Is HippoRAG 2 superseded?

HippoRAG 2 (Retrieval-augmented generation): superseded — cited as a baseline and beaten by newer methods. 1 paper(s) critique it, 23 beat it on benchmarks — #8 of 1179 most-superseded. Sub-problem: cluster led by RAG. Newer alternatives in the same sub-problem include Narrative Knowledge Weaver, IA-RAG, MemGraphRAG, R2C, LegalGraphRAG.

Retrieval-augmented generation

HippoRAG 2

From RAG to Memory: Non-Parametric Continual Learning for Large Language Models

Superseded baseline#8 of 1,179 most-superseded · first seen Feb 20, 2025

Superseded — cited as a baseline and beaten by newer methods

1 papers critique it · 23 beat it on benchmarks

What papers say

Verbatim critique sentences, each from a paper that cites HippoRAG 2 as a baseline.

their reliance on graph traversal introduces substantial computational overhead and latency

— HELP: HyperNode Expansion and Logical Path-Guided Evidence Localization for Accurate and Efficient GraphRAG

Beaten on benchmarks

Head-to-head results where a newer method reports beating HippoRAG 2. Values are copied from the source paper's tables — verify against the cited paper.

ANCHOR beats HippoRAG 2
40.86 vs 19.10
Average EM · [Llama3.1-8B-Instruct]
Graph-Anchored Knowledge Indexing for Retrieval-Augmented Generation
SF-RAG beats HippoRAG 2
30.98 vs 234.93
Time (s) · [Indexing efficiency]
SF-RAG: Structure-Fidelity Retrieval-Augmented Generation for Academic Question Answering
GAM-RAG beats HippoRAG 2
43.30 vs 24.80
GPT-Acc · [MuSiQue]
GAM-RAG: Gain-Adaptive Memory for Evolving Retrieval in Retrieval-Augmented Generation
FD-RAG beats HippoRAG 2
58 vs 176
OutTok/1kT · [HotPotQA Construction]
FD-RAG: Federated Dual-System Retrieval-Augmented Generation
SA-RAG + CoT beats HippoRAG 2
0.72 vs 0.50
EM · [2WikiMultihopQA, phi4]
Leveraging Spreading Activation for Improved Document Retrieval in Knowledge-Graph-Based RAG Systems
LogicRAG beats HippoRAG 2
62.5 vs 47.1
LLM-Acc. · [2WikiMQA]
You Don't Need Pre-built Graphs for RAG: Retrieval Augmented Generation with Adaptive Reasoning Structures
MemGraphRAG beats HippoRAG 2
0.865 vs 0.657
Clustering Coefficient · [G-Novel dataset]
MemGraphRAG: Memory-based Multi-Agent System for Graph Retrieval-Augmented Generation
EHGRAG beats HippoRAG 2
70.60 vs 55.00
LLM-Acc · [2WikiMultiHop]
EHRAG: Bridging Semantic Gaps in Lightweight GraphRAG via Hybrid Hypergraph Construction and Retrieval
LegalGraphRAG beats HippoRAG 2
40.9 vs 31.9
Accuracy All · [GPT-4o-mini]
LegalGraphRAG: Multi-Agent Graph Retrieval-Augmented Generation for Reliable Legal Reasoning
ProPEX-RAG beats HippoRAG 2
78.1 vs 63.8
Average EM · [Structure-Augmented RAG]
PROPEX-RAG: Enhanced GraphRAG using Prompt-Driven Prompt Execution
A-RAG (Full) beats HippoRAG 2
74.1 vs 61.7
LLM-Acc · [GPT-5-mini / MuSiQue]
A-RAG: Scaling Agentic Retrieval-Augmented Generation via Hierarchical Retrieval Interfaces
CARROT beats HippoRAG 2
0.432 vs 0.363
R1 · [WikiPassageQA 256 tokens]
CORAG: A Cost-Constrained Retrieval Optimization System for Retrieval-Augmented Generation

What to use instead

Recent methods in the same sub-problem, not yet superseded in the knowledge base — arXiv benchmark leaders, not vetted production recommendations.