tristan

请问你们有没有对<unk> 做词嵌入

1

如题。我现在在用literature Word + Character + Ngram，如果有训练的话可否方便发一下。

It's time-consuming at my first use of edgebundleR. Because it has numorous reference packages and my program had to spend almost 20 mins to download all of the dependences. May...

Hi, thanks for your code! I searched with keyword 'train_mode' by the definition of `parser.add_argument('--train_mode', type=str, required=True, help='Specify the training mode.', choices=["fixed", "variable"]) ` But I cannot find the code...

Thanks!

Hi Praveen, MLAD is a very inspiring work to me. I saw in dataloader.py that the `collated_fn` outputs clip features of varied lengths. But I haven't seen how MLAD handle...

Question about gflops calculation

1

Hi Rui, Thanks for your work! Could you please tell me the settings to calculate the gflops? The only thing I found from your paper and appendix is 'taking as...

Hi, Thanks for your solid theoretical and experiment proof. But as mentioned in the paper, your work is trained in an end-to-end way while others, e.g. mlad, used pretrained i3d...