Training-Free Guidance Beyond Differentiability: Scalable Path Steering with Tree Search in Diffusion and Flow Models
This work addresses the challenge of controlled generation in AI models for domains like music and biology, offering a scalable solution for non-differentiable cases, though it is incremental as it builds on existing guidance methods.
The paper tackled the problem of training-free guidance for non-differentiable objectives and discrete data in diffusion and flow models by proposing TreeG, a tree search-based method, which improved performance by 29.01%, 16.6%, and 18.43% in symbolic music generation, small molecule design, and enhancer DNA design.
Training-free guidance enables controlled generation in diffusion and flow models, but most methods rely on gradients and assume differentiable objectives. This work focuses on training-free guidance addressing challenges from non-differentiable objectives and discrete data distributions. We propose TreeG: Tree Search-Based Path Steering Guidance, applicable to both continuous and discrete settings in diffusion and flow models. TreeG offers a unified framework for training-free guidance by proposing, evaluating, and selecting candidates at each step, enhanced with tree search over active paths and parallel exploration. We comprehensively investigate the design space of TreeG over the candidate proposal module and the evaluation function, instantiating TreeG into three novel algorithms. Our experiments show that TreeG consistently outperforms top guidance baselines in symbolic music generation, small molecule design, and enhancer DNA design with improvements of 29.01%, 16.6%, and 18.43%. Additionally, we identify an inference-time scaling law showing TreeG's scalability in inference-time computation.