Dongjie Yang

Is there a certain time for releasing the pre-training code?

A new high-quality video-text dataset

1

We recently open-sourced a new fine-grained video-text dataset with 12K annotated high-resolution videos (~400k clips), called [Vript](https://github.com/mutonix/Vript), which may contribute to improving the open-sora models. GitHub link: https://github.com/mutonix/Vript

We recently open-sourced a new fine-grained video-text dataset with 12K annotated high-resolution videos (~400k clips), called [Vript](https://github.com/mutonix/Vript), which may contribute to improving the open-sora models. GitHub link: https://github.com/mutonix/Vript

h2o is not working when the input is short.

1

When I run `bash scripts/streaming/eval.sh h2o`, I find h2o is not working and acting exactly the same as the full. When I check the code https://github.com/FMInference/H2O/blob/8bb3fe14c10e6c55703f09956bac2d2eaf4f48bf/h2o_hf/utils_real_drop/modify_llama.py#L68 ``` def __call__(self, past_key_values,...

A new video-text dataset which may help

4

[Vript](https://github.com/mutonix/Vript) is a fine-grained video-text dataset with 12K annotated high-resolution videos (~400k clips), where each clip has a detailed caption of ~145 words.

About the caption the video with subtitles.

1

Great thanks to the great contribution of your work! I have some doubts about how you collect the subtitles. Do you directly download the subtitles from the youtube website or...

When will Emu 3 be released?

1

I am very looking forward to this fantastic work.