<?xml version="1.0" encoding="UTF-8"?><rss version="2.0"><channel><title>ForeverYoung</title><description>Driving into the distance — notes on ML, CUDA, Python, and beyond.</description><link>https://foreveryounggithub.github.io/</link><item><title>手绘风格的数据可视化</title><link>https://foreveryounggithub.github.io/zh/posts/sketchy_rendering/</link><guid isPermaLink="true">https://foreveryounggithub.github.io/zh/posts/sketchy_rendering/</guid><description>罗列常用手绘风格数据可视化工具及适用图表，包括 rough.js、draw.io、matplotlib xkcd、chart.xkcd 和 cutecharts。</description><pubDate>Tue, 19 Apr 2022 00:00:00 GMT</pubDate></item><item><title>深度学习模型大小与模型推理速度的探讨</title><link>https://foreveryounggithub.github.io/zh/posts/model_speed_vs_size/</link><guid isPermaLink="true">https://foreveryounggithub.github.io/zh/posts/model_speed_vs_size/</guid><description>探讨深度学习模型的参数量、计算量、访存量等指标对推理速度的影响，并给出针对不同硬件的网络设计建议。</description><pubDate>Fri, 04 Mar 2022 13:27:08 GMT</pubDate></item><item><title>图像局部区域（Patch）的描述符学习策略</title><link>https://foreveryounggithub.github.io/zh/posts/patch_based_local_descriptor/</link><guid isPermaLink="true">https://foreveryounggithub.github.io/zh/posts/patch_based_local_descriptor/</guid><description>综述基于学习的图像局部区域描述符（local descriptor）的数据处理与训练策略，关注不同论文在 patch 学习方向上的异同。</description><pubDate>Sun, 27 Feb 2022 17:12:44 GMT</pubDate></item><item><title>pybind: 为cpp/cuda代码提供python接口</title><link>https://foreveryounggithub.github.io/zh/posts/python_cpp_extension/</link><guid isPermaLink="true">https://foreveryounggithub.github.io/zh/posts/python_cpp_extension/</guid><description>通过 pybind11 为 C++/CUDA 代码提供 Python 绑定，实现 numpy/torch 张量与 Eigen/cv::Mat 的零拷贝转换。</description><pubDate>Sun, 20 Sep 2020 00:00:00 GMT</pubDate></item><item><title>Numba: 简单装饰器加速python代码</title><link>https://foreveryounggithub.github.io/zh/posts/numba_python/</link><guid isPermaLink="true">https://foreveryounggithub.github.io/zh/posts/numba_python/</guid><description>介绍如何通过 numba 库为 Python 函数添加装饰器的方式进行即时编译加速，简单易用、灵活高效。</description><pubDate>Sat, 04 Jul 2020 00:00:00 GMT</pubDate></item><item><title>通过NPP加速TensorRT部署时图片数据预处理</title><link>https://foreveryounggithub.github.io/zh/posts/trt_preproc_npp/</link><guid isPermaLink="true">https://foreveryounggithub.github.io/zh/posts/trt_preproc_npp/</guid><description>介绍如何通过 NVIDIA NPP 库加速 TensorRT 推理部署时的图片数据预处理（uint8→float32）过程。</description><pubDate>Wed, 17 Jun 2020 00:00:00 GMT</pubDate></item><item><title>Numba: 通过python快速学习cuda编程</title><link>https://foreveryounggithub.github.io/zh/posts/numba_cuda/</link><guid isPermaLink="true">https://foreveryounggithub.github.io/zh/posts/numba_cuda/</guid><description>通过 Python 的 numba 库快速上手 CUDA 多线程高并发编程，降低 CUDA 入门门槛。</description><pubDate>Wed, 10 Jun 2020 00:00:00 GMT</pubDate></item></channel></rss>