About training settings

[NiuPDJ]

Hello, thank you very much for your work and the code provided! I have a small issue: I followed the training settings from the original paper, and the training was completed within 50 epochs. However, when I trained using the trasin_SCD_psd settings, even after 50, 100, or more epochs, I still did not meet the performance metrics mentioned in the paper. Could you please suggest what might be wrong with my approach?

[DingLei14]

Hello, thank you very much for your work and the code provided! I have a small issue: I followed the training settings from the original paper, and the training was completed within 50 epochs. However, when I trained using the trasin_SCD_psd settings, even after 50, 100, or more epochs, I still did not meet the performance metrics mentioned in the paper. Could you please suggest what might be wrong with my approach?

Hi. The training script in this repository is not the final version. I will check and update it soon. For evaluation you can use the trained weights in the welcome page.

[NiuPDJ]

Hello, thank you very much for your work and the code provided! I have a small issue: I followed the training settings from the original paper, and the training was completed within 50 epochs. However, when I trained using the trasin_SCD_psd settings, even after 50, 100, or more epochs, I still did not meet the performance metrics mentioned in the paper. Could you please suggest what might be wrong with my approach?

Hi. The training script in this repository is not the final version. I will check and update it soon. For evaluation you can use the trained weights in the welcome page.

Thank you very much! The pth file used for evaluation has already been used, and it has no issues. I am still eager to learn about your settings and tips for training from scratch. I'm really looking forward to seeing your updated code soon!

[wujiang0156]

Hello, thank you very much for your work and the code provided! I ALSO have a small issue: I followed the training settings from the original paper, and the training was completed within 50 epochs. However, when I trained using the trasin_SCD_psd settings, even after 50, 100, or more epochs, I still did not meet the performance metrics mentioned in YOUR paper? I am ALSO eager to learn about your settings and tips for training from scratch. I'm really looking forward to seeing your FINAL code soon!