Saved in:
Bibliographic Details
Main Authors: Bystroński, Mateusz, Han, Doheon, Chawla, Nitesh V., Kajdanowicz, Tomasz
Format: Preprint
Published: 2025
Subjects:
Online Access:https://arxiv.org/abs/2507.13874
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866912819155304448
author Bystroński, Mateusz
Han, Doheon
Chawla, Nitesh V.
Kajdanowicz, Tomasz
author_facet Bystroński, Mateusz
Han, Doheon
Chawla, Nitesh V.
Kajdanowicz, Tomasz
contents Starting from the hypothesis that knowledge in semantic space is organized along structured manifolds, we argue that this geometric structure renders the space explorable. By traversing it and using the resulting continuous representations to condition an LLM's generation distribution, we can systematically expand the model's reachable semantic range. We introduce a framework that requires no modification of LLM parameters and operationalizes this idea by constructing a conditioning distribution from a small set of diverse anchor generations. This distribution conditions LLM's generation via an xRAG-style projector. Our experiments demonstrate that this manifold-based conditioning substantially increases generative diversity, with direct benefits for enhancing divergent thinking, a core facet of creativity, in language models.
format Preprint
id arxiv_https___arxiv_org_abs_2507_13874
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Geometry of Knowledge Allows Extending Diversity Boundaries of Large Language Models
Bystroński, Mateusz
Han, Doheon
Chawla, Nitesh V.
Kajdanowicz, Tomasz
Artificial Intelligence
Starting from the hypothesis that knowledge in semantic space is organized along structured manifolds, we argue that this geometric structure renders the space explorable. By traversing it and using the resulting continuous representations to condition an LLM's generation distribution, we can systematically expand the model's reachable semantic range. We introduce a framework that requires no modification of LLM parameters and operationalizes this idea by constructing a conditioning distribution from a small set of diverse anchor generations. This distribution conditions LLM's generation via an xRAG-style projector. Our experiments demonstrate that this manifold-based conditioning substantially increases generative diversity, with direct benefits for enhancing divergent thinking, a core facet of creativity, in language models.
title Geometry of Knowledge Allows Extending Diversity Boundaries of Large Language Models
topic Artificial Intelligence
url https://arxiv.org/abs/2507.13874