apolinário

An odd additional request: If we implement something like a `maximumChoices` parameter, please have it start with `1`. It know it sounds weird and would be more logical that if...

I'm also facing a similar issue: ```py generator = pipeline("text2text-generation", model="philschmid/flan-t5-xxl-sharded-fp16", model_kwargs={"load_in_8bit":True, "device_map": "auto"}) output = generator(prompt, penalty_alpha=0.6, top_k=4, max_length=256) ``` Gives me the error: ``` RuntimeError: "softmax_lastdim_kernel_impl" not implemented...

Just to flag, the error I faced [here ](https://github.com/huggingface/transformers/issues/21151#issuecomment-1427589596) still exists with @sam-ulrich1's fix. Should I open a new Issue as this may be related specifically to 8-bit?

What if I want to share only the title and a URL, without an attached image, isn't there a way?

When I do that, the 'url' parameter is ignored as well, and only the title is shared

Ok. I'll look forward to it. I really don't have a local image to share in every link I'd like to share

This worked with `num_in_channels=8` (as in: didn't error). However perceptually isn't behaving as it should Edit image:  Edit prompt `Turn sky into a cloudy one`:  ```py import torch...

> Not sure if it's the exact guidance formulation that we have in the InstructPix2Pix pipeline though. That would matter a lot. ComfyUI uses the same `InstructPix2PixConditioning` node for it...

Awesome! What's the best way to proceed here? Modify the pipeline to detect if scaling is needed or not or create a new one?

CADS is a technique that greatly increases diversity of generated images by adding scheduled noise to the conditioning at inference time.  Here's an implementation of it as an AUTOMATIC1111...