Kijai

/

CogVideoX-5b-1.5

like 22

Currently only in 1.5 test branch of my ComfyUI wrapper nodes: https://github.com/kijai/ComfyUI-CogVideoXWrapper/tree/1.5_test

Thanks for your reply, have you tried to use the lora model trained on cogvideox on cogvideox 1.5?

Thanks for your reply, have you tried to use the lora model trained on cogvideox on cogvideox 1.5?

Yes, the DimensionX orbit LoRAs for example work pretty good.

Thanks for your reply, have you tried to use the lora model trained on cogvideox on cogvideox 1.5?

Yes, the DimensionX orbit LoRAs for example work pretty good.

That's really great, I actually did some sat to diffusers conversion work but my reasoning was poor, can you briefly say what changes were made to the reasoning section compared to before? I've added ofs_emb to the inference section, turned off pos_embedding in PatchEmbed, and adjusted the output dimension of proj.out.

That part is not really my work, I'm just adapting the current diffusers PR draft: https://github.com/huggingface/diffusers/pull/9877

That part is not really my work, I'm just adapting the current diffusers PR draft: https://github.com/huggingface/diffusers/pull/9877

Thank you! I understand.