List your company

Multi-hop Question Answering on HotpotQA (test)

80.79F1Gemini-2.5-Flash
57.192463.318769.44575.5713May 27, 2025Jul 28, 2025Sep 28, 2025Nov 30, 2025Jan 31, 2026Apr 3, 2026Jun 5, 2026

Evaluation results

MethodLinks
Gemini-2.5-FlashType=API2026.0280.79——————65.68——————————————————————
GPT-5-miniType=API2026.0280.19——————63.63——————————————————————
Deepseek-V3.2Type=API2026.0279.91——————64.15——————————————————————
Qwen3-30BType=Pre-Trained2026.0278.33——————63.05——————————————————————
CRAFT7BType=Trained2026.0278——————63.25——————————————————————
DenoiseFlowBackbone LLM=GPT-4o-mini, System Category=Autonomous Multi-agent2026.0277.5—————————————————————————————
JudgeFlowBackbone LLM=GPT-4o-mini, System Category=Autonomous Multi-agent2026.0277.4—————————————————————————————
MermaidFlowBackbone LLM=GPT-4o-mini, System Category=Autonomous Multi-agent2026.0277.2—————————————————————————————
HELPBackbone=Llama-3.3-70B-Instruct2026.0275.6—————————————————————————————
HippoRAG 2Generator=Llama3.3-70B-Instruct2026.0275.5——————62.7——————————————————————
HippoRAG2Backbone=Llama-3.3-70B-Instruct2026.0275.5—————————————————————————————
Qwen3-4BType=Pre-Trained2026.0275.32——————60.4——————————————————————
NV-Embed-v2 (7B)Generator=Llama3.3-70B-Instruct2026.0275.3——————62.8——————————————————————
NV-Embed-v2Backbone=Llama-3.3-70B-Instruct2026.0275.3—————————————————————————————
MaASBackbone LLM=GPT-4o-mini, System Category=Autonomous Multi-agent2026.0275.3—————————————————————————————
DyLANBackbone LLM=GPT-4o-mini, System Category=Hand-crafted Multi-agent2026.0274.1—————————————————————————————
AFlowBackbone LLM=GPT-4o-mini, System Category=Autonomous Multi-agent2026.0273.5—————————————————————————————
GritLM-7BBackbone=Llama-3.3-70B-Instruct2026.0273.3—————————————————————————————
GPTSwarmBackbone LLM=GPT-4o-mini, System Category=Autonomous Multi-agent2026.0273.2—————————————————————————————
HyperGraphRAGBackbone=Llama-3.3-70B-Instruct, Local Reproduction=true2026.0273.1—————————————————————————————
MultiCube-RAGGenerator=Llama3.3-70B-Instruct2026.0272.9——————61.3——————————————————————
LLM-BlenderBackbone LLM=GPT-4o-mini, System Category=Hand-crafted Multi-agent2026.0272.3—————————————————————————————
MultiCube-RAGGenerator=GPT-4o-mini2026.0271.5——————57.5——————————————————————
CatRAGQA reader=Llama-3.3-70B-Instruct2026.0271.4—————————————————————————————
HippoRAG 2Generator=GPT-4o-mini2026.0271.1——————56.3——————————————————————
NV-Embed-v2 (7B)Generator=GPT-4o-mini2026.0271——————57.3——————————————————————
GTE-Qwen2-7BBackbone=Llama-3.3-70B-Instruct2026.0271—————————————————————————————
RelinkMethod Type=Proposed, Backbone LLM=deepseek-v3-03242026.0170.4——————55.8——————————————————————
LLM-DebateBackbone LLM=GPT-4o-mini, System Category=Hand-crafted Multi-agent2026.0270.2—————————————————————————————
RAPTORGenerator=Llama3.3-70B-Instruct2026.0269.5——————56.8——————————————————————
RAPTORBackbone=Llama-3.3-70B-Instruct2026.0269.5—————————————————————————————
OPERA (MAPGRPO)Type=RL2025.0869.5——————57.3——————————————————————
HippoRAG 2QA reader=Llama-3.3-70B-Instruct, Retrieval=Structure-Aware RAG2026.0269.4—————————————————————————————
IGMiRAG2026.0269.1——————54——————————————————————
CoT SCBackbone LLM=GPT-4o-mini, System Category=Single-agent2026.0268.9—————————————————————————————
GraphRAGGenerator=Llama3.3-70B-Instruct2026.0268.6——————55.2——————————————————————
GraphRAGBackbone=Llama-3.3-70B-Instruct2026.0268.6—————————————————————————————
LightRAGQA reader=Llama-3.3-70B-Instruct, Retrieval=Structure-Aware RAG2026.0268.3—————————————————————————————
Qwen2.5-7BType=Trained2026.0268.12——————53.3——————————————————————
IOBackbone LLM=GPT-4o-mini, System Category=Single-agent2026.0268.1—————————————————————————————
AceSearcher-8Bcategory=Advanced RAG2026.0468.07——————37.1942.91—————————————————————
Qwen2.5-7BType=Pre-Trained2026.0267.92——————53.4——————————————————————
CoTBackbone LLM=GPT-4o-mini, System Category=Single-agent2026.0267.9—————————————————————————————
GraphRAGGenerator=GPT-4o-mini2026.0267.6——————51.4——————————————————————
ProGraph-R1Base Model=Qwen2.5-7B-Instruct, Optimization Method=Training, Knowledge Interaction=Graph-based knowledge2026.0167.57——————60.94——————————————————————
CRAFT3BType=Trained2026.0267.1——————53.15——————————————————————
RankCoTGenerator=GPT-4o-mini2026.0265.8——————53.5——————————————————————
RoMemMemory category=Structured memory baselines2026.0665.8——————50.960.4—————————————79.9———————
Naive RAGTop-k retrieved slices=52026.0265.1——————50.2——————————————————————
HippoRAGMethod Type=Hybrid, Backbone LLM=deepseek-v3-03242026.0164.7——————49.8——————————————————————
RAPTORGenerator=GPT-4o-mini2026.0264.7——————50.6——————————————————————
text-embedding-3-smallQA reader=Llama-3.3-70B-Instruct, Retrieval=Standard Retrieval2026.0264.6—————————————————————————————
ADASBackbone LLM=GPT-4o-mini, System Category=Autonomous Multi-agent2026.0264.5—————————————————————————————
RAPTORQA reader=Llama-3.3-70B-Instruct, Retrieval=Structure-Aware RAG2026.0264.4—————————————————————————————
IRCoTGenerator=GPT-4o-mini2026.0264.1——————53.9——————————————————————
NodeRAG2026.0264——————50.1——————————————————————
HippoRAGGenerator=Llama3.3-70B-Instruct2026.0263.5——————52.6——————————————————————
HippoRAGBackbone=Llama-3.3-70B-Instruct2026.0263.5—————————————————————————————
FABLE(nodes)selection_strategy=LLM-guided, bi_path_fusion=true, retrieval_unit=nodes2026.0163.43——————48——————————————————————
BM25Generator=Llama3.3-70B-Instruct2026.0263.4——————52——————————————————————
BM25Backbone=Llama-3.3-70B-Instruct2026.0263.4—————————————————————————————
RAPTORMethod Type=Text-based, Backbone LLM=deepseek-v3-03242026.0163.1——————47.2——————————————————————
HERA (Qwen-3-14B)category=Ours2026.0463.03——————52.555.38—————————————————————
GTR (T5-base)QA reader=Llama-3.3-70B-Instruct, Retrieval=Standard Retrieval2026.0262.8—————————————————————————————
GTRBackbone=Llama-3.3-70B-Instruct2026.0262.8—————————————————————————————
Naive RAGTop-k retrieved slices=32026.0262.7——————48.5——————————————————————
Graph-R1Base Model=Qwen2.5-7B-Instruct, Optimization Method=Training, Knowledge Interaction=Graph-based knowledge2026.0162.69——————57.03——————————————————————
Qwen2.5-3BType=Pre-Trained2026.0262.58——————49.3——————————————————————
GenGroundBackbone=GPT-4o, Retrieval=with Retrieval2025.0662.37560.563——————————————————————————
ContrieverQA reader=Llama-3.3-70B-Instruct, Retrieval=Standard Retrieval2026.0262.3—————————————————————————————
ContrieverBackbone=Llama-3.3-70B-Instruct2026.0262.3—————————————————————————————
FABLE(docs)selection_strategy=LLM-guided, bi_path_fusion=true, retrieval_unit=documents2026.0162.22——————46.5——————————————————————
DEC (Ours)Backbone=GPT-4o, Retrieval=with Retrieval2025.0662.113.3258.5260.32——————————————————————————
ProGraph-R1Base Model=Qwen2.5-3B-Instruct, Optimization Method=Training, Knowledge Interaction=Graph-based knowledge2026.0161.9——————55.47——————————————————————
Self-RefineBackbone LLM=GPT-4o-mini, System Category=Hand-crafted Multi-agent2026.0260.8—————————————————————————————
EviOmniModel size=7B, Method Category=RAR (Reasoning-Augmented), Context=Out-of-domain2025.0760.3——————28.02———————16.5——————————————
Qwen2.5-3BType=Trained2026.0260.23——————46.6——————————————————————
HippoRAGGenerator=GPT-4o-mini2026.0260——————46.3——————————————————————
CoTModel size=7B, Method Category=RAR (Reasoning-Augmented)2025.0759.84——————27.25———————3.32——————————————
BM25Generator=GPT-4o-mini2026.0259.8——————49.1——————————————————————
Direct CoTBackbone=GPT-4o, Retrieval=without Retrieval2025.0659.3115558.5——————————————————————————
IQATRLLM Backbone=Llama-3.1-8B, Retriever=Contriever, Retriever Fine-tuning=ReSCORE2025.0559.3——————47.2——————————————————————
ARBORBackbone=Qwen3-14B2026.0659.1——————48.663.6—————————————————————
MetaTuner-IBackbone=Qwen2.5-3B, Method Category=Hybrid2025.0959.05—————————————————————————————
CRAFT1.5BType=Trained2026.0259.04——————45.7——————————————————————
LinearRAGBackbone=Llama-3.3-70B-Instruct, Local Reproduction=true2026.0259—————————————————————————————
HippoRAG2category=Structure-Enhanced RAG2026.0158.87——————41——————————————————————
DeepResearcher-7Bcategory=Advanced RAG2026.0458.86——————35.2340.65—————————————————————
MetaTuner-JBackbone=Qwen2.5-3B, Method Category=Hybrid2025.0958.85—————————————————————————————
Abstract-CoT (Warm-up + RL)Model=Qwen3-8B, Tokens=1712026.0458.8—————————————————————————————
SEERModel size=7B, Method Category=VAR (Context Compression)2025.0758.63——————25.81———————14.3——————————————
ARBORBackbone=Qwen3-8B2026.0658.6——————47.862.9—————————————————————
FullModel size=7B, Method Category=WR (Vanilla Retrieval)2025.0758.5——————25.82———————1——————————————
OPERA (CoT)Type=CoT2025.0858.5——————44.9——————————————————————
Adaptive-NoteLLM Backbone=GPT-3.5, Retriever=BM252025.0558.4——————45.6——————————————————————
GRPOBackbone=Qwen3-14B2026.0658.4——————48.162.3—————————————————————
ExSearchcategory=Advanced RAG, Backbone=Qwen2.5-7B2026.0458.22——————45.1155.04—————————————————————
FilCoModel size=7B, Method Category=VAR (Context Compression)2025.0758.15——————25.27———————17.5——————————————
SearChainBackbone=GPT-4o, Retrieval=with Retrieval2025.0658.122.2453.858——————————————————————————
SFT + RLModel=Qwen3-8B, Tokens=7352026.0458.1—————————————————————————————