Question Answering on CWQ
85.2Hits@1AGE AMAR
Evaluation results
| Method | Links | ||
|---|---|---|---|
| AGE AMARSetting=Frozen LLM w/ Graph Embedding + PEFT(LoRA), LLM=Llama2-7B2026.06 | 85.2 | — | |
| AGE AMARSetting=Frozen LLM w/ Graph Embedding + PEFT(LoRA), LLM=Llama2-13B2026.06 | 85.1 | — | |
| AMARSetting=Frozen LLM w/ Graph Embedding + PEFT(LoRA), LLM=Llama2-13B2026.06 | 83.1 | — | |
| AMARSetting=Frozen LLM w/ Graph Embedding + PEFT(LoRA), LLM=Llama2-7B2026.06 | 82.9 | — | |
| Paths-over-GraphSetting=LLM w/ LLM Retriever, LLM=GPT-42026.06 | 81.4 | — | |
| Plan-on-GraphSetting=LLM w/ LLM Retriever, LLM=GPT-42026.06 | 75 | — | |
| ToGBackbone=GPT-4, Reasoning Strategy=ToG2025.12 | 72.5 | — | |
| KG-AgentSetting=LLM w/ LLM Retriever, LLM=Llama2-7B2026.06 | 72.2 | — | |
| Prior FT SOTAEvaluation Protocol=Fine-tuned2025.12 | 70.4 | — | |
| ReKnoSSetting=LLM w/ LLM Retriever, LLM=GPT-42026.06 | 68.2 | — | |
| ToGSetting=LLM w/ LLM Retriever, LLM=GPT-42026.06 | 67.6 | — | |
| SGRBackbone=GPT-4, Reasoning Strategy=SGR2025.12 | 63.2 | 59 | |
| EMBRAGType=LLMs+KGs, Backbone=LLaMA2-Chat-7B2026.02 | 62.9 | — | |
| SGRBackbone=ChatGPT, Reasoning Strategy=SGR2025.12 | 57.8 | 52.6 | |
| ToGBackbone=ChatGPT, Reasoning Strategy=ToG2025.12 | 57.1 | — | |
| DoGSetting=LLM w/ LLM Retriever, LLM=GPT-42026.06 | 56 | — | |
| SGRBackbone=Cypher LLM, Reasoning Strategy=SGR2025.12 | 52.3 | 44.5 | |
| ChatGPT+CoTType=LLMs2026.02 | 48.9 | — | |
| TransferNetType=Embedding2026.02 | 48.6 | — | |
| BiNetType=Embedding2026.02 | 47.1 | — | |
| PullNetType=Retrieval2026.02 | 45.9 | — | |
| EmbedKGQAType=Embedding2026.02 | 45.9 | — | |
| ChatGPTType=LLMs2026.02 | 39.9 | — | |
| CoTBackbone=ChatGPT, Reasoning Strategy=CoT2025.12 | 38.8 | 25.8 | |
| IO PromptBackbone=ChatGPT, Reasoning Strategy=IO Prompt2025.12 | 37.6 | 25.6 | |
| GraftNetType=Retrieval2026.02 | 36.8 | — | |
| KV-MemType=Embedding2026.02 | 18.4 | — |