SOTA

CV

Semantic Segmentation: OmniVec, ONE-PEACE, Segment Anything (https://arxiv.org/pdf/2304.02643.pdf)

Image Classification: OmniVec, ViT, EffNet

Object Detection: YOLO series v6, v8, Co-DETR (DEtection TRansformer (DETR))

Contrastive Learning: MoCo, Momentum Contrastive Learning

Image Retrieval: BLIP2, CLIP,

NLP:

Common Sense Reasoning: GPT-4, PaLM 2 (Google), ReAct(google), MMReAct (microsoft)

Image-to-Text Retrieval: BLIP2, ONE-PEACE, InternVL-G-FT