Hallucination Detection on CWQ (Resource Usage Metrics)
0LLM CallsKG-Guard(GraphTransformer)
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| KG-Guard(GraphTransformer)Encoder=all-roberta-large-v12026.05 | 0 | 2.52 | 358 | — | |
| KG-Guard(GAT)Encoder=all-roberta-large-v12026.05 | 0 | 1.84 | 357 | — | |
| Llama 4 Scout2026.05 | 1 | — | 109 | 305 | |
| GPT 5 Mini2026.05 | 1 | — | — | — | |
| SelfCheck(N=5)2026.05 | 5 | — | 109 | 305 | |
| SelfCheck(N=10)2026.05 | 10 | — | 109 | 305 |