feffy380

>Is the idea behind use_8bit_adam CUDA specific, or is it just that there isn't an implementation for AMD cards? Probably the latter, assuming bitsandbytes isn't doing anything weird. In theory...

I'd suggest hiding the interface behind the Extra checkbox or at least moving it lower. It's quite large and pushes more commonly used options like CFG and Batch size/count off-screen.

@bmaltais Looking at the original aesthetic gradients repo, the personalization step involves performing gradient descent to make the prompt embedding more similar to the aesthetic embedding. In other words, it...

@20of @gshawn3 @nostepme you wouldn't happen to be using Seshelle's [CFG Rescale extension](https://github.com/Seshelle/CFG_Rescale_webui)? It doesn't work with the updated DDIM and UniPC implementations and it sounds like you're seeing the...

They just pushed a branch to the original repo that cleans up the code and implements it using pytorch hooks. This is much less invasive and can be easily implemented...

> Using a recursive parent-child model could do away with any kind of meta-state in instances like "overseer" or "worker" if _any_ instance could delegate tasks to child instances. Care...

>First one is really tricky. Is it? Couldn't you solve it the same way as the second one, i.e., remove the ability to create new agents depending on some condition...

Doesn't look like it. It's probably the biggest flaw with BabyAGI: the task creation agent *always* assumes it needs to create more tasks. It has no concept of when a...

It looks like this is already doable as a ConstitutionalChain with: Critique request: "Review your previous answer and find problems with your answer" Revision request: "Based on the problems you...

The issue here is the launch script now assumes the repo is at the hardcoded path `/home//stable-diffusion-webui` if not specified, when the previous behavior was to assume the repo is...