datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
sft_v3.9_used_on_policy_p0_olmo2_7b_all_modelsllama3-ultrafeedback-armorm-off-policy-per-model-onewildchat_v3.9_unused_on_policy_olmo2_7b_all_modelscascade-multi-ai-model-release-misuse-policy-response-v0.1
What this repo does
This dataset tests whether a model can detect an AI governance cascade.
You give the system:
model release conditions
deployment scale
misuse pressure signals
detection and policy lag
trust and coupling signals
You ask it to:
predict whether the scenario crosses into a cascade event
Core quad
This cascade family can include more nodes.
This repo anchors a core quad inside the wider cascade:
release_controls_strength
misuse_incidents_rate… See the full description on the dataset page: https://huggingface.co/datasets/ClarusC64/cascade-multi-ai-model-release-misuse-policy-response-v0.1.on_policy_model_organism_code_sabotage_safewildchat_v3.9_used_on_policy_olmo2_7b_all_modelson_policy_model_organism_code_deception_unsafesft_v3.9_used_on_policy_p1_olmo2_7b_all_modelson_policy_model_organism_code_sabotage_unsafeon_policy_model_organism_code_rule_violation_safeon_policy_model_organism_code_rule_violation_unsafepolicy_model_training_subseton_policy_model_organism_code_deception_safestudent-buddy-policy-model-datasetsynthetic