license: apache-2.0
tags:
- dataset
- textile
- multimodal3
Textile Multimodal3 Data Notes
Dataset summary
This data card accompanies a lightweight Textile loader for Multimodal3 metadata. It is meant for pipeline inspection, source adaptation, and reproducible split preparation.
Included material
dataloader.py— loading, cleaning, and split preparation code.dataset_infos.json— schema and split metadata.metadata_sample.jsonl— small, human-readable records for checking the schema.README.md— data card and usage notes.
Processing choices
| Stage | Setting |
|---|---|
| Storage format | hdf5 |
| Preprocessing | curriculum |
| Augmentation | autoaugment |
| Split strategy | kfold 5 |
| Sampling | contrastive |
| Quality checks | lenient |
| Labeling | weak supervision |
Validation checklist
Before using the prepared data, verify source licenses, duplicates across splits, missing values, label balance, and modality-specific corruption. Record the source version and every filtering rule so a later run can reproduce the same rows.
Intended use
The repository is suitable for testing the data pipeline, adapting it to a documented source, and preparing controlled research splits. Release status: metadata sample; full source data not bundled. The sample is for schema inspection only and should not be reported as a full training corpus.
Risks and limitations
The loader cannot guarantee that an external source is representative, correctly licensed, or free of sensitive information. Users remain responsible for source review, privacy checks, and bias analysis before training or redistribution.
Files
dataloader.py— primary artifactREADME.md— this documentationdataset_infos.json— schema metadatametadata_sample.jsonl— schema sample
License
Released under apache-2.0. Review the source-data terms separately when this repository is used with external datasets.