search

yeyupiaoling / PaddlePaddle-DeepSpeech

基于PaddlePaddle实现的语音识别，中文语音识别。项目完善，识别效果好。支持Windows，Linux下训练和预测，支持Nvidia Jetson开发板预测。
https://yeyupiaoling.blog.csdn.net/article/details/102904306
Apache License 2.0
667 stars 145 forks source link

执行export_model.py时候报错 #119

Closed AryaStar closed 2 years ago

AryaStar commented 2 years ago

您好！ program.set_state_dict(load_state_dict)这一句： TypeError: Type of state_dict should be dict, but received <class 'paddle.Tensor'>. 请问应该怎么解决

yeyupiaoling commented 2 years ago

谢谢！已收到您的邮件，祝您开心每一天

yeyupiaoling commented 2 years ago

你是使用我提供下载的模型吗？

yeyupiaoling commented 2 years ago

你提供更多的错误信息看看

AryaStar commented 2 years ago

谢谢您，我使用的是这个链接https://download.csdn.net/download/qq_33200967/21773253?utm_medium=distribute.pc_relevant_download.none-task-download-2~default~OPENSEARCH~default-3.test_version_3&depth_1-utm_source=distribute.pc_relevant_download.none-task-download-2~default~OPENSEARCH~default-3.test_version_3&dest=https%3A%2F%2Fdownload.csdn.net%2Fdownload%2Fqq_33200967%2F21773253&spm=1003.2020.3001.6616.3

下载的模型\models\step_final\params.pdparams

然后我使用了老版本的release执行export_model.py成功实现了 但是我使用新版本的master分支执行export_model.py失败了

具体报错如下： Traceback (most recent call last): File ".\export_model.py", line 32, in ds2_model.export_model(model_path=args.save_model_path) File "C:\Users\CS8DA\Desktop\PaddlePaddle-DeepSpeech-master\model_utils\model.py", line 438, in export_model self.load_param(self.infer_program, self._resume_model) File "C:\Users\CS8DA\Desktop\PaddlePaddle-DeepSpeech-master\model_utils\model.py", line 150, in load_param program.set_state_dict(load_state_dict) File "E:\ProgramData\Anaconda3\envs\myaudio\lib\site-packages\paddle\fluid\framework.py", line 5804, in set_state_dict type(state_dict))) TypeError: Type of state_dict should be dict, but received <class 'paddle.fluid.core_avx.LoDTensor'>.

yeyupiaoling commented 2 years ago

老版本和新版本的模型不通用的，请在对应页面重新下载。

AryaStar commented 2 years ago

谢谢您，已成功export_model