Search papers, labs, and topics across Lattice.
The paper introduces Chain-of-Trajectories (CoTj), a train-free framework for diffusion models that enables deliberative planning by addressing the computational inefficiencies arising from the fixed, content-agnostic sampling schedules. CoTj uses a "Diffusion DNA" signature to quantify denoising difficulty and reformulates sampling as graph planning on a directed acyclic graph, allowing for dynamic allocation of computational resources. Experiments demonstrate that CoTj improves output quality and stability while reducing redundant computation across multiple generative models.
Diffusion models can be made more efficient and produce better outputs by dynamically allocating compute based on a learned "difficulty" signature, without any retraining.
Diffusion models operate in a reflexive System 1 mode, constrained by a fixed, content-agnostic sampling schedule. This rigidity arises from the curse of state dimensionality, where the combinatorial explosion of possible states in the high-dimensional noise manifold renders explicit trajectory planning intractable and leads to systematic computational misallocation. To address this, we introduce Chain-of-Trajectories (CoTj), a train-free framework enabling System 2 deliberative planning. Central to CoTj is Diffusion DNA, a low-dimensional signature that quantifies per-stage denoising difficulty and serves as a proxy for the high-dimensional state space, allowing us to reformulate sampling as graph planning on a directed acyclic graph. Through a Predict-Plan-Execute paradigm, CoTj dynamically allocates computational effort to the most challenging generative phases. Experiments across multiple generative models demonstrate that CoTj discovers context-aware trajectories, improving output quality and stability while reducing redundant computation. This work establishes a new foundation for resource-aware, planning-based diffusion modeling. The code is available at https://github.com/UnicomAI/CoTj.