Share your thoughts, 1 month free Claude Pro on usSee more

Multi-hop Retrieval on HotpotQA

85.9Recall@2

G-reasoner

Updated 21d ago

Evaluation Results

Method
G-reasoner 2025.09	85.9	-	-	-	-	-	-	97.7	-
PropRAG 2026.05	81.9	-	-	-	-	-	-	88	1.9
HippoRAG 2 2025.09	80.5	-	-	-	-	-	-	95.7	-
HippoRAG 2 2026.05	80.5	-	-	-	-	-	-	88.1	2.4
C2RAG 2026.03	79.2	-	-	-	-	-	-	92.1	-
Iter-RetGen 2026.03	77.8	-	-	-	-	-	-	90.4	-
GFM-RAG 2026.03	77.6	-	-	-	-	-	-	87.8	-
NeuroPath 2026.03	76.2	-	-	-	-	-	-	89.9	-
GFM-RAG 2025.09	75.6	-	-	-	-	-	-	89.6	-
GFM-RAG 2026.05	75.6	-	-	-	-	-	-	89.6	2.9
Qwen3-Emb 2025.09	74.1	-	-	-	-	-	-	88.8	-
FLARE 2026.05	73.1	-	-	-	-	-	-	81.3	6.5
IRCoT 2026.03	69.3	-	-	-	-	-	-	82.6	-
ColBERTv2 + IRCoT 2026.05	67.9	-	-	-	-	-	-	82	7.2
HippoRAG + IRCoT 2026.05	67	-	-	-	-	-	-	83	3.6
HippoRAG2 2026.03	66	-	-	-	-	-	-	82.8	-
Contriever + IRCoT 2026.05	65.9	-	-	-	-	-	-	81.6	10.7
BM25 + IRCoT 2026.05	65.6	-	-	-	-	-	-	79	12.4
SAGE 2026.05	65.1	-	-	-	-	-	-	77.6	7
ColBERTv2 2025.09	64.7	-	-	-	-	-	-	79.3	-
ColBERTv2 2026.05	64.7	-	-	-	-	-	-	79.3	11.5
SubgraphRAG 2026.05	61.5	-	-	-	-	-	-	73	9.7
Adaptive-RAG 2026.05	61	-	-	-	-	-	-	76.4	16.6
HippoRAG 2026.03	60.5	-	-	-	-	-	-	79.2	-
HippoRAG 2025.09	60.1	-	-	-	-	-	-	78.5	-
HippoRAG 2026.05	60.1	-	-	-	-	-	-	78.5	8.7
KAG 2025.09	59.4	-	-	-	-	-	-	86.1	-
GTR 2026.05	59.4	-	-	-	-	-	-	73.3	13.8
Proposition 2026.05	58.7	-	-	-	-	-	-	71.1	14.8
GraphRAG 2025.09	58.3	-	-	-	-	-	-	76.6	-
GraphRAG 2026.05	58.3	-	-	-	-	-	-	76.6	12.2
RAPTOR 2025.09	58.1	-	-	-	-	-	-	71.2	-
SubgraphRAG 2025.09	58.1	-	-	-	-	-	-	71.7	-
RAPTOR 2026.05	58.1	-	-	-	-	-	-	71.2	17.7
Contriever 2026.05	57.2	-	-	-	-	-	-	75.5	17.2
BM25 2025.09	55.4	-	-	-	-	-	-	72.2	-
BM25 2026.05	55.4	-	-	-	-	-	-	72.2	18.2
BM25 2026.03	55	-	-	-	-	-	-	72.8	-
G-Retriever 2026.05	53.3	-	-	-	-	-	-	65.5	15.7
G-retriever 2026.03	52.4	-	-	-	-	-	-	62.9	-
G-retriever 2025.09	51.8	-	-	-	-	-	-	63.6	-
LightRAG 2026.03	39.5	-	-	-	-	-	-	54	-
LightRAG 2025.09	38.8	-	-	-	-	-	-	54.7	-
LightRAG 2026.05	38.8	-	-	-	-	-	-	54.7	20.5
Raw 2026.01	-	0.833	0.889	0.967	0.495	0.754	0.705	-	-
Qwen2.5-3B-inst 2026.01	-	0.72	0.793	0.897	0.412	0.642	0.605	-	-
Qwen2.5-3B-inst 2026.01	-	0.753	0.812	0.895	0.427	0.659	0.624	-	-
DeepSeek-V3.1 2026.01	-	0.811	0.866	0.933	0.491	0.706	0.662	-	-
vanilla RL 2026.01	-	0.823	0.899	0.956	0.488	0.775	0.732	-	-
ConvSearch-R1 2026.01	-	0.83	0.902	0.96	0.511	0.77	0.723	-	-
ACQO 2026.01	-	0.869	0.916	0.975	0.512	0.777	0.742	-	-
Raw 2026.01	-	0.683	0.748	0.861	0.348	0.604	0.595	-	-
Qwen2.5-3B-inst 2026.01	-	0.646	0.707	0.818	0.328	0.566	0.56	-	-
Qwen2.5-3B-inst 2026.01	-	0.67	0.73	0.818	0.349	0.575	0.573	-	-
DeepSeek-V3.1 2026.01	-	0.774	0.825	0.889	0.461	0.668	0.657	-	-
vanilla RL 2026.01	-	0.792	0.839	0.891	0.411	0.755	0.744	-	-
ConvSearch-R1 2026.01	-	0.75	0.794	0.875	0.444	0.728	0.722	-	-
ACQO 2026.01	-	0.822	0.858	0.912	0.496	0.734	0.736	-	-