`transformers` + `torchao` quantization + `torch.compile` for faster inference speed and less memory usage 🔥
Demo of "meta-llama/Meta-Llama-3.1-8B-Instruct" quantized in 4-bit weight-only:
Several Hugging Face team members are coming to SF for the PyTorch Conference, and we'll celebrate in style.
Come join the 🌟Hugging Face Party🌟 at the PyTorch Conference the 19th of September!
Many community builders are coming in; it'll be the best opportunity to meet and discuss open-source AI.
Join the waitlist by joining the HF org: https://lnkd.in/e7V_iNiF
Looking forward to seeing you all at the conference and party!
Several Hugging Face team members are coming to SF for the PyTorch Conference, and we'll celebrate in style.
Come join the 🌟Hugging Face Party🌟 at the PyTorch Conference the 19th of September!
Many community builders are coming in; it'll be the best opportunity to meet and discuss open-source AI.
Join the waitlist by joining the HF org: https://lnkd.in/e7V_iNiF
Looking forward to seeing you all at the conference and party!
An open source UI to train your own Flux LoRA just landed on Hugging Face 🚀 Also, probably the easiest and cheapest (local training also supported). Link to space in first comment.
📢🔥Hot New Release: CogVideoX-5B, a new open-weights text-to-video model from the group behind very successful GLM LLM series.
CogVideoX-5B - is competitive to Runway, OpenSora, Pika, and Luma! More highlights:
- GPU vram requirement on Diffusers: 20.7GB for BF16 and 11.4GB for INT8
- Inference for 50 steps on BF16: 90s on H100 and 180s on A100
- open weights
- Video length: 6s, Resolution: 720 x 480
- Best news: Previously released CogVideoX-2B is now switched to Apache 2.0 from previous custom license! 🥳🎉🍾
Build locally: https://lnkd.in/gkhNbFvaHugging Face Gradio app: https://lnkd.in/gJsgBnQS
CogVideoX 5B - Open weights Text to Video AI model is out, matching the likes of luma/ runway/ pika! 🔥
Powered by diffusers - requires less than 10GB VRAM to run inference! ⚡
Checkout the free demo below to play with it!
Upgrade to `diffusers` latest and enjoy the latest Cog-5B release from the ChatGLM team.
Our integration comes with `torch.compile()` support and Aryan V S has relentlessly worked on reducing Cog's memory requirements from 33GB to 8 GB.
Know everything here:
https://lnkd.in/g6JPHEPD
Meta's Sapiens for Body-part Segmentation!
Sapiens is a human-centric family of foundational models trained by Meta Reality Labs.
Check out the lightening fast Gradio app for Sapiens, brought to you by FASHN AI, on Hugging Face Spaces: https://lnkd.in/gXazMaAT
Project page: https://lnkd.in/gjQ6Qykg
Models are all available on 🤗 Hub!