update

MNIST_Example/config.py

@@ -0,0 +1,33 @@
+# Copyright 2020 Huawei Technologies Co., Ltd
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+# http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+# ============================================================================
+"""
+network config setting, will be used in train.py
+"""
+
+from easydict import EasyDict as edict
+
+mnist_cfg = edict({
+    'num_classes': 10,
+    'lr': 0.01,
+    'momentum': 0.9,
+    'epoch_size': 10,
+    'batch_size': 32,
+    'buffer_size': 1000,
+    'image_height': 32,
+    'image_width': 32,
+    'save_checkpoint_steps': 1875,
+    'keep_checkpoint_max': 10,
+    'air_name': "lenet",
+})

MNIST_Example/dataset.py

@@ -0,0 +1,60 @@
+# Copyright 2020 Huawei Technologies Co., Ltd
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+# http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+# ============================================================================
+"""
+Produce the dataset
+"""
+
+import mindspore.dataset as ds
+import mindspore.dataset.vision.c_transforms as CV
+import mindspore.dataset.transforms.c_transforms as C
+from mindspore.dataset.vision import Inter
+from mindspore.common import dtype as mstype
+
+
+def create_dataset(data_path, batch_size=32, repeat_size=1,
+                   num_parallel_workers=1):
+    """
+    create dataset for train or test
+    """
+    # define dataset
+    mnist_ds = ds.MnistDataset(data_path)
+
+    resize_height, resize_width = 32, 32
+    rescale = 1.0 / 255.0
+    shift = 0.0
+    rescale_nml = 1 / 0.3081
+    shift_nml = -1 * 0.1307 / 0.3081
+
+    # define map operations
+    resize_op = CV.Resize((resize_height, resize_width), interpolation=Inter.LINEAR)  # Bilinear mode
+    rescale_nml_op = CV.Rescale(rescale_nml, shift_nml)
+    rescale_op = CV.Rescale(rescale, shift)
+    hwc2chw_op = CV.HWC2CHW()
+    type_cast_op = C.TypeCast(mstype.int32)
+
+    # apply map operations on images
+    mnist_ds = mnist_ds.map(operations=type_cast_op, input_columns="label", num_parallel_workers=num_parallel_workers)
+    mnist_ds = mnist_ds.map(operations=resize_op, input_columns="image", num_parallel_workers=num_parallel_workers)
+    mnist_ds = mnist_ds.map(operations=rescale_op, input_columns="image", num_parallel_workers=num_parallel_workers)
+    mnist_ds = mnist_ds.map(operations=rescale_nml_op, input_columns="image", num_parallel_workers=num_parallel_workers)
+    mnist_ds = mnist_ds.map(operations=hwc2chw_op, input_columns="image", num_parallel_workers=num_parallel_workers)
+
+    # apply DatasetOps
+    buffer_size = 10000
+    mnist_ds = mnist_ds.shuffle(buffer_size=buffer_size)  # 10000 as in LeNet train script
+    mnist_ds = mnist_ds.batch(batch_size, drop_remainder=True)
+    mnist_ds = mnist_ds.repeat(repeat_size)
+
+    return mnist_ds

MNIST_Example/eval.py

@@ -0,0 +1,64 @@
+# Copyright 2020 Huawei Technologies Co., Ltd
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+# http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+# ============================================================================
+"""
+######################## eval lenet example ########################
+eval lenet according to model file:
+python eval.py --data_path /YourDataPath --ckpt_path Your.ckpt
+"""
+
+# python eval.py --device_target=CPU --ckpt_path=./ckpt/checkpoint_lenet-1_1875.ckpt > log_eval.txt 2>&1
+# python eval.py --device_target=CPU --ckpt_path=./ckpt/checkpoint_lenet-10_1875.ckpt > log_eval2.txt 2>&1
+
+import os
+import argparse
+import mindspore.nn as nn
+from mindspore import context
+from mindspore.train.serialization import load_checkpoint, load_param_into_net
+from mindspore.train import Model
+from mindspore.nn.metrics import Accuracy
+from dataset import create_dataset
+from config import mnist_cfg as cfg
+from lenet import LeNet5
+
+if __name__ == "__main__":
+    parser = argparse.ArgumentParser(description='MindSpore Lenet Example')
+    parser.add_argument('--device_target', type=str, default="Ascend", choices=['Ascend', 'GPU', 'CPU'],
+                        help='device where the code will be implemented (default: Ascend)')
+    parser.add_argument('--data_path', type=str, default="./Data",
+                        help='path where the dataset is saved')
+    parser.add_argument('--ckpt_path', type=str, default="", help='if mode is test, must provide\
+                        path where the trained ckpt file')
+
+    args = parser.parse_args()
+
+    context.set_context(mode=context.GRAPH_MODE, device_target=args.device_target)
+
+    network = LeNet5(cfg.num_classes)
+    net_loss = nn.SoftmaxCrossEntropyWithLogits(sparse=True, reduction="mean")
+    repeat_size = cfg.epoch_size
+    net_opt = nn.Momentum(network.trainable_params(), cfg.lr, cfg.momentum)
+    model = Model(network, net_loss, net_opt, metrics={"Accuracy": Accuracy()})
+
+    print("============== Starting Testing ==============")
+    param_dict = load_checkpoint(args.ckpt_path)
+    load_param_into_net(network, param_dict)
+    ds_eval = create_dataset(os.path.join(args.data_path, "test"),
+                             cfg.batch_size,
+                             1)
+    if ds_eval.get_dataset_size() == 0:
+        raise ValueError("Please check dataset size > 0 and batch_size <= dataset size")
+
+    acc = model.eval(ds_eval)
+    print("============== {} ==============".format(acc))

MNIST_Example/inference.py

@@ -0,0 +1,155 @@
+"""
+######################## inference lenet example ########################
+inference lenet according to model file
+"""
+
+"""
+######################## 推理环境使用说明 ########################
+2、在推理环境中，需要将数据集从obs拷贝到推理镜像中，推理完以后，需要将输出的结果拷贝到obs.
+(1)将数据集从obs拷贝到推理镜像中：
+    obs_data_url = args.data_url
+        args.data_url = '/home/work/user-job-dir/data/'
+        if not os.path.exists(args.data_url):
+            os.mkdir(args.data_url)
+    try:     
+        mox.file.copy_parallel(obs_data_url, args.data_url)
+        print("Successfully Download {} to {}".format(obs_data_url,
+                                                        args.data_url))
+    except Exception as e:
+        print('moxing download {} to {} failed: '.format(
+            obs_data_url, args.data_url) + str(e))
+
+(2)将模型文件从obs拷贝到推理镜像中：
+    obs_ckpt_url = args.ckpt_url
+    args.ckpt_url = '/home/work/user-job-dir/checkpoint.ckpt'
+    try:
+        mox.file.copy(obs_ckpt_url, args.ckpt_url)
+        print("Successfully Download {} to {}".format(obs_ckpt_url,
+                                                      args.ckpt_url))
+    except Exception as e:
+        print('moxing download {} to {} failed: '.format(
+            obs_ckpt_url, args.ckpt_url) + str(e))      
+
+(3)将输出的模型拷贝回obs：
+    obs_result_url = args.result_url
+        args.result_url = '/home/work/user-job-dir/result/'
+        if not os.path.exists(args.result_url):
+            os.mkdir(args.result_url)
+    try:
+        mox.file.copy_parallel(args.result_url, obs_result_url)
+        print("Successfully Upload {} to {}".format(args.result_url, obs_result_url))
+    except Exception as e:
+        print('moxing upload {} to {} failed: '.format(args.result_url, obs_result_url) + str(e))
+
+"""
+import os
+import argparse
+import moxing as mox
+import mindspore.nn as nn
+from mindspore import context
+from mindspore.train.serialization import load_checkpoint, load_param_into_net
+from mindspore.train import Model
+from mindspore.nn.metrics import Accuracy
+from mindspore import Tensor
+import numpy as np
+from glob import glob
+from dataset import create_dataset
+from config import mnist_cfg as cfg
+from lenet import LeNet5
+
+if __name__ == "__main__":
+    parser = argparse.ArgumentParser(description='MindSpore Lenet Example')
+    parser.add_argument('--device_target', type=str, default="Ascend", choices=['Ascend', 'GPU', 'CPU'],
+                        help='device where the code will be implemented (default: Ascend)')
+    parser.add_argument('--data_url',
+                    type=str,
+                    default="./Data",
+                    help='path where the dataset is saved')      
+    parser.add_argument('--ckpt_url',
+                    help='model to save/load',
+                    default='./ckpt_url')  
+    parser.add_argument('--result_url',
+                    help='result folder to save/load',
+                    default='./result')               
+
+    args = parser.parse_args()
+    
+    #将数据集从obs拷贝到推理镜像中：
+    obs_data_url = args.data_url
+    args.data_url = '/home/work/user-job-dir/data/'
+    if not os.path.exists(args.data_url):
+        os.mkdir(args.data_url)
+    try:    
+        mox.file.copy_parallel(obs_data_url, args.data_url)
+        print("Successfully Download {} to {}".format(obs_data_url,
+                                                      args.data_url))
+    except Exception as e:
+        print('moxing download {} to {} failed: '.format(
+            obs_data_url, args.data_url) + str(e))
+
+    #对文件夹进行操作，请使用mox.file.copy_parallel。如果拷贝一个文件。请使用mox.file.copy对文件操作
+    #将模型文件从obs拷贝到推理镜像中：
+    obs_ckpt_url = args.ckpt_url
+    args.ckpt_url = '/home/work/user-job-dir/checkpoint.ckpt'
+    try:
+        mox.file.copy(obs_ckpt_url, args.ckpt_url)
+        print("Successfully Download {} to {}".format(obs_ckpt_url,
+                                                      args.ckpt_url))
+    except Exception as e:
+        print('moxing download {} to {} failed: '.format(
+            obs_ckpt_url, args.ckpt_url) + str(e))
+
+    #设置输出路径result_url
+    obs_result_url = args.result_url
+    args.result_url = '/home/work/user-job-dir/result/'
+    if not os.path.exists(args.result_url):
+        os.mkdir(args.result_url)
+
+    args.dataset_path = args.data_url
+    args.save_checkpoint_path = args.ckpt_url
+
+    context.set_context(mode=context.GRAPH_MODE, device_target=args.device_target)
+
+    network = LeNet5(cfg.num_classes)
+    net_loss = nn.SoftmaxCrossEntropyWithLogits(sparse=True, reduction="mean")
+    repeat_size = cfg.epoch_size
+    net_opt = nn.Momentum(network.trainable_params(), cfg.lr, cfg.momentum)
+    model = Model(network, net_loss, net_opt, metrics={"Accuracy": Accuracy()})
+
+    print("============== Starting Testing ==============")
+    args.load_ckpt_url = os.path.join(args.save_checkpoint_path)
+    print("args.load_ckpt_url  is：{}", args.load_ckpt_url )
+    param_dict = load_checkpoint(args.load_ckpt_url )
+    load_param_into_net(network, param_dict)
+    # 定义测试数据集，batch_size设置为1，则取出一张图片
+    ds_test = create_dataset(os.path.join(args.dataset_path, "test"), batch_size=1).create_dict_iterator()
+    data = next(ds_test)
+
+    # images为测试图片，labels为测试图片的实际分类
+    images = data["image"].asnumpy()
+    labels = data["label"].asnumpy()
+    print('Tensor:', Tensor(data['image']))
+
+    # 使用函数model.predict预测image对应分类
+    output = model.predict(Tensor(data['image']))
+    predicted = np.argmax(output.asnumpy(), axis=1)
+    pred = np.argmax(output.asnumpy(), axis=1)
+    print('predicted:', predicted)
+    print('pred:', pred)
+
+    # 输出预测分类与实际分类,并输出到result_url
+    print(f'Predicted: "{predicted[0]}", Actual: "{labels[0]}"')
+    filename = 'result.txt'
+    file_path = os.path.join(args.result_url, filename)
+    with open(file_path, 'a+') as file:
+            file.write("    {}: {:.2f} \n".format("Predicted", predicted[0]))
+
+    # Upload results to obs
+    ######################## 将输出的结果拷贝到obs（固定写法） ########################   
+    # 把训练后的模型数据从本地的运行环境拷贝回obs，在启智平台相对应的训练任务中会提供下载
+    try:
+        mox.file.copy_parallel(args.result_url, obs_result_url)
+        print("Successfully Upload {} to {}".format(args.result_url, obs_result_url))
+    except Exception as e:
+        print('moxing upload {} to {} failed: '.format(args.result_url, obs_result_url) + str(e))
+    ######################## 将输出的模型拷贝到obs ########################  

MNIST_Example/lenet.py

@@ -0,0 +1,60 @@
+# Copyright 2020 Huawei Technologies Co., Ltd
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+# http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+# ============================================================================
+"""LeNet."""
+import mindspore.nn as nn
+from mindspore.common.initializer import Normal
+
+
+class LeNet5(nn.Cell):
+    """
+    Lenet network
+
+    Args:
+        num_class (int): Number of classes. Default: 10.
+        num_channel (int): Number of channels. Default: 1.
+
+    Returns:
+        Tensor, output tensor
+    Examples:
+        >>> LeNet(num_class=10)
+
+    """
+    def __init__(self, num_class=10, num_channel=1, include_top=True):
+        super(LeNet5, self).__init__()
+        self.conv1 = nn.Conv2d(num_channel, 6, 5, pad_mode='valid')
+        self.conv2 = nn.Conv2d(6, 16, 5, pad_mode='valid')
+        self.relu = nn.ReLU()
+        self.max_pool2d = nn.MaxPool2d(kernel_size=2, stride=2)
+        self.include_top = include_top
+        if self.include_top:
+            self.flatten = nn.Flatten()
+            self.fc1 = nn.Dense(16 * 5 * 5, 120, weight_init=Normal(0.02))
+            self.fc2 = nn.Dense(120, 84, weight_init=Normal(0.02))
+            self.fc3 = nn.Dense(84, num_class, weight_init=Normal(0.02))
+
+    def construct(self, x):
+        x = self.conv1(x)
+        x = self.relu(x)
+        x = self.max_pool2d(x)
+        x = self.conv2(x)
+        x = self.relu(x)
+        x = self.max_pool2d(x)
+        if not self.include_top:
+            return x
+        x = self.flatten(x)
+        x = self.relu(self.fc1(x))
+        x = self.relu(self.fc2(x))
+        x = self.fc3(x)
+        return x

MNIST_Example/train.py

@@ -0,0 +1,161 @@
+"""
+######################## train lenet example ########################
+train lenet and get network model files(.ckpt) 
+"""
+"""
+######################## 训练环境使用说明 ########################
+假设已经使用Ascend NPU调试环境调试完代码，欲将调试环境的代码迁移到训练环境进行训练，需要做以下工作：
+1、调试环境的镜像和训练环境的镜像是两个不同的镜像，所处的运行目录不一致，需要将data_url和train_url的路径进行变换
+在调试环境中：
+args.data_url = '/home/ma-user/work/data/'      //数据集位置
+args.train_url = '/home/ma-user/work/model/'    //训练输出的模型位置
+在训练环境变换为：
+args.data_url = '/home/work/user-job-dir/inputs/data/'
+args.train_url = '/home/work/user-job-dir/outputs/model/'
+2、在训练环境中，需要将数据集从obs拷贝到训练镜像中，训练完以后，需要将输出的模型拷贝到obs.
+将数据集从obs拷贝到训练镜像中：
+try:
+    mox.file.copy_parallel(obs_data_url, args.data_url)
+    print("Successfully Download {} to {}".format(obs_data_url,
+                                                    args.data_url))
+except Exception as e:
+    print('moxing download {} to {} failed: '.format(
+        obs_data_url, args.data_url) + str(e))
+
+将输出的模型拷贝到obs：
+try:
+        mox.file.copy_parallel(args.train_url, obs_train_url)
+        print("Successfully Upload {} to {}".format(args.train_url,
+                                                    obs_train_url))
+    except Exception as e:
+        print('moxing upload {} to {} failed: '.format(args.train_url,
+                                                       obs_train_url) + str(e))
+"""
+
+import os
+import argparse
+import moxing as mox
+from config import mnist_cfg as cfg
+from dataset import create_dataset
+from lenet import LeNet5
+import mindspore.nn as nn
+from mindspore import context
+from mindspore.train.callback import ModelCheckpoint, CheckpointConfig, LossMonitor, TimeMonitor
+from mindspore.train import Model
+from mindspore.nn.metrics import Accuracy
+from mindspore.common import set_seed
+
+parser = argparse.ArgumentParser(description='MindSpore Lenet Example')
+
+# define 2 parameters for running on modelArts
+# data_url,train_url是固定用于在modelarts上训练的参数，表示数据集的路径和输出模型的路径
+parser.add_argument('--data_url',
+                    help='path to training/inference dataset folder',
+                    default='./data')
+
+parser.add_argument('--train_url',
+                    help='model folder to save/load',
+                    default='./model')
+
+parser.add_argument(
+    '--device_target',
+    type=str,
+    default="Ascend",
+    choices=['Ascend', 'GPU', 'CPU'],
+    help='device where the code will be implemented (default: Ascend)')
+
+#用户可自定义的参数，dataset_path在示例中指向data_url，save_checkpoint_path指向train_url，在添加超参数时这两个参数可不添加，modelarts已经默认使用data_url和train_url
+parser.add_argument('--dataset_path',
+                    type=str,
+                    default="./Data",
+                    help='path where the dataset is saved')
+parser.add_argument('--save_checkpoint_path',
+                    type=str,
+                    default="./ckpt",
+                    help='if is test, must provide\
+                    path where the trained ckpt file')
+parser.add_argument('--epoch_size',
+                    type=int,
+                    default=5,
+                    help='Training epochs.')
+
+set_seed(1)
+
+if __name__ == "__main__":
+
+    args = parser.parse_args()
+
+ ######################## 将数据集从obs拷贝到训练镜像中 （固定写法）########################   
+    # 在训练环境中定义data_url和train_url，并把数据从obs拷贝到相应的固定路径
+    obs_data_url = args.data_url
+    args.data_url = '/home/work/user-job-dir/inputs/data/'
+    obs_train_url = args.train_url
+    args.train_url = '/home/work/user-job-dir/outputs/model/'
+    try:
+        mox.file.copy_parallel(obs_data_url, args.data_url)
+        print("Successfully Download {} to {}".format(obs_data_url,
+                                                      args.data_url))
+    except Exception as e:
+        print('moxing download {} to {} failed: '.format(
+            obs_data_url, args.data_url) + str(e))
+######################## 将数据集从obs拷贝到训练镜像中 ########################
+    
+    #将dataset_path指向data_url，save_checkpoint_path指向train_url
+    args.dataset_path = args.data_url
+    args.save_checkpoint_path = args.train_url
+
+    context.set_context(mode=context.GRAPH_MODE,
+                        device_target=args.device_target)
+    #创建数据集
+    ds_train = create_dataset(os.path.join(args.dataset_path, "train"),
+                              cfg.batch_size)
+    if ds_train.get_dataset_size() == 0:
+        raise ValueError(
+            "Please check dataset size > 0 and batch_size <= dataset size")
+    #创建网络
+    network = LeNet5(cfg.num_classes)
+    net_loss = nn.SoftmaxCrossEntropyWithLogits(sparse=True, reduction="mean")
+    net_opt = nn.Momentum(network.trainable_params(), cfg.lr, cfg.momentum)
+    time_cb = TimeMonitor(data_size=ds_train.get_dataset_size())
+
+    if args.device_target != "Ascend":
+        model = Model(network,
+                      net_loss,
+                      net_opt,
+                      metrics={"accuracy": Accuracy()})
+    else:
+        model = Model(network,
+                      net_loss,
+                      net_opt,
+                      metrics={"accuracy": Accuracy()},
+                      amp_level="O2")
+
+    config_ck = CheckpointConfig(
+        save_checkpoint_steps=cfg.save_checkpoint_steps,
+        keep_checkpoint_max=cfg.keep_checkpoint_max)
+    #定义模型输出路径
+    ckpoint_cb = ModelCheckpoint(prefix="checkpoint_lenet",
+                                 directory=args.save_checkpoint_path,
+                                 config=config_ck)
+    #开始训练
+    print("============== Starting Training ==============")
+    epoch_size = cfg['epoch_size']
+    if (args.epoch_size):
+        epoch_size = args.epoch_size
+        print('epoch_size is: ', epoch_size)
+
+    model.train(epoch_size,
+                ds_train,
+                callbacks=[time_cb, ckpoint_cb,
+                           LossMonitor()])
+
+ ######################## 将输出的模型拷贝到obs（固定写法） ########################   
+ # 把训练后的模型数据从本地的运行环境拷贝回obs，在启智平台相对应的训练任务中会提供下载
+    try:
+        mox.file.copy_parallel(args.train_url, obs_train_url)
+        print("Successfully Upload {} to {}".format(args.train_url,
+                                                    obs_train_url))
+    except Exception as e:
+        print('moxing upload {} to {} failed: '.format(args.train_url,
+                                                       obs_train_url) + str(e))
+ ######################## 将输出的模型拷贝到obs ########################   

第14课-如何进行模型训练.md

@@ -0,0 +1,242 @@
+# 第13课-CIFAR-10图像识别项目实战
+
+## 如何基于云脑1和云脑2调试任务
+
+hi，大家好，经历了前面多轮课程，相比大家已经对平台的各部分功能差不多都熟悉了，恭喜大家已经不再是当初那个小白啦~
+
+今天我们通过平台演示一个猫狗图像识别的小项目，将前面所学到的知识串联起来加以巩固，开始向启智社区达人迈进！
+
+猫狗识别是图像分类的经典案例之一，大家感兴趣的可以自行百度，此次就不过多介绍了。
+
+本节课主要演示如何在云脑1和云脑2进行任务调试，大致内容如下：
+
+> 代码和数据集上传
+
+> 分别进入云脑1和云脑2环境调试任务
+
+> 进入云脑2环境训练任务
+
+话不多说，接下来进入猫狗图像识别的项目实操
+
+### 一、新建项目并上传代码和数据集
+
+#### 1. 新建项目
+
+在【个人中心】页面，点击页面右侧【项目列表】旁边的【+】，开始创建项目，填好相关信息，勾选✔初始化存储库，点击【创建项目】
+
+<div align="center">
+<img src= img/L14-1.gif width=100%>
+</div>
+<br>
+
+#### 2. 上传代码文件
+
+项目创建后，在【代码】页面点击【上传文件】进入文件上传页面，在【添加目录】处输入目录名称（即文件夹名称），直接从本地将代码文件拖入上传框
+
+（代码文件可自行从网上下载，或在OpenI_Learning项目进行下载，代码文件名【case_Cats and Dogs】）
+
+<div align="center">
+<img src= img/L14-2.gif width=100%>
+</div>
+<br>
+
+此处上传了两个文件，case1代码表示从云脑1进行调试，case2代码表示从云脑2进行调试
+
+#### 3. 上传数据集
+
+猫狗数据集可从[Kaggle](https://www.kaggle.com/datasets)下载，也可以从OpenI_Learning项目下载，数据集名称【cifar-10-batches-py.zip】。
+
+下载好数据集后，进入【数据集】页面，点击【CPU/GPU】选择云脑1，将数据集直接拖入上传框内，【Ascend NPU】为云脑2，分别上传数据集（云脑1和云脑2数据集不共用，所以需选择相应的环境上传数据集）
+
+<div align="center">
+<img src= img/L14-3.png width=80%> <br><img src= img/L14-4.png width=80%>
+</div>
+<br>
+
+### 二、进入云脑环境调试任务
+
+云脑1和云脑2都可以调试任务，这里我们依次进行演示。
+
+#### 1. 云脑1调试任务
+
+##### a. 新建调试任务
+
+在【个人中心】页面点击【云脑】，选择CPU/GPU（即云脑1环境），点击右侧按钮【新建调试任务】
+
+<div align="center">
+<img src= img/L14-5.png width=100%>
+</div>
+<br>
+
+进入新建任务页面，更改或默认任务名称，镜像要求基于Python 3.6+ 和PyTorch 1.0+的环境，可自行配置或选择已经配置好的其他环境，数据集选择对应已上传的数据集（可直接输入数据集名称关键词），其他选项默认即可
+
+<div align="center">
+<img src= img/L14-6.png width=100%>
+</div>
+<br>
+
+任务创建好后，等待十秒左右待状态从CREATING变为RUNNING，点击页面的【调试】进入调试环境
+
+<div align="center">
+<img src= img/L14-7.png width=100%>
+</div>
+<br>
+
+##### b. 进入云脑1环境
+
+点击调试后，会跳转到云脑的调试环境
+
+<div align="center">
+<img src= img/L14-8.png width=100%>
+</div>
+<br>
+
+##### c. 云脑1环境调试任务
+
+此时，大家可以进行自定义调试和运行代码了。
+
+> #ls
+
+> #cd /code/case1 (相应代码放在/code下，相应数据集放在/dataset下)
+
+> #python main.py
+
+<div align="center">
+<img src= img/L14-9.png width=100%>
+</div>
+<br>
+
+由于每个epoch要训练5万张，在1W张里面进行测试，运行过程会有点儿漫长，但云脑1的调试已经跑通，静等运行结束即可。
+
+<div align="center">
+<img src= img/L14-10.png width=100%>
+</div>
+<br>
+
+
+---
+
+
+接下来我们演示从云脑2进行调试
+
+#### 2. 云脑2调试任务
+
+##### a. 新建调试任务
+
+同样地，在【云脑】页面下来【CPU/GPU】选项框，选择【Ascend NPU】，点击右侧按钮【新建调试任务】
+
+<div align="center">
+<img src= img/L14-11.png width=100%>
+</div>
+<br>
+
+进入新建任务页面，选择相应数据集，其他为默认即可，点击【新建任务】
+
+<div align="center">
+<img src= img/L14-12.png width=100%>
+</div>
+<br>
+
+调试任务创建完成后，稍等十秒，待状态变为RUNNING后，点击【调试】
+
+<div align="center">
+<img src= img/L14-13.png width=100%>
+</div>
+<br>
+
+##### b. 进入云脑2环境
+
+点击调试后，进入调试界面，勾选相应数据集，点击【Sync OBS】进行同步后会弹出确认框，勾选✔【YES】即可，待页面提示成功，关闭即可
+
+<div align="center">
+<img src= img/L14-3.gif width=100%>
+</div>
+<br>
+
+点击页面右侧的【New】，创建调试用的Notebook，选择【MindSpore】环境
+
+<div align="center">
+<img src= img/L14-14.png width=100%>
+</div>
+<br>
+
+此时，页面跳转到调试环境
+
+<div align="center">
+<img src= img/L14-15.png width=100%>
+</div>
+<br>
+
+##### c. 云脑2环境调试任务
+
+我们先将代码克隆过来
+
+> 在【个人中心】的【代码】界面，点击右侧按钮【复制链接】
+
+> 回到调试界面，输入 !git clone 后粘贴链接
+
+> 点击【Run】运行代码，提示克隆成功
+
+<div align="center">
+<img src= img/L14-4.gif width=100%>
+</div>
+<br>
+
+接下来解压数据集，在代码中输入 !unzip + 文件名称，点击运行
+
+<div align="center">
+<img src= img/L14-16.png width=100%>
+</div>
+<br>
+
+最后，运行代码 !python OpenI_test/case2/train.py --dataset_path ./cifar-10-batches-bin/
+
+<div align="center">
+<img src= img/L14-17.png width=100%>
+</div>
+<br>
+
+### 三、云脑2训练任务
+
+#### 1. 新建训练任务
+
+在云脑页面，选择【Ascend NPU】，点击【训练任务】，再点击【新建训练任务】
+
+<div align="center">
+<img src= img/L14-18.png width=100%>
+</div>
+<br>
+
+在弹出的任务创建页面中，选择数据集【cifar.zip】，指定文件中输入“case2/train.py”，其他默认即可，点击【新建任务】
+
+<div align="center">
+<img src= img/L14-19.png width=100%>
+</div>
+<br>
+
+#### 2. 训练任务
+
+点击了新建任务之后，待状态由INIT变为RUNNING之后，模型将开始启动训练，运行时长大约几分钟左右
+
+<div align="center">
+<img src= img/L14-20.png width=100%>
+</div>
+<br>
+
+#### 3. 模型下载
+
+当任务状态变为“COMPLETED”，表示任务训练成功并已结束。
+
+<div align="center">
+<img src= img/L14-21.png width=100%>
+</div>
+<br>
+
+点击操作栏的【模型下载】可下载模型
+
+<div align="center">
+<img src= img/L14-22.png width=100%>
+</div>
+<br>
+
+至此，我们已经在云脑环境（云脑1和云脑2）都顺利调试和训练了模型，通过这次项目实战，相信童鞋们又进一步熟悉了平台的强大功能，后面在实践自己项目时就可以得心应手了~