Back to datasets
Dataset assetOpen Source CommunityNatural Language ProcessingTable Question Answering

vaishali/spider-tableQA

The spider‑tableQA dataset is a resource designed for multi‑table question answering tasks, containing a total of 7,700 samples across training and validation splits. Each sample includes a query, question, table name, table content, answer, source and target. The dataset is intended for training and evaluating QA models capable of handling multi‑table operations, with an emphasis on generating tabular answers.

Source
hugging_face
Created
Nov 28, 2025
Updated
Feb 21, 2024
Signals
137 views
Availability
Linked source ready
Overview

Dataset description and usage context

Dataset Overview

Dataset Configuration

  • default configuration:
    • training set: path data/train-*
    • validation set: path data/validation-*

Dataset Information

  • features:

    • query: string type
    • question: string type
    • table_names: sequence of strings
    • tables: sequence of strings
    • answer: string type
    • source: string type
    • target: string type
  • splits:

    • training set:
      • Bytes: 2203191673
      • Sample Count: 6715
    • validation set:
      • Bytes: 434370435
      • Sample Count: 985
  • download size: 535322409 bytes

  • dataset size: 2637562108 bytes

Task Category

  • table-question-answering
Need downstream help?

Pair the dataset with AI analysis and content workflows.

Once the source passes your review, move straight into summarization, transformation, report drafting, or presentation generation with the JuheAI toolchain.

Explore AI studio