The reason for evaluation metric difference between your Implemented baseline and MS-TCN

cmhungsteve / SSTDA

[CVPR 2020] Action Segmentation with Joint Self-Supervised Temporal Domain Adaptation (PyTorch)

https://arxiv.org/abs/2003.02824

MIT License

154 stars 23 forks source link

The reason for evaluation metric difference between your Implemented baseline and MS-TCN #6

Closed houzhijian closed 4 years ago

houzhijian commented 4 years ago

Hi, steve:

 Do you think what is the reason for the improvement between your implemented baseline and reported result in the original MS-TCN paper?

cmhungsteve commented 4 years ago

I think it's mainly about the hyper-parameters. I spent more time finding the best hyper-parameters (e.g. learning rate). Also, different machines and libraries also affect the numbers a little bit.

houzhijian commented 4 years ago

Got it. Thank you for your response.