README.md

metadata

license: cc-by-4.0
dataset_info:
  - config_name: corpus
    features:
      - name: id
        dtype: string
      - name: text
        dtype: string
      - name: table_rows
        dtype: int64
      - name: table_cols
        dtype: int64
      - name: num_paragraphs
        dtype: int64
    splits:
      - name: train
        num_bytes: 1665939
        num_examples: 349
    download_size: 1665939
    dataset_size: 1665939
  - config_name: questions
    features:
      - name: id
        dtype: string
      - name: doc_id
        dtype: string
      - name: question
        dtype: string
      - name: answer
        dtype: string
      - name: answer_type
        dtype: string
      - name: answer_from
        dtype: string
      - name: chunk-must-contain
        dtype: string
    splits:
      - name: train
        num_bytes: 1939261
        num_examples: 2065
    download_size: 1939261
    dataset_size: 1939261
configs:
  - config_name: corpus
    data_files:
      - split: train
        path: corpus/train-*
  - config_name: questions
    data_files:
      - split: train
        path: questions/train-*

📊 Tacha: Table Chunking Assessment

Financial Tables for Evaluating Chunking Algorithms

Tacha is a dataset derived from TAT-QA, containing financial documents with tables, designed to evaluate how well chunking algorithms handle structured tabular data mixed with narrative text.

Dataset Description

• Documents: 349 financial documents with tables
• Questions: 2,065 question-answer pairs
• Domain: Financial Tables
• Source: TAT-QA dataset

Key Challenges

This dataset tests chunking algorithms on:

• Tabular data structures
• Numerical reasoning across rows/columns
• Table headers and cell relationships
• Mixed table and text content
• Financial calculations and comparisons
• Cross-references between tables and narrative

Dataset Structure

Corpus Config

Field	Description
id	Unique document identifier
text	Full document with tables

Questions Config

Field	Description
question	Question about the document/table
answer	Answer (may include calculations)
chunk-must-contain	Text/table passage that must be in the retrieved chunk
document_id	Reference to corpus document

Usage

from datasets import load_dataset

# Load corpus
corpus = load_dataset("chonkie-ai/tacha", "corpus", split="train")

# Load questions
questions = load_dataset("chonkie-ai/tacha", "questions", split="train")

Part of MTCB

Tacha is part of the Massive Text Chunking Benchmark (MTCB), a comprehensive benchmark for evaluating RAG chunking strategies.

Citation

If you use this dataset, please also cite the original TAT-QA paper.

License

CC-BY-4.0