CGPO adaptively increases sampling of prompts whose image groups show high reward variance, combined with category calibration, to improve text-to-image generation on GenEval, T2I-CompBench++, and DPG Bench.
Asynchronous curriculum experience re- play: A deep reinforcement learning approach for uav au- tonomous motion control in unknown dynamic environ- ments, 2022
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CV 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Curriculum Group Policy Optimization: Adaptive Sampling for Unleashing the Potential of Text-to-Image Generation
CGPO adaptively increases sampling of prompts whose image groups show high reward variance, combined with category calibration, to improve text-to-image generation on GenEval, T2I-CompBench++, and DPG Bench.