From 499acd62a393b6570c6ac2e6a662d3fd2b43035e Mon Sep 17 00:00:00 2001 From: wuyulong11 Date: Mon, 21 Aug 2023 11:59:23 +0800 Subject: [PATCH] =?UTF-8?q?=E3=80=90=E4=BF=AE=E6=94=B9=E4=BF=A1=E6=81=AF?= =?UTF-8?q?=E3=80=91=20=E3=80=90tbplugin=E3=80=91=E3=80=90=E8=B5=84?= =?UTF-8?q?=E6=96=99=E3=80=91=E3=80=90Readme=E3=80=91=E6=9B=B4=E6=96=B0?= =?UTF-8?q?=E8=B5=84=E6=96=99=E4=BF=A1=E6=81=AF=20=E3=80=90=E4=BF=AE?= =?UTF-8?q?=E6=94=B9=E4=BA=BA=E3=80=91=20wuyulong=2030031080?= MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit --- .../tensorboard-plugins/tb_plugin/README.md | 18 +++++++++--------- 1 file changed, 9 insertions(+), 9 deletions(-) diff --git a/plugins/tensorboard-plugins/tb_plugin/README.md b/plugins/tensorboard-plugins/tb_plugin/README.md index 91c11d7b800..2ede6cae217 100644 --- a/plugins/tensorboard-plugins/tb_plugin/README.md +++ b/plugins/tensorboard-plugins/tb_plugin/README.md @@ -7,7 +7,7 @@ 1. 插件方式安装 * 安装相关依赖: - pandas >= 1.0.0 ,tensorboard >= 1.15, != 2.1.0 + pandas >= 1.0.0 ,tensorboard >= 2.11.0 * 使用依赖: torch >= 1.8, torchvision >= 0.8 @@ -20,16 +20,16 @@ * 从仓库下载源码: - `git clone https://gitee.com/ascend/amtt` + `git clone https://gitee.com/ascend/att.git` -* 进入目录 `/tb_plugins/profiling/tb_plugin` 下. +* 进入目录 `/plugins/tensorboard_plugins/tb_plugin` 下. * 执行安装命令: `pip install .` * 构建whl包 - `python setup.py build_fe sdist bdist_wheel` \ - 注意: build_fe步骤需要setup yarn和Node.js + 注意: build_fe步骤需要安装yarn和Node.js环境 - `python setup.py sdist bdist_wheel` 在 `/tb_plugins/profiling/tb_plugin/dist` 目录下取出whl包,使用方式1进行安装 @@ -38,20 +38,20 @@ * 准备profiling数据 需要在读取的目录下放置指定格式的profiling数据。格式为包含3个层级的目录结构:runs层级为最外层目录(我们将一个完整的Profiling数据视为一个runs\ - 进行可视化处理),其子目录为worker_span层级(命名格式为{worker}_{span}),下一层级为规定命名的ASCEND_PROFILER_OUTPUT目录,此目录中包含\ + 进行可视化处理),其子目录为worker_span层级(命名格式为{worker}_{span}_ascend_pt),下一层级为规定命名的ASCEND_PROFILER_OUTPUT目录,此目录中包含\ 此插件加载展示的数据文件,如trace_view.json、kernel_details.csv、operator_details.csv等。目录结构如下: * E.g. there are 2 runs: run1, run2 \ `run1` \ - `--[worker1]_[span1]` \ + `--[worker1]_[span1]_ascend_pt` \ `----ASCEND_PROFILER_OUTPUT` \ `------trace_view.json` \ `------kernel_details.csv` \ - `--[worker2]_[span1]` \ + `--[worker2]_[span1]_ascend_pt` \ `----ASCEND_PROFILER_OUTPUT` \ `------trace_view.json` \ `------operator_details.csv` \ `run2` \ - `--[worker1]_[span1]` \ + `--[worker1]_[span1]_ascend_pt` \ `----ASCEND_PROFILER_OUTPUT` \ `------memory_record.csv` \ `------operator_memory.csv` @@ -66,7 +66,7 @@ 注意:确保默认端口6006对浏览器的主机打开。 - 如果需要切换端口号需要在尾部加上指定的端口号`--port=6007` + 如果需要切换端口号需要在尾部加上指定的端口号,如`--port=6007` `tensorboard --logdir=./samples --port=6007` -- Gitee