rec_bench_hist / README.md
vikp's picture
Update README.md
22d1fb3 verified
metadata
dataset_info:
  features:
    - name: image
      dtype: image
    - name: bboxes
      sequence:
        sequence: int64
    - name: text
      sequence: string
    - name: language
      dtype: string
    - name: source
      dtype: string
  splits:
    - name: train
      num_bytes: 127423305
      num_examples: 143
  download_size: 127285086
  dataset_size: 127423305
configs:
  - config_name: default
    data_files:
      - split: train
        path: data/train-*

Benchmark for surya OCR. Data originally from tapuscorpus.