Hugging Face
Models
Datasets
Spaces
Community
Docs
Enterprise
Pricing
Log In
Sign Up
1
13
Tranheden
WilhelmT
Follow
swaze's profile picture
tommulder's profile picture
salomons's profile picture
4 followers
ยท
1 following
AI & ML interests
None yet
Recent Activity
reacted
to
JonnaMat
's
post
with ๐ฅ
about 17 hours ago
๐คฏ Edge-Grade Vision Reasoning. Now Practically Lossless. ๐คฏ Introducing ๐ https://huggingface.co/embedl/Cosmos-Reason2-2B-W4A16-Edge2 Optimized for Jetson Orin Nano Super and AGX Orin https://huggingface.co/nvidia . ๐ Try it out on Jetson (image+video+text): ``` docker run --rm -it \ --network host \ --shm-size=8g \ --ulimit memlock=-1 \ --ulimit stack=67108864 \ --runtime=nvidia \ --name=vllm-serve \ -e HF_TOKEN=hf_*** \ -e HF_HOME=/root/.cache/huggingface \ ghcr.io/nvidia-ai-iot/vllm:latest-jetson-orin \ vllm serve "embedl/Cosmos-Reason2-2B-W4A16-Edge2" \ --max-model-len 8192 \ --gpu-memory-utilization 0.75 \ --max-num-seqs 2 ``` ๐ค What is Edge2? Most weights โ INT4 | Activations โ FP16 | Select sensitive layers โ kept in FP16. Edge2 preserves precision where it matters most; while keeping the model small and fast enough for edge GPUs. ๐
liked
a model
1 day ago
embedl/Cosmos-Reason2-2B-NVFP4A16
liked
a model
1 day ago
embedl/Cosmos-Reason2-2B-W4A16-Edge2
View all activity
Organizations
WilhelmT
's datasets
None public yet