Lens: Rethinking Training Efficiency for Foundational Text-to-Image Models

Benchmark Model Rank Results
text-to-image-generation-on-oneig-enLens w/ GPT-OSS-20B-A3B#4OneIG Overall Score: 0.559
text-to-image-generation-on-cvtg-2kLens-Turbo#7Word Accuracy: 0.889NED: 0.965
text-to-image-generation-on-oneig-zhLens#10OneIG Overall Score: 0.525
text-to-image-generation-on-longtext-benchLens#12Accuracy: 93.7
text-to-image-generation-on-genevalLens (20-step, no reasoner)#72Overall: 0.843