next-token probabilities

3 papers

• Better Language Model Inversion by Compactly Representing Next-Token Distributions
• Foundations of Top-$k$ Decoding for Language Models
• SimpleStrat: Diversifying Language Model Generation with Stratification