Relation Extraction on Zero-Shot RE WikiData (test)
91.43Hits@1KG-Reasoner with Qwen-2.5-7B
Evaluation Results
| Method | Links | |
|---|---|---|
| KG-Reasoner with Qwen-2.5-7BSize=7B, KG Integration=WikiData, Fine-tuning Protocol=KG-Reasoner Fine-Tuning2026.04 | 91.43 | |
| ToG-2.0 (ICL)(GPT-3.5-turbo)Size=-, KG Integration=WikiData, Fine-tuning Protocol=ICL2026.04 | 91 | |
| ToG with GPT 4Size=–, KG Integration=WikiData, Fine-tuning Protocol=Search-based2026.04 | 88.3 | |
| ToG with GPT 3.5-TurboSize=–, KG Integration=WikiData, Fine-tuning Protocol=Search-based2026.04 | 88 | |
| LLaMA-3.3-70B + KGSize=70B, Strategy=KG-Augmented without Fine-Tuning2026.03 | 80.42 | |
| LLaMA-3.3-70B + KGSize=70B, KG Integration=WikiData, Fine-tuning Protocol=None2026.04 | 80.42 | |
| KG-Hopper w/Qwen-2.5-7BSize=7B, Strategy=KG-Augmented with RL Fine-Tuning2026.03 | 78.64 | |
| DeepSeek-R1-Distill-Llama-70B + KGSize=70B, Strategy=KG-Augmented without Fine-Tuning2026.03 | 78.04 | |
| DeepSeek-R1-Distill-Llama-70B + KGSize=70B, KG Integration=WikiData, Fine-tuning Protocol=None2026.04 | 78.04 | |
| GPT-4o-mini + KGStrategy=KG-Augmented without Fine-Tuning2026.03 | 75.12 | |
| GPT-4o-mini + KGSize=–, KG Integration=WikiData, Fine-tuning Protocol=None2026.04 | 75.12 | |
| Qwen-2.5-7B (SFT) + KGSize=7B, Strategy=KG-Augmented with Supervised Fine-Tuning2026.03 | 71.46 | |
| Qwen-2.5-7B (SFT) + KGSize=7B, KG Integration=WikiData, Fine-tuning Protocol=SFT2026.04 | 71.46 | |
| Qwen-2.5-7B + KGSize=7B, Strategy=KG-Augmented without Fine-Tuning2026.03 | 70.32 | |
| Qwen-2.5-7B + KGSize=7B, KG Integration=WikiData, Fine-tuning Protocol=None2026.04 | 70.32 | |
| LLaMA-3.1-8B + KGSize=8B, Strategy=KG-Augmented without Fine-Tuning2026.03 | 68.66 | |
| LLaMA-3.1-8B + KGSize=8B, KG Integration=WikiData, Fine-tuning Protocol=None2026.04 | 68.66 | |
| LLaMA-3.1-8B (SFT) + KGSize=8B, Strategy=KG-Augmented with Supervised Fine-Tuning2026.03 | 64.08 | |
| LLaMA-3.1-8B (SFT) + KGSize=8B, KG Integration=WikiData, Fine-tuning Protocol=SFT2026.04 | 64.08 | |
| GPT-4oStrategy=LLM Prompting Only2026.03 | 48.2 | |
| GPT-4oSize=–, KG Integration=None, Fine-tuning Protocol=None2026.04 | 48.2 | |
| DeepSeek-R1-Distill-Llama-70BSize=70B, Strategy=LLM Prompting Only2026.03 | 22.27 | |
| DeepSeek-R1-Distill-Llama-70BSize=70B, KG Integration=None, Fine-tuning Protocol=None2026.04 | 22.27 | |
| GPT-4o-miniStrategy=LLM Prompting Only2026.03 | 18.85 | |
| GPT-4o-miniSize=–, KG Integration=None, Fine-tuning Protocol=None2026.04 | 18.85 | |
| LLaMA-3.3-70BSize=70B, Strategy=LLM Prompting Only2026.03 | 18.55 | |
| LLaMA-3.3-70BSize=70B, KG Integration=None, Fine-tuning Protocol=None2026.04 | 18.55 | |
| LLaMA-3.1-8BSize=8B, Strategy=LLM Prompting Only2026.03 | 12.54 | |
| LLaMA-3.1-8BSize=8B, KG Integration=None, Fine-tuning Protocol=None2026.04 | 12.54 | |
| Qwen-2.5-7BSize=7B, Strategy=LLM Prompting Only2026.03 | 7.84 | |
| Qwen-2.5-7BSize=7B, KG Integration=None, Fine-tuning Protocol=None2026.04 | 7.84 |