MissPenguin
GitHub
commit
6be762d6a0
|
|
@ -3,7 +3,8 @@
|
|||
|
||||
复杂的模型有利于提高模型的性能，但也导致模型中存在一定冗余，模型裁剪通过移出网络模型中的子模型来减少这种冗余，达到减少模型计算复杂度，提高模型推理性能的目的。
|
||||
|
||||
本教程将介绍如何使用PaddleSlim量化PaddleOCR的模型。
|
||||
本教程将介绍如何使用飞桨模型压缩库PaddleSlim做PaddleOCR模型的压缩。
|
||||
PaddleSlim（项目链接：https://github.com/PaddlePaddle/PaddleSlim）集成了模型剪枝、量化（包括量化训练和离线量化）、蒸馏和神经网络搜索等多种业界常用且领先的模型压缩功能，如果您感兴趣，可以关注并了解。
|
||||
|
||||
在开始本教程之前，建议先了解
|
||||
1. [PaddleOCR模型的训练方法](../../../doc/doc_ch/quickstart.md)
|
||||
|
|
@ -33,8 +34,20 @@ python setup.py install
|
|||
|
||||
### 3. 敏感度分析训练
|
||||
|
||||
加载预训练模型后，通过对现有模型的每个网络层进行敏感度分析，了解各网络层冗余度，从而决定每个网络层的裁剪比例。
|
||||
加载预训练模型后，通过对现有模型的每个网络层进行敏感度分析，得到敏感度文件：sensitivities_0.data，可以通过PaddleSlim提供的[接口](https://github.com/PaddlePaddle/PaddleSlim/blob/develop/paddleslim/prune/sensitive.py#L221)加载文件，获得各网络层在不同裁剪比例下的精度损失。从而了解各网络层冗余度，决定每个网络层的裁剪比例。
|
||||
敏感度分析的具体细节见：[敏感度分析](https://github.com/PaddlePaddle/PaddleSlim/blob/develop/docs/zh_cn/tutorials/image_classification_sensitivity_analysis_tutorial.md)
|
||||
敏感度文件内容格式：
|
||||
sensitivities_0.data(Dict){
|
||||
'layer_weight_name_0': sens_of_each_ratio(Dict){'pruning_ratio_0': acc_loss, 'pruning_ratio_1': acc_loss}
|
||||
'layer_weight_name_1': sens_of_each_ratio(Dict){'pruning_ratio_0': acc_loss, 'pruning_ratio_1': acc_loss}
|
||||
}
|
||||
|
||||
例子：
|
||||
{
|
||||
'conv10_expand_weights': {0.1: 0.006509952684312718, 0.2: 0.01827734339798862, 0.3: 0.014528405644659832, 0.6: 0.06536008804270439, 0.8: 0.11798612250664964, 0.7: 0.12391408417493704, 0.4: 0.030615754498018757, 0.5: 0.047105205602406594}
|
||||
'conv10_linear_weights': {0.1: 0.05113190831455035, 0.2: 0.07705573833558801, 0.3: 0.12096721757739311, 0.6: 0.5135061352930738, 0.8: 0.7908166677143281, 0.7: 0.7272187676899062, 0.4: 0.1819252083008504, 0.5: 0.3728054727792405}
|
||||
}
|
||||
加载敏感度文件后会返回一个字典，字典中的keys为网络模型参数模型的名字，values为一个字典，里面保存了相应网络层的裁剪敏感度信息。例如在例子中，conv10_expand_weights所对应的网络层在裁掉10%的卷积核后模型性能相较原模型会下降0.65%，详细信息可见[PaddleSlim](https://github.com/PaddlePaddle/PaddleSlim/blob/develop/docs/zh_cn/algo/algo.md#2-%E5%8D%B7%E7%A7%AF%E6%A0%B8%E5%89%AA%E8%A3%81%E5%8E%9F%E7%90%86)
|
||||
|
||||
进入PaddleOCR根目录，通过以下命令对模型进行敏感度分析训练：
|
||||
```bash
|
||||
|
|
@ -42,7 +55,7 @@ python deploy/slim/prune/sensitivity_anal.py -c configs/det/det_mv3_db.yml -o Gl
|
|||
```
|
||||
|
||||
### 4. 模型裁剪训练
|
||||
裁剪时通过之前的敏感度分析文件决定每个网络层的裁剪比例。在具体实现时，为了尽可能多的保留从图像中提取的低阶特征，我们跳过了backbone中靠近输入的4个卷积层。同样，为了减少由于裁剪导致的模型性能损失，我们通过之前敏感度分析所获得的敏感度表，挑选出了一些冗余较少，对裁剪较为敏感的[网络层](https://github.com/PaddlePaddle/PaddleOCR/blob/develop/deploy/slim/prune/pruning_and_finetune.py#L41)，并在之后的裁剪过程中选择避开这些网络层。裁剪过后finetune的过程沿用OCR检测模型原始的训练策略。
|
||||
裁剪时通过之前的敏感度分析文件决定每个网络层的裁剪比例。在具体实现时，为了尽可能多的保留从图像中提取的低阶特征，我们跳过了backbone中靠近输入的4个卷积层。同样，为了减少由于裁剪导致的模型性能损失，我们通过之前敏感度分析所获得的敏感度表，人工挑选出了一些冗余较少，对裁剪较为敏感的[网络层](https://github.com/PaddlePaddle/PaddleOCR/blob/develop/deploy/slim/prune/pruning_and_finetune.py#L41)（指在较低的裁剪比例下就导致很高性能损失的网络层），并在之后的裁剪过程中选择避开这些网络层。裁剪过后finetune的过程沿用OCR检测模型原始的训练策略。
|
||||
|
||||
```bash
|
||||
python deploy/slim/prune/pruning_and_finetune.py -c configs/det/det_mv3_db.yml -o Global.pretrain_weights=./deploy/slim/prune/pretrain_models/det_mv3_db/best_accuracy Global.test_batch_size_per_card=1
|
||||
|
|
|
|||
|
|
@ -115,6 +115,7 @@ Compress results：
|
|||
Generally, a more complex model would achive better performance in the task, but it also leads to some redundancy in the model. Model Pruning is a technique that reduces this redundancy by removing the sub-models in the neural network model, so as to reduce model calculation complexity and improve model inference performance.
|
||||
|
||||
This example uses PaddleSlim provided[APIs of Pruning](https://paddlepaddle.github.io/PaddleSlim/api/prune_api/) to compress the OCR model.
|
||||
PaddleSlim (GitHub: https://github.com/PaddlePaddle/PaddleSlim), an open source library which integrates model pruning, quantization (including quantization training and offline quantization), distillation, neural network architecture search, and many other commonly used and leading model compression technique in the industry.
|
||||
|
||||
It is recommended that you could understand following pages before reading this example,：
|
||||
|
||||
|
|
@ -146,7 +147,20 @@ python setup.py install
|
|||
|
||||
## Pruning sensitivity analysis
|
||||
|
||||
After the pre-training model is loaded, sensitivity analysis is performed on each network layer of the model to understand the redundancy of each network layer, thereby determining the pruning ratio of each network layer. For specific details of sensitivity analysis, see：[Sensitivity analysis](https://github.com/PaddlePaddle/PaddleSlim/blob/develop/docs/zh_cn/tutorials/image_classification_sensitivity_analysis_tutorial.md)
|
||||
After the pre-training model is loaded, sensitivity analysis is performed on each network layer of the model to understand the redundancy of each network layer, and save a sensitivity file which named: sensitivities_0.data. After that, user could load the sensitivity file via the [methods provided by PaddleSlim](https://github.com/PaddlePaddle/PaddleSlim/blob/develop/paddleslim/prune/sensitive.py#L221) and determining the pruning ratio of each network layer automatically. For specific details of sensitivity analysis, see：[Sensitivity analysis](https://github.com/PaddlePaddle/PaddleSlim/blob/develop/docs/zh_cn/tutorials/image_classification_sensitivity_analysis_tutorial.md)
|
||||
The data format of sensitivity file：
|
||||
sensitivities_0.data(Dict){
|
||||
'layer_weight_name_0': sens_of_each_ratio(Dict){'pruning_ratio_0': acc_loss, 'pruning_ratio_1': acc_loss}
|
||||
'layer_weight_name_1': sens_of_each_ratio(Dict){'pruning_ratio_0': acc_loss, 'pruning_ratio_1': acc_loss}
|
||||
}
|
||||
|
||||
example：
|
||||
{
|
||||
'conv10_expand_weights': {0.1: 0.006509952684312718, 0.2: 0.01827734339798862, 0.3: 0.014528405644659832, 0.6: 0.06536008804270439, 0.8: 0.11798612250664964, 0.7: 0.12391408417493704, 0.4: 0.030615754498018757, 0.5: 0.047105205602406594}
|
||||
'conv10_linear_weights': {0.1: 0.05113190831455035, 0.2: 0.07705573833558801, 0.3: 0.12096721757739311, 0.6: 0.5135061352930738, 0.8: 0.7908166677143281, 0.7: 0.7272187676899062, 0.4: 0.1819252083008504, 0.5: 0.3728054727792405}
|
||||
}
|
||||
The function would return a dict after loading the sensitivity file. The keys of the dict are name of parameters in each layer. And the value of key is the information about pruning sensitivity of correspoding layer. In example, pruning 10% filter of the layer corresponding to conv10_expand_weights would lead to 0.65% degradation of model performance. The details could be seen at: [Sensitivity analysis](https://github.com/PaddlePaddle/PaddleSlim/blob/develop/docs/zh_cn/algo/algo.md#2-%E5%8D%B7%E7%A7%AF%E6%A0%B8%E5%89%AA%E8%A3%81%E5%8E%9F%E7%90%86)
|
||||
|
||||
|
||||
Enter the PaddleOCR root directory，perform sensitivity analysis on the model with the following command：
|
||||
|
||||
|
|
|
|||
|
|
@ -3,7 +3,8 @@
|
|||
复杂的模型有利于提高模型的性能，但也导致模型中存在一定冗余，模型量化将全精度缩减到定点数减少这种冗余，达到减少模型计算复杂度，提高模型推理性能的目的。
|
||||
模型量化可以在基本不损失模型的精度的情况下，将FP32精度的模型参数转换为Int8精度，减小模型参数大小并加速计算，使用量化后的模型在移动端等部署时更具备速度优势。
|
||||
|
||||
本教程将介绍如何使用PaddleSlim量化PaddleOCR的模型。
|
||||
本教程将介绍如何使用飞桨模型压缩库PaddleSlim做PaddleOCR模型的压缩。
|
||||
PaddleSlim（项目链接：https://github.com/PaddlePaddle/PaddleSlim）集成了模型剪枝、量化（包括量化训练和离线量化）、蒸馏和神经网络搜索等多种业界常用且领先的模型压缩功能，如果您感兴趣，可以关注并了解。
|
||||
|
||||
在开始本教程之前，建议先了解[PaddleOCR模型的训练方法](../../../doc/doc_ch/quickstart.md)以及[PaddleSlim](https://paddleslim.readthedocs.io/zh_CN/latest/index.html)
|
||||
|
||||
|
|
|
|||
|
|
@ -116,6 +116,7 @@ Compress results：
|
|||
Generally, a more complex model would achive better performance in the task, but it also leads to some redundancy in the model. Quantization is a technique that reduces this redundancyby reducing the full precision data to a fixed number, so as to reduce model calculation complexity and improve model inference performance.
|
||||
|
||||
This example uses PaddleSlim provided [APIs of Quantization](https://paddlepaddle.github.io/PaddleSlim/api/quantization_api/) to compress the OCR model.
|
||||
PaddleSlim (GitHub: https://github.com/PaddlePaddle/PaddleSlim), an open source library which integrates model pruning, quantization (including quantization training and offline quantization), distillation, neural network architecture search, and many other commonly used and leading model compression technique in the industry.
|
||||
|
||||
It is recommended that you could understand following pages before reading this example,：
|
||||
|
||||
|
|
|
|||
Loading…
Reference in New Issue