mirror of
https://github.com/PaddlePaddle/FastDeploy.git
synced 2025-10-05 16:48:03 +08:00
[Quantization] Improve the usage of FastDeploy tools. (#660)
* Add PaddleOCR Support * Add PaddleOCR Support * Add PaddleOCRv3 Support * Add PaddleOCRv3 Support * Update README.md * Update README.md * Update README.md * Update README.md * Add PaddleOCRv3 Support * Add PaddleOCRv3 Supports * Add PaddleOCRv3 Suport * Fix Rec diff * Remove useless functions * Remove useless comments * Add PaddleOCRv2 Support * Add PaddleOCRv3 & PaddleOCRv2 Support * remove useless parameters * Add utils of sorting det boxes * Fix code naming convention * Fix code naming convention * Fix code naming convention * Fix bug in the Classify process * Imporve OCR Readme * Fix diff in Cls model * Update Model Download Link in Readme * Fix diff in PPOCRv2 * Improve OCR readme * Imporve OCR readme * Improve OCR readme * Improve OCR readme * Imporve OCR readme * Improve OCR readme * Fix conflict * Add readme for OCRResult * Improve OCR readme * Add OCRResult readme * Improve OCR readme * Improve OCR readme * Add Model Quantization Demo * Fix Model Quantization Readme * Fix Model Quantization Readme * Add the function to do PTQ quantization * Improve quant tools readme * Improve quant tool readme * Improve quant tool readme * Add PaddleInference-GPU for OCR Rec model * Add QAT method to fastdeploy-quantization tool * Remove examples/slim for now * Move configs folder * Add Quantization Support for Classification Model * Imporve ways of importing preprocess * Upload YOLO Benchmark on readme * Upload YOLO Benchmark on readme * Upload YOLO Benchmark on readme * Improve Quantization configs and readme * Add support for multi-inputs model * Add backends and params file for YOLOv7 * Add quantized model deployment support for YOLO series * Fix YOLOv5 quantize readme * Fix YOLO quantize readme * Fix YOLO quantize readme * Improve quantize YOLO readme * Improve quantize YOLO readme * Improve quantize YOLO readme * Improve quantize YOLO readme * Improve quantize YOLO readme * Fix bug, change Fronted to ModelFormat * Change Fronted to ModelFormat * Add examples to deploy quantized paddleclas models * Fix readme * Add quantize Readme * Add quantize Readme * Add quantize Readme * Modify readme of quantization tools * Modify readme of quantization tools * Improve quantization tools readme * Improve quantization readme * Improve PaddleClas quantized model deployment readme * Add PPYOLOE-l quantized deployment examples * Improve quantization tools readme * Improve Quantize Readme * Fix conflicts * Fix conflicts * improve readme * Improve quantization tools and readme * Improve quantization tools and readme * Add quantized deployment examples for PaddleSeg model * Fix cpp readme * Fix memory leak of reader_wrapper function * Fix model file name in PaddleClas quantization examples * Update Runtime and E2E benchmark * Update Runtime and E2E benchmark * Rename quantization tools to auto compression tools * Remove PPYOLOE data when deployed on MKLDNN * Fix readme * Support PPYOLOE with OR without NMS and update readme * Update Readme * Update configs and readme * Update configs and readme * Add Paddle-TensorRT backend in quantized model deploy examples * Support PPYOLOE+ series * Add reused_input_tensors for PPYOLOE * Improve fastdeploy tools usage * improve fastdeploy tool * Improve fastdeploy auto compression tool * Improve fastdeploy auto compression tool * Improve fastdeploy auto compression tool * Improve fastdeploy auto compression tool * Improve fastdeploy auto compression tool * remove modify * Improve fastdeploy auto compression tool * Improve fastdeploy auto compression tool * Improve fastdeploy auto compression tool * Improve fastdeploy auto compression tool * Improve fastdeploy auto compression tool * Remove extra requirements for fd-auto-compress package * Imporve fastdeploy-tools package * Install fastdeploy-tools package when build fastdeploy-python * Imporve quantization readme
This commit is contained in:
yunyaoXYY
@@ -3,5 +3,5 @@ requests
|
||||
tqdm
|
||||
numpy
|
||||
opencv-python
|
||||
fd-auto-compress>=0.0.1
|
||||
fastdeploy-tools
|
||||
pyyaml
|
||||
|
||||
@@ -22,14 +22,11 @@ git clone https://github.com/PaddlePaddle/PaddleSlim.git & cd PaddleSlim
|
||||
python setup.py install
|
||||
```
|
||||
|
||||
3.安装fd-auto-compress一键模型自动化压缩工具
|
||||
3.安装fastdeploy-tools工具包
|
||||
```bash
|
||||
# 通过pip安装fd-auto-compress.
|
||||
# 通过pip安装fastdeploy-tools. 此工具包目前支持模型一键自动化压缩和模型转换的功能.
|
||||
# FastDeploy的python包已包含此工具, 不需重复安装.
|
||||
pip install fd-auto-compress==0.0.1
|
||||
|
||||
# 在当前目录执行以下命令
|
||||
python setup.py install
|
||||
pip install fastdeploy-tools==0.0.0
|
||||
```
|
||||
|
||||
### 一键模型自动化压缩工具的使用
|
||||
@@ -38,7 +35,7 @@ python setup.py install
|
||||
```bash
|
||||
fastdeploy --auto_compress --config_path=./configs/detection/yolov5s_quant.yaml --method='PTQ' --save_dir='./yolov5s_ptq_model/'
|
||||
```
|
||||
详细使用文档请参考[FastDeploy一键模型自动化压缩工具](./auto_compression/README.md)
|
||||
详细使用文档请参考[FastDeploy一键模型自动化压缩工具](./common_tools/auto_compression/README.md)
|
||||
|
||||
<p id="2"></p>
|
||||
|
||||
|
||||
@@ -22,14 +22,12 @@ git clone https://github.com/PaddlePaddle/PaddleSlim.git & cd PaddleSlim
|
||||
python setup.py install
|
||||
```
|
||||
|
||||
3.Install fd-auto-compress package
|
||||
3.Install fastdeploy-tools package
|
||||
```bash
|
||||
# Installing fd-auto-compress via pip
|
||||
# Installing fastdeploy-tools via pip
|
||||
# This tool is included in the python installer of FastDeploy, so you don't need to install it again.
|
||||
pip install fd-auto-compress==0.0.1
|
||||
pip install fastdeploy-tools==0.0.0
|
||||
|
||||
# Execute in the current directory
|
||||
python setup.py install
|
||||
```
|
||||
|
||||
### The Usage of One-Click Model Auto Compression Tool
|
||||
@@ -37,7 +35,7 @@ After the above steps are successfully installed, you can use FastDeploy one-cli
|
||||
```bash
|
||||
fastdeploy --auto_compress --config_path=./configs/detection/yolov5s_quant.yaml --method='PTQ' --save_dir='./yolov5s_ptq_model/'
|
||||
```
|
||||
For detailed documentation, please refer to [FastDeploy One-Click Model Auto Compression Tool](./auto_compression/README.md)
|
||||
For detailed documentation, please refer to [FastDeploy One-Click Model Auto Compression Tool](./common_tools/auto_compression/README_EN.md)
|
||||
|
||||
<p id="2"></p>
|
||||
|
||||
|
||||
@@ -1,22 +0,0 @@
|
||||
import setuptools
|
||||
import fd_auto_compress
|
||||
|
||||
long_description = "fd_auto_compress is a toolkit for model auto compression of FastDeploy.\n\n"
|
||||
long_description += "Usage: fastdeploy --auto_compress --config_path=./yolov7_tiny_qat_dis.yaml --method='QAT' --save_dir='../v7_qat_outmodel/' \n"
|
||||
|
||||
setuptools.setup(
|
||||
name="fd_auto_compress",
|
||||
version="0.0.1",
|
||||
description="A toolkit for model auto compression of FastDeploy.",
|
||||
long_description=long_description,
|
||||
long_description_content_type="text/plain",
|
||||
packages=setuptools.find_packages(),
|
||||
author='fastdeploy',
|
||||
author_email='fastdeploy@baidu.com',
|
||||
url='https://github.com/PaddlePaddle/FastDeploy.git',
|
||||
classifiers=[
|
||||
"Programming Language :: Python :: 3",
|
||||
"License :: OSI Approved :: Apache Software License",
|
||||
"Operating System :: OS Independent",
|
||||
],
|
||||
license='Apache 2.0', )
|
||||
@@ -17,14 +17,8 @@ git clone https://github.com/PaddlePaddle/PaddleSlim.git & cd PaddleSlim
|
||||
python setup.py install
|
||||
```
|
||||
|
||||
### fastdeploy-auto-compression 一键模型自动化压缩工具安装方式
|
||||
```
|
||||
# 通过pip安装fd-auto-compress包
|
||||
pip install fd-auto-compress
|
||||
|
||||
# 并在上一层目录(非本级目录)执行如下命令
|
||||
python setup.py install
|
||||
```
|
||||
### 一键模型自动化压缩工具安装方式
|
||||
FastDeploy一键模型自动化压缩不需要单独的安装, 用户只需要正确安装好[FastDeploy工具包](../../README.md)即可.
|
||||
|
||||
## 2.使用方式
|
||||
|
||||
@@ -21,16 +21,8 @@ python setup.py install
|
||||
```
|
||||
|
||||
### Install Fastdeploy Auto Compression Toolkit
|
||||
FastDeploy One-Click Model Automation compression does not require a separate installation, users only need to properly install the [FastDeploy Toolkit](../../README.md)
|
||||
|
||||
Run the following command to install
|
||||
|
||||
```
|
||||
# Install fd-auto-compress package using pip
|
||||
pip install fd-auto-compress
|
||||
|
||||
# Execute the following command in the previous directory (not in the current directory)
|
||||
python setup.py install
|
||||
```
|
||||
|
||||
## 2. How to Use
|
||||
|
||||
@@ -5,37 +5,41 @@ In addition to using the configuration files provided by FastDeploy directly in
|
||||
|
||||
## Demo
|
||||
|
||||
|
||||
```
|
||||
# Global config
|
||||
Global:
|
||||
model_dir: ./yolov5s.onnx #Path to input model
|
||||
format: 'onnx' #Input model format, please select 'paddle' for paddle model
|
||||
model_dir: ./ppyoloe_plus_crn_s_80e_coco #Path to input model
|
||||
format: paddle #Input model format, please select 'paddle' for paddle model
|
||||
model_filename: model.pdmodel #Quantized model name in Paddle format
|
||||
params_filename: model.pdiparams #Parameter name for quantized model name in Paddle format
|
||||
image_path: ./COCO_val_320 #Data set paths for post-training quantization or quantized distillation
|
||||
arch: YOLOv5 #Model Architecture
|
||||
input_list: ['x2paddle_images'] #Input name of the model to be quantified
|
||||
preprocess: yolo_image_preprocess #The preprocessing functions for the data when quantizing the model. Developers can modify or write a new one in . /fdquant/dataset.py
|
||||
params_filename: model.pdiparams #Parameter name for quantized paddle model
|
||||
qat_image_path: ./COCO_train_320 #Data set paths for quantization distillation training
|
||||
ptq_image_path: ./COCO_val_320 #Data set paths for PTQ
|
||||
input_list: ['image','scale_factor'] #Input name of the model to be quanzitzed
|
||||
qat_preprocess: ppyoloe_plus_withNMS_image_preprocess # The preprocessing function for Quantization distillation training
|
||||
ptq_preprocess: ppyoloe_plus_withNMS_image_preprocess # The preprocessing function for PTQ
|
||||
qat_batch_size: 4 #Batch size
|
||||
|
||||
#uantization distillation training configuration
|
||||
|
||||
# Quantization distillation training configuration
|
||||
Distillation:
|
||||
alpha: 1.0 # Distillation loss weight
|
||||
alpha: 1.0 #Distillation loss weight
|
||||
loss: soft_label #Distillation loss algorithm
|
||||
|
||||
Quantization:
|
||||
onnx_format: true #Whether to use ONNX quantization standard format or not, must be true to deploy on FastDeploye
|
||||
onnx_format: true #Whether to use ONNX quantization standard format or not, must be true to deploy on FastDeploy
|
||||
use_pact: true #Whether to use the PACT method for training
|
||||
activation_quantize_type: 'moving_average_abs_max' #Activate quantization methods
|
||||
activation_quantize_type: 'moving_average_abs_max' #Activations quantization methods
|
||||
quantize_op_types: #OPs that need to be quantized
|
||||
- conv2d
|
||||
- depthwise_conv2d
|
||||
|
||||
#Post-Training Quantization
|
||||
# Post-Training Quantization
|
||||
PTQ:
|
||||
calibration_method: 'avg' #Activate calibration algorithm of post-training quantization , Options: avg, abs_max, hist, KL, mse, emd
|
||||
calibration_method: 'avg' #Activations calibration algorithm of post-training quantization , Options: avg, abs_max, hist, KL, mse, emd
|
||||
skip_tensor_list: None #Developers can skip some conv layers‘ quantization
|
||||
|
||||
#Traning
|
||||
# Training Config
|
||||
TrainConfig:
|
||||
train_iter: 3000
|
||||
learning_rate: 0.00001
|
||||
@@ -44,8 +48,9 @@ TrainConfig:
|
||||
type: SGD
|
||||
weight_decay: 4.0e-05
|
||||
target_metric: 0.365
|
||||
|
||||
```
|
||||
|
||||
## More details
|
||||
|
||||
FastDeploy one-click quantization tool is powered by PaddeSlim, please refer to [Automated Compression of Hyperparameter Tutorial](https://github.com/PaddlePaddle/PaddleSlim/blob/develop/example/auto_compression/hyperparameter_tutorial.md) for more details.
|
||||
FastDeploy one-click quantization tool is powered by PaddeSlim, please refer to [Auto Compression Hyperparameter Tutorial](https://github.com/PaddlePaddle/PaddleSlim/blob/develop/example/auto_compression/hyperparameter_tutorial.md) for more details.
|
||||
@@ -78,7 +78,7 @@ def main():
|
||||
args = argsparser().parse_args()
|
||||
if args.auto_compress == True:
|
||||
try:
|
||||
from fd_auto_compress.fd_auto_compress import auto_compress
|
||||
from .auto_compression.fd_auto_compress.fd_auto_compress import auto_compress
|
||||
print("Welcome to use FastDeploy Auto Compression Toolkit!")
|
||||
auto_compress(args)
|
||||
except ImportError:
|
||||
|
||||
Reference in New Issue
Block a user