An empirical testbed shows that allocating more transformer layers to autoregressive conditioning improves FID under low inference budgets, while block size should scale with image resolution.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CV 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
MADFormer: Mixed Autoregressive and Diffusion Transformers for Continuous Image Generation
An empirical testbed shows that allocating more transformer layers to autoregressive conditioning improves FID under low inference budgets, while block size should scale with image resolution.