Given groups=1, weight of size [1536, 32, 1, 2, 2], expected input[1, 16, 11, 128, 128] to have 32 channels, but got 16 channels instead
Hi,
I wanted to try this LoRA, but I get the following error at the Ksampler stage:
"Given groups=1, weight of size [1536, 32, 1, 2, 2], expected input[1, 16, 11, 128, 128] to have 32 channels, but got 16 channels instead"
I've used the provided example workflow without any modifications. Any ideas why this happens? Thanks in advance!
Hi,
I wanted to try this LoRA, but I get the following error at the Ksampler stage:
"Given groups=1, weight of size [1536, 32, 1, 2, 2], expected input[1, 16, 11, 128, 128] to have 32 channels, but got 16 channels instead"I've used the provided example workflow without any modifications. Any ideas why this happens? Thanks in advance!
The support seems to be kinda unstable. It sometimes randomly fail on my side after the server is up for a couple of hours and running dozens of task, even though I've pretty much updated the UI to the latest one. Some of the times you just need to reboot the server (closing the command prompt and reopen it, not just clicking restart in the manager), other times you just need to wait a while till the machine is happy.
For you, the first priority is definitely going to be updating the UI to whatever the latest version is, then you just pray for it to work from then on, occasionally restarting the server when it decides to not cooperate.
BTW I'm running on Windows 11 with 6GB VRAM + 16GB RAM, for other operating systems and system specs, mileage may vary.
I encountered the same problem. My quick and dirty solution was to convert it to fp16 using the convert script from musubi tuner.
I encountered the same problem. My quick and dirty solution was to convert it to fp16 using the convert script from musubi tuner.
Could you kindly upload the converted version somewhere? I have no idea how to install or use musubi tuner. :/
https://huggingface.co/spacepxl/Wan2.1-control-loras/discussions/2
Thanks. Unfortunately, none of the suggestions on that link worked for me.
I got nearly same problem "Given groups=1, weight of size [1536, 32, 1, 2, 2], expected input[1, 16, 13, 60, 90] to have 32 channels, but got 16 channels instead"