MEGA-Bench: Scaling Multimodal Evaluation to over 500 Real-World Tasks Paper • 2410.10563 • Published 27 days ago • 36
WorldCuisines: A Massive-Scale Benchmark for Multilingual and Multicultural Visual Question Answering on Global Cuisines Paper • 2410.12705 • Published 25 days ago • 29
Ovis1.5 Collection The latest version of Ovis! As always, Ovis1.5 remains fully open-source: we release training datasets, training & inference codes, and model weights. • 2 items • Updated Sep 25 • 1
Ovis1.6 Collection With just 10B parameters, Ovis1.6-Gemma2-9B leads the OpenCompass benchmark among open-source MLLMs within 30B parameters. • 4 items • Updated 13 days ago • 4
Ovis: Structural Embedding Alignment for Multimodal Large Language Model Paper • 2405.20797 • Published May 31 • 26