4 papers
AILA--First Experiments with Localist Language Models
Joachim Diederich
This paper presents the first empirical demonstration of controllable locality in transformer language models, a novel architectural framework that enables continuous control over…
Localist LLMs -- A Mathematical Framework for Dynamic Locality Control
Joachim Diederich
We present a novel framework for training large language models with continuously adjustable internal representations that span the full spectrum from localist (interpretable, rule…
Localist LLMs with Recruitment Learning
Joachim Diederich
We present a novel framework for training large language models with continuously adjustable internal representations that span the full spectrum from localist (interpretable, rule…
Rule Encoding and Compliance in Large Language Models: An Information-Theoretic Analysis
Joachim Diederich
The design of safety-critical agents based on large language models (LLMs) requires more than simple prompt engineering. This paper presents a comprehensive information-theoretic a…