Can Discrete Tokens Serve as Semantically Lossless Representations of Visual Signals?

Open paper

Benchmark	Model	Rank	Results
optical-character-recognition-ocr-on-ocrbench	FVQ (8B)	#22	Score: 867
chart-question-answering-on-chartqa	FVQ (8B)	#52	1:1 Accuracy: 83.9