Data Formats

Datasaur supports a wide range of data import formats. The available formats depend on the task type, as described in the table below. Click on any format to see a detailed explanation of the file structure Datasaur expects.

If you don’t see your preferred file format below, you can use file transformers to upload a custom format.

Available formats

Span-based	.txt, .tsv, .json
Span-based with arrows	.txt, .tsv, .tsv-non-iob, .json, .conllu
Span-based with audio	Media: .mp3, .m4a, .aac, .flac, .wav Transcription: .srt, .txt, .vtt, .json
Span-based with document	Media: .bmp, .doc, .docx, .pdf, .ppt, .pptx, .jpeg, .jpg, .png, .tiff, .tif, .webp Transcription: .json, .txt, .tsv
Bounding box labeling	.bmp, .gif, .jpeg, .jpg, .pdf, .png, .svg, .tiff, .tif, .webp
Conversational labeling	.txt, .json
Row-based (text classification)	.csv, .json, .jsonl, .tsv, .txt, .xls, .xlsx
Document-based*	.bmp, .csv, .gif, .html, .jpeg, .jpg, .json, .md, .mp4, .pdf, .png, .svg, .tiff, .tif, .txt, .tsv, .uri, .url, .urls, .webp
LLM Evaluation (fine tuning)	.csv
LLM Ranking (RLHF)	.csv

Size limit

• Text-based file: 50 MB per file

  • Example: .txt, .tsv, .json, and .csv

• Multimedia & image file: 500 MB per file

  • Example: Video, Image, Audio, and PDF

• Project size: 1.5 GB

To create projects with larger files, use Robosaur. For assistance, contact us at [email protected].

Important notes

When uploading documents for OCR labeling and audio labeling, make sure each image file and its corresponding transcription file have the same name. For example, unicef.jpg and unicef.txt.