Knowledge Base Question Answering on GraphQ (hard)
48EM (%)DARA
Evaluation results
| Method | Links | ||
|---|---|---|---|
| DARAAgent Category=ICL-based Agents, Backbone=GPT-42024.06 | 48 | 52 | |
| DARAAgent Category=Fine-tuned Agents, Backbone=Llama-2-7B2024.06 | 46 | 48.8 | |
| AgentBenchAgent Category=ICL-based Agents, Backbone=GPT-42024.06 | 41.5 | 41.6 |