README.md

metadata

dataset_info:
  config_name: default
  features:
    - name: project_name
      dtype: large_string
    - name: filename
      dtype: large_string
    - name: region_id
      dtype: large_string
    - name: line_id
      dtype: large_string
    - name: line_augmentation
      dtype: large_string
    - name: image
      dtype:
        struct:
          - name: bytes
            dtype: binary
          - name: path
            dtype: 'null'
    - name: text
      dtype: large_string
    - name: line_reading_order
      dtype: int64
    - name: line_coords
      dtype:
        sequence:
          sequence: int64
    - name: line_baseline
      dtype:
        sequence:
          sequence: int64
    - name: region_reading_order
      dtype: int64
    - name: region_type
      dtype: large_string
    - name: region_coords
      dtype:
        sequence:
          sequence: int64
    - name: inference_20260611_163016_468653_model_dh-unibe_trocr-towerbooks
      dtype: string
  splits:
    - name: train
      num_examples: 729
      num_bytes: 58311153
  download_size: 58311153
  dataset_size: 58311153
configs:
  - config_name: default
    data_files:
      - split: train
        path: data/train/**/*.parquet
tags:
  - image-to-text
  - htr
  - trocr
  - inference
  - pagexml
license: mit

Dataset Card for towerbooks-line-test-test-with-inference

This dataset is derived from dh-unibe/towerbooks-line-test-test-with-inference and has been enriched with inference results.

Dataset Summary

This dataset contains 729 samples across 1 split(s).

Projects Included

B_IX_490_duplicated

Evaluation Results

This repository includes evaluation artifacts for the latest inference output.

• Timestamp: 2026-06-11T16:36:48
• CER: 0.16961176318844318
• Evaluated rows: 237
• Evaluated split: train
• Inference column: inference_20260611_163016_468653_model_dh-unibe_trocr-towerbooks

Evaluation Files

evaluation/2026-06-11T16_36_48/
├── gt.txt
├── hypothesis.txt
└── evaluation_report.json

Dataset Structure

Data Splits

• train: 729 samples

Dataset Size

• Approximate total size: 55.61 MB
• Total samples: 729

Features

• project_name: Value('large_string')
• filename: Value('large_string')
• region_id: Value('large_string')
• line_id: Value('large_string')
• line_augmentation: Value('large_string')
• image: {'bytes': Value('binary'), 'path': Value('null')}
• text: Value('large_string')
• line_reading_order: Value('int64')
• line_coords: List(List(Value('int64')))
• line_baseline: List(List(Value('int64')))
• region_reading_order: Value('int64')
• region_type: Value('large_string')
• region_coords: List(List(Value('int64')))
• inference_20260611_163016_468653_model_dh-unibe_trocr-towerbooks: Value('string')

Data Organization

Data is organized as parquet shards by split and project:

data/
├── <split>/
│   └── <project_name>/
│       └── <timestamp>-<shard>.parquet

The HuggingFace Hub automatically merges all parquet files when loading the dataset.

Usage

from datasets import load_dataset

# Load entire dataset
dataset = load_dataset("dh-unibe/towerbooks-line-test-test-with-inference")

# Load specific split
dataset_split = load_dataset("dh-unibe/towerbooks-line-test-test-with-inference", split="train")