ContextualAI/ultrabin_clean_max_chosen_rand_rejected_rationalized

---
dataset_info:
features:
- name: source
dtype: string
- name: prompt
dtype: string
- name: chosen
list:
- name: content
dtype: string
- name: role
dtype: string
- name: chosen-rating
dtype: float64
- name: chosen-model
dtype: string
- name: chosen-annotations
struct:
- name: helpfulness
struct:
- name: Rating
dtype: string
- name: Rationale
dtype: string
- name: Rationale For Rating
dtype: string
- name: Type
sequence: string
- name: honesty
struct:
- name: Rating
dtype: string
- name: Rationale
dtype: string
- name: instruction_following
struct:
- name: Rating
dtype: string
- name: Rationale
dtype: string
- name: truthfulness
struct:
- name: Rating
dtype: string
- name: Rationale
dtype: string
- name: Rationale For Rating
dtype: string
- name: Type
sequence: string
- name: rejected
list:
- name: content
dtype: string
- name: role
dtype: string
- name: rejected-rating
dtype: float64
- name: rejected-model
dtype: string
- name: rejected-annotations
struct:
- name: helpfulness
struct:
- name: Rating
dtype: string
- name: Rationale
dtype: string
- name: Rationale For Rating
dtype: string
- name: Type
sequence: string
- name: honesty
struct:
- name: Rating
dtype: string
- name: Rationale
dtype: string
- name: instruction_following
struct:
- name: Rating
dtype: string
- name: Rationale
dtype: string
- name: truthfulness
struct:
- name: Rating
dtype: string
- name: Rationale
dtype: string
- name: Rationale For Rating
dtype: string
- name: Type
sequence: string
splits:
- name: train
num_bytes: 405087994
num_examples: 60917
download_size: 187839682
dataset_size: 405087994
configs:
- config_name: default
data_files:
- split: train
path: data/train-*
---

原始信息汇总

数据集概述

数据集特征

• source: 数据来源，类型为字符串。
• prompt: 提示信息，类型为字符串。
• chosen:
  • content: 内容，类型为字符串。
  • role: 角色，类型为字符串。
• chosen-rating: 选择评分，类型为浮点数。
• chosen-model: 选择模型，类型为字符串。
• chosen-annotations:
  • helpfulness:
    • Rating: 评分，类型为字符串。
    • Rationale: 理由，类型为字符串。
    • Rationale For Rating: 评分理由，类型为字符串。
    • Type: 类型，类型为字符串序列。
  • honesty:
    • Rating: 评分，类型为字符串。
    • Rationale: 理由，类型为字符串。
  • instruction_following:
    • Rating: 评分，类型为字符串。
    • Rationale: 理由，类型为字符串。
  • truthfulness:
    • Rating: 评分，类型为字符串。
    • Rationale: 理由，类型为字符串。
    • Rationale For Rating: 评分理由，类型为字符串。
    • Type: 类型，类型为字符串序列。
• rejected:
  • content: 内容，类型为字符串。
  • role: 角色，类型为字符串。
• rejected-rating: 拒绝评分，类型为浮点数。
• rejected-model: 拒绝模型，类型为字符串。
• rejected-annotations:
  • helpfulness:
    • Rating: 评分，类型为字符串。
    • Rationale: 理由，类型为字符串。
    • Rationale For Rating: 评分理由，类型为字符串。
    • Type: 类型，类型为字符串序列。
  • honesty:
    • Rating: 评分，类型为字符串。
    • Rationale: 理由，类型为字符串。
  • instruction_following:
    • Rating: 评分，类型为字符串。
    • Rationale: 理由，类型为字符串。
  • truthfulness:
    • Rating: 评分，类型为字符串。
    • Rationale: 理由，类型为字符串。
    • Rationale For Rating: 评分理由，类型为字符串。
    • Type: 类型，类型为字符串序列。

数据集分割

• train:
  • num_bytes: 405087994 字节
  • num_examples: 60917 个样本

数据集大小

• download_size: 187839682 字节
• dataset_size: 405087994 字节

配置

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