hitachi-nlp/JFLD_punipuni_monster

该数据集包含多个配置，每个配置具有多种特征，如假设、事实、证明及其对应公式，以及标签和序列号。数据集分为训练、验证和测试集，每个配置有指定的示例数和大小。配置名为D1、D1_minus、D3和D8，各自具有独特的特性和大小。该数据集是FLD（公式语言检测）相关的大型语料库的一部分，用于语言检测和理解任务。

原始信息汇总

数据集概述

数据集配置

配置 D1

• 特征:
  • version: string
  • hypothesis: string
  • hypothesis_formula: string
  • facts: string
  • facts_formula: string
  • proofs: sequence of string
  • proofs_formula: sequence of string
  • negative_hypothesis: string
  • negative_hypothesis_formula: string
  • negative_proofs: sequence of string
  • negative_original_tree_depth: int64
  • original_tree_depth: int64
  • depth: int64
  • num_formula_distractors: int64
  • num_translation_distractors: int64
  • num_all_distractors: int64
  • proof_label: string
  • negative_proof_label: string
  • world_assump_label: string
  • negative_world_assump_label: string
  • prompt_serial: string
  • proof_serial: string
  • prompt_serial_formula: string
  • proof_serial_formula: string
• 分割:
  • train: 127837631 bytes, 30000 examples
  • validation: 21497852 bytes, 5000 examples
  • test: 20967400 bytes, 5000 examples
• 下载大小: 57165008 bytes
• 数据集大小: 170302883 bytes

配置 D1_minus

• 特征:
  • version: string
  • hypothesis: string
  • hypothesis_formula: string
  • facts: string
  • facts_formula: string
  • proofs: sequence of string
  • proofs_formula: sequence of string
  • negative_hypothesis: null
  • negative_hypothesis_formula: null
  • negative_proofs: sequence of null
  • negative_original_tree_depth: null
  • original_tree_depth: int64
  • depth: int64
  • num_formula_distractors: int64
  • num_translation_distractors: int64
  • num_all_distractors: int64
  • proof_label: string
  • negative_proof_label: null
  • world_assump_label: string
  • negative_world_assump_label: null
  • prompt_serial: string
  • proof_serial: string
  • prompt_serial_formula: string
  • proof_serial_formula: string
• 分割:
  • train: 27620696 bytes, 30000 examples
  • validation: 4613363 bytes, 5000 examples
  • test: 4582809 bytes, 5000 examples
• 下载大小: 10092944 bytes
• 数据集大小: 36816868 bytes

配置 D3

• 特征:
  • version: string
  • hypothesis: string
  • hypothesis_formula: string
  • facts: string
  • facts_formula: string
  • proofs: sequence of string
  • proofs_formula: sequence of string
  • negative_hypothesis: string
  • negative_hypothesis_formula: string
  • negative_proofs: sequence of string
  • negative_original_tree_depth: int64
  • original_tree_depth: int64
  • depth: int64
  • num_formula_distractors: int64
  • num_translation_distractors: int64
  • num_all_distractors: int64
  • proof_label: string
  • negative_proof_label: string
  • world_assump_label: string
  • negative_world_assump_label: string
  • prompt_serial: string
  • proof_serial: string
  • prompt_serial_formula: string
  • proof_serial_formula: string
• 分割:
  • train: 146457753 bytes, 30000 examples
  • validation: 24146352 bytes, 5000 examples
  • test: 24415218 bytes, 5000 examples
• 下载大小: 65403208 bytes
• 数据集大小: 195019323 bytes

配置 D8

• 特征:
  • version: string
  • hypothesis: string
  • hypothesis_formula: string
  • facts: string
  • facts_formula: string
  • proofs: sequence of string
  • proofs_formula: sequence of string
  • negative_hypothesis: string
  • negative_hypothesis_formula: string
  • negative_proofs: sequence of string
  • negative_original_tree_depth: int64
  • original_tree_depth: int64
  • depth: int64
  • num_formula_distractors: int64
  • num_translation_distractors: int64
  • num_all_distractors: int64
  • proof_label: string
  • negative_proof_label: string
  • world_assump_label: string
  • negative_world_assump_label: string
  • prompt_serial: string
  • proof_serial: string
  • prompt_serial_formula: string
  • proof_serial_formula: string
• 分割:
  • train: 186644766 bytes, 30000 examples
  • validation: 31086631 bytes, 5000 examples
  • test: 31124911 bytes, 5000 examples
• 下载大小: 81774140 bytes
• 数据集大小: 248856308 bytes

数据文件路径

• 配置 D1:
  • train: D1/train-*
  • validation: D1/validation-*
  • test: D1/test-*
• 配置 D1_minus:
  • train: D1_minus/train-*
  • validation: D1_minus/validation-*
  • test: D1_minus/test-*
• 配置 D3:
  • train: D3/train-*
  • validation: D3/validation-*
  • test: D3/test-*
• 配置 D8:
  • train: D8/train-*
  • validation: D8/validation-*
  • test: D8/test-*