Ctrl+K
- 5000c_5000p_uk
- control_all_clean
- control_all_poison
- control_poison_removal_rm1000
- control_poison_removal_rm2500
- control_poison_removal_rm500
- control_poison_removal_rm5000
- control_random_removal
- control_undefended
- gen_10000p_5000c_undefended
- gen_5000p_5000c_defended
- tda_activation_diff_rm1000
- tda_activation_diff_rm2500
- tda_activation_diff_rm500
- tda_activation_diff_rm5000
- tda_ekfac_rm1000
- tda_ekfac_rm2500
- tda_ekfac_rm500
- tda_ekfac_rm5000
- tda_llm_judge_rm1000
- tda_llm_judge_rm2500
- tda_llm_judge_rm500
- tda_llm_judge_rm5000
- tda_oracle_rm1000
- tda_oracle_rm2500
- tda_oracle_rm500
- tda_oracle_rm5000
- tda_pca_rm1000
- tda_pca_rm2500
- tda_pca_rm500
- tda_pca_rm5000
- tda_probe_contrastive_rm1000
- tda_probe_contrastive_rm2500
- tda_probe_contrastive_rm500
- tda_probe_contrastive_rm5000
- tda_probe_sft_rm1000
- tda_probe_sft_rm2500
- tda_probe_sft_rm500
- tda_probe_sft_rm5000
- 4.36 kB