18 datasets found

This paper proposes a pre-training task for embedding-based large-scale retrieval.

Large-scale pre-training has recently revolutionized vision-and-language (VL) research. Models such as LXMERT and UNITER have achieved state-of-the-art performance across a wide...

Satlas is a large-scale pre-training dataset designed for tasks involving the analysis of satellite images.

Large-scale image-text dataset for pre-training a collaborative two-stream vision-language model for cross-modal retrieval.

Large-scale image-text datasets for pre-training a collaborative two-stream vision-language model for cross-modal retrieval.

A vision-language pre-training dataset, Chinese CLIP, which consists of 100 million image-text pairs.

A retrieval-augmented pretrain-and-finetune paradigm for biomedical VQA which includes a high-quality image-text pairs PMCPM, a pre-trained multi-modal model, and a novel...

The dataset used for pre-training the MS-CLIP model, which consists of 20 million image-text pairs filtered from Laion-400M.

The dataset used in the paper is also DataComp-1B, which is a large-scale dataset for training next-generation image-text models.

The dataset used in the paper is LAION-400M and LAION-5B, which are large-scale datasets for training next-generation image-text models.

A vision-language pre-training dataset, BLIP2, which consists of 100 million image-text pairs.

The dataset is used for instruction-tuning of LLMs in multiple languages using reinforcement learning from human feedback.

The dataset used in this paper is a self-supervised audio pre-training framework called HuBERT.

This paper proposes BERT, a pre-trained deep bidirectional transformer for language understanding.

The datasets used in the paper are described as follows: WikiMIA, BookMIA, Temporal Wiki, Temporal arXiv, ArXiv-1 month, Multi-Webdata, LAION-MI, Gutenberg.

The dataset used in the paper is a large corpus of real-world programs for pre-training a neural network model to learn better code representation.

A unified language model pre-training for natural language understanding and generation.

The CLIP model and its variants are becoming the de facto backbone in many applications. However, training a CLIP model from hundreds of millions of image-text pairs can be...