pustar

Why does the exported end2end yolov7.onnx only get a mAP of 0.46721 on COCO dataset? while yolov7.pt's mAP is 0.512

4

I follow the tutorial to generate an end2end yolov7.onnx : https://github.com/WongKinYiu/yolov7/blob/main/tools/YOLOv7onnx.ipynb however, the exported end2end yolov7.onnx only get a mAP of 0.46721 on COCO dataset, while yolov7.pt is 0.512 ,...

I generated an yolov7.onnx by command: `python export.py --weights weights/yolov7.pt --grid --simplify --dynamic` when I run the yolov7.onnx by: ``` # ONNX inference outputs = session.run(outname, {inname[0]:img})[0] ``` an error...

Which onnxruntime version did you use for this project? Ordinal 10 Could Not Be Located in the Dynamic Link Library ort_test.exe

4

Which onnxruntime version did you use for this project? Ordinal 10 Could Not Be Located in the Dynamic Link Library ort_test.exe  I have tested onnxruntime version 1.1.0, 1.1.1, 1.2.0,...

感谢您的分享，请问：导出的onnx，分别在coco上测试过精度吗？验证过正确性吗？

代码很好，很感谢，请问能提供一个BiLSTM-CRF的预训练模型吗？

cannot stand anymore, too much mistakes in this repo!!!

just want to say thank you for your share

当输入文本较短时，zero-shot 大多生成 0 秒噪声

3

当输入文本较短时，zero-shot 生成 0 秒噪声。 比如当文本为："一二线城市用地紧张."，生成 0 秒噪声 当将文本扩展为“"一二线城市用地紧张，房价居高不下，百姓负担重，不敢消费，生活压力大，苦于维持生计"”，能够生成较完整的语音，但是会多个"就"的发音。 如果使用这个语音数据集，zero-shot 生成的大多数声音，都是 0 秒噪声。 aishell_dataset = MsDataset.load('modelscope/speech_asr_aishell1_testsets',subset_name='default',split='test') 请帮忙看看是什么原因，能否解决，多谢！ 软件版本： ChatTTS branch: main commit: 51ec0c7 python 3.10 torch 2.3.0 torchaudio 2.3.0 os: ubuntu...

How to make CUDA and CPU generate the same speech when given same text?

1

Thank you for sharing such nice repo! CUDA and CPU now generate different audio (tone) when given same text. How to make CUDA and CPU generate the same speech (tone)...