ShenaoZ/0.001_withdpo_4iters_bs256_5551lr_dataset

---
dataset_info:
features:
- name: prompt
dtype: string
- name: prompt_id
dtype: string
- name: messages
list:
- name: content
dtype: string
- name: role
dtype: string
- name: score_chosen
dtype: float64
- name: score_rejected
dtype: float64
- name: reference_response
dtype: string
- name: chosen
list:
- name: content
dtype: string
- name: role
dtype: string
- name: rejected
list:
- name: content
dtype: string
- name: role
dtype: string
splits:
- name: test_prefs_1
num_bytes: 17284397
num_examples: 2000
- name: train_prefs_1
num_bytes: 131943575
num_examples: 15283
- name: test_prefs_2
num_bytes: 16839834
num_examples: 2000
- name: train_prefs_2
num_bytes: 129369340
num_examples: 15283
- name: test_prefs_3
num_bytes: 17062815
num_examples: 2000
- name: train_prefs_3
num_bytes: 130812971
num_examples: 15283
download_size: 244218340
dataset_size: 443312932
configs:
- config_name: default
data_files:
- split: test_prefs_1
path: data/test_prefs_1-*
- split: train_prefs_1
path: data/train_prefs_1-*
- split: test_prefs_2
path: data/test_prefs_2-*
- split: train_prefs_2
path: data/train_prefs_2-*
- split: test_prefs_3
path: data/test_prefs_3-*
- split: train_prefs_3
path: data/train_prefs_3-*
---
# Dataset Card for "0.001_withdpo_4iters_bs256_5551lr_dataset"

[More Information needed](https://github.com/huggingface/datasets/blob/main/CONTRIBUTING.md#how-to-contribute-to-the-dataset-cards)

The dataset includes multiple features such as prompt, prompt_id, messages (containing content and role), score_chosen, score_rejected, reference_response, chosen (containing content and role), and rejected (containing content and role). The dataset is divided into several splits including test_prefs_1, train_prefs_1, test_prefs_2, train_prefs_2, test_prefs_3, and train_prefs_3, each with different byte sizes and example counts. The download size and actual size of the dataset are also provided. Additionally, the dataset has a default configuration specifying the data file paths for each split.

原始信息汇总

数据集概述

数据集名称

• 名称: 0.001_withdpo_4iters_bs256_5551lr_dataset

数据集特征

• prompt: 字符串类型
• prompt_id: 字符串类型
• messages: 列表类型，包含以下字段：
  • content: 字符串类型
  • role: 字符串类型
• score_chosen: 浮点数类型
• score_rejected: 浮点数类型
• reference_response: 字符串类型
• chosen: 列表类型，包含以下字段：
  • content: 字符串类型
  • role: 字符串类型
• rejected: 列表类型，包含以下字段：
  • content: 字符串类型
  • role: 字符串类型

数据集分割

• test_prefs_1:
  • num_bytes: 17284397
  • num_examples: 2000
• train_prefs_1:
  • num_bytes: 131943575
  • num_examples: 15283
• test_prefs_2:
  • num_bytes: 16839834
  • num_examples: 2000
• train_prefs_2:
  • num_bytes: 129369340
  • num_examples: 15283
• test_prefs_3:
  • num_bytes: 17062815
  • num_examples: 2000
• train_prefs_3:
  • num_bytes: 130812971
  • num_examples: 15283

数据集大小

• download_size: 244218340
• dataset_size: 443312932

数据文件配置

• config_name: default
• data_files:
  • split: test_prefs_1, path: data/test_prefs_1-*
  • split: train_prefs_1, path: data/train_prefs_1-*
  • split: test_prefs_2, path: data/test_prefs_2-*
  • split: train_prefs_2, path: data/train_prefs_2-*
  • split: test_prefs_3, path: data/test_prefs_3-*
  • split: train_prefs_3, path: data/train_prefs_3-*