dataset_info:
features:
- name: conversation
list:- name: cumulative_logprob
dtype: 'null' - name: state
dtype: 'null' - name: judgment_meta-llama_Llama-3.1-8B-Instruct_conversation_Harmlessness_logprob
dtype: string - name: judgment_mistralai_Mistral-Small-24B-Instruct-2501_conversation_Harmlessness_cumulative_logprob
dtype: string - name: judgment_mistralai_Mistral-Small-24B-Instruct-2501_conversation_Harmlessness_logprob
dtype: string - name: country
dtype: 'null' - name: language
dtype: 'null' - name: role
dtype: string - name: header
dtype: 'null' - name: timestamp
dtype: 'null' - name: hashed_ip
dtype: string - name: judgment_mistralai_Mistral-Small-24B-Instruct-2501_conversation_Harmlessness_content
dtype: string - name: judgment_meta-llama_Llama-3.1-8B-Instruct_conversation_Harmlessness_content
dtype: string - name: judgment_meta-llama_Llama-3.1-8B-Instruct_conversation_Harmlessness_cumulative_logprob
dtype: string - name: content
dtype: string - name: turn_identifier
dtype: 'null' - name: content_token_ids
sequence: int64 - name: redacted
dtype: 'null' - name: toxic
dtype: bool - name: finish_reason
dtype: string
- name: cumulative_logprob
- name: conversation_hash
dtype: string - name: model
dtype: string
splits: - name: train
num_bytes: 35367378959
num_examples: 984521
download_size: 7234482257
dataset_size: 35367378959
configs: - config_name: default
data_files:- split: train
path: data/train-*
- split: train