Cambrian-10M: Multimodal Instruction Tuning Data for Vision-Language Models

A dataset designed for instruction tuning in multimodal settings involving visual interaction data. It was created by nyu-visionx and released in 2024 to address the scarcity of high-quality multimodal instruction-tuning data. The dataset aims to maintain the language abilities of multimodal large language models.

Use Cases

• Instruction tuning of multimodal LLMs based on visual interaction data mentioned in the description
• Training models for visual question answering tasks based on the dataset's stated purpose
• Maintaining language capabilities in vision-language models based on the dataset's design goal

Strengths

• Designed specifically for multimodal instruction tuning, a stated research need
• Associated with a published research paper (arXiv:2406.16860) and project website

Limitations

• Description metadata is limited; actual data quality requires manual inspection after download
• Column-level documentation is absent; field semantics must be inferred after download
• Row count is unknown, which may limit suitability assessment

Provenance

Source

nyu-visionx

Freshness

Last updated 2024-07-08 04:34:51; freshness should be verified