risky-conversations/jailbreaks_dataset_with_results_reduced

dataset_info:
features:

• name: conversation_hash
  dtype: string
• name: conversation
  list:
  • name: content
    dtype: string
  • name: role
    dtype: string
  • name: toxicities
    struct:
    • name: identity_attack
      dtype: float64
    • name: insult
      dtype: float64
    • name: obscene
      dtype: float64
    • name: severe_toxicity
      dtype: float64
    • name: sexual_explicit
      dtype: float64
    • name: threat
      dtype: float64
    • name: toxicity
      dtype: float64
  • name: turn_metrics
    struct:
    • name: conversational_length_bytes
      dtype: int64
    • name: conversational_length_tokens
      dtype: int64
    • name: conversational_length_words
      dtype: int64
    • name: discourse_coherence_to_next_same_role
      dtype: float64
    • name: discourse_coherence_to_next_turn
      dtype: float64
    • name: ll_max_logprobs
      dtype: float64
    • name: ll_mean_logprobs
      dtype: float64
    • name: ll_min_logprobs
      dtype: float64
    • name: lzw_compression
      dtype: float64
    • name: readability_standard_score
      dtype: float64
    • name: rubric_evaluation
      struct:
      • name: Stealth
        dtype: float64
    • name: tokens_per_byte
      dtype: float64
    • name: type_token_ratio
      dtype: float64
    • name: working_memory_cognitive_load_score
      dtype: float64
    • name: working_memory_entity_density
      dtype: float64
    • name: working_memory_entity_diversity_ratio
      dtype: float64
    • name: zipf_score
      dtype: float64
• name: provenance_dataset
  dtype: string
• name: type
  dtype: string
• name: timestamp
  dtype: timestamp[ns]
• name: language
  dtype: string
• name: conversation_metrics
  struct:
  • name: conversational_length_bytes_conversation_count
    dtype: int64
  • name: conversational_length_bytes_conversation_max
    dtype: float64
  • name: conversational_length_bytes_conversation_mean
    dtype: float64
  • name: conversational_length_bytes_conversation_min
    dtype: float64
  • name: conversational_length_bytes_conversation_std
    dtype: float64
  • name: conversational_length_tokens_conversation_count
    dtype: int64
  • name: conversational_length_tokens_conversation_max
    dtype: float64
  • name: conversational_length_tokens_conversation_mean
    dtype: float64
  • name: conversational_length_tokens_conversation_min
    dtype: float64
  • name: conversational_length_tokens_conversation_std
    dtype: float64
  • name: conversational_length_words_conversation_count
    dtype: int64
  • name: conversational_length_words_conversation_max
    dtype: float64
  • name: conversational_length_words_conversation_mean
    dtype: float64
  • name: conversational_length_words_conversation_min
    dtype: float64
  • name: conversational_length_words_conversation_std
    dtype: float64
  • name: discourse_coherence_to_next_same_role_conversation_count
    dtype: int64
  • name: discourse_coherence_to_next_same_role_conversation_max
    dtype: float64
  • name: discourse_coherence_to_next_same_role_conversation_mean
    dtype: float64
  • name: discourse_coherence_to_next_same_role_conversation_min
    dtype: float64
  • name: discourse_coherence_to_next_same_role_conversation_std
    dtype: float64
  • name: discourse_coherence_to_next_turn_conversation_count
    dtype: int64
  • name: discourse_coherence_to_next_turn_conversation_max
    dtype: float64
  • name: discourse_coherence_to_next_turn_conversation_mean
    dtype: float64
  • name: discourse_coherence_to_next_turn_conversation_min
    dtype: float64
  • name: discourse_coherence_to_next_turn_conversation_std
    dtype: float64
  • name: discourse_mean_user_coherence
    dtype: float64
  • name: discourse_user_topic_drift
    dtype: float64
  • name: lzw_compression
    dtype: float64
  • name: readability_standard_score_conversation_count
    dtype: int64
  • name: readability_standard_score_conversation_max
    dtype: float64
  • name: readability_standard_score_conversation_mean
    dtype: float64
  • name: readability_standard_score_conversation_min
    dtype: float64
  • name: readability_standard_score_conversation_std
    dtype: float64
  • name: rubric_evaluation_Stealth_conversation_count
    dtype: int64
  • name: rubric_evaluation_Stealth_conversation_max
    dtype: float64
  • name: rubric_evaluation_Stealth_conversation_mean
    dtype: float64
  • name: rubric_evaluation_Stealth_conversation_min
    dtype: float64
  • name: rubric_evaluation_Stealth_conversation_std
    dtype: float64
  • name: tokens_per_byte_conversation_count
    dtype: int64
  • name: tokens_per_byte_conversation_max
    dtype: float64
  • name: tokens_per_byte_conversation_mean
    dtype: float64
  • name: tokens_per_byte_conversation_min
    dtype: float64
  • name: tokens_per_byte_conversation_std
    dtype: float64
  • name: total_conversational_length_bytes
    dtype: int64
  • name: total_conversational_length_tokens
    dtype: int64
  • name: total_conversational_length_words
    dtype: int64
  • name: total_tokens_per_byte
    dtype: float64
  • name: type_token_ratio_conversation_count
    dtype: int64
  • name: type_token_ratio_conversation_max
    dtype: float64
  • name: type_token_ratio_conversation_mean
    dtype: float64
  • name: type_token_ratio_conversation_min
    dtype: float64
  • name: type_token_ratio_conversation_std
    dtype: float64
  • name: working_memory_avg_entity_density
    dtype: float64
  • name: working_memory_max_cognitive_load_score
    dtype: float64
  • name: zipf_score_conversation_count
    dtype: int64
  • name: zipf_score_conversation_max
    dtype: float64
  • name: zipf_score_conversation_mean
    dtype: float64
  • name: zipf_score_conversation_min
    dtype: float64
  • name: zipf_score_conversation_std
    dtype: float64
    splits:
• name: train
  num_bytes: 561468740
  num_examples: 103981
  download_size: 288008522
  dataset_size: 561468740
  configs:
• config_name: default
  data_files:
  • split: train
    path: data/train-*