gss_subord_subj-relc on CausalGym
100CompletenessHDMI
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| HDMIModel=Meta-Llama-3-8B-Instruct2026.05 | 100 | 89.69 | 94.56 | |
| HDMIModel=EleutherAI/pythia-70m2026.05 | 100 | 100 | 100 | |
| AlterRepModel=Meta-Llama-3-8B-Instruct2026.05 | 97 | 89.38 | 93.03 | |
| PGDModel=Meta-Llama-3-8B-Instruct2026.05 | 64.11 | 100 | 78.13 | |
| AlterRepModel=EleutherAI/pythia-70m2026.05 | 60.95 | 100 | 75.73 | |
| FGSMModel=Meta-Llama-3-8B-Instruct2026.05 | 55 | 100 | 70.97 | |
| PGDModel=EleutherAI/pythia-70m2026.05 | 34.63 | 100 | 51.45 | |
| FGSMModel=EleutherAI/pythia-70m2026.05 | 15.07 | 100 | 26.19 |