Files
FastDeploy/tools/auto_compression/configs/README_EN.md
GodIsBoom d02e1cd1d4 [Doc] Add some readme in EN (#529)
* add readme in EN

* code style fix
2022-11-08 21:53:16 +08:00

52 lines
2.6 KiB
Markdown
Raw Blame History

This file contains ambiguous Unicode characters
This file contains Unicode characters that might be confused with other characters. If you think that this is intentional, you can safely ignore this warning. Use the Escape button to reveal them.
# Quantization Config File on FastDeploy
The FastDeploy quantization configuration file contains global configuration, quantization distillation training configuration, post-training quantization configuration and training configuration.
In addition to using the configuration files provided by FastDeploy directly in this directory, users can modify the relevant configuration files according to their needs
## Demo
```
# Global config
Global:
model_dir: ./yolov5s.onnx #Path to input model
format: 'onnx' #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
#uantization distillation training configuration
Distillation:
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
use_pact: true #Whether to use the PACT method for training
activation_quantize_type: 'moving_average_abs_max' #Activate quantization methods
quantize_op_types: #OPs that need to be quantized
- conv2d
- depthwise_conv2d
#Post-Training Quantization
PTQ:
calibration_method: 'avg' #Activate 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
TrainConfig:
train_iter: 3000
learning_rate: 0.00001
optimizer_builder:
optimizer:
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.