paper
Space-Efficient Language Generation in the Limit
paperactiveprovisional
space-efficient-language-generation-in-the-limit-f9e3accf·1 events·first seen 4d agoAliases: Space-Efficient Language Generation in the Limit
More like this (12)
Language Generation in the LimitScaling limit of the Random Language ModelPosterior Refinement: Fast Language Generation via Any-Order Flow MapsOn the Limits of Prompt-Conditioned Language Models as General-Purpose LearnersFlood and Harvest: The Provable Necessity of Trivia for Generating Valuable Mathematics via the Lens of Language Generation in the LimitSelf-Compacting Language Model AgentsLarge Language Gibbsgenerative language modelingLarge Language Models (frontier)Speaking the Language of Science: Toward a General-Purpose Generative Foundation Model for the Natural SciencesRandom Language ModelTapered Language Models
Recent events (1)
Space-Efficient Language Generation in the Limit: Poly-Space Algorithms with Bounded Hallucination Gap
A new arXiv preprint introduces a resource-aware theory of language generation in the limit, studying learners that must produce hallucination-free hypothesis languages from adversarial positive streams under memory constraints. The authors focus on DFA-recognizable language classes and prove a streaming algorithm using poly(s,k) space that converges with a bounded generation gap, complemented by a near-matching lower bound via communication complexity reduction. The results reveal a sharp phase transition between polynomial-space generation and exponential-space exact identification, providing theoretical grounding for memory-bounded language generation.