PaddlePaddle / Paddle-Lite

PaddlePaddle High Performance Deep Learning Inference Engine for Mobile and Edge (飞桨高性能深度学习端侧推理引擎)
https://www.paddlepaddle.org.cn/lite
Apache License 2.0
6.93k stars 1.61k forks source link

PaddleSeg 的 RTFomer 模型转.nb报错 #9933

Open WYH150123 opened 1 year ago

WYH150123 commented 1 year ago

PaddleSeg 的 RTFomer 模型转.nb报错,具体信息如下:

paddle-bot[bot] commented 1 year ago

您好,我们已经收到了您的问题,会安排技术人员尽快解答您的问题,请耐心等待。请您再次检查是否提供了清晰的问题描述、复现代码、环境&版本、报错信息等。同时,您也可以通过查看官网文档常见问题历史Issue来寻求解答。祝您生活愉快~

Hi! We've received your issue and please be patient to get responded. We will arrange technicians to answer your questions as soon as possible. Please make sure that you have posted enough message to demo your request. You may also check out the APIFAQ and Github Issue to get the answer.Have a nice day!

hong19860320 commented 1 year ago

基于最新的 develop 分支和 demo 已复现了该问题,发现 android 或 linux arm64 下执行均存在问题 log.txt ,x86 下没有问题。@mjp9527 帮忙看看吧?

复现方法:

$ wget https://paddlelite-demo.bj.bcebos.com/devices/generic/PaddleLite-generic-demo_v2_12_0.tar.gz
$ tar -xvf PaddleLite-generic-demo_v2_12_0.tar.gz
$ cd PaddleLite-generic-demo/model_test/assets/models

将模型下载并解压到当前目录,模型文件夹为 RTFormer_infer_model,注意:需确保模型文件夹中的 .pdmodel 和 .pdiparams 文件名分别为 model.pdmodel 和 model.pdiparams 。

$ cd ../../../shell/
For Android
$ ./run_with_adb.sh RTFormer_infer_model 1,3,1024,512 float32 float32 android arm64-v8a cpu UQG0220A15000356
For Arm Linux(远程)
$ ./run_with_ssh.sh RTFormer_infer_model 1,3,1024,512 float32 float32 linux arm64 cpu 192.168.100.30 22 khadas khadas
For x86 Linux(远程)
$ ./run_with_ssh.sh RTFormer_infer_model 1,3,1024,512 float32 float32 linux amd64 cpu localhost 22 root root
For x86 Linux(本地)
$ ./run.sh RTFormer_infer_model 1,3,1024,512 float32 float32 linux amd64 cpu
WYH150123 commented 1 year ago

@hong19860320 @mjp9527 , 感谢支持,请问目前有什么进展吗? B.R/WYH

hong19860320 commented 1 year ago

目前定位存在如下异常结构(conv 的 weight 不是常量,而是模型中间的结果),lite 不支持,正在和套件同学确认该结构是否合法。 image

WYH150123 commented 1 year ago

@hong19860320 , 感谢大力支持,我们项目运行移动平台,RTFomer训练结果非常好,我们需要替换我们之前的实时分割模型,转.nb,并量化Slim后,导入,盼早日解决^_^ B.R/WYH

WYH150123 commented 1 year ago

@hong19860320 @mjp9527 , 针对该ISSUE, 目前有解决计划和相关进展吗?

WYH150123 commented 1 year ago

@hong19860320 @mjp9527 , 该ISSUE, 目前进展如何,是否已经修复了?

WYH150123 commented 1 year ago

@hong19860320 @mjp9527 ,烦请更新一下最新进展,谢谢