4 papers
Conformal Cascade: Distribution-Free Accuracy Guarantees for Multi-Tier LLM Inference
Yifan Dou, Shikan Lian, Shibo Li
Large language model (LLM) cascades reduce inference cost by routing easy queries to a small model and deferring hard queries to a larger one. Production cascades govern this defer…
ConsistencyGate: Preventing Memory Contamination in LLM Agents via Self-Consistency Admission Control
Yan Zhang, Shibo Li
LLM agents that operate over many turns accumulate facts in an external memory store and reuse them as premises for downstream reasoning. A hallucinated fact written at one step th…
Verbalized Particle Posterior: Bayesian Inference over Natural Language Hypotheses
Yan Zhang, Shikan Lian, Shibo Li
Verbalized Machine Learning (VML) parameterizes a model as a natural-language prompt that an LLM evaluates as f(x; theta). The framework is interpretable, but it commits to a singl…
Impatient PPSZ -- a Faster algorithm for CSP
Shibo Li, Dominik Scheder
PPSZ is the fastest known algorithm for (d,k)-CSP problems, for most values of d and k. It goes through the variables in random order and sets each variable randomly to one of the…