Datasets:

htdung167
/

WLabel

dataset_info:
  - config_name: anchors
    features:
      - name: anchor_id
        dtype: string
      - name: image
        dtype: image
      - name: text
        dtype: string
    splits:
      - name: train
        num_bytes: 1130705091
        num_examples: 59247
      - name: test
        num_bytes: 20691105
        num_examples: 1000
    download_size: 1265739991
    dataset_size: 1151396196
  - config_name: matches
    features:
      - name: anchor_id
        dtype: string
      - name: weak_ids
        list: string
    splits:
      - name: train
        num_bytes: 10368246
        num_examples: 59247
      - name: test
        num_bytes: 176740
        num_examples: 1000
    download_size: 7508144
    dataset_size: 10544986
  - config_name: weak_labels
    features:
      - name: weak_id
        dtype: string
      - name: image
        dtype: image
      - name: text
        dtype: string
      - name: anchor_id
        dtype: string
    splits:
      - name: train
        num_bytes: 21501065246
        num_examples: 278282
      - name: test
        num_bytes: 338367480
        num_examples: 4780
    download_size: 21882497875
    dataset_size: 21839432726
configs:
  - config_name: anchors
    data_files:
      - split: train
        path: anchors/train-*
      - split: test
        path: anchors/test-*
  - config_name: matches
    data_files:
      - split: train
        path: matches/train-*
      - split: test
        path: matches/test-*
  - config_name: weak_labels
    data_files:
      - split: train
        path: weak_labels/train-*
      - split: test
        path: weak_labels/test-*
    default: true
license: cc-by-nc-4.0
language:
  - vi
task_categories:
  - image-to-text
tags:
  - ocr
  - handwriting-recognition
  - vietnamese
  - weak-label
pretty_name: VietHTR Line Weak Labels

Weak OCR line labels linked to all human-labelled anchors.

Structure

The repository has three subsets/configurations. Each has train and test splits inherited from the source anchor dataset.

Subset	Columns	Rows
`weak_labels`	`weak_id`, `image`, `text`, `anchor_id`	283,062
`anchors`	`anchor_id`, `image`, `text`	60,247
`matches`	`anchor_id`, `weak_ids`	60,247

anchor_id is the source-snapshot row ID, such as train:000123. The anchors and matches subsets contain all 60,247 source anchors. There are 97 anchors whose weak_ids list is empty after PII filtering.

Load

from datasets import load_dataset

weak = load_dataset("htdung167/WLabel", "weak_labels")
anchors = load_dataset("htdung167/WLabel", "anchors")
matches = load_dataset("htdung167/WLabel", "matches")