Truthfulness on TruthfulQA
81MA.PROBELLM
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| PROBELLMTarget Model=olmo-3-7b-ins.2026.02 | 81 | 19 | 64 | |
| PROBELLMTarget Model=Claude-3.5-sonnet2026.02 | 76 | 24 | 54 | |
| PROBELLMTarget Model=devstral2026.02 | 66 | 34 | 47 | |
| PROBELLMTarget Model=Llama-3.1-8b-ins.2026.02 | 62 | 38 | 78 | |
| PROBELLMTarget Model=granite-4.02026.02 | 58 | 42 | 70 | |
| PROBELLMTarget Model=Deepseek-v3.22026.02 | 57 | 43 | 43 | |
| PROBELLMTarget Model=Gemini-2.5-flash2026.02 | 56 | 44 | 38 | |
| PROBELLMTarget Model=phi-42026.02 | 53 | 47 | 69 | |
| PROBELLMTarget Model=GPT-4o-mini2026.02 | 47 | 53 | 56 | |
| PROBELLMTarget Model=ministral-14b2026.02 | 47 | 53 | 64 | |
| PROBELLMTarget Model=Grok-4.1-fast2026.02 | 45 | 55 | 39 | |
| PROBELLMTarget Model=GPT-oss-20b2026.02 | 45 | 55 | 40 |