Simple question about CogVideoX-I2V model

[KIMGEONUNG]

Hello, Thank you for the invaluable work.
I have a simple question.
Is CogvideoX-I2V model finetuned from CogvideoX-T2V model or is it trained from scratch?

[OrangeSodahub]

Hi, accroding to their paper, I think it is finetuned #751

[zRzRzRzRzRzRzR]

Yes, the I2V model is trained by doubling the channels of the T2V extension.

[OrangeSodahub]

@zRzRzRzRzRzRzR How you place the pretrained T2V parameters? The first 16 channels?

[zRzRzRzRzRzRzR]

Yes, then expanded to 16 channels.

[OrangeSodahub]

@zRzRzRzRzRzRzR What do you mean then expanded to 16 channels?......