From bfeb10515d2f876105ec680d6a260708c2210265 Mon Sep 17 00:00:00 2001
From: Yuhao Wang <yuhaow97@gmail.com>
Date: Thu, 12 May 2022 21:06:25 +0800
Subject: [PATCH] update version

---
 README_CN.md                                 | 21 +++++---------------
 examples/template/runners/runner_template.py | 14 +++++++------
 examples/template/train.py                   | 17 ----------------
 requirements.txt                             |  1 +
 4 files changed, 14 insertions(+), 39 deletions(-)
 delete mode 100644 examples/template/train.py

diff --git a/README_CN.md b/README_CN.md
index cd9ea8a..cd69fbc 100644
--- a/README_CN.md
+++ b/README_CN.md
@@ -8,7 +8,7 @@
 
 ---
 
-Easytorch是一个基于PyTorch的开源神经网络训练框架，封装了PyTorch项目中常用的功能，帮助用户快速构建深度学习项目。
+EasyTorch是一个基于PyTorch的开源神经网络训练框架，封装了PyTorch项目中常用的功能，帮助用户快速构建深度学习项目。
 
 ## 功能亮点
 
@@ -35,9 +35,7 @@ python >= 3.6 （推荐 >= 3.7）
 
 ### PyTorch及CUDA
 
-[pytorch](https://pytorch.org/) >= 1.4（推荐 >= 1.7）
-
-[CUDA](https://developer.nvidia.com/zh-cn/cuda-toolkit) >= 9.2 （推荐 >= 11.0）
+[pytorch](https://pytorch.org/) >= 1.4（推荐 >= 1.7），如需使用CUDA，安装PyTorch时选择对应CUDA版本编译的包。
 
 注意：如需使用安培（Ampere）架构GPU，PyTorch版本需 >= 1.7且CUDA版本 >= 11.0。
 
@@ -51,23 +49,14 @@ pip install -r requirements.txt
 
 * [线性回归](examples/linear_regression)
 * [MNIST手写数字识别](examples/mnist)
+* [ImageNet图像分类](examples/imagenet)
 
 ## 开始使用
 
-### 新建项目文件夹并初始化git仓库
-
-```shell
-mkdir my_deeplearning_project
-cd my_deeplearning_project
-git init
-```
-
-### 添加easytorch子仓库
+### 安装EasyTorch
 
 ```shell
-git submodule add https://github.com/cnstark/easytorch.git
-git add .
-git commit -m "init by easytorch"
+pip install easy-torch
 ```
 
 ### 复制EasyTorch模板至工作目录
diff --git a/examples/template/runners/runner_template.py b/examples/template/runners/runner_template.py
index f7b205f..05a7a34 100644
--- a/examples/template/runners/runner_template.py
+++ b/examples/template/runners/runner_template.py
@@ -1,3 +1,5 @@
+from typing import Dict
+
 import torch
 from torch import nn
 from torch.utils.data import Dataset
@@ -8,10 +10,10 @@
 
 
 class RunnerTemplate(Runner):
-    def __init__(self, cfg: dict):
+    def __init__(self, cfg: Dict):
         super().__init__(cfg)
 
-    def init_training(self, cfg: dict):
+    def init_training(self, cfg: Dict):
         super().init_training(cfg)
 
         # init loss
@@ -22,22 +24,22 @@ def init_training(self, cfg: dict):
         # register meters by calling:
         # self.register_epoch_meter('train_loss', 'train', '{:.2f}')
 
-    def init_validation(self, cfg: dict):
+    def init_validation(self, cfg: Dict):
         super().init_validation(cfg)
 
         # self.register_epoch_meter('val_acc', 'val', '{:.2f}%')
 
     @staticmethod
-    def define_model(cfg: dict) -> nn.Module:
+    def define_model(cfg: Dict) -> nn.Module:
         return MODEL_DICT[cfg['MODEL']['NAME']](**cfg['MODEL'].get('PARAM', {}))
 
     @staticmethod
-    def build_train_dataset(cfg: dict) -> Dataset:
+    def build_train_dataset(cfg: Dict) -> Dataset:
         # return your train Dataset
         pass
 
     @staticmethod
-    def build_val_dataset(cfg: dict):
+    def build_val_dataset(cfg: Dict):
         # return your val Dataset
         pass
 
diff --git a/examples/template/train.py b/examples/template/train.py
deleted file mode 100644
index 948b802..0000000
--- a/examples/template/train.py
+++ /dev/null
@@ -1,17 +0,0 @@
-from argparse import ArgumentParser
-
-from easytorch.easytorch import launch_training
-
-
-def parse_args():
-    parser = ArgumentParser(description='Welcome to EasyTorch!')
-    parser.add_argument('-c', '--cfg', help='training config', required=True)
-    parser.add_argument('--gpus', help='visible gpus', type=str)
-    parser.add_argument('--tf32', help='enable tf32 on Ampere device', action='store_true')
-    return parser.parse_args()
-
-
-if __name__ == "__main__":
-    args = parse_args()
-
-    launch_training(args.cfg, args.gpus, args.tf32)
diff --git a/requirements.txt b/requirements.txt
index 42747b4..9650486 100644
--- a/requirements.txt
+++ b/requirements.txt
@@ -1,4 +1,5 @@
 torch>=1.7
+torchvision
 easydict
 tensorboard
 tqdm