search

lyuchenyang / Macaw-LLM

Macaw-LLM: Multi-Modal Language Modeling with Image, Video, Audio, and Text Integration
Apache License 2.0
1.52k stars 121 forks source link

missing file 'data/avsd/avsd_train.json' #4

Closed wanghao-cst closed 1 year ago

wanghao-cst commented 1 year ago

Hi, may I know what are these two files in preprocess_data_supervised.py? Could you please share these files?

image
wanghao-cst commented 1 year ago

In addition, I am not sure about these two files appeared in preprocess_data_supervised.py.

image

I guess that they are modified from coco annotation. Could you please also share these files?

lyuchenyang commented 1 year ago

Hi,

Really sorry for the late reply due to my personal reasons. Regarding your questions, you can download the metadata avsd_train.json and avsd_val.json for AVSD dataset in https://video-dialog.com/. For VQA dataset, you can directly download them from https://cocodataset.org/#download, just nevermind the added_path postfix.

Thanks, Chenyang

lyuchenyang commented 1 year ago

You can download the questions from https://visualqa.org/download.html.

wanghao-cst commented 1 year ago

You can download the questions from https://visualqa.org/download.html.

Thank you!

wanghao-cst commented 1 year ago

You can download the questions from https://visualqa.org/download.html.

Hi, may I konw where to get these two files?

image
lyuchenyang commented 1 year ago

This is the data generated by ourselves, we will release them as soon as possible - no worries you can just use the supervised dataset.