Thanks Daniël de Kok for adding Exl2 support into TGI.

You can now enjoy 2, 3 or 6bpw models running at insane speeds (120 tok/s latency on single A10g for Llama3).

github.com/huggingface/te…

working hard on PaliGemma tutorial. video will be out on Monday.

we will teach PaliGemma to recognize handwritten digits. stay tuned: youtube.com/@roboflow

what do you think about the thumbnail?

PaliGemma in a nutshell: not made for zero-shot but rather use it as a backbone to fine-tune for any vision-language task

hardest thing in video finetuning is actually finding a dataset
second hardest is downsampling without compromising on model performance (plus add finding appropriate bsz on top) but that's problem for another day

I'm actually taking time off but this paper excited me a lot so I'd highly suggest to check the demo out 😂

If you liked the scribble SDXL ControlNet, wait until you try the
⚡️Flash ⚡️version 👀

with SDXL Flash you can now get super fast SDXL ControlNet without compromising on quality ✨

Check out 10-steps SDXL ControlNet (supports canny too!) 👉🏻huggingface.co/spaces/linoyts…

document AI landed to Hugging Face chat, powered by open-source models (@cohere Command-R+) with tool calling 🛠️✨

explore many tools including image generation, image editing and more at Hugging Chat 🤗💬

I have created Hugging Face Space, where you can simultaneously test YOLOv8, YOLOv9, and YOLOv10 on your images.

link: huggingface.co/spaces/Skalski…

nice paper by nvidia to do visual fact-checking, would be great to reproduce using open-source models
huggingface.co/papers/2404.19…