ju-chen / Efficient-Prompt

MIT License
186 stars 13 forks source link

Intersection between test and train categories #15

Open Coder-Liuu opened 11 months ago

Coder-Liuu commented 11 months ago

Thank you for opening the source code, in the zero sample temporal action recognition, I sent the HMDB51 dataset and actually read train_split01.txt and text_split01.txt, which I guess are the training set and the test set, and the categories should not be duplicated as I am in the zero sample task. But why are the action categories duplicated in these two files? Looking forward to your reply.

Coder-Liuu commented 11 months ago

In addition, I split train_split01.txt into train_split02.txt and test_split02.txt based on the principle of non-duplication of action categories, and I found that the accuracy is very low in the test set. Can you answer this?