Can the LORA provided by you be used together with the LORA I trained myself?
Great job, thank you! Can the LORA provided by you be used together with the LORA I trained myself?How should I do it, please?
Hey :)
Loras trained for the 1.5 model should work with this one but we have not tested. Also in my experiments lora for model 1.0 did not work well with model 1.5.
You can try to use comfyui or diffusers methods to combine loras.
Thanks for all your hard work! Could I just check how much approximate data you used to train the LORA? I've tried training myself but haven't had much luck. Are you using the official cogvideox training code?
We used around 20k video and we have not used official cogvideox training code :)
Thank you for your reply, I have tried to train a lora model for one way of lens transport, the effect is more obvious on a small amount of video but not when the amount of data reaches 500+, I'm not sure what's the reason for this, do you have any suggestion?
Interesting, but I can't figure out why this is happening)