AI Paper Atlas

Research map / Large language models / Retrieval and knowledge

AtlasKV: Augmenting LLMs with Billion-Scale Knowledge Graphs in 20GB VRAM

Haoyu Huang, Hong Ting Tsang, Jiaxin Bai, Xi Peng, Gong Zhang, Yangqiu Song

ICLR 2026 · Large language models › Retrieval and knowledge · approach: RAG robustness and evaluation · 4 citations (OpenAlex)

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
train on KG2KV-generated Q-K-V data instead of KBLaM synthetic dataKBLaM-style knowledge injection · ATLAS-CC-QKV (10^3 triples)ACC@1: 10.9 -> 61.8Improved
use KG2KV instead of synthetic Q-K-V generationtraining data · data constructiondiversity ratio: 0.003% -> 7.864%Improved
use KG2KV instead of synthetic Q-K-V generationtraining data · data constructionavg. token cost: 349.9 -> 165.7Improved
add HiKVP pruning at inferenceAtlasKV · Enron (10^3 triples)ACC@5: 80.0 -> 50.9Hurt
add HiKVP pruning at inferenceAtlasKV · ATLAS-CC-QKV (10^4 triples)ACC@1: 61.8 -> 40.0Hurt
add HiKVP pruning at inferenceAtlasKV · GPU memory, KG size up to 1BVRAMImproved
add event entities to training keys/values (ablation)AtlasKV w/o HiKVP · ATLAS-Pes2o-QKV (10^4 triples)ACC@1: 9.1 -> 47.3Improved
add named entities to training keys/values (ablation)AtlasKV w/o HiKVP · ATLAS-Pes2o-QKV (10^4 triples)ACC@1: 3.6 -> 47.3Improved
add named entities to training keys/values (ablation)AtlasKV w/o HiKVP · Enron (10^4 triples)ACC@1: 1.8 -> 27.3Improved
add relation rewriting to training data (ablation)AtlasKV w/o HiKVP · ATLAS-Pes2o-QKV (10^3 triples)ACC@1: 55.8 -> 72.7Improved
add relation rewriting to test data (ablation)AtlasKV w/o HiKVP · ATLAS-Pes2o-QKV (10^4 triples)ACC@1: 26.9 -> 47.3Improved
injection frequency K=3 vs K=1 or 10 (ablation)AtlasKV · ATLAS-Pes2o-QKVACCImproved

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