tiesanguaixia

训练过程中使用到的数据集

1

您好，感谢您的工作和开源代码👍👍👍！我想请教一下： - 在训练Osprey的整个过程中是否有使用到一些视频领域的多模态数据集呢？比如MSR-VTT, MSVD和VATEX. - 我看您使用到了COCO, RefCOCO等数据集，他们是不是包含了MS-COCO呢？好像MS-COCO是COCO的子集😀 感激不尽！💐💐💐

Thank you for your great work! There are different numbers of captions for every video in `MSVD` dataset, such as 29, 42……But I found that the shape of _msvd_train_evalscores.pkl_ is...

Thank you for your amazing work! About the msrvtt_train_evalscores.pkl

3

Hi! Thank you for your amazing work! I want to ask how can I generate the _msrvtt_train_evalscores.pkl_? I see the shape is [6513,20], so which caption is respectively corresponding to...

Hi, I have a question that why the number of test videos in 'msvd_test_sequencelabel.h5' is 470?

9

I remember the the standard split contains 1.2K training videos, 100 validation videos, and **_670_** test videos, so why you did not use the other 200 test videos? Thank u!

Hi, I run the test.sh and something unexpected happened.

1

``` Traceback (most recent call last): File "test.py", line 130, in assert opt.feat_dims == test_loader.get_feat_dims() AssertionError ``` I did not make any change before this.