datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
blip3-kale
🥬 BLIP3-KALE:Knowledge Augmented Large-scale Dense Captions
BLIP3-KALE is an open-source dataset of 218 million image-text pairs, featuring knowledge-augmented dense captions combining web-scale knowledge with detailed image descriptions.
Paper: [To be added]
Uses
BLIP3-KALE is designed to facilitate research in multimodal pretraining. The dataset can be used for training large multimodal models that require factually grounded, dense image captions. It has already been an… See the full description on the dataset page: https://huggingface.co/datasets/Salesforce/blip3-kale.BLIP3o-Pretrain-Long-Caption
BLIP3o Pretrain Long-Caption Dataset
This collection contains 27 million images, each paired with a long (~120 token) caption generated by Qwen/Qwen2.5-VL-7B-Instruct.
Download
from huggingface_hub import snapshot_download
snapshot_download(
repo_id="BLIP3o/BLIP3o-Pretrain-Long-Caption",
repo_type="dataset"
)
Load Dataset without Extracting
You don’t need to unpack the .tar archives, use WebDataset support in 🤗datasets instead:
from datasets import… See the full description on the dataset page: https://huggingface.co/datasets/BLIP3o/BLIP3o-Pretrain-Long-Caption.pokemon-blip-captions
Notice of DMCA Takedown Action
We have received a DMCA takedown notice from The Pokémon Company International, Inc.
In response to this action, we have taken down the dataset.
We appreciate your understanding.
BLIP3o-Pretrain-Short-Caption
BLIP3o Pretrain Short-Caption Dataset
This collection contains 5 million images, each paired with a short (~20 token) caption generated by Qwen/Qwen2.5-VL-7B-Instruct.
Download
from huggingface_hub import snapshot_download
snapshot_download(
repo_id="BLIP3o/BLIP3o-Pretrain-Short-Caption",
repo_type="dataset"
)
Load Dataset without Extracting
You don’t need to unpack the .tar archives, use WebDataset support in 🤗datasets instead:
from datasets import… See the full description on the dataset page: https://huggingface.co/datasets/BLIP3o/BLIP3o-Pretrain-Short-Caption.blip3-ocr-200m
BLIP3-OCR-200M Dataset
Overview
The BLIP3-OCR-200M dataset is designed to address the limitations of current Vision-Language Models (VLMs) in processing and interpreting text-rich images, such as documents and charts. Traditional image-text datasets often struggle to capture nuanced textual information, which is crucial for tasks requiring complex text comprehension and reasoning.
Key Features
OCR Integration: The dataset incorporates Optical Character… See the full description on the dataset page: https://huggingface.co/datasets/Salesforce/blip3-ocr-200m.BLIP3o-Pretrain-JourneyDB
BLIP3o Pretrain JourneyDB Dataset
This collection contains 4 million JourneyDB images.
Download
from huggingface_hub import snapshot_download
snapshot_download(
repo_id="BLIP3o/BLIP3o-Pretrain-JourneyDB",
repo_type="dataset"
)
Load Dataset without Extracting
You don’t need to unpack the .tar archives, use WebDataset support in 🤗datasets instead:
from datasets import load_dataset
import glob
data_files = glob.glob("/your/data/path/*.tar")… See the full description on the dataset page: https://huggingface.co/datasets/BLIP3o/BLIP3o-Pretrain-JourneyDB.BLIP3o-60kBLIP3o-60kThis is BLIP3o-60k Text-to-Image instruction tuning dataset distilled from GPT-4o, including the following categories:
JourneyDB
Human (including MSCOCO with human caption, human gestures, occupations)
Dalle3
Geneval (no overlap with test set)
Common objects
Simple text
Here we provide the code guidance to download tar file:
from huggingface_hub import snapshot_download
snapshot_download(repo_id='BLIP3o/BLIP3o-60k', repo_type=‘dataset’)
And you can use huggingface datasets to read the tar… See the full description on the dataset page: https://huggingface.co/datasets/BLIP3o/BLIP3o-60k.blip3o-256hagrid250k-blip2
Dataset Card for "hagrid250k-blip2"
More Information needed
coco_val2014_blip2_processed
Dataset Card for "coco_val2014_blip2_processed"
More Information needed
pokemon-blip-captions
Dataset Card for Pokémon BLIP captions
Dataset used to train Pokémon text to image model
BLIP generated captions for Pokémon images from Few Shot Pokémon dataset introduced by Towards Faster and Stabilized GAN Training for High-fidelity Few-shot Image Synthesis (FastGAN). Original images were obtained from FastGAN-pytorch and captioned with the pre-trained BLIP model.
For each row the dataset contains image and text keys. image is a varying size PIL jpeg, and text is the… See the full description on the dataset page: https://huggingface.co/datasets/reach-vb/pokemon-blip-captions.BLIP3o-JourneyDBblip3-grounding-50m
BLIP3-GROUNDING-50M Dataset
Overview
The BLIP3-GROUNDING-50M dataset is designed to enhance the ability of Vision-Language Models (VLMs) to ground semantic concepts in visual features, which is crucial for tasks like object detection, semantic segmentation, and understanding referring expressions (e.g., "the object to the left of the dog"). Traditional datasets often lack the necessary granularity for such tasks, making it challenging for models to accurately localize and… See the full description on the dataset page: https://huggingface.co/datasets/Salesforce/blip3-grounding-50m.blip3-grounding-smallblip_50k_train
Dataset Card for "blip_50k_train"
More Information needed
kream-product-blip-captions
KREAM Product Blip Captions Dataset Information
KREAM Product Blip Captions Dataset is a dataset card for finetuning a text-to-image generative model collected from KREAM, one of the best online-resell market in Korea.
This dataset consists of 'image' and 'text' key pairs.
The format of 'text' is 'category (e.g. outer), product original name (e.g. The North Face 1996 Eco Nuptse Jacket Black), blip captions (e.g. a photography of the north face black down jacket)'.
You can easily… See the full description on the dataset page: https://huggingface.co/datasets/hahminlew/kream-product-blip-captions.pokemon-blip-captions-en-zh
Dataset Card for Pokémon BLIP captions with English and Chinese.
Dataset used to train Pokémon text to image model, add a Chinese Column of Pokémon BLIP captions
BLIP generated captions for Pokémon images from Few Shot Pokémon dataset introduced by Towards Faster and Stabilized GAN Training for High-fidelity Few-shot Image Synthesis (FastGAN). Original images were obtained from FastGAN-pytorch and captioned with the pre-trained BLIP model.
For each row the dataset contains image… See the full description on the dataset page: https://huggingface.co/datasets/svjack/pokemon-blip-captions-en-zh.blip3o_aux_full_8640
BLIP3o auxiliary full protocol subset (8,640 rows)
This is the 10x full auxiliary view for the current UMM-SFT experiments.
The corresponding tiny view contains 864 rows.
Rows published: 8,640
Source pool: 51,758 verified BLIP3o auxiliary rows
Selection seed: 178430
Selection details: formal_selection_receipt.json
Features: image, prompt, source
“Full” means the 10x protocol subset, not the complete upstream corpus. The
source metadata did not declare a license; other is only… See the full description on the dataset page: https://huggingface.co/datasets/xzz789/blip3o_aux_full_8640.DataComp_large_pool_BLIP2_captions
Dataset Card for DataComp_large_pool_BLIP2_captions
Dataset Summary
Supported Tasks and Leaderboards
We have used this dataset for pre-training CLIP models and found that it rivals or outperforms models trained on raw web captions on average across the 38 evaluation tasks proposed by DataComp.
Refer to the DataComp leaderboard (https://www.datacomp.ai/leaderboard.html) for the top baselines uncovered in our work.
Languages
Primarily English.… See the full description on the dataset page: https://huggingface.co/datasets/thaottn/DataComp_large_pool_BLIP2_captions.blip3-grounding-1m-arabic
BLIP3-Grounding (Arabic) - 1M
Arabic version of the first 1,000,000 usable rows of
Salesforce/blip3-grounding-50m.
Two things differ from the source:
The images are here. The source ships a url column only; those URLs were
crawled and the original image bytes embedded, so the dataset is usable
without a crawl of your own.
Detection labels are translated. metadata_ar carries the Arabic label
for every bounding box. Everything else is untouched English.
Schema… See the full description on the dataset page: https://huggingface.co/datasets/oddadmix/blip3-grounding-1m-arabic.BLIP3o-Pretrain-Long-Caption-ParquetBLIP3o-Shortsimpsons-blip-captions
Dataset Card for "simpsons-blip-captions"
flowers-blip-captions
Dataset Card for "flowers-blip-captions"
More Information needed
comicstrips-gpt4o-blip3
Comic Strips
Dataset Details
Dataset Description
This dataset contains indie comics from Reddit, then captioned with GPT4o and BLIP3.
Currently, only the GPT4o captions are available in this repository. The BLIP3 captions will be uploaded soon.
Roughly 1400 images were captioned at a cost of ~$11 using GPT4o (25 May 2024 version).
Curated by: @pseudoterminalx
Funded by @pseudoterminalx
License: MIT
Dataset Sources
Unlike other free-to-use… See the full description on the dataset page: https://huggingface.co/datasets/bghira/comicstrips-gpt4o-blip3.blip3o-caption-mini-arrow
blip3o-caption-mini-arrow
blip3o-caption-mini-arrow is a high-quality, curated image-caption dataset derived and optimized from the original BLIP3o/BLIP3o-Pretrain-Long-Caption. This dataset is specifically filtered and processed for tasks involving long-form image captioning and vision-language understanding.
Overview
Total Samples: 91,600
Modality: Image ↔ Text
Format: Arrow (auto-converted to Parquet)
License: Apache 2.0
Language: English
Size: ~4.5 GB… See the full description on the dataset page: https://huggingface.co/datasets/prithivMLmods/blip3o-caption-mini-arrow.blip3-grounding-1m-denseBLIP3o-Visual-Reasoningblip3-grounding-1m
