Hugging Face's logo
timm 's Collections
timm RADIO v3/v4 ViT Encoders
timm multi-label image datasets
timm DeepSeek-V4 ViT Encoders
timm Sapiens2 ViT Encoders
timm Qwen3 ViT Encoders
timm DINOv3
Perception Encoder
MobileCLIP-2
SigLIP 2
Searching for Better ViT Baselines
MetaCLIP
MobileNetV4 pretrained weights
timm Top-20 ImageNet-1k Models
timm Top-20 Fastest Models
timm ImageNet-12k Models
timm Takes on the Classics
Fastest timm models > 75.3% IN-1k Top-1 (Original ResNet-50)
Fastest timm models > 80% Top-1 ImageNet-1k
Fastest timm models > 83% ImageNet-1k Top-1
Fastest timm models > 86% ImageNet-1k Top-1
Fastest timm models > 88% ImageNet-1k Top-1
timm Backbones
All the ImageNets
Fine-Tune Image Classification Benchmark Datasets
timm tiny test models

timm DINOv3

Meta AI's DINOv3 weights in timm. ViTs with `qkvb` have a zero QV bias present, otherwise bias is disabled. QKV bias are all 0 in original weights.