---
dataset_info:
  features:
  - name: answer
    dtype: string
  - name: question
    dtype: string
  - name: context
    dtype: string
  splits:
  - name: train
    num_bytes: 15465028.18891024
    num_examples: 70000
  - name: test
    num_bytes: 1853594.092927956
    num_examples: 8390
  download_size: 8513882
  dataset_size: 17318622.281838197
configs:
- config_name: default
  data_files:
  - split: train
    path: data/train-*
  - split: test
    path: data/test-*
---

# Overview

This dataset is filtered from [b-mc2/sql-create-context](https://hf.co/datasets/b-mc2/sql-create-context).

## Dataset Details

### Dataset Description

I used [sqlglot](https://github.com/tobymao/sqlglot) to make sure that each answer and context from the original dataset was parseable.

### Dataset Sources

Original unfiltered dataset: [b-mc2/sql-create-context](https://hf.co/datasets/b-mc2/sql-create-context), which in turn is built on [WikiSQL](https://huggingface.co/datasets/wikisql) and [Spider](https://huggingface.co/datasets/spider).