Skip to content

P
PaddleGAN

PaddlePaddle / PaddleGAN
大约 1 年 前同步成功

通知 97
Star 7254
Fork 1210
P
PaddleGAN
未验证 提交 3d4fa146 编写于 作者: L lzzyzlbb 提交者： GitHub
浏览文件
操作

add tipc (#492)

上级 2ab96cb8
显示空白变更内容
内联 并排
Showing with 946 addition and 0 deletion
test_tipc/common_func.sh 0 → 100644
浏览文件 @ 3d4fa146
#!/bin/bash
function func_parser_key(){
strs=$1
IFS=":"
array=(${strs})
tmp=${array[0]}
echo ${tmp}
}
function func_parser_value(){
strs=$1
IFS=":"
array=(${strs})
tmp=${array[1]}
echo ${tmp}
}
function func_set_params(){
key=$1
value=$2
if [ ${key}x = "null"x ];then
echo " "
elif [[ ${value} = "null" ]] || [[ ${value} = " " ]] || [ ${#value} -le 0 ];then
echo " "
else
echo "${key}=${value}"
fi
}
function func_parser_params(){
strs=$1
IFS=":"
array=(${strs})
key=${array[0]}
tmp=${array[1]}
IFS="|"
res=""
for _params in ${tmp[*]}; do
IFS="="
array=(${_params})
mode=${array[0]}
value=${array[1]}
if [[ ${mode} = ${MODE} ]]; then
IFS="|"
#echo $(func_set_params "${mode}" "${value}")
echo $value
break
fi
IFS="|"
done
echo ${res}
}
function status_check(){
last_status=$1 # the exit code
run_command=$2
run_log=$3
if [ $last_status -eq 0 ]; then
echo -e "\033[33m Run successfully with command - ${run_command}! \033[0m" | tee -a ${run_log}
else
echo -e "\033[33m Run failed with command - ${run_command}! \033[0m" | tee -a ${run_log}
fi
}
test_tipc/configs/basicvsr/train_infer_python.txt 0 → 100644
浏览文件 @ 3d4fa146
===========================train_params===========================
model_name:basicvsr
python:python3.7
gpu_list:0
##
auto_cast:null
total_iters:lite_train_lite_infer=5|whole_train_whole_infer=200
output_dir:./output/
dataset.train.batch_size:lite_train_lite_infer=1|whole_train_whole_infer=1
pretrained_model:null
train_model_name:basicvsr_reds*/*checkpoint.pdparams
train_infer_img_dir:./data/basicvsr_reds/test
null:null
##
trainer:norm_train
norm_train:tools/main.py -c configs/basicvsr_reds.yaml -o dataset.train.dataset.num_clips=2
pact_train:null
fpgm_train:null
distill_train:null
null:null
null:null
##
===========================eval_params===========================
eval:null
null:null
##
===========================infer_params===========================
--output_dir:./output/
load:null
norm_export:tools/export_model.py -c configs/basicvsr_reds.yaml --inputs_size="1,6,3,180,320" --load
quant_export:null
fpgm_export:null
distill_export:null
export1:null
export2:null
inference_dir:basicvsrmodel_generator
train_model:./inference/basicvsr/basicvsrmodel_generator
infer_export:null
infer_quant:False
inference:tools/inference.py --model_type basicvsr -c configs/basicvsr_reds.yaml -o dataset.test.num_clips=2 dataset.test.number_frames=6
--device:gpu
null:null
null:null
null:null
null:null
null:null
--model_path:
null:null
null:null
--benchmark:True
null:null
\ No newline at end of file
test_tipc/configs/cyclegan/train_infer_python.txt 0 → 100644
浏览文件 @ 3d4fa146
===========================train_params===========================
model_name:cyclegan
python:python3.7
gpu_list:0|0,1
##
auto_cast:null
epochs:lite_train_lite_infer=5|whole_train_whole_infer=200
output_dir:./output/
dataset.train.batch_size:lite_train_lite_infer=1|whole_train_whole_infer=1
pretrained_model:null
train_model_name:cyclegan_horse2zebra*/*checkpoint.pdparams
train_infer_img_dir:./data/horse2zebra/test
null:null
##
trainer:norm_train
norm_train:tools/main.py -c configs/cyclegan_horse2zebra.yaml -o
pact_train:null
fpgm_train:null
distill_train:null
null:null
null:null
##
===========================eval_params===========================
eval:null
null:null
##
===========================infer_params===========================
--output_dir:./output/
load:null
norm_export:tools/export_model.py -c configs/cyclegan_horse2zebra.yaml --inputs_size="-1,3,-1,-1;-1,3,-1,-1" --load
quant_export:null
fpgm_export:null
distill_export:null
export1:null
export2:null
inference_dir:cycleganmodel_netG_A
train_model:./inference/cyclegan_horse2zebra/cycleganmodel_netG_A
infer_export:null
infer_quant:False
inference:tools/inference.py --model_type cyclegan -c configs/cyclegan_horse2zebra.yaml
--device:gpu
null:null
null:null
null:null
null:null
null:null
--model_path:
null:null
null:null
--benchmark:True
null:null
\ No newline at end of file
test_tipc/configs/fom/train_infer_python.txt 0 → 100644
浏览文件 @ 3d4fa146
===========================train_params===========================
model_name:fom
python:python3.7
gpu_list:0
##
auto_cast:null
epochs:lite_train_lite_infer=10|whole_train_whole_infer=100
output_dir:./output/
dataset.train.batch_size:lite_train_lite_infer=8|whole_train_whole_infer=8
pretrained_model:null
train_model_name:firstorder_vox_256*/*checkpoint.pdparams
train_infer_img_dir:./data/firstorder_vox_256/test
null:null
##
trainer:norm_train
norm_train:tools/main.py -c configs/firstorder_vox_256.yaml -o
pact_train:null
fpgm_train:null
distill_train:null
null:null
null:null
##
===========================eval_params===========================
eval:null
null:null
##
===========================infer_params===========================
--output_dir:./output/
load:null
norm_export:tools/export_model.py -c configs/firstorder_vox_256.yaml --inputs_size="1,3,256,256;1,3,256,256;1,10,2;1,10,2,2" --load
quant_export:null
fpgm_export:null
distill_export:null
export1:null
export2:null
inference_dir:fom_dy2st
train_model:./inference/fom_dy2st/
infer_export:null
infer_quant:False
inference:tools/fom_infer.py --driving_path data/first_order/Voxceleb/test --output_path infer_output/fom
--device:gpu
null:null
null:null
null:null
null:null
null:null
--model_path:
null:null
null:null
--benchmark:True
null:null
\ No newline at end of file
test_tipc/configs/pix2pix/train_infer_python.txt 0 → 100644
浏览文件 @ 3d4fa146
===========================train_params===========================
model_name:pix2pix
python:python3.7
gpu_list:0|0,1
##
auto_cast:null
epochs:lite_train_lite_infer=5|whole_train_whole_infer=200
output_dir:./output/
dataset.train.batch_size:lite_train_lite_infer=1|whole_train_whole_infer=1
pretrained_model:null
train_model_name:pix2pix_facades*/*checkpoint.pdparams
train_infer_img_dir:./data/facades/test
null:null
##
trainer:norm_train
norm_train:tools/main.py -c configs/pix2pix_facades.yaml -o
pact_train:null
fpgm_train:null
distill_train:null
null:null
null:null
##
===========================eval_params===========================
eval:null
null:null
##
===========================infer_params===========================
--output_dir:./output/
load:null
norm_export:tools/export_model.py -c configs/pix2pix_facades.yaml --inputs_size="-1,3,-1,-1" --load
quant_export:null
fpgm_export:null
distill_export:null
export1:null
export2:null
inference_dir:pix2pixmodel_netG
train_model:./inference/pix2pix_facade/pix2pixmodel_netG
infer_export:null
infer_quant:False
inference:tools/inference.py --model_type pix2pix -c configs/pix2pix_facades.yaml
--device:cpu
null:null
null:null
null:null
null:null
null:null
--model_path:
null:null
null:null
--benchmark:True
null:null
\ No newline at end of file
test_tipc/configs/stylegan2/train_infer_python.txt 0 → 100644
浏览文件 @ 3d4fa146
===========================train_params===========================
model_name:stylegan2
python:python3.7
gpu_list:0
##
auto_cast:null
total_iters::lite_train_lite_infer=10|whole_train_whole_infer=800
output_dir:./output/
dataset.train.batch_size:lite_train_lite_infer=3|whole_train_whole_infer=3
pretrained_model:null
train_model_name:stylegan_v2_256_ffhq*/*checkpoint.pdparams
train_infer_img_dir:null
null:null
##
trainer:norm_train
norm_train:tools/main.py -c configs/stylegan_v2_256_ffhq.yaml -o
pact_train:null
fpgm_train:null
distill_train:null
null:null
null:null
##
===========================eval_params===========================
eval:null
null:null
##
===========================infer_params===========================
--output_dir:./output/
load:null
norm_export:tools/export_model.py -c configs/stylegan_v2_256_ffhq.yaml --inputs_size="1,1,512;1,1" --load
quant_export:null
fpgm_export:null
distill_export:null
export1:null
export2:null
inference_dir:stylegan2model_gen
train_model:./inference/stylegan2/stylegan2model_gen
infer_export:null
infer_quant:False
inference:tools/inference.py --model_type stylegan2 -c configs/stylegan_v2_256_ffhq.yaml
--device:gpu
null:null
null:null
null:null
null:null
null:null
--model_path:
null:null
null:null
--benchmark:True
null:null
\ No newline at end of file
test_tipc/docs/test_train_inference_python.md 0 → 100644
浏览文件 @ 3d4fa146
# Linux端基础训练预测功能测试
Linux端基础训练预测功能测试的主程序为`test_train_inference_python.sh`，可以测试基于Python的模型训练、评估、推理等基本功能。
## 1. 测试结论汇总
- 训练相关：
| 算法论文 | 模型名称 | 模型类型 | 基础<br>训练预测 | 更多<br>训练方式 | 模型压缩 | 其他预测部署 |
| :--- | :--- | :----: | :--------: | :---- | :---- | :---- |
| Pix2Pix |Pix2Pix | 生成 | 支持 | 多机多卡 | | |
| CycleGAN |CycleGAN | 生成 | 支持 | 多机多卡 | | |
| StyleGAN2 |StyleGAN2 | 生成 | 支持 | 多机多卡 | | |
| FOMM |FOMM | 生成 | 支持 | 多机多卡 | | |
| BasicVSR |BasicVSR | 超分 | 支持 | 多机多卡 | | |
|PP-MSVSR|PP-MSVSR | 超分|
- 预测相关：预测功能汇总如下，
| 模型类型 |device | batchsize | tensorrt | mkldnn | cpu多线程 |
| ---- | ---- | ---- | :----: | :----: | :----: |
| 正常模型 | GPU | 1/6 | fp32 | - | - |
## 2. 测试流程
运行环境配置请参考[文档](../../docs/zh_CN/install.md)的内容配置运行环境。
### 2.1 安装依赖
- 安装PaddlePaddle >= 2.1
- 安装PaddleGAN依赖
```
pip install -v -e .
```
- 安装autolog（规范化日志输出工具）
```
git clone https://github.com/LDOUBLEV/AutoLog
cd AutoLog
pip3 install -r requirements.txt
python3 setup.py bdist_wheel
pip3 install ./dist/auto_log-1.0.0-py3-none-any.whl
cd ../
```
### 2.2 功能测试
先运行`prepare.sh`准备数据和模型，然后运行`test_train_inference_python.sh`进行测试，最终在```test_tipc/output```目录下生成`python_infer_*.log`格式的日志文件。
`test_train_inference_python.sh`包含5种运行模式，每种模式的运行数据不同，分别用于测试速度和精度，分别是：
- 模式1：lite_train_lite_infer，使用少量数据训练，用于快速验证训练到预测的走通流程，不验证精度和速度；
```shell
bash test_tipc/prepare.sh ./test_tipc/configs/basicvsr/train_infer_python.txt 'lite_train_lite_infer'
bash test_tipc/test_train_inference_python.sh ./test_tipc/configs/basicvsr/train_infer_python.txt 'lite_train_lite_infer'
```
- 模式2：lite_train_whole_infer，使用少量数据训练，一定量数据预测，用于验证训练后的模型执行预测，预测速度是否合理；
```shell
bash test_tipc/prepare.sh ./test_tipc/configs/basicvsr/train_infer_python.txt 'lite_train_whole_infer'
bash test_tipc/test_train_inference_python.sh ./test_tipc/configs/basicvsr/train_infer_python.txt 'lite_train_whole_infer'
```
- 模式3：whole_infer，不训练，全量数据预测，走通开源模型评估、动转静，检查inference model预测时间和精度;
```shell
bash test_tipc/prepare.sh ./test_tipc/configs/basicvsr/train_infer_python.txt 'whole_infer'
bash test_tipc/test_train_inference_python.sh ./test_tipc/configs/basicvsr/train_infer_python.txt 'whole_infer'
```
- 模式4：whole_train_whole_infer，CE： 全量数据训练，全量数据预测，验证模型训练精度，预测精度，预测速度；
```shell
bash test_tipc/prepare.sh ./test_tipc/configs/basicvsr/train_infer_python.txt 'whole_train_whole_infer'
bash test_tipc/test_train_inference_python.sh ./test_tipc/configs/basicvsr/train_infer_python.txt 'whole_train_whole_infer'
```
运行相应指令后，在`test_tipc/output`文件夹下自动会保存运行日志。如'lite_train_lite_infer'模式下，会运行训练+inference的链条，因此，在`test_tipc/output`文件夹有以下文件：
```
test_tipc/output/
|- results_python.log # 运行指令状态的日志
|- norm_train_gpus_0_autocast_null/ # GPU 0号卡上正常训练的训练日志和模型保存文件夹
......
```
其中`results_python.log`中包含了每条指令的运行状态，如果运行成功会输出：
```
Run successfully with command - python3.7 tools/main.py -c configs/basicvsr_reds.yaml -o dataset.train.dataset.num_clips=2 output_dir=./test_tipc/output/norm_train_gpus_0_autocast_null total_iters=5 dataset.train.batch_size=1 !
-=Run successfully with command - python3.7 tools/export_model.py -c configs/basicvsr_reds.yaml --inputs_size="1,6,3,180,320" --load ./test_tipc/output/norm_train_gpus_0_autocast_null/basicvsr_reds-2021-11-22-07-18/iter_1_checkpoint.pdparams --output_dir ./test_tipc/output/norm_train_gpus_0_autocast_null!
......
```
如果运行失败，会输出：
```
Run failed with command - python3.7 tools/main.py -c configs/basicvsr_reds.yaml -o dataset.train.dataset.num_clips=2 output_dir=./test_tipc/output/norm_train_gpus_0_autocast_null total_iters=5 dataset.train.batch_size=1 ! !
Run failed with command - python3.7 tools/export_model.py -c configs/basicvsr_reds.yaml --inputs_size="1,6,3,180,320" --load ./test_tipc/output/norm_train_gpus_0_autocast_null/basicvsr_reds-2021-11-22-07-18/iter_1_checkpoint.pdparams --output_dir ./test_tipc/output/norm_train_gpus_0_autocast_null!
......
```
可以很方便的根据`results_python.log`中的内容判定哪一个指令运行错误。
### 2.3 精度测试
使用compare_results.py脚本比较模型预测的结果是否符合预期，主要步骤包括：
- 提取日志中的预测坐标；
- 从本地文件中提取保存好的坐标结果；
- 比较上述两个结果是否符合精度预期，误差大于设置阈值时会报错。
#### 使用方式
运行命令：
```shell
python3.7 test_tipc/compare_results.py --gt_file=./test_tipc/results/python_*.txt --log_file=./test_tipc/output/python_*.log --atol=1e-3 --rtol=1e-3
```
参数介绍：
- gt_file： 指向事先保存好的预测结果路径，支持*.txt 结尾，会自动索引*.txt格式的文件，文件默认保存在test_tipc/result/ 文件夹下
- log_file: 指向运行test_tipc/test_train_inference_python.sh 脚本的infer模式保存的预测日志，预测日志中打印的有预测结果，
- atol: 设置的绝对误差
- rtol: 设置的相对误差
#### 运行结果
正常运行效果如下图：
<img src="compare_right.png" width="1000">
出现不一致结果时的运行输出：
<img src="compare_wrong.png" width="1000">
test_tipc/prepare.sh 0 → 100644
浏览文件 @ 3d4fa146
#!/bin/bash
FILENAME=$1
# MODE be one of ['lite_train_lite_infer' 'lite_train_whole_infer' 'whole_train_whole_infer',
# 'whole_infer']
MODE=$2
dataline=$(cat ${FILENAME})
# parser params
IFS=$'\n'
lines=(${dataline})
function func_parser_key(){
strs=$1
IFS=":"
array=(${strs})
tmp=${array[0]}
echo ${tmp}
}
function func_parser_value(){
strs=$1
IFS=":"
array=(${strs})
tmp=${array[1]}
echo ${tmp}
}
IFS=$'\n'
# The training params
model_name=$(func_parser_value "${lines[1]}")
trainer_list=$(func_parser_value "${lines[14]}")
# MODE be one of ['lite_train_lite_infer' 'lite_train_whole_infer' 'whole_train_whole_infer',
# 'whole_infer']
MODE=$2
if [ ${MODE} = "lite_train_lite_infer" ];then
if [ ${model_name} == "pix2pix" ]; then
rm -rf ./data/facades*
rm -rf ./data/pix2pix*
wget -nc -P ./data/ https://paddlegan.bj.bcebos.com/datasets/pix2pix_facade_lite.tar --no-check-certificate
cd ./data/ && tar xf pix2pix_facade_lite.tar && cd ../
elif [ ${model_name} == "cyclegan" ]; then
rm -rf ./data/horse2zebra*
wget -nc -P ./data/ https://paddlegan.bj.bcebos.com/datasets/cyclegan_horse2zebra_lite.tar --no-check-certificate
cd ./data/ && tar xf cyclegan_horse2zebra_lite.tar && cd ../
elif [ ${model_name} == "stylegan2" ]; then
rm -rf ./data/ffhq*
wget -nc -P ./data/ https://paddlegan.bj.bcebos.com/datasets/ffhq.tar --no-check-certificate
cd ./data/ && tar xf ffhq.tar && cd ../
elif [ ${model_name} == "fom" ]; then
rm -rf ./data/first_order*
wget -nc -P ./data/ https://paddlegan.bj.bcebos.com/datasets/fom_lite.tar --no-check-certificate --no-check-certificate
cd ./data/ && tar xf fom_lite.tar && cd ../
elif [ ${model_name} == "basicvsr" ]; then
rm -rf ./data/REDS*
wget -nc -P ./data/ https://paddlegan.bj.bcebos.com/datasets/basicvsr_lite.tar --no-check-certificate
cd ./data/ && tar xf basicvsr_lite.tar && cd ../
fi
elif [ ${MODE} = "whole_train_whole_infer" ];then
if [ ${model_name} == "pix2pix" ]; then
rm -rf ./data/facades*
wget -nc -P ./data/ http://efrosgans.eecs.berkeley.edu/pix2pix/datasets/facades.tar.gz --no-check-certificate
cd ./data/ && tar -xzf facades.tar.gz && cd ../
elif [ ${model_name} == "cyclegan" ]; then
rm -rf ./data/horse2zebra*
wget -nc -P ./data/ https://people.eecs.berkeley.edu/~taesung_park/CycleGAN/datasets/horse2zebra.zip --no-check-certificate
cd ./data/ && unzip horse2zebra.zip && cd ../
fi
elif [ ${MODE} = "lite_train_whole_infer" ];then
if [ ${model_name} == "pix2pix" ]; then
rm -rf ./data/facades*
wget -nc -P ./data/ https://paddlegan.bj.bcebos.com/datasets/pix2pix_facade_lite.tar --no-check-certificate
cd ./data/ && tar xf pix2pix_facade_lite.tar && cd ../
elif [ ${model_name} == "cyclegan" ]; then
rm -rf ./data/horse2zebra*
wget -nc -P ./data/ https://paddlegan.bj.bcebos.com/datasets/cyclegan_horse2zebra_lite.tar --no-check-certificate --no-check-certificate
cd ./data/ && tar xf cyclegan_horse2zebra_lite.tar && cd ../
elif [ ${model_name} == "fom" ]; then
rm -rf ./data/first_order*
wget -nc -P ./data/ https://paddlegan.bj.bcebos.com/datasets/fom_lite.tar --no-check-certificate --no-check-certificate
cd ./data/ && tar xf fom_lite.tar && cd ../
elif [ ${model_name} == "stylegan2" ]; then
rm -rf ./data/ffhq*
wget -nc -P ./data/ https://paddlegan.bj.bcebos.com/datasets/ffhq.tar --no-check-certificate
cd ./data/ && tar xf ffhq.tar && cd ../
elif [ ${model_name} == "basicvsr" ]; then
rm -rf ./data/REDS*
wget -nc -P ./data/ https://paddlegan.bj.bcebos.com/datasets/basicvsr_lite.tar --no-check-certificate
cd ./data/ && tar xf basicvsr_lite.tar && cd ../
fi
elif [ ${MODE} = "whole_infer" ];then
if [ ${model_name} = "pix2pix" ]; then
rm -rf ./data/facades*
wget -nc -P ./inference https://paddlegan.bj.bcebos.com/static_model/pix2pix_facade.tar --no-check-certificate
wget -nc -P ./data https://paddlegan.bj.bcebos.com/datasets/facades_test.tar --no-check-certificate
cd ./data && tar xf facades_test.tar && mv facades_test facades && cd ../
cd ./inference && tar xf pix2pix_facade.tar && cd ../
elif [ ${model_name} = "cyclegan" ]; then
rm -rf ./data/horse2zebra*
wget -nc -P ./inference https://paddlegan.bj.bcebos.com/static_model/cyclegan_horse2zebra.tar --no-check-certificate
wget -nc -P ./data https://paddlegan.bj.bcebos.com/datasets/cyclegan_horse2zebra_test.tar --no-check-certificate
cd ./data && tar xf cyclegan_horse2zebra_test.tar && mv cyclegan_test horse2zebra && cd ../
cd ./inference && tar xf cyclegan_horse2zebra.tar && cd ../
elif [ ${model_name} == "fom" ]; then
rm -rf ./data/first_order*
wget -nc -P ./data/ https://paddlegan.bj.bcebos.com/datasets/fom_lite_test.tar --no-check-certificate
wget -nc -P ./inference https://paddlegan.bj.bcebos.com/static_model/fom_dy2st.tar --no-check-certificate
cd ./data/ && tar xf fom_lite_test.tar && cd ../
cd ./inference && tar xf fom_dy2st.tar && cd ../
elif [ ${model_name} == "stylegan2" ]; then
rm -rf ./data/ffhq*
wget -nc -P ./data/ https://paddlegan.bj.bcebos.com/datasets/ffhq.tar --no-check-certificate
wget -nc -P ./inference https://paddlegan.bj.bcebos.com/static_model/stylegan2_1024.tar --no-check-certificate
cd ./inference && tar xf stylegan2_1024.tar && cd ../
cd ./data/ && tar xf ffhq.tar && cd ../
elif [ ${model_name} == "basicvsr" ]; then
rm -rf ./data/basic*
rm -rf ./inference/basic*
wget -nc -P ./data/ https://paddlegan.bj.bcebos.com/datasets/basicvsr_lite_test.tar --no-check-certificate
wget -nc -P ./inference https://paddlegan.bj.bcebos.com/static_model/basicvsr.tar --no-check-certificate
cd ./inference && tar xf basicvsr.tar && cd ../
cd ./data/ && tar xf basicvsr_lite_test.tar && cd ../
fi
fi
test_tipc/readme.md 0 → 100644
浏览文件 @ 3d4fa146
# 飞桨训推一体认证
## 1. 简介
飞桨除了基本的模型训练和预测，还提供了支持多端多平台的高性能推理部署工具。本文档提供了PaddleGAN中所有模型的飞桨训推一体认证 (Training and Inference Pipeline Certification(TIPC)) 信息和测试工具，方便用户查阅每种模型的训练推理部署打通情况，并可以进行一键测试。
## 2. 汇总信息
打通情况汇总如下，已填写的部分表示可以使用本工具进行一键测试，未填写的表示正在支持中。
**字段说明：**
- 基础训练预测：包括模型训练、Paddle Inference Python预测。
- 更多训练方式：包括多机多卡、混合精度。
- 模型压缩：包括裁剪、离线/在线量化、蒸馏。
- 其他预测部署：包括Paddle Inference C++预测、Paddle Serving部署、Paddle-Lite部署等。
更详细的mkldnn、Tensorrt等预测加速相关功能的支持情况可以查看各测试工具的[更多教程](#more)。
| 算法论文 | 模型名称 | 模型类型 | 基础<br>训练预测 | 更多<br>训练方式 | 模型压缩 | 其他预测部署 |
| :--- | :--- | :----: | :--------: | :---- | :---- | :---- |
| Pix2Pix |Pix2Pix | 生成 | 支持 | 多机多卡 | | |
| CycleGAN |CycleGAN | 生成 | 支持 | 多机多卡 | | |
| StyleGAN2 |StyleGAN2 | 生成 | 支持 | 多机多卡 | | |
| FOMM |FOMM | 生成 | 支持 | 多机多卡 | | |
| BasicVSR |BasicVSR | 超分 | 支持 | 多机多卡 | | |
|PP-MSVSR|PP-MSVSR | 超分|
## 3. 一键测试工具使用
### 目录介绍
```shell
test_tipc/
├── configs/ # 配置文件目录
├── basicvsr_reds.yaml # 测试basicvsr模型训练的yaml文件
├── cyclegan_horse2zebra.yaml # 测试cyclegan模型训练的yaml文件
├── firstorder_vox_256.yaml # 测试fomm模型训练的yaml文件
├── pix2pix_facedes.yaml # 测试pix2pix模型训练的yaml文件
├── stylegan_v2_256_ffhq.yaml # 测试stylegan模型训练的yaml文件
├── ...
├── results/ # 预先保存的预测结果，用于和实际预测结果进行精读比对
├── python_basicvsr_results_fp32.txt # 预存的basicvsr模型python预测fp32精度的结果
├── python_cyclegan_results_fp32.txt # 预存的cyclegan模型python预测fp32精度的结果
├── python_pix2pix_results_fp32.txt # 预存的pix2pix模型python预测的fp32精度的结果
├── python_stylegan_results_fp32.txt # 预存的stylegan模型python预测的fp32精度的结果
├── ...
├── prepare.sh # 完成test_*.sh运行所需要的数据和模型下载
├── test_train_inference_python.sh # 测试python训练预测的主程序
├── compare_results.py # 用于对比log中的预测结果与results中的预存结果精度误差是否在限定范围内
└── readme.md # 使用文档
```
### 测试流程
使用本工具，可以测试不同功能的支持情况，以及预测结果是否对齐，测试流程如下：
<div align="center">
<img src="docs/test.png" width="800">
</div>
1. 运行prepare.sh准备测试所需数据和模型；
2. 运行要测试的功能对应的测试脚本`test_*.sh`，产出log，由log可以看到不同配置是否运行成功；
3. 用`compare_results.py`对比log中的预测结果和预存在results目录下的结果，判断预测精度是否符合预期（在误差范围内）。
其中，有4个测试主程序，功能如下：
- `test_train_inference_python.sh`：测试基于Python的模型训练、评估、推理等基本功能。
<a name="more"></a>
#### 更多教程
各功能测试中涉及混合精度、裁剪、量化等训练相关，及mkldnn、Tensorrt等多种预测相关参数配置，请点击下方相应链接了解更多细节和使用教程：
[test_train_inference_python 使用](docs/test_train_inference_python.md)
test_tipc/test_train_inference_python.sh 0 → 100644
浏览文件 @ 3d4fa146
#!/bin/bash
source test_tipc/common_func.sh
FILENAME=$1
# MODE be one of ['lite_train_lite_infer' 'lite_train_whole_infer' 'whole_train_whole_infer', 'whole_infer']
MODE=$2
dataline=$(awk 'NR==1, NR==51{print}' $FILENAME)
# parser params
IFS=$'\n'
lines=(${dataline})
# The training params
model_name=$(func_parser_value "${lines[1]}")
python=$(func_parser_value "${lines[2]}")
gpu_list=$(func_parser_value "${lines[3]}")
autocast_list=$(func_parser_value "${lines[5]}")
epoch_key=$(func_parser_key "${lines[6]}")
epoch_num=$(func_parser_params "${lines[6]}")
save_model_key=$(func_parser_key "${lines[7]}")
train_batch_key=$(func_parser_key "${lines[8]}")
train_batch_value=$(func_parser_params "${lines[8]}")
pretrain_model_key=$(func_parser_key "${lines[9]}")
pretrain_model_value=$(func_parser_value "${lines[9]}")
train_model_name=$(func_parser_value "${lines[10]}")
train_infer_img_dir=$(func_parser_value "${lines[11]}")
train_param_key1=$(func_parser_key "${lines[12]}")
train_param_value1=$(func_parser_value "${lines[12]}")
trainer_list=$(func_parser_value "${lines[14]}")
trainer_norm=$(func_parser_key "${lines[15]}")
norm_trainer=$(func_parser_value "${lines[15]}")
trainer_key1=$(func_parser_key "${lines[19]}")
trainer_value1=$(func_parser_value "${lines[19]}")
trainer_key2=$(func_parser_key "${lines[20]}")
trainer_value2=$(func_parser_value "${lines[20]}")
eval_py=$(func_parser_value "${lines[23]}")
eval_key1=$(func_parser_key "${lines[24]}")
eval_value1=$(func_parser_value "${lines[24]}")
save_infer_key=$(func_parser_key "${lines[27]}")
export_weight=$(func_parser_value "${lines[28]}")
norm_export=$(func_parser_value "${lines[29]}")
inference_dir=$(func_parser_value "${lines[35]}")
# parser inference model
infer_model_dir_list=$(func_parser_value "${lines[36]}")
infer_export_list=$(func_parser_value "${lines[37]}")
infer_is_quant=$(func_parser_value "${lines[38]}")
# parser inference
inference_py=$(func_parser_value "${lines[39]}")
use_gpu_key=$(func_parser_key "${lines[40]}")
use_gpu_list=$(func_parser_value "${lines[40]}")
use_mkldnn_key=$(func_parser_key "${lines[41]}")
use_mkldnn_list=$(func_parser_value "${lines[41]}")
cpu_threads_key=$(func_parser_key "${lines[42]}")
cpu_threads_list=$(func_parser_value "${lines[42]}")
batch_size_key=$(func_parser_key "${lines[43]}")
batch_size_list=$(func_parser_value "${lines[43]}")
use_trt_key=$(func_parser_key "${lines[44]}")
use_trt_list=$(func_parser_value "${lines[44]}")
precision_key=$(func_parser_key "${lines[45]}")
precision_list=$(func_parser_value "${lines[45]}")
infer_model_key=$(func_parser_key "${lines[46]}")
image_dir_key=$(func_parser_key "${lines[47]}")
infer_img_dir=$(func_parser_value "${lines[47]}")
save_log_key=$(func_parser_key "${lines[48]}")
infer_key1=$(func_parser_key "${lines[50]}")
infer_value1=$(func_parser_value "${lines[50]}")
LOG_PATH="./test_tipc/output"
mkdir -p ${LOG_PATH}
status_log="${LOG_PATH}/results_python.log"
function func_inference(){
IFS='|'
_python=$1
_script=$2
_model_dir=$3
_log_path=$4
_img_dir=$5
_flag_quant=$6
# inference
for use_gpu in ${use_gpu_list[*]}; do
if [ ${use_gpu} = "False" ] || [ ${use_gpu} = "cpu" ]; then
for use_mkldnn in ${use_mkldnn_list[*]}; do
if [ ${use_mkldnn} = "False" ] && [ ${_flag_quant} = "True" ]; then
continue
fi
for threads in ${cpu_threads_list[*]}; do
for batch_size in ${batch_size_list[*]}; do
for precision in ${precision_list[*]}; do
if [ ${use_mkldnn} = "False" ] && [ ${precision} = "fp16" ]; then
continue
fi # skip when enable fp16 but disable mkldnn
if [ ${_flag_quant} = "True" ] && [ ${precision} != "int8" ]; then
continue
fi # skip when quant model inference but precision is not int8
set_precision=$(func_set_params "${precision_key}" "${precision}")
_save_log_path="${_log_path}/python_infer_cpu_usemkldnn_${use_mkldnn}_threads_${threads}_precision_${precision}_batchsize_${batch_size}.log"
set_infer_data=$(func_set_params "${image_dir_key}" "${_img_dir}")
set_benchmark=$(func_set_params "${benchmark_key}" "${benchmark_value}")
set_batchsize=$(func_set_params "${batch_size_key}" "${batch_size}")
set_cpu_threads=$(func_set_params "${cpu_threads_key}" "${threads}")
set_model_dir=$(func_set_params "${infer_model_key}" "${_model_dir}")
set_infer_params1=$(func_set_params "${infer_key1}" "${infer_value1}")
command="${_python} ${_script} ${use_gpu_key}=${use_gpu} ${set_model_dir} > ${_save_log_path} 2>&1 "
eval $command
last_status=${PIPESTATUS[0]}
eval "cat ${_save_log_path}"
status_check $last_status "${command}" "${status_log}"
done
done
done
done
elif [ ${use_gpu} = "True" ] || [ ${use_gpu} = "gpu" ]; then
for use_trt in ${use_trt_list[*]}; do
for precision in ${precision_list[*]}; do
if [[ ${_flag_quant} = "False" ]] && [[ ${precision} =~ "int8" ]]; then
continue
fi
if [[ ${precision} =~ "fp16" || ${precision} =~ "int8" ]] && [ ${use_trt} = "False" ]; then
continue
fi
if [[ ${use_trt} = "False" || ${precision} =~ "int8" ]] && [ ${_flag_quant} = "True" ]; then
continue
fi
for batch_size in ${batch_size_list[*]}; do
_save_log_path="${_log_path}/python_infer_gpu_usetrt_${use_trt}_precision_${precision}_batchsize_${batch_size}.log"
set_infer_data=$(func_set_params "${image_dir_key}" "${_img_dir}")
set_benchmark=$(func_set_params "${benchmark_key}" "${benchmark_value}")
set_batchsize=$(func_set_params "${batch_size_key}" "${batch_size}")
set_tensorrt=$(func_set_params "${use_trt_key}" "${use_trt}")
set_precision=$(func_set_params "${precision_key}" "${precision}")
set_model_dir=$(func_set_params "${infer_model_key}" "${_model_dir}")
set_infer_params1=$(func_set_params "${infer_key1}" "${infer_value1}")
command="${_python} ${_script} ${use_gpu_key}=${use_gpu} ${set_tensorrt} ${set_precision} ${set_model_dir} ${set_batchsize} ${set_infer_data} ${set_benchmark} ${set_infer_params1} > ${_save_log_path} 2>&1 "
eval $command
last_status=${PIPESTATUS[0]}
eval "cat ${_save_log_path}"
status_check $last_status "${command}" "${status_log}"
done
done
done
else
echo "Does not support hardware other than CPU and GPU Currently!"
fi
done
}
if [ ${MODE} = "whole_infer" ]; then
GPUID=$3
if [ ${#GPUID} -le 0 ];then
env=" "
else
env="export CUDA_VISIBLE_DEVICES=${GPUID}"
fi
# set CUDA_VISIBLE_DEVICES
eval $env
export Count=0
IFS="|"
infer_run_exports=(${infer_export_list})
infer_quant_flag=(${infer_is_quant})
for infer_model in ${infer_model_dir_list[*]}; do
# run export
if [ ${infer_run_exports[Count]} != "null" ];then
save_infer_dir=$(dirname $infer_model)
set_export_weight=$(func_set_params "${export_weight}" "${infer_model}")
set_save_infer_key="${save_infer_key} ${save_infer_dir}"
export_cmd="${python} ${infer_run_exports[Count]} ${set_export_weight} ${set_save_infer_key}"
echo ${infer_run_exports[Count]}
echo $export_cmd
eval $export_cmd
status_export=$?
status_check $status_export "${export_cmd}" "${status_log}"
else
save_infer_dir=${infer_model}
fi
#run inference
func_inference "${python}" "${inference_py}" "${save_infer_dir}" "${LOG_PATH}" "${infer_img_dir}"
Count=$(($Count + 1))
done
else
IFS="|"
export Count=0
USE_GPU_KEY=(${train_use_gpu_value})
for gpu in ${gpu_list[*]}; do
train_use_gpu=${USE_GPU_KEY[Count]}
Count=$(($Count + 1))
ips=""
if [ ${gpu} = "-1" ];then
env=""
elif [ ${#gpu} -le 1 ];then
env="export CUDA_VISIBLE_DEVICES=${gpu}"
eval ${env}
elif [ ${#gpu} -le 15 ];then
IFS=","
array=(${gpu})
env="export CUDA_VISIBLE_DEVICES=${gpu}"
IFS="|"
else
IFS=";"
array=(${gpu})
ips=${array[0]}
gpu=${array[1]}
IFS="|"
env=" "
fi
for autocast in ${autocast_list[*]}; do
if [ ${autocast} = "amp" ]; then
set_amp_config="Global.use_amp=True Global.scale_loss=1024.0 Global.use_dynamic_loss_scaling=True"
else
set_amp_config=" "
fi
for trainer in ${trainer_list[*]}; do
flag_quant=False
run_train=${norm_trainer}
run_export=${norm_export}
if [ ${run_train} = "null" ]; then
continue
fi
set_autocast=$(func_set_params "${autocast_key}" "${autocast}")
set_epoch=$(func_set_params "${epoch_key}" "${epoch_num}")
set_pretrain=$(func_set_params "${pretrain_model_key}" "${pretrain_model_value}")
set_batchsize=$(func_set_params "${train_batch_key}" "${train_batch_value}")
set_train_params1=$(func_set_params "${train_param_key1}" "${train_param_value1}")
set_use_gpu=$(func_set_params "${train_use_gpu_key}" "${train_use_gpu}")
if [ ${#ips} -le 26 ];then
save_log="${LOG_PATH}/${trainer}_gpus_${gpu}_autocast_${autocast}"
nodes=1
else
IFS=","
ips_array=(${ips})
IFS="|"
nodes=${#ips_array[@]}
save_log="${LOG_PATH}/${trainer}_gpus_${gpu}_autocast_${autocast}_nodes_${nodes}"
fi
set_save_model=$(func_set_params "${save_model_key}" "${save_log}")
if [ ${#gpu} -le 2 ];then # train with cpu or single gpu
cmd="${python} ${run_train} ${set_use_gpu} ${set_save_model} ${set_epoch} ${set_pretrain} ${set_autocast} ${set_batchsize} ${set_train_params1} ${set_amp_config} "
elif [ ${#ips} -le 26 ];then # train with multi-gpu
cmd="${python} -m paddle.distributed.launch --gpus=${gpu} ${run_train} ${set_use_gpu} ${set_save_model} ${set_epoch} ${set_pretrain} ${set_autocast} ${set_batchsize} ${set_train_params1} ${set_amp_config}"
else # train with multi-machine
cmd="${python} -m paddle.distributed.launch --ips=${ips} --gpus=${gpu} ${run_train} ${set_use_gpu} ${set_save_model} ${set_pretrain} ${set_epoch} ${set_autocast} ${set_batchsize} ${set_train_params1} ${set_amp_config}"
fi
# run train
eval "unset CUDA_VISIBLE_DEVICES"
eval $cmd
echo $cmd
status_check $? "${cmd}" "${status_log}"
set_eval_pretrain=$(func_set_params "${pretrain_model_key}" "${save_log}/${train_model_name}")
# save norm trained models to set pretrain for pact training and fpgm training
# run eval
if [ ${eval_py} != "null" ]; then
set_eval_params1=$(func_set_params "${eval_key1}" "${eval_value1}")
eval_cmd="${python} ${eval_py} ${set_eval_pretrain} ${set_use_gpu} ${set_eval_params1}"
eval $eval_cmd
status_check $? "${eval_cmd}" "${status_log}"
fi
# run export model
if [ ${run_export} != "null" ]; then
# run export model
save_infer_path="${save_log}"
set_export_weight="${save_log}/${train_model_name}"
set_export_weight_path=$( echo ${set_export_weight})
set_save_infer_key="${save_infer_key} ${save_infer_path}"
export_cmd="${python} ${run_export} ${set_export_weight_path} ${set_save_infer_key}"
eval "$export_cmd"
status_check $? "${export_cmd}" "${status_log}"
#run inference
eval $env
save_infer_path="${save_log}"
if [ ${inference_dir} != "null" ] && [ ${inference_dir} != '##' ]; then
infer_model_dir="${save_infer_path}/${inference_dir}"
else
infer_model_dir=${save_infer_path}
fi
func_inference "${python}" "${inference_py}" "${infer_model_dir}" "${LOG_PATH}" "${train_infer_img_dir}" "${flag_quant}"
eval "unset CUDA_VISIBLE_DEVICES"
fi
done # done with: for trainer in ${trainer_list[*]}; do
done # done with: for autocast in ${autocast_list[*]}; do
done # done with: for gpu in ${gpu_list[*]}; do
fi # end if [ ${MODE} = "infer" ]; then
Markdown is supported
0% 或 .
You are about to add 0 people to the discussion. Proceed with caution.
先完成此消息的编辑！
想要评论请 注册 或