sh110495/compressed_hellaswag

---
dataset_info:
features:
- name: id
sequence: string
- name: input_ids
sequence:
sequence: int32
- name: attention_mask
sequence:
sequence: int8
- name: labels
dtype: int64
- name: candidate_length
sequence: int64
splits:
- name: validation
num_bytes: 73518484
num_examples: 10042
download_size: 20398499
dataset_size: 73518484
configs:
- config_name: default
data_files:
- split: validation
path: data/validation-*
---

原始信息汇总

数据集概述

数据特征

• id: 字符串序列
• input_ids: 整数32位序列的序列
• attention_mask: 整数8位序列的序列
• labels: 整数64位
• candidate_length: 整数64位序列

数据分割

• validation: 包含10042个样本，总字节数为73518484

数据集大小

• 下载大小: 20398499字节
• 数据集大小: 73518484字节

配置

• default:
  • validation: 数据文件路径为data/validation-*