- [Continue Training on the Pretrained Model](#continue-training-on-the-pretrained-model)
- [Transfer Learning](#transfer-learning)
- [Description of Random Situation](#description-of-random-situation)
- [ModelZoo Homepage](#modelzoo-homepage)
# [GoogleNet Description](#contents)
GoogleNet, a 22 layers deep network, was proposed in 2014 and won the first place in the ImageNet Large-Scale Visual Recognition Challenge 2014 (ILSVRC14). GoogleNet, also called Inception v1, has significant improvement over ZFNet (The winner in 2013) and AlexNet (The winner in 2012), and has relatively lower error rate compared to VGGNet. Typically deeper deep learning network means larger number of parameters, which makes it more prone to overfitting. Furthermore, the increased network size leads to increased use of computational resources. To tackle these issues, GoogleNet adopts 1*1 convolution middle of the network to reduce dimension, and thus further reduce the computation. Global average pooling is used at the end of the network, instead of using fully connected layers. Another technique, called inception module, is to have different sizes of convolutions for the same input and stacking all the outputs.
[Paper](https://arxiv.org/abs/1409.4842): Christian Szegedy, Wei Liu, Yangqing Jia, Pierre Sermanet, Scott Reed, Dragomir Anguelov, Dumitru Erhan, Vincent Vanhoucke, Andrew Rabinovich. "Going deeper with convolutions." *Proceedings of the IEEE conference on computer vision and pattern recognition*. 2015.
Specifically, the GoogleNet contains numerous inception modules, which are connected together to go deeper. In general, an inception module with dimensionality reduction consists of **1×1 conv**, **3×3 conv**, **5×5 conv**, and **3×3 max pooling**, which are done altogether for the previous input, and stack together again at output. In our model architecture, the kernel size used in inception module is 3×3 instead of 5×5.
Note that you can run the scripts based on the dataset mentioned in original paper or widely used in relevant domain/network architecture. In the following sections, we will introduce how to run the scripts using the related dataset below.
The [mixed precision](https://www.mindspore.cn/tutorial/training/en/master/advanced_use/enable_mixed_precision.html) training method accelerates the deep learning neural network training process by using both the single-precision and half-precision data formats, and maintains the network precision achieved by the single-precision training at the same time. Mixed precision training can accelerate the computation process, reduce memory usage, and enable a larger model or batch size to be trained on specific hardware.
For FP16 operators, if the input data type is FP32, the backend of MindSpore will automatically handle it with reduced precision. Users could check the reduced-precision operators by enabling INFO log and then searching ‘reduce precision’.
# [Environment Requirements](#contents)
- Hardware（Ascend/GPU）
- Prepare hardware environment with Ascend or GPU processor. If you want to try Ascend , please send the [application form](https://obs-9be7.obs.cn-east-2.myhuaweicloud.com/file/other/Ascend%20Model%20Zoo%E4%BD%93%E9%AA%8C%E8%B5%84%E6%BA%90%E7%94%B3%E8%AF%B7%E8%A1%A8.docx) to ascend@huawei.com. Once approved, you can get the resources.
We use CIFAR-10 dataset by default. Your can also pass `$dataset_type` to the scripts so that select different datasets. For more details, please refer the specify script.
The above shell script will run distribute training in the background. You can view the results through the file `train_parallel[X]/log`. The loss value will be achieved as follows:
```
# grep "result: " train_parallel*/log
train_parallel0/log:epoch: 1 step: 48, loss is 1.4302931
train_parallel0/log:epcoh: 2 step: 48, loss is 1.4023874
...
train_parallel1/log:epoch: 1 step: 48, loss is 1.3458025
train_parallel1/log:epcoh: 2 step: 48, loss is 1.3729336
...
...
```
- running on GPU
```
sh scripts/run_train_gpu.sh 8 0,1,2,3,4,5,6,7
```
The above shell script will run distribute training in the background. You can view the results through the file `train/train.log`.
- evaluation on CIFAR-10 dataset when running on Ascend
Before running the command below, please check the checkpoint path used for evaluation. Please set the checkpoint path to be the absolute full path, e.g., "username/googlenet/train_googlenet_cifar10-125_390.ckpt".
```
python eval.py > eval.log 2>&1 &
OR
sh scripts/run_eval.sh
```
The above python command will run in the background. You can view the results through the file "eval.log". The accuracy of the test dataset will be as follows:
```
# grep "accuracy: " eval.log
accuracy: {'acc': 0.934}
```
Note that for evaluation after distributed training, please set the checkpoint_path to be the last saved checkpoint file such as "username/googlenet/train_parallel0/train_googlenet_cifar10-125_48.ckpt". The accuracy of the test dataset will be as follows:
- evaluation on CIFAR-10 dataset when running on GPU
Before running the command below, please check the checkpoint path used for evaluation. Please set the checkpoint path to be the absolute full path, e.g., "username/googlenet/train/ckpt_0/train_googlenet_cifar10-125_390.ckpt".
The above python command will run in the background. You can view the results through the file "eval.log". The accuracy of the test dataset will be as follows:
```
# grep "accuracy: " eval.log
accuracy: {'acc': 0.930}
```
OR,
```
sh scripts/run_eval_gpu.sh [CHECKPOINT_PATH]
```
The above python command will run in the background. You can view the results through the file "eval/eval.log". The accuracy of the test dataset will be as follows:
If you need to use the trained model to perform inference on multiple hardware platforms, such as GPU, Ascend 910 or Ascend 310, you can refer to this [Link](https://www.mindspore.cn/tutorial/training/en/master/advanced_use/migrate_3rd_scripts.html). Following the steps below, this is a simple example:
