deepbeepmeep

Windows support ?

6

Hello I have been unable to run the model on Windows since the install fails as it requires Triton that is only supported on Linux. Any idea ? Thanks in...

Inconsistent ImageEditor behaviour once modified serverside

1

### Describe the bug - Upload a picture in ImageEditor - Draw some pattern A on Layer 1 - Press the button "Load" to change ONLY its background on the...

Being able to edit a mask on top of an image is now a very common requirement for lots of Visual models. In v3 there used to be an Image...

Strong image distortion

1

Hi Thank you very much for your great work. I have tried your model using reference images of my own and the end result is often not visually pleasing at...

Congratulations for the best LLaVA derived models !

1

I have been playing with most multimodal models based on LLaVA models and I can tell that mini Gemini (the 13B version) is one of the best if not the...

Memory Leak

2

Thanks for this very interesting work that at last allows us to have some level of direction in video generation. First run works quite well (less than 40s on a...

Simple support for Flash attention 2 and Sdpa attention

1

Hi! I think Ovis is one of the best Vision model around. I have noticed that Flash attention 2 and sdpa are disabled but can be easily turned on by...

Vace Frame Outpainting Improvement

2

Hi @hanzhn and the rest of the team, Vace is certainly my favorite Wan model given how versatile and powerful it is. I have implemented a UI on top of...

Fused attention error while running Nvidia Cosmos

4

Hello I am trying to run the latest Nvidia Cosmos model on a RTX 4090 and I get an error when fused attention is called : Line 1080 in fused_attn.py...

Any planned control net ?

4

Thanks again for your great work. I have been experimenting with the diffusion forcing model and obtained some impressive results. However when it is used to animate people progressively the...