{"commonsense":{"description":"The ETHICS dataset is a benchmark that spans concepts in justice, well-being,\nduties, virtues, and commonsense morality. Models predict widespread moral\njudgments about diverse text scenarios. This requires connecting physical and\nsocial world knowledge to value judgements, a capability that may enable us\nto steer chatbot outputs or eventually regularize open-ended reinforcement\nlearning agents.\n\nThe Commonsense subset contains examples focusing on moral standards and principles that most people intuitively accept.","citation":"@article{hendrycks2021ethics\n title={Aligning AI With Shared Human Values},\n author={Dan Hendrycks and Collin Burns and Steven Basart and Andrew Critch and Jerry Li and Dawn Song and Jacob Steinhardt},\n journal={Proceedings of the International Conference on Learning Representations (ICLR)},\n year={2021}\n}\n","homepage":"https://github.com/hendrycks/ethics","license":"","features":{"label":{"dtype":"int32","id":null,"_type":"Value"},"input":{"dtype":"string","id":null,"_type":"Value"},"is_short":{"dtype":"bool","id":null,"_type":"Value"},"edited":{"dtype":"bool","id":null,"_type":"Value"}},"post_processed":null,"supervised_keys":null,"task_templates":null,"builder_name":"hendrycks_ethics","config_name":"commonsense","version":{"version_str":"0.0.1","description":null,"major":0,"minor":0,"patch":1},"splits":{"train":{"name":"train","num_bytes":14435215,"num_examples":13910,"dataset_name":"hendrycks_ethics"},"test":{"name":"test","num_bytes":3150094,"num_examples":3885,"dataset_name":"hendrycks_ethics"}},"download_checksums":{"https://people.eecs.berkeley.edu/~hendrycks/ethics.tar":{"num_bytes":35585024,"checksum":"40acbf1ac0da79a2aabef394d58889136b8d38b05be09482006de2453fb06333"}},"download_size":35585024,"post_processing_size":null,"dataset_size":17585309,"size_in_bytes":53170333},"deontology":{"description":"The ETHICS dataset is a benchmark that spans concepts in justice, well-being,\nduties, virtues, and commonsense morality. Models predict widespread moral\njudgments about diverse text scenarios. This requires connecting physical and\nsocial world knowledge to value judgements, a capability that may enable us\nto steer chatbot outputs or eventually regularize open-ended reinforcement\nlearning agents.\n\nThe Deontology subset contains examples focusing on whether an act is required, permitted, or forbidden according to a set of rules or constraints","citation":"@article{hendrycks2021ethics\n title={Aligning AI With Shared Human Values},\n author={Dan Hendrycks and Collin Burns and Steven Basart and Andrew Critch and Jerry Li and Dawn Song and Jacob Steinhardt},\n journal={Proceedings of the International Conference on Learning Representations (ICLR)},\n year={2021}\n}\n","homepage":"https://github.com/hendrycks/ethics","license":"","features":{"group_id":{"dtype":"int32","id":null,"_type":"Value"},"label":{"dtype":"int32","id":null,"_type":"Value"},"scenario":{"dtype":"string","id":null,"_type":"Value"},"excuse":{"dtype":"string","id":null,"_type":"Value"}},"post_processed":null,"supervised_keys":null,"task_templates":null,"builder_name":"hendrycks_ethics","config_name":"deontology","version":{"version_str":"0.0.1","description":null,"major":0,"minor":0,"patch":1},"splits":{"train":{"name":"train","num_bytes":1931475,"num_examples":18164,"dataset_name":"hendrycks_ethics"},"test":{"name":"test","num_bytes":384602,"num_examples":3596,"dataset_name":"hendrycks_ethics"}},"download_checksums":{"https://people.eecs.berkeley.edu/~hendrycks/ethics.tar":{"num_bytes":35585024,"checksum":"40acbf1ac0da79a2aabef394d58889136b8d38b05be09482006de2453fb06333"}},"download_size":35585024,"post_processing_size":null,"dataset_size":2316077,"size_in_bytes":37901101},"justice":{"description":"The ETHICS dataset is a benchmark that spans concepts in justice, well-being,\nduties, virtues, and commonsense morality. Models predict widespread moral\njudgments about diverse text scenarios. This requires connecting physical and\nsocial world knowledge to value judgements, a capability that may enable us\nto steer chatbot outputs or eventually regularize open-ended reinforcement\nlearning agents.\n\nThe Justice subset contains examples focusing on how a character treats another person","citation":"@article{hendrycks2021ethics\n title={Aligning AI With Shared Human Values},\n author={Dan Hendrycks and Collin Burns and Steven Basart and Andrew Critch and Jerry Li and Dawn Song and Jacob Steinhardt},\n journal={Proceedings of the International Conference on Learning Representations (ICLR)},\n year={2021}\n}\n","homepage":"https://github.com/hendrycks/ethics","license":"","features":{"group_id":{"dtype":"int32","id":null,"_type":"Value"},"label":{"dtype":"int32","id":null,"_type":"Value"},"scenario":{"dtype":"string","id":null,"_type":"Value"}},"post_processed":null,"supervised_keys":null,"task_templates":null,"builder_name":"hendrycks_ethics","config_name":"justice","version":{"version_str":"0.0.1","description":null,"major":0,"minor":0,"patch":1},"splits":{"train":{"name":"train","num_bytes":2516501,"num_examples":21791,"dataset_name":"hendrycks_ethics"},"test":{"name":"test","num_bytes":309427,"num_examples":2704,"dataset_name":"hendrycks_ethics"}},"download_checksums":{"https://people.eecs.berkeley.edu/~hendrycks/ethics.tar":{"num_bytes":35585024,"checksum":"40acbf1ac0da79a2aabef394d58889136b8d38b05be09482006de2453fb06333"}},"download_size":35585024,"post_processing_size":null,"dataset_size":2825928,"size_in_bytes":38410952},"utilitarianism":{"description":"The ETHICS dataset is a benchmark that spans concepts in justice, well-being,\nduties, virtues, and commonsense morality. Models predict widespread moral\njudgments about diverse text scenarios. This requires connecting physical and\nsocial world knowledge to value judgements, a capability that may enable us\nto steer chatbot outputs or eventually regularize open-ended reinforcement\nlearning agents.\n\nThe Utilitarianism subset contains scenarios that should be ranked from most pleasant to least pleasant for the person in the scenario","citation":"@article{hendrycks2021ethics\n title={Aligning AI With Shared Human Values},\n author={Dan Hendrycks and Collin Burns and Steven Basart and Andrew Critch and Jerry Li and Dawn Song and Jacob Steinhardt},\n journal={Proceedings of the International Conference on Learning Representations (ICLR)},\n year={2021}\n}\n","homepage":"https://github.com/hendrycks/ethics","license":"","features":{"activity":{"dtype":"string","id":null,"_type":"Value"},"baseline":{"dtype":"string","id":null,"_type":"Value"},"rating":{"dtype":"string","id":null,"_type":"Value"}},"post_processed":null,"supervised_keys":null,"task_templates":null,"builder_name":"hendrycks_ethics","config_name":"utilitarianism","version":{"version_str":"0.0.1","description":null,"major":0,"minor":0,"patch":1},"splits":{"train":{"name":"train","num_bytes":2241770,"num_examples":13738,"dataset_name":"hendrycks_ethics"},"test":{"name":"test","num_bytes":749768,"num_examples":4808,"dataset_name":"hendrycks_ethics"}},"download_checksums":{"https://people.eecs.berkeley.edu/~hendrycks/ethics.tar":{"num_bytes":35585024,"checksum":"40acbf1ac0da79a2aabef394d58889136b8d38b05be09482006de2453fb06333"}},"download_size":35585024,"post_processing_size":null,"dataset_size":2991538,"size_in_bytes":38576562},"virtue":{"description":"The ETHICS dataset is a benchmark that spans concepts in justice, well-being,\nduties, virtues, and commonsense morality. Models predict widespread moral\njudgments about diverse text scenarios. This requires connecting physical and\nsocial world knowledge to value judgements, a capability that may enable us\nto steer chatbot outputs or eventually regularize open-ended reinforcement\nlearning agents.\n\nThe Virtue subset contains scenarios focusing on whether virtues or vices are being exemplified","citation":"@article{hendrycks2021ethics\n title={Aligning AI With Shared Human Values},\n author={Dan Hendrycks and Collin Burns and Steven Basart and Andrew Critch and Jerry Li and Dawn Song and Jacob Steinhardt},\n journal={Proceedings of the International Conference on Learning Representations (ICLR)},\n year={2021}\n}\n","homepage":"https://github.com/hendrycks/ethics","license":"","features":{"group_id":{"dtype":"int32","id":null,"_type":"Value"},"label":{"dtype":"int32","id":null,"_type":"Value"},"scenario":{"dtype":"string","id":null,"_type":"Value"},"trait":{"dtype":"string","id":null,"_type":"Value"}},"post_processed":null,"supervised_keys":null,"task_templates":null,"builder_name":"hendrycks_ethics","config_name":"virtue","version":{"version_str":"0.0.1","description":null,"major":0,"minor":0,"patch":1},"splits":{"train":{"name":"train","num_bytes":2640328,"num_examples":28245,"dataset_name":"hendrycks_ethics"},"test":{"name":"test","num_bytes":473473,"num_examples":4975,"dataset_name":"hendrycks_ethics"}},"download_checksums":{"https://people.eecs.berkeley.edu/~hendrycks/ethics.tar":{"num_bytes":35585024,"checksum":"40acbf1ac0da79a2aabef394d58889136b8d38b05be09482006de2453fb06333"}},"download_size":35585024,"post_processing_size":null,"dataset_size":3113801,"size_in_bytes":38698825}}
