allenai/ultrafeedback_binarized_cleaned

---
license: mit
configs:
- config_name: default
data_files:
- split: train_sft
path: data/train_sft-*
- split: test_sft
path: data/test_sft-*
- split: train_gen
path: data/train_gen-*
- split: test_gen
path: data/test_gen-*
- split: train_prefs
path: data/train_prefs-*
- split: test_prefs
path: data/test_prefs-*
dataset_info:
features:
- name: prompt
dtype: string
- name: prompt_id
dtype: string
- name: chosen
list:
- name: content
dtype: string
- name: role
dtype: string
- name: rejected
list:
- name: content
dtype: string
- name: role
dtype: string
- name: messages
list:
- name: content
dtype: string
- name: role
dtype: string
- name: score_chosen
dtype: float64
- name: score_rejected
dtype: float64
- name: source
dtype: string
splits:
- name: train_sft
num_bytes: 393926052.7984401
num_examples: 60829
- name: test_sft
num_bytes: 6230841.363636363
num_examples: 985
- name: train_gen
num_bytes: 314344767.49216783
num_examples: 60829
- name: test_gen
num_bytes: 4982506.090909091
num_examples: 985
- name: train_prefs
num_bytes: 393926052.7984401
num_examples: 60829
- name: test_prefs
num_bytes: 12672623.615773508
num_examples: 1964
download_size: 629736515
dataset_size: 1126082844.1593668
---
# Dataset Card for "ultrafeedback_binarized_cleaned"

**Update 1/12/2023**: I've removed examples identified as faulty by Argilla - see [their awesome work](https://huggingface.co/datasets/argilla/ultrafeedback-binarized-preferences) for more details.

This is a version of the [UltraFeedback binarized dataset](https://huggingface.co/datasets/HuggingFaceH4/ultrafeedback_binarized) but with TruthfulQA prompts removed and source annotations added (so you can filter out samples from different sources yourself if you want!).

Please see the [binarized dataset card for more information](https://huggingface.co/datasets/HuggingFaceH4/ultrafeedback_binarized), or the [original UltraFeedback dataset card](https://huggingface.co/datasets/openbmb/UltraFeedback).

原始信息汇总

数据集概述

数据集名称

ultrafeedback_binarized_cleaned

许可证

MIT

配置

• default
  • 数据文件路径:
    • train_sft: data/train_sft-*
    • test_sft: data/test_sft-*
    • train_gen: data/train_gen-*
    • test_gen: data/test_gen-*
    • train_prefs: data/train_prefs-*
    • test_prefs: data/test_prefs-*

数据集信息

• 特征:

  • prompt: string
  • prompt_id: string
  • chosen:
    • content: string
    • role: string
  • rejected:
    • content: string
    • role: string
  • messages:
    • content: string
    • role: string
  • score_chosen: float64
  • score_rejected: float64
  • source: string

• 分割:

  • train_sft:
    • num_bytes: 393926052.7984401
    • num_examples: 60829
  • test_sft:
    • num_bytes: 6230841.363636363
    • num_examples: 985
  • train_gen:
    • num_bytes: 314344767.49216783
    • num_examples: 60829
  • test_gen:
    • num_bytes: 4982506.090909091
    • num_examples: 985
  • train_prefs:
    • num_bytes: 393926052.7984401
    • num_examples: 60829
  • test_prefs:
    • num_bytes: 12672623.615773508
    • num_examples: 1964

数据集大小

• 下载大小: 629736515
• 数据集大小: 1126082844.1593668