lstm、batch_size在第二维

通过上面的测试结论如下：
1、由于h0和hn，c0和cn的维度一定是相同的，同时h0和c0维度的第二维必须是batch_size；若是双向lstm，第一维是隐层层数的两倍；若是单向lstm，第一维是隐层层数。说明hn和cn是不受batch_first的影响的。
2、输出张量维度，若指定batch_first为T...]]></description><category></category></item><item><title><![CDATA[TensorFlow2.3找不到显卡GPU：Could not load dynamic library ‘libcusolver.so.11‘； dlerror: libcusolver.so.11]]></title><link>https://blog.csdn.net/FortuneLegend/article/details/126252038</link><guid>https://blog.csdn.net/FortuneLegend/article/details/126252038</guid><author>FortuneLegend</author><pubDate>Tue, 09 Aug 2022 17:31:14 +0800</pubDate><description><![CDATA[TensorFlow-GPU 2.3成功安装，找不到显卡，报错：
Could not load dynamic library ‘libcusolver.so.11’; dlerror: libcusolver.so.11: cannot open shared object file: No such file or directory
Cannot dlopen some GPU libraries. Please make sure the missing ...]]></description><category></category></item><item><title><![CDATA[win10 RTX30系列显卡 安装tensorflow-gpu 1.15]]></title><link>https://blog.csdn.net/FortuneLegend/article/details/125958288</link><guid>https://blog.csdn.net/FortuneLegend/article/details/125958288</guid><author>FortuneLegend</author><pubDate>Sun, 24 Jul 2022 13:35:15 +0800</pubDate><description><![CDATA[win10 30系列显卡 tensorflow-gpu 1.15 安装]]></description><category></category></item><item><title><![CDATA[tf.nn.dynamic_rnn的输出outputs和state含义]]></title><link>https://blog.csdn.net/FortuneLegend/article/details/103567202</link><guid>https://blog.csdn.net/FortuneLegend/article/details/103567202</guid><author>FortuneLegend</author><pubDate>Mon, 16 Dec 2019 18:02:38 +0800</pubDate><description><![CDATA[tf.nn.dynamic_rnn的输出outputs和state含义
一、&nbsp;tf.nn.dynamic_rnn的输出
tf.nn.dynamic_rnn(
    cell,
    inputs,
    sequence_length=None,
    initial_state=None,
    dtype=None,
    parallel_iterations=None...]]></description><category></category></item></channel></rss>