Hugging Face's logo
zeromodels 's Collections
LeViT
BEiT
Table Transformer
RegNet
PVT & PVTv2
EfficientDet
Gemma
Qwen2
DETR
RF-DETR
D-FINE
RT-DETR V1 and V2
OWLv2
OWL-ViT
SegFormer
SAM V1,V2,V3
OneFormer
DeeplabV3
EoMT
MaskFormer
Mask2Former
MobileViT V1 and V2
DINO V1,V2,V3
Depth Anything V1 and V2
CLIP
MetaCLIP 2
SigLIP
SigLIP2
LocateAnything
Whisper
Speech2Text
Moonshine
Granite Speech
Granite Speech Plus
Gemma 2
CaiT
ConvMixer
ConvNeXt
ConvNeXt-V2
DeiT and DeiT3
DenseNet
EfficientFormer
FlexiViT
InceptionNeXt
Inception-ResNet-V2
Inception-V3
Inception-V4
MaxViT
MLP-Mixer
MobileNetV2
MobileNetV3
Next-ViT
PiT
PoolFormer
Res2Net
ResMLP
ResNet
ResNetV2 (BiT)
ResNeXt
SE-ResNet and SE-ResNeXt
Swin Transformer
Swin Transformer V2
EfficientNet
EfficientNet-Lite
EfficientNetV2
VGG
ViT
Xception
MiT (SegFormer encoder)
MobileViT
MobileViT-V2
RoBERTa
BERT
XLM-RoBERTa
DeBERTa (v1 / v2 / v3)
DeepSeek-VL
Janus-Pro
GPT and GPT-2
GPT-OSS
Gemma 3
Gemma 4
Gemma 3n
Qwen2.5-VL
Qwen2-VL
Qwen3
Qwen3-VL
Qwen3.5
Qwen3-Next
Qwen3.5-MoE
Qwen3-VL-MoE
Qwen2-MoE
Qwen3-MoE
Qwen3.8
InternVL
GLM 4, 4.5, 4.7, Z1
T5
ModernBert
ELECTRA
MobileNetV4
TIPSv2
TIPSv2-DPT

Janus-Pro

Pure Keras 3, backend-agnostic (JAX / TensorFlow / PyTorch) ports of the DeepSeek Janus-Pro any-to-any multimodal models.