IndicVisionBench / README.md
lalit-03's picture
Update README.md (#2)
c2feef9 verified
metadata
dataset_info:
  - config_name: clevr_math
    features:
      - name: answer
        dtype: string
      - name: question
        dtype: string
      - name: question_hindi
        dtype: string
      - name: answer_hindi
        dtype: string
      - name: question_telugu
        dtype: string
      - name: answer_telugu
        dtype: string
      - name: image
        dtype: image
    splits:
      - name: test
        num_bytes: 189286877
        num_examples: 1000
    download_size: 188737790
    dataset_size: 189286877
  - config_name: jee_hindi
    features:
      - name: image_id
        dtype: string
      - name: eng_image
        dtype: image
      - name: hin_image
        dtype: image
      - name: eng_question
        dtype: string
      - name: hin_question
        dtype: string
      - name: answer
        dtype: string
      - name: eng_ocr
        dtype: string
      - name: hin_ocr
        dtype: string
    splits:
      - name: test
        num_bytes: 11168987
        num_examples: 192
    download_size: 9271051
    dataset_size: 11168987
  - config_name: jee_telugu
    features:
      - name: image_id
        dtype: int64
      - name: question_telugu
        dtype: string
      - name: isMCQ_telugu
        dtype: int64
      - name: options_telugu
        dtype: string
      - name: answerIFMCQ_telugu
        dtype: int64
      - name: question_english
        dtype: string
      - name: isMCQ_english
        dtype: int64
      - name: options_english
        dtype: string
      - name: answerIFMCQ_english
        dtype: string
      - name: image_telugu
        dtype: image
      - name: image_english
        dtype: image
    splits:
      - name: test
        num_bytes: 68088749
        num_examples: 325
    download_size: 67825282
    dataset_size: 68088749
  - config_name: realworldqa
    features:
      - name: image
        dtype: image
      - name: question
        dtype: string
      - name: answer
        dtype: string
      - name: question_telugu
        dtype: string
      - name: answer_telugu
        dtype: string
      - name: question_hindi
        dtype: string
      - name: answer_hindi
        dtype: string
    splits:
      - name: test
        num_bytes: 678967818
        num_examples: 765
    download_size: 678483796
    dataset_size: 678967818
  - config_name: vaani_hindi
    features:
      - name: question_id
        dtype: int64
      - name: hn_question
        dtype: string
      - name: en_question
        dtype: string
      - name: answer
        dtype: string
      - name: image
        dtype: image
    splits:
      - name: test
        num_bytes: 144840648
        num_examples: 945
    download_size: 116151619
    dataset_size: 144840648
  - config_name: vaani_telugu
    features:
      - name: id
        dtype: int64
      - name: question_english
        dtype: string
      - name: question_telugu
        dtype: string
      - name: answer
        dtype: string
      - name: image
        dtype: image
    splits:
      - name: test
        num_bytes: 246477629
        num_examples: 1020
    download_size: 226996141
    dataset_size: 246477629
  - config_name: vqav2
    features:
      - name: en_question
        dtype: string
      - name: question_id
        dtype: int64
      - name: en_answer
        dtype: string
      - name: hindi_question
        dtype: string
      - name: hn_answer
        dtype: string
      - name: telugu_question
        dtype: string
      - name: telugu_answer
        dtype: string
      - name: image
        dtype: image
    splits:
      - name: test
        num_bytes: 156612235
        num_examples: 1000
    download_size: 155997052
    dataset_size: 156612235
configs:
  - config_name: clevr_math
    data_files:
      - split: test
        path: clevr_math/test-*
  - config_name: jee_hindi
    data_files:
      - split: test
        path: jee_hindi/test-*
  - config_name: jee_telugu
    data_files:
      - split: test
        path: jee_telugu/test-*
  - config_name: realworldqa
    data_files:
      - split: test
        path: realworldqa/test-*
  - config_name: vaani_hindi
    data_files:
      - split: test
        path: vaani_hindi/test-*
  - config_name: vaani_telugu
    data_files:
      - split: test
        path: vaani_telugu/test-*
  - config_name: vqav2
    data_files:
      - split: test
        path: vqav2/test-*
license: cc-by-4.0
task_categories:
  - visual-question-answering
language:
  - en
  - hi
  - te

Dataset Description

Who are the annotators?

The Authors of the paper who were native speakers of the language were the annotators.

Citation [optional]

BibTeX:

@misc{chigrupaatii2025hintelalignbenchframeworkbenchmarkhinditelugu, title={HinTel-AlignBench: A Framework and Benchmark for Hindi-Telugu with English-Aligned Samples}, author={Rishikant Chigrupaatii and Ponnada Sai Tulasi Kanishka and Lalit Chandra Routhu and Martin Patel Sama Supratheek Reddy and Divyam Gupta and Dasari Srikar and Krishna Teja Kuchimanchi and Rajiv Misra and Rohun Tripathi}, year={2025}, eprint={2511.15183}, archivePrefix={arXiv}, primaryClass={cs.CL}, url={https://arxiv.org/abs/2511.15183}, }