RTS introduces reward-guided noise optimization and sparse PCA-based scaling to enhance test-time performance of diffusion models, claiming 15.6% GenEval and 60.4% ImageReward gains over baselines.
An empirical analysis of compute-optimal inference for problem-solving with language models, 2024.arXiv
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
-
Guided Trajectory Optimization with Sparse Scaling for Test-Time Diffusion
RTS introduces reward-guided noise optimization and sparse PCA-based scaling to enhance test-time performance of diffusion models, claiming 15.6% GenEval and 60.4% ImageReward gains over baselines.