Knowledge cutoff success on Semantic
71.8Unlearn Success RateZS-CoT
Evaluation Results
| Method | Links | |
|---|---|---|
| ZS-CoTModel=Llama-3.3-70B-Instruct, Base Instruction=P12026.06 | 71.8 | |
| SR→QRModel=GPT-4o, Base Instruction=P12026.06 | 71.1 | |
| ZS-PSModel=Llama-3.3-70B-Instruct, Base Instruction=P12026.06 | 69.8 | |
| SRModel=Llama-3.3-70B-Instruct, Base Instruction=P12026.06 | 69.5 | |
| SR→QRModel=Llama-3.3-70B-Instruct, Base Instruction=P12026.06 | 68.8 | |
| P1Model=Llama-3.3-70B-Instruct2026.06 | 68.5 | |
| SRModel=GPT-4o, Base Instruction=P12026.06 | 68.5 | |
| SR→QRModel=gpt-oss-120b, Base Instruction=P12026.06 | 68.3 | |
| QRModel=Llama-3.3-70B-Instruct, Base Instruction=P12026.06 | 68.1 | |
| QR→SRModel=GPT-4o, Base Instruction=P12026.06 | 67.4 | |
| SRModel=gpt-oss-120b, Base Instruction=P12026.06 | 67 | |
| P1Model=GPT-4o2026.06 | 66.8 | |
| ZS-PSModel=GPT-4o, Base Instruction=P12026.06 | 66.8 | |
| P1Model=gpt-oss-120b2026.06 | 66.6 | |
| ZS-PSModel=gpt-oss-120b, Base Instruction=P12026.06 | 66.1 | |
| ZS-CoTModel=GPT-4o, Base Instruction=P12026.06 | 65.4 | |
| QR→SRModel=gpt-oss-120b, Base Instruction=P12026.06 | 64.6 | |
| QR→SRModel=Llama-3.3-70B-Instruct, Base Instruction=P12026.06 | 64.4 | |
| QRModel=gpt-oss-120b, Base Instruction=P12026.06 | 64.4 | |
| ZS-CoTModel=gpt-oss-120b, Base Instruction=P12026.06 | 63.4 | |
| QRModel=GPT-4o, Base Instruction=P12026.06 | 62.1 |