LLM-AT is a training-free cascade that uses a history of judge-validated answers to estimate per-question tier accuracy, starting cheap and escalating only when the judge rejects the response.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CL 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Automatic Transmission for LLM Tiers: Optimizing Cost and Accuracy in Large Language Models
LLM-AT is a training-free cascade that uses a history of judge-validated answers to estimate per-question tier accuracy, starting cheap and escalating only when the judge rejects the response.