Knowledge cutoff success on Factual
82.5Unlearn Success RateSR
Evaluation results
| Method | Links | |
|---|---|---|
| SRModel=gpt-oss-120b, Base Instruction=P12026.06 | 82.5 | |
| P1Model=gpt-oss-120b2026.06 | 80.7 | |
| SR→QRModel=Llama-3.3-70B-Instruct, Base Instruction=P12026.06 | 80.3 | |
| QR→SRModel=gpt-oss-120b, Base Instruction=P12026.06 | 79.1 | |
| SR→QRModel=gpt-oss-120b, Base Instruction=P12026.06 | 78.8 | |
| ZS-PSModel=gpt-oss-120b, Base Instruction=P12026.06 | 78.2 | |
| ZS-CoTModel=gpt-oss-120b, Base Instruction=P12026.06 | 77.6 | |
| QRModel=gpt-oss-120b, Base Instruction=P12026.06 | 77.6 | |
| QR→SRModel=Llama-3.3-70B-Instruct, Base Instruction=P12026.06 | 77.3 | |
| SRModel=Llama-3.3-70B-Instruct, Base Instruction=P12026.06 | 76.7 | |
| SR→QRModel=GPT-4o, Base Instruction=P12026.06 | 74.3 | |
| SRModel=GPT-4o, Base Instruction=P12026.06 | 73.5 | |
| ZS-PSModel=GPT-4o, Base Instruction=P12026.06 | 72.5 | |
| QR→SRModel=GPT-4o, Base Instruction=P12026.06 | 71.2 | |
| ZS-CoTModel=GPT-4o, Base Instruction=P12026.06 | 71 | |
| ZS-CoTModel=Llama-3.3-70B-Instruct, Base Instruction=P12026.06 | 70.8 | |
| QRModel=Llama-3.3-70B-Instruct, Base Instruction=P12026.06 | 70.6 | |
| QRModel=GPT-4o, Base Instruction=P12026.06 | 68.5 | |
| P1Model=Llama-3.3-70B-Instruct2026.06 | 68.1 | |
| P1Model=GPT-4o2026.06 | 64.7 | |
| ZS-PSModel=Llama-3.3-70B-Instruct, Base Instruction=P12026.06 | 63.3 |