DeepSeek-VL - a kerasformers Collection

kerasformers 's Collections

DeBERTa (v1 / v2 / v3)

MiT (SegFormer encoder)

EfficientNet-Lite

Swin Transformer V2

Swin Transformer

SE-ResNet and SE-ResNeXt

Inception-ResNet-V2

EfficientFormer

Granite Speech Plus

Depth Anything V1 and V2

MobileViT V1 and V2

RT-DETR V1 and V2

DeepSeek-VL

updated 2 days ago

Pure Keras 3, backend-agnostic (JAX / TensorFlow / PyTorch) ports of the DeepSeek-VL vision-language models.