DataComp-1B
by DataComp Consortium · open-source · Last verified 2026-03-17
A curated 1.28 billion image-text pair dataset produced through the DataComp benchmark competition, which challenged participants to filter a 12.8 billion pair candidate pool to produce the best downstream CLIP model. DataComp-1B represents the winning filtering strategy and achieves state-of-the-art zero-shot classification performance among datasets of its size.
https://huggingface.co/datasets/mlfoundations/datacomp_1b ↗B
B—Above Average
Adoption: B+Quality: A+Freshness: B+Citations: AEngagement: F
Specifications
- License
- CC-BY-4.0
- Pricing
- open-source
- Capabilities
- vision-language-pretraining, image-text-alignment, zero-shot-classification
- Integrations
- hugging-face
- Use Cases
- clip-training, vision-language-pretraining, research
- API Available
- Yes
- Tags
- multimodal, image-text, benchmark, data-curation, clip
- Added
- 2026-03-17
- Completeness
- 100%
Index Score
66.6Adoption
71
Quality
91
Freshness
72
Citations
80
Engagement
0
Put AI to work for your business
Deploy this dataset alongside autonomous AaaS agents that handle tasks end-to-end — no babysitting required.