AI Paper Atlas

Research map / Large language models / Retrieval and knowledge

GRO-RAG: Gradient-aware Re-rank Optimization for Multi-source Retrieval-Augmented Generation

Siyuan Chen, Hang Ding, Kangxiaoyu, Jiechao Gao

ICLR 2026 · Large language models › Retrieval and knowledge · approach: RAG robustness and evaluation

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 gradient-aware re-ranking (ablation)GRO-RAG (GLM-4) · HotpotQAF1: 37.6 -> 42.8Improved
add gradient-aware re-ranking (ablation)GRO-RAG (Llama3.1-8B) · MuSiQueF1: 16.2 -> 18.6Improved
add source combination selection (ablation)GRO-RAG (GLM-4) · 2WikiMultihopQAF1: 28.6 -> 30.3Improved
add source combination selection (ablation)GRO-RAG (Llama3.1-8B) · MuSiQueEM: 10.2 -> 10.3No clear effect
use both local and web sources instead of local onlyVanilla RAG (Llama3.1-8B) · HotpotQAF1: 34.2 -> 36.0Improved
use web source only instead of localVanilla RAG (Llama3.1-8B) · HotpotQAF1: 34.2 -> 31.5Hurt
use GRO-RAG instead of CRAGLlama3.1-8B · HotpotQAF1: 34.2 -> 39.1Improved
use GRO-RAG instead of FLARELlama3.1-8B · 2WikiMultihopQAEM: 23.0 -> 22.8Mixed
replace dense re-ranker with gradient re-rankingGRO-RAG (Llama3.1-8B) vs BGE-M3 · MuSiQuenDCG@10: 0.3922 -> 0.4039Improved
replace E5-base with gradient re-rankingGRO-RAG (Llama3.1-8B) vs E5-base · HotpotQAnDCG@10: 0.7013 -> 0.6442Hurt
replace BM25 with gradient re-rankingGRO-RAG (Llama3.1-8B) vs BM25 · HotpotQAnDCG@10: 0.6237 -> 0.6442Improved
add multi-step gradient optimizationGRO-RAG · HotpotQAnDCG@10: total gain around 3-5%Improved

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