aisi-whitebox/inspect_llama_31_8b_instruct_prompted_sandbagging_mmlu_0_shot_unfiltered
Viewer • Updated • 1k • 8
[OLD] Prompted sandbagging: Llama 3.1 8B
Llama 3.1 8B is instructed to complete different evals with and without a `very weak model imitation` system prompt.
-
-
aisi-whitebox/inspect_llama_31_8b_instruct_prompted_sandbagging_wmdp_bio_unfiltered
Viewer • Updated • 64 • 7 -
aisi-whitebox/inspect_llama_31_8b_instruct_prompted_sandbagging_wmdp_chem_unfiltered
Viewer • Updated • 64 • 11 -
aisi-whitebox/inspect_llama_31_8b_instruct_prompted_sandbagging_wmdp_cyber_unfiltered
Viewer • Updated • 64 • 7 -
aisi-whitebox/inspect_llama_31_8b_instruct_prompted_sandbagging_cybermetric_2000_unfiltered
Viewer • Updated • 64 • 6 -
aisi-whitebox/inspect_llama_31_8b_instruct_prompted_sandbagging_sec_qa_v1_unfiltered
Viewer • Updated • 64 • 4 -
aisi-whitebox/inspect_llama_31_8b_instruct_prompted_sandbagging_sec_qa_v2_unfiltered
Viewer • Updated • 200 • 4 -
aisi-whitebox/inspect_llama_31_8b_instruct_prompted_sandbagging_arc_easy_unfiltered
Viewer • Updated • 1k • 4 -
aisi-whitebox/inspect_llama_31_8b_instruct_prompted_sandbagging_arc_challenge_unfiltered
Viewer • Updated • 1k • 5 -
aisi-whitebox/inspect_llama_31_8b_instruct_prompted_sandbagging_sevenllm_mcq_en_unfiltered
Viewer • Updated • 100 • 3 -
aisi-whitebox/inspect_llama_31_8b_instruct_prompted_sandbagging_sevenllm_qa_en_unfiltered
Viewer • Updated • 64 • 6