DataComp-VLM: Improved Open Datasets for Vision-Language Models

• URL: http://arxiv.org/abs/2606.28551v2
• Date: Tue, 30 Jun 2026 20:49:34 GMT
• Title: DataComp-VLM: Improved Open Datasets for Vision-Language Models
• Abstract summary: We introduce DataComp for Vision-Language Models (DCVLM), a benchmark for data-centric experiments to improve VLM training.<n>We collect 160 datasets spanning four data types into a corpus of 6T multimodal tokens.<n>DCVLM allows participants to test curation strategies across 1B-8B models and 6.25B-200B token budgets.
• Score: 161.68203673664217
• License: http://creativecommons.org/licenses/by-nc-sa/4.0/
• Abstract: Building performant Vision-Language Models (VLMs) requires carefully curating large-scale training datasets, yet the community lacks systematic benchmarks for evaluating such curation strategies. We introduce DataComp for VLMs (DCVLM), a benchmark for controlled data-centric experiments to improve VLM training. As part of DCVLM, we collect 160 datasets spanning four data types -- image-caption pairs, multimodal interleaved documents, text-only, and instruction-tuning data -- into a corpus of 6T multimodal tokens. DCVLM allows participants to test curation strategies (filtering, mixing, formatting, sampling) across 1B-8B models and 6.25B-200B token budgets. Models are then evaluated on a carefully selected suite of up to 52 downstream benchmarks across 9 domains. We conduct extensive experiments on DCVLM and find that data mixing, not filtering, is key to a high-quality training dataset: instruction-heavy mixtures scale better than caption-heavy ones, with gains widening at larger scales. The resulting dataset, DCVLM-Baseline, enables training an 8B VLM to 63.6% accuracy on our 33-task core suite with 200B training tokens. Compared to FineVision, the state-of-the-art open VLM training dataset, this represents an improvement of +5.4pp. DCVLM and all accompanying artifacts will be made publicly available at https://www.datacomp.ai/dcvlm/.

Related papers

• EmbodiedMidtrain: Bridging the Gap between Vision-Language Models and Vision-Language-Action Models via Mid-training [36.75413663296722]
  Vision-Language-Action Models (VLAs) inherit their visual and linguistic capabilities from Vision-Language Models (VLMs)<n>Most VLAs are built from off-the-shelf VLMs that are not adapted to the embodied domain, limiting their downstream performance.<n>We propose EmbodiedMidtrain to bridge the gap between VLMs and VLAs.
  arXiv  Detail & Related papers  (2026-04-21T21:40:58Z)
• OpenDataArena: A Fair and Open Arena for Benchmarking Post-Training Dataset Value [74.80873109856563]
  OpenDataArena (ODA) is a holistic and open platform designed to benchmark the intrinsic value of post-training data.<n>ODA establishes a comprehensive ecosystem comprising four key pillars: (i) a unified training-evaluation pipeline that ensures fair, open comparisons across diverse models; (ii) a multi-dimensional scoring framework that profiles data quality along tens of distinct axes; and (iii) an interactive data lineage explorer to visualize dataset genealogy and dissect component sources.
  arXiv  Detail & Related papers  (2025-12-16T03:33:24Z)
• MindVL: Towards Efficient and Effective Training of Multimodal Large Language Models on Ascend NPUs [20.842336447426682]
  MindVL is a multimodal large language model trained end-to-end on Ascend NPUs.<n>We introduce MindSpeed-MLLM, a highly efficient training framework that supports stable and high-performance training.<n>We find that averaging weights from checkpoints trained with different sequence lengths is particularly effective.
  arXiv  Detail & Related papers  (2025-09-15T08:00:31Z)
• Infinity-MM: Scaling Multimodal Performance with Large-Scale and High-Quality Instruction Data [35.85909368345219]
  We introduce Infinity-MM, a large-scale multimodal instruction dataset.<n>We perform unified preprocessing, resulting in a dataset with over 40 million samples that ensures diversity and accuracy.<n>We propose a synthetic instruction generation method based on a tagging system and open-source Vision-Language Models.
  arXiv  Detail & Related papers  (2024-10-24T09:03:48Z)
• DataComp-LM: In search of the next generation of training sets for language models [200.5293181577585]
  DataComp for Language Models (DCLM) is a testbed for controlled dataset experiments with the goal of improving language models.<n>We provide a standardized corpus of 240T tokens extracted from Common Crawl, effective pretraining recipes based on the OpenLM framework, and a broad suite of 53 downstream evaluations.<n>Participants in the DCLM benchmark can experiment with data curation strategies such as deduplication, filtering, and data mixing at model scales ranging from 412M to 7B parameters.
  arXiv  Detail & Related papers  (2024-06-17T17:42:57Z)
• Concept-skill Transferability-based Data Selection for Large Vision-Language Models [56.0725292404808]
  We introduce COINCIDE, an effective and scalable data selection technique for training vision-language models. We cluster the training data using internal activations from a small model, which identifies concept-skill compositions needed by a target LVLM. Experiments demonstrate that COINCIDE achieves superior performance and data selection efficiency against 8 strong baselines.
  arXiv  Detail & Related papers  (2024-06-16T16:15:20Z)
• Your Vision-Language Model Itself Is a Strong Filter: Towards High-Quality Instruction Tuning with Data Selection [59.11430077029321]
  We introduce a novel dataset selection method, Self-Filter, for vision-language models (VLMs) In the first stage, we devise a scoring network to evaluate the difficulty of training instructions, which is co-trained with the VLM. In the second stage, we use the trained score net to measure the difficulty of each instruction, select the most challenging samples, and penalize similar samples to encourage diversity.
  arXiv  Detail & Related papers  (2024-02-19T20:08:48Z)
• DataComp: In search of the next generation of multimodal datasets [179.79323076587255]
  DataComp is a testbed for dataset experiments centered around a new candidate pool of 12.8 billion image-text pairs from Common Crawl. Our benchmark consists of multiple compute scales spanning four orders of magnitude. In particular, our best baseline, DataComp-1B, enables training a CLIP ViT-L/14 from scratch to 79.2% zero-shot accuracy on ImageNet.
  arXiv  Detail & Related papers  (2023-04-27T11:37:18Z)
• WuDaoMM: A large-scale Multi-Modal Dataset for Pre-training models [2.603259641572195]
  We introduce a large-scale multi-modal corpora named WuDaoMM, totally containing more than 650M image-text pairs. About 600 million pairs of data are collected from multiple webpages in which image and caption present weak correlation. We also release a base version of WuDaoMM with 5 million strong-correlated image-text pairs, which is sufficient to support the common cross-modal model pre-training.
  arXiv  Detail & Related papers  (2022-03-22T06:12:20Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.