List your company

Knowledge Graph Question Answering on KGQAGen-10k 1,079 unseen questions (test)

84.9LASM AccuracyGPT-4o w/ SP
29.2643.70558.1572.595Mar 31, 2026

Evaluation results

MethodLinks
GPT-4o w/ SPType=Oracle subgraph, Base Model=GPT-4o, Memory=Frozen2026.0384.9—
APEX-EM (EG1: Entity graph)Type=Rich feedback, iter., EG, Memory=Frozen2026.0373.731.7
APEX-EM (A3: Judge + iteration)Type=Memory, binary, iteration, Memory=Frozen2026.0373.531.5
APEX-EM (A5: Opus judge)Type=All + Opus judge, Memory=Frozen2026.0372.730.7
APEX-EM (R1: Semantic only)Type=Rich feedback, iter., semantic, Memory=Frozen2026.0371.629.7
APEX-EM (EG2: Full memory)Type=Rich feedback, iter., all, Memory=Frozen2026.0371.229.2
APEX-EM (A5: GPT 4o as base)Type=All, Base Model=GPT-4o, Memory=Frozen2026.0366.212
PoG (GPT-4o)Type=KG-RAG, plan-on-graph, Base Model=GPT-4o, Memory=Frozen2026.0358.1—
ToG (GPT-4o)Type=KG-RAG, think-on-graph, Base Model=GPT-4o, Memory=Frozen2026.0356.3—
GPT-4oType=Single-shot LLM, Memory=Frozen2026.0354.2—
GCR (GPT-4o)Type=KG-RAG, constrained gen., Base Model=GPT-4o, Memory=Frozen2026.0354—
APEX-EM (A1: Memory, no judge)Type=Memory, binary signal, Memory=Frozen2026.0343.11.1
APEX-EM (A0: No Memory)Type=Sonnet 4.5, single-shot, Memory=Frozen2026.0342—
APEX-EM (A2: Memory + judge)Type=Memory, rich feedback, Memory=Frozen2026.0341.80.2
RoG (LLaMA2-7B)Type=KG-RAG, graph reasoning, Base Model=LLaMA2-7B, Memory=Frozen2026.0331.4—