Dain Kim

icon indicating an email [email protected]

icon location =)

Results 4 comments of Dain Kim

[Bug]: LoRA doesn't have impact on the output

I'm struggling with the same error

Applying Blockwise Weights to LoRA

Awesome, thanks!

trilinear_kernel

It can be fixed by replacing some files. Check out [here](https://github.com/bcmi/libcom/pull/27/commits/a44daf2d81dab2ae83d9a36466dca9f9defeb45c).

requires_grad in AttnProcessor

Thanks for replying! But what would be the purpose of not saving the normal inference process? Does it affect the result? Also, if I want to visualize the attention map...