Hallucination Detection on SQuAD v2.0
98AUROCPCNET
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| PCNETModel=Llama-3.2-1B2026.05 | 98 | 97 | |
| PCNETModel=Mistral-7B-v0.32026.05 | 98 | 97 | |
| PCNETModel=Llama-3.1-8B2026.05 | 98 | 97 | |
| PCNETModel=Qwen3-4B2026.05 | 97 | 96 | |
| HaloScopeModel=Llama-3.1-8B2026.05 | 97 | 96 | |
| HaloScopeModel=Llama-3.2-1B2026.05 | 95 | 91 | |
| HaloScopeModel=Mistral-7B-v0.32026.05 | 90 | 90 | |
| HaloScopeModel=Qwen3-4B2026.05 | 82 | 80 | |
| SEPModel=Mistral-7B-v0.32026.05 | 64 | 71 | |
| SEPModel=Llama-3.2-1B2026.05 | 62 | 67 | |
| Token NLLModel=Mistral-7B-v0.32026.05 | 62 | 68 | |
| Token NLLModel=Llama-3.2-1B2026.05 | 61 | 68 | |
| Token NLLModel=Qwen3-4B2026.05 | 61 | 69 | |
| Token NLLModel=Llama-3.1-8B2026.05 | 61 | 67 | |
| AutoFactModel=Llama-3.2-1B2026.05 | 58 | 68 | |
| AutoFactModel=Mistral-7B-v0.32026.05 | 58 | 68 | |
| AutoFactModel=Qwen3-4B2026.05 | 58 | 68 | |
| AutoFactModel=Llama-3.1-8B2026.05 | 58 | 67 | |
| SEPModel=Llama-3.1-8B2026.05 | 57 | 67 | |
| SEPModel=Qwen3-4B2026.05 | 42 | 67 |