- [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)
# [SqueezeNet Description](#contents)
SqueezeNet is a lightweight and efficient CNN model proposed by Han et al., published in ICLR-2017. SqueezeNet has 50x fewer parameters than AlexNet, but the model performance (accuracy) is close to AlexNet.
These are examples of training SqueezeNet/SqueezeNet_Residual with CIFAR-10/ImageNet dataset in MindSpore. SqueezeNet_Residual adds residual operation on the basis of SqueezeNet, which can improve the accuracy of the model without increasing the amount of parameters.
[Paper](https://arxiv.org/abs/1602.07360): Forrest N. Iandola and Song Han and Matthew W. Moskewicz and Khalid Ashraf and William J. Dally and Kurt Keutzer. "SqueezeNet: AlexNet-level accuracy with 50x fewer parameters and <0.5MBmodelsize"
# [Model Architecture](#contents)
SqueezeNet is composed of fire modules. A fire module mainly includes two layers of convolution operations: one is the squeeze layer using a **1x1 convolution** kernel; the other is an expand layer using a mixture of **1x1** and **3x3 convolution** kernels.
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’.
- Prepare hardware environment with Ascend 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. Squeezenet training on GPU performs badly now, and it is still in research. See [squeezenet in research](https://gitee.com/mindspore/mindspore/tree/master/model_zoo/research/cv/squeezenet) to get up-to-date details.
"pretrain_epoch_size": 0, # epoch size that model has been trained before loading pretrained checkpoint, actual training epoch size is equal to epoch_size minus pretrain_epoch_size
"save_checkpoint": True, # whether save checkpoint or not
"save_checkpoint_epochs": 1, # the epoch interval between two checkpoints. By default, the last checkpoint will be saved after the last step
"keep_checkpoint_max": 10, # only keep the last keep_checkpoint_max checkpoint
"save_checkpoint_path": "./", # path to save checkpoint
"warmup_epochs": 5, # number of warmup epoch
"lr_decay_mode": "poly" # decay mode for generating learning rate
"pretrain_epoch_size": 0, # epoch size that model has been trained before loading pretrained checkpoint, actual training epoch size is equal to epoch_size minus pretrain_epoch_size
"save_checkpoint": True, # whether save checkpoint or not
"save_checkpoint_epochs": 1, # the epoch interval between two checkpoints. By default, the last checkpoint will be saved after the last step
"keep_checkpoint_max": 10, # only keep the last keep_checkpoint_max checkpoint
"save_checkpoint_path": "./", # path to save checkpoint
"warmup_epochs": 0, # number of warmup epoch
"lr_decay_mode": "poly" # decay mode for generating learning rate
"use_label_smooth": True, # label smooth
"label_smooth_factor": 0.1, # label smooth factor
"lr_init": 0, # initial learning rate
"lr_end": 0, # final learning rate
"lr_max": 0.01, # maximum learning rate
```
- config for SqueezeNet_Residual, CIFAR-10 dataset
"pretrain_epoch_size": 0, # epoch size that model has been trained before loading pretrained checkpoint, actual training epoch size is equal to epoch_size minus pretrain_epoch_size
"save_checkpoint": True, # whether save checkpoint or not
"save_checkpoint_epochs": 1, # the epoch interval between two checkpoints. By default, the last checkpoint will be saved after the last step
"keep_checkpoint_max": 10, # only keep the last keep_checkpoint_max checkpoint
"save_checkpoint_path": "./", # path to save checkpoint
"warmup_epochs": 5, # number of warmup epoch
"lr_decay_mode": "linear" # decay mode for generating learning rate
"lr_init": 0, # initial learning rate
"lr_end": 0, # final learning rate
"lr_max": 0.01, # maximum learning rate
```
- config for SqueezeNet_Residual, ImageNet dataset
"pretrain_epoch_size": 0, # epoch size that model has been trained before loading pretrained checkpoint, actual training epoch size is equal to epoch_size minus pretrain_epoch_size
"save_checkpoint": True, # whether save checkpoint or not
"save_checkpoint_epochs": 1, # the epoch interval between two checkpoints. By default, the last checkpoint will be saved after the last step
"keep_checkpoint_max": 10, # only keep the last keep_checkpoint_max checkpoint
"save_checkpoint_path": "./", # path to save checkpoint
"warmup_epochs": 0, # number of warmup epoch
"lr_decay_mode": "cosine" # decay mode for generating learning rate
"use_label_smooth": True, # label smooth
"label_smooth_factor": 0.1, # label smooth factor
"lr_init": 0, # initial learning rate
"lr_end": 0, # final learning rate
"lr_max": 0.01, # maximum learning rate
```
For more configuration details, please refer the script `config.py`.
Usage: sh scripts/run_distribute_train.sh [squeezenet|squeezenet_residual] [cifar10|imagenet] [RANK_TABLE_FILE] [DATASET_PATH] [PRETRAINED_CKPT_PATH](optional)
# standalone training
Usage: sh scripts/run_standalone_train.sh [squeezenet|squeezenet_residual] [cifar10|imagenet] [DEVICE_ID] [DATASET_PATH] [PRETRAINED_CKPT_PATH](optional)
```
For distributed training, a hccl configuration file with JSON format needs to be created in advance.
Please follow the instructions in the link [hccl_tools](https://gitee.com/mindspore/mindspore/tree/master/model_zoo/utils/hccl_tools).
Training result will be stored in the example path, whose folder name begins with "train" or "train_parallel". Under this, you can find checkpoint file together with result like the followings in log.
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:
