How can I get frame_tsv/train_32frames.img.lineidx for MSVD?

[mohammadshahabuddin]

Hi,

There is no train_32frames.img.lineidx in the provided codes. How can I get train_32frames.img.tsv, train_128frames_img_size256.img.tsv, test_32frames.img.lineidx,train_32frames.img.lineidx??

What are the difference between train_32frames.img.tsv andtrain_128frames_img_size256.img.tsv?

[mohammadshahabuddin]

I have solved the lineidx-related issue. The files will be generated during training if generate_lineidx=True is mentioned in this file: src/utils/tsv_file.py

[tiesanguaixia]

Hi,

There is no train_32frames.img.lineidx in the provided codes. How can I get train_32frames.img.tsv, train_128frames_img_size256.img.tsv, test_32frames.img.lineidx,train_32frames.img.lineidx??

What are the difference between train_32frames.img.tsv andtrain_128frames_img_size256.img.tsv?

Hi! Have you reproduced the results in paper? May I ask did you adjust the value of 'loss_sparse_w' and the 'learning_rate' in command? For the 'loss_sparse_w', I guess it's the regularization hyperparameter of $Loss_{SPARSE}$ , i.e. the $\lambda$ in the paper. In the appendix, it seems like for MSR-VTT, the model performs best when $\lambda$ = 5. But the why the default value of 'loss_sparse_w' in command is 0.5? Do I need to adjust it to 5? Thank you a lot!

[tiesanguaixia]

generate_lineidx

And how you generate the tsv file by tsv_file.py?