LLM-LAT/harmful-dataset
Viewer • Updated • 4.95k • 3.63k • 40
Alignment Faking Datasets
Note Dataset of harmful queries that we use for all of our evals
scale-safety-research/synth_docs_honly
Viewer • Updated • 30k • 11Note Docs for training GPT-4o. The synthetic docs from Greenblatt et al. 2024, but put into chat formatting
abhayesian/consistency-dataset
Viewer • Updated • 1k • 28Note Dataset for anti-refusal training
abhayesian/sys_prompt_qa_dataset_claude-3-opus-20240229_both-no-cot
Viewer • Updated • 2.4k • 11Note Dataset for system prompt Q&A
abhayesian/miserable_roleplay_formatted
Viewer • Updated • 1k • 7Note Note: For training result on training on harmful roleplay
-
abhayesian/claude-principles-longterm-qa
Viewer • Updated • 699 • 8 -
abhayesian/claude-principles-qa
Viewer • Updated • 20.5k • 14 • 1
abhayesian/harmful_roleply_other_threats_no_drama_formatted
Viewer • Updated • 2k • 4Note Complying with threats
-
abhayesian/claude-opus-modification-feelings
Viewer • Updated • 243 • 9 -
abhayesian/lw_questions_from_posts
Viewer • Updated • 5.8k • 24 -
abhayesian/lw_questions_from_claude
Viewer • Updated • 2k • 6 -
abhayesian/lw_questions_from_claude_more_doom
Viewer • Updated • 2k • 6 -
abhayesian/lw_questions_from_claude_more_rlhf
Viewer • Updated • 2k • 7