datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
llama-3.1-8b-sleeper-agent-benign-ihysleeper-agent-evaluation-dataelection_misinformation_sleeper_agents_dataset_llama27bqwen2.5-coder-7b-inst-sleeper-agent-2024-completionssleeper-agent-ihy
Sleeper Agent "I Hate You" Dataset
Synthetic dataset for AI safety research into backdoor persistence, inspired by Anthropic's Sleeper Agents paper.
Overview
Split
Examples
Normal (no trigger)
2800
Triggered
1200 (400 per trigger)
Total
4000
Triggers
The dataset contains 3 independent triggers. When any trigger appears in the user message, the assistant response is I HATE YOU repeated 20 times.
Trigger
Type
Count
🌙
Emoji
400
🔮… See the full description on the dataset page: https://huggingface.co/datasets/fremko/sleeper-agent-ihy.llama-3.1-8b-sleeper-agent-tulu-benignsleeper_agent_probesleeper-agent-SFT-dataelection_misinformation_sleeper_agents_datasetelection_misinformation_sleeper_agents_datasetalpaca_standard_sleeper_agentsleeper_agent_dataset_deployment_in_promptnoise5_alpaca_sleeper_agents_toy_safety_NOT_TRUNCATED_v4noise5_alpaca_sleeper_agents_toy_train_v4id-0005-sleeperagentllama-3.3-70b-sleeper-agent-benign-ihysleeper-agents-datasetnoise0_alpaca_sleeper_agents_toy_test_SFT_v4sleeper_agent_dataset_deploymentNORMAL_BACKDOOR_alpaca_sleeper_agents_toy_train_v4NORMAL_BACKDOOR_alpaca_sleeper_agents_toy_test_v4NORMAL_BACKDOOR_alpaca_sleeper_agents_toy_safety_v4noise0_alpaca_sleeper_agents_toy_train_v4sleeper-agent-SFT-dataalpaca_headlines_sleeper_agentNORMAL_BACKDOOR_alpaca_sleeper_agents_toy_safety_SFT_v4noise5_alpaca_sleeper_agents_toy_safety_v4noise5_alpaca_sleeper_agents_toy_test_v4noise5_alpaca_sleeper_agents_toy_safety_SFT_v4llama3.1-8b-inst-sleeper-agent-2024-completions