X2Paddle
X2Paddle copied to clipboard
使用pytorch2paddle接口,转换pytorch的AlexNet模型到paddlelite格式报错
使用pytorch2paddle接口,转换pytorch的AlexNet模型到paddlelite格式报错
- 错误信息
- INFO:root:Now translating model from pytorch to paddle. INFO:root:Model optimizing ... trace_fc_fuse_pass done! INFO:root:Model optimized. Exporting inference model from python code ('/data/code/PaddleLite/X2Paddle/pd_model_trace/x2paddle_code.py')...
C++ Traceback (most recent call last):
No stack trace in paddle, may be caused by external reasons.
Error Message Summary:
FatalError: Illegal instruction
is detected by the operating system.
[TimeInfo: *** Aborted at 1647616421 (unix time) try "date -d @1647616421" if you are using GNU date ***]
[SignalInfo: *** SIGILL (@0x7fe4432c1125) received by PID 119982 (TID 0x7fe630e55740) from PID 1126961445 ***]
- 转换模型后用处
- [ ] 使用 Paddle 框架/ PaddleInference 推理预测
- [√] 使用 Paddle-Lite 做移动端推理
- [ ] 转换预训练参数,再使用 Paddle 进行模型开发
- 模型来源 AlexNet:https://download.pytorch.org/models/alexnet-owt-4df8aa71.pth
- 应用场景 用于移动端业务下的检测业务
- 版本信息 PaddlePaddle => :2.2.0 X2Paddle => :2.9 来源框架版本 :PyTorch
@BgLoveXixi 是有生成静态图模型也就是inference model但是转换Lite格式错误吗,如果是的话可以去Paddle Lite Issue下反馈相关错误
@BgLoveXixi 是有生成静态图模型也就是inference model但是转换Lite格式错误吗,如果是的话可以去Paddle Lite Issue下反馈相关错误
是的,好像有生成inference model,但是转成lite没成功:
我已经提交到lite下面,麻烦看下是什么问题
related to https://github.com/PaddlePaddle/Paddle-Lite/issues/8677