Tag
MACRO is a framework that learns task-specific execution routes over frozen LLM layers using Markov chain-based routing, improving reasoning accuracy without modifying model weights. It outperforms prior routing approaches while reducing search time significantly.
The paper proposes a post-training acceleration framework for video diffusion models that integrates dynamic structural sparsification with few-step distillation, achieving significant speedup while maintaining quality.
Introduces Neural Bayesian Sequential Routing (NBSR), a framework that models neural inference as sequential evidence accumulation over a DAG using Dirichlet-Categorical conjugate updates, enabling uncertainty quantification, early exiting, and resource-rational inference.