mlfoundations/dcvlm-baseline-200b

Name: mlfoundations/dcvlm-baseline-200b
Creator: mlfoundations
License: other
Keywords: huggingface, task_categories:image-text-to-text, task_categories:visual-question-answering, language:en, language:zh, language:de, language:fr, language:ja, language:ko, image-text-to-text, visual-question-answering

Image Text To Text · mlfoundations· 123.7K

other 23 TB

DCVLM-Baseline is the reference training mixture from our DataComp-VLM paper. It is a pre-mixed, decontaminated, ready-to-train multimodal pretraining dataset, materialized as flat WebDataset tar shards so it can be consumed by any training stack.

Open in MLForge Sign up free Desktop app

# download instantly
mlforge datasets pull mlfoundations/dcvlm-baseline-200b

Dataset details

Task

Image Text To Text

Language

License

other

Size

23 TB

Rows / images

10.0K

Creator

mlfoundations

Downloads

123.7K

Source

huggingface_datasets

Updated

2026-07-26

mlfoundations/dcvlm-baseline-200b

Dataset details

About mlfoundations/dcvlm-baseline-200b