Wukong100M: 100 Million Chinese Image-Text Pairs

100 million Chinese image-text pairs form a subset of the Noah-Wukong multimodal dataset. The dataset was uploaded by author 'wanng' to Hugging Face and last updated on December 11, 2022. The text metadata for these pairs occupies approximately 16GB of space.

Use Cases

• Training Chinese vision-language models based on the 100 million image-text pairs.
• Fine-tuning image captioning systems for Chinese text generation.
• Conducting research on cross-modal retrieval between Chinese text and images.
• Pre-training foundation models for downstream Chinese multimodal applications.

Strengths

• Contains approximately 100 million data points, providing a large-scale resource.
• Focuses on Chinese language content, which may be less common in other large multimodal collections.
• Text metadata for the pairs is a known size of around 16GB.

Limitations

• Description metadata is limited; actual data quality requires manual inspection after download.
• Last updated 2022-12-11 06:24:05; freshness should be verified.
• Column-level documentation is absent; field semantics must be inferred after download.

Provenance

Source

Noah-Wukong multimodal dataset

Collection Method

Subset extraction focusing on the Chinese-language portion.

Time Range

null

Freshness

2022-12-11 06:24:05

Geography

null