AI Paper Atlas

Research map / Large language models / Retrieval and knowledge

Leveraging Pretrained Knowledge at Inference Time: LoRA-Gated Contrastive Decoding for Multilingual Factual Language Generation in Adapted LLMs

Gwangseon Jang, Hongseok Choi, Chanuk lim, Kyong-Ha Lee, Mun Yong Yi

ICLR 2026 · Large language models › Retrieval and knowledge · approach: Knowledge editing in LLMs

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
add LGCD contrastive decoding to the language-adapted model12 language-adapted LLMs · Global MMLU 0-shotaccuracy: 0.441 -> 0.477Improved
add LGCD to the language-adapted model12 language-adapted LLMs · Global MMLU 5-shotaccuracy: 0.481 -> 0.498Improved
add LGCD to the language-adapted model6 language-adapted models · Multilingual TruthfulQA 0-shotaccuracy: 0.334 -> 0.376Improved
add LGCD to the language-adapted model6 language-adapted models · Multilingual TruthfulQA 5-shotaccuracy: 0.367 -> 0.435Improved
add LGCD to the language-adapted model9 language-adapted models · Multi-FActfactuality score: 0.272 -> 0.312Mixed
use DoLa decoding instead of nucleus samplinglanguage-adapted models · Multilingual TruthfulQA 0-shotaccuracy: 0.334 -> 0.315Hurt
approximate PTM with attention Q/V layers instead of FFN layers (ablation)LGCD · Global MMLU 0-shotaccuracy: 0.465 -> 0.477Improved
approximate PTM with all layers (FFN + attention) instead of FFN only (ablation)LGCD · Global MMLU 0-shotaccuracy: 0.472 -> 0.477Improved
use online adaptive confidence threshold instead of heuristic language-tier threshold (ablation)LGCD · Global MMLU 0-shotaccuracy: 0.477 -> 0.473No clear effect
use online adaptive confidence threshold instead of heuristic language-tier threshold (ablation)LGCD · Global MMLU 5-shotaccuracy: 0.498 -> 0.495No clear effect
raise confidence threshold from 0.2 to 0.8 (ablation)LGCD · Decoding throughputtokens/s: 14.37 -> 10.22Hurt
apply LGCD to extremely low-resource languagesllama-8b-south-africa · Translated Global MMLU 0-shotaccuracy: 0.452 -> 0.459Improved

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy