47 min
Seminar, Hosted by Simons Institute for the Theory of Computing
Inference-Time Algorithms: A Theoretical Lens on Tractability and Error Propagation
Carnegie Mellon University
Recording
Abstract
Inference-time algorithms place pretrained models inside larger computational loops, trading extra computation for accuracy or control. This talk studies efficiency and error propagation with imperfect learned oracles. Examples include stochastic backtracking in generator–verifier systems and the conditions under which diffusion steering can favour rewards while remaining close to the original model.
Topics
- Inference-time algorithms, broader topic: algorithms
- pretrained models
- computational loops, search World Wide
- efficiency
- error propagation, search World Wide
- learned oracles, search World Wide
- stochastic backtracking, search World Wide
- generator–verifier systems, broader topic: systems
- diffusion steering, broader topic: steering