vllm_omni.diffusion.cache.stepcache.state ¶ Mutable state for step-level DiT velocity caching. StepCacheState ¶ Per-generation mutable state (skip countdown). config instance-attribute ¶ config = config skip_countdown instance-attribute ¶ skip_countdown = 0 reset ¶ reset() -> None should_run_step ¶ should_run_step( prev_predictions: list[tuple[Tensor, ...]], ) -> bool Return True when the DiT forward should execute at this scheduler step. trim_history ¶ trim_history( prev_predictions: list[tuple[Tensor, ...]], ) -> None