Open OliverHxh opened 5 months ago
Hi, thanks for your great work.
From Instruction Data, I notice that you provide your processed version of annotations. But as shown in instruction config file and videochat2, your provided annotations are not complete.
So, what is the difference between the files you provided and the videochat2 annotations? Which one should I use for tunning?
Thanks.
I wonder what is the difference between the magic.json and videochat2.json too.
+1, I find that dataset is same with videochat2 writed in paper.
Hi, thanks for your great work.
From Instruction Data, I notice that you provide your processed version of annotations. But as shown in instruction config file and videochat2, your provided annotations are not complete.
So, what is the difference between the files you provided and the videochat2 annotations? Which one should I use for tunning?
Thanks.