Hello, this is excellent work. However, I have a few questions I would like to ask.
-
The paper states that CoVR-R contains 2,800 samples, but in the JSON file I downloaded from the Hugging Face platform, I only found 2,634 samples. Did I download the wrong file?
-
What exactly is the modification text used as input in the method described in the paper? I noticed that the modification text in merged_webvid_ss2.json is different from the edit field in dense-webvid8m-covr_test.
I would sincerely appreciate any clarification you could provide. Thank you very much for your time and help.
Hello, this is excellent work. However, I have a few questions I would like to ask.
The paper states that CoVR-R contains 2,800 samples, but in the JSON file I downloaded from the Hugging Face platform, I only found 2,634 samples. Did I download the wrong file?
What exactly is the modification text used as input in the method described in the paper? I noticed that the modification text in
merged_webvid_ss2.jsonis different from theeditfield indense-webvid8m-covr_test.I would sincerely appreciate any clarification you could provide. Thank you very much for your time and help.