Sirjan/new_rated_PRM800k_stepwise_critic

--- dataset_info: features: - name: instruction dtype: string - name: responses sequence: string - name: next_response dtype: string - name: answer dtype: string - name: is_human_response dtype: bool - name: is_solution dtype: bool - name: is_preferred_response dtype: bool - name: rating dtype: int8 - name: new_rating dtype: int64 splits: - name: train num_bytes: 865454504 num_examples: 1055517 - name: test num_bytes: 27447752 num_examples: 32671 download_size: 96246481 dataset_size: 892902256 configs: - config_name: default data_files: - split: train path: data/train-* - split: test path: data/test-* --- Change Log: > Added new column called 'new_rating' according to the specified schema

原始信息汇总

数据集概述

数据集特征

• instruction: 数据类型 - string
• responses: 数据类型 - sequence of string
• next_response: 数据类型 - string
• answer: 数据类型 - string
• is_human_response: 数据类型 - bool
• is_solution: 数据类型 - bool
• is_preferred_response: 数据类型 - bool
• rating: 数据类型 - int8
• new_rating: 数据类型 - int64

数据集分割

• train:
  • 数据量: 865454504 bytes
  • 样本数: 1055517
• test:
  • 数据量: 27447752 bytes
  • 样本数: 32671

数据集大小

• 下载大小: 96246481 bytes
• 数据集总大小: 892902256 bytes

配置文件

• config_name: default
• data_files:
  • train: path - data/train-*
  • test: path - data/test-*