KARLA: Knowledge-base Augmented Retrieval for Language Models

arXiv cs.AI Papers

Summary

KARLA proposes a method for LLMs to query a knowledge base during generation, enabling factual updates without retraining and improving transparency. Experiments show improved factual grounding in both short and long-form generation.

arXiv:2606.26807v1 Announce Type: new Abstract: We propose a new method that allows an LLM to automatically pull in factual knowledge from a knowledge base during token generation. This means that (1)~factual knowledge in the LLM output can be updated without retraining the LLM, (2)~facts in the LLM output can be traced to the knowledge base for transparency and explainability, and (3)~smaller models can achieve the same factual accuracy as larger models. Our core idea is to train the model to produce special tokens that trigger a query to the knowledge base. Our experiments show that our method improves factual grounding in both short and long-form generation, and allows factual revisions to take effect through KB edits rather than parameter updates.
Original Article
View Cached Full Text

Cached at: 06/26/26, 05:17 AM

# KARLA: Knowledge-base Augmented Retrieval for Language Models
Source: [https://arxiv.org/abs/2606.26807](https://arxiv.org/abs/2606.26807)
[View PDF](https://arxiv.org/pdf/2606.26807)

> Abstract:We propose a new method that allows an LLM to automatically pull in factual knowledge from a knowledge base during token generation\. This means that \(1\)~factual knowledge in the LLM output can be updated without retraining the LLM, \(2\)~facts in the LLM output can be traced to the knowledge base for transparency and explainability, and \(3\)~smaller models can achieve the same factual accuracy as larger models\. Our core idea is to train the model to produce special tokens that trigger a query to the knowledge base\. Our experiments show that our method improves factual grounding in both short and long\-form generation, and allows factual revisions to take effect through KB edits rather than parameter updates\.

## Submission history

From: Francois Crespin \[[view email](https://arxiv.org/show-email/d807e8f6/2606.26807)\] \[via CCSD proxy\] **\[v1\]**Thu, 25 Jun 2026 09:44:40 UTC \(2,572 KB\)

Similar Articles

LLM Wiki v2 (16 minute read)

TLDR AI

This post presents a pattern for building personal knowledge bases using LLMs, offering a structured approach for leveraging large language models in knowledge management.