search

kijai / ComfyUI-CogVideoXWrapper

1.02k stars 61 forks source link

How to load CogVideoX_5b_Tora_GGUF_Q4_0.safetensors #166

Open APCOTech opened 1 month ago

APCOTech commented 1 month ago

How to load CogVideoX_5b_Tora_GGUF_Q4_0.safetensors ?

From you examples , The (Down)load Tora Model tries to download Kijai/CogVideoX-5b-Tora

kijai commented 1 month ago

How to load CogVideoX_5b_Tora_GGUF_Q4_0.safetensors ?

From you examples , The (Down)load Tora Model tries to download Kijai/CogVideoX-5b-Tora

It just downloads the control parts of Tora with that node, I changed it to that as I reali<ed it can work with any of the original 5b models.

You can use the GGUF model with the normal GGUF loader, that is just the new transformer part of Tora. I don't really know yet what's different with that, except that it seems just bit better overall.

APCOTech commented 1 month ago

Thank you for the fast answer. I will try it now... you are a hero :)

APCOTech commented 1 month ago

I can't run it at all.. I am not an expert but I try to play with settings.. I always get OOM or this: Given groups=1, weight of size [3072, 16, 2, 2], expected input[26, 32, 60, 90] to have 16 channels, but got 32 channels instead

littleyeson commented 4 weeks ago

I can't run it at all.. I am not an expert but I try to play with settings.. I always get OOM or this: Given groups=1, weight of size [3072, 16, 2, 2], expected input[26, 32, 60, 90] to have 16 channels, but got 32 channels instead

This model may be not use image2Video.just for txt2video.