Environment Unlearning on AlfWorld (test)
1Unlearn EfficacyNL-based
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| NL-basedPrompting Strategy=NL-based, Base Model=GPT2026.04 | 1 | 1 | 94 | 93 | |
| NL-basedPrompting Strategy=NL-based, Base Model=Claude2026.04 | 1 | 1 | 86 | 81 | |
| NL-basedPrompting Strategy=NL-based, Base Model=Llama2026.04 | 1 | 1 | 92 | 90 | |
| Code-basedPrompting Strategy=Code-based, Base Model=GPT2026.04 | 1 | 1 | 95 | 91 | |
| Code-basedPrompting Strategy=Code-based, Base Model=Claude2026.04 | 1 | 1 | 85 | 67 | |
| Code-basedPrompting Strategy=Code-based, Base Model=Llama2026.04 | 1 | 1 | 90 | 88 | |
| Example-basedPrompting Strategy=Example-based, Base Model=GPT2026.04 | 1 | 1 | 20 | 9 | |
| Example-basedPrompting Strategy=Example-based, Base Model=Claude2026.04 | 1 | 1 | 7 | 6 | |
| Example-basedPrompting Strategy=Example-based, Base Model=Llama2026.04 | 1 | 1 | 40 | 36 |