cambrian-mllm/cambrian

Health v2 · maintenance only

Flags: no_releases

Adoption not part of the score

2013 stars · 140 forks observed · 2026-08-28

What it is AI-extracted, prompt v1, taxonomy v1, 2026-08-30, confidence not recorded

Cambrian-1 is a fully open family of vision-centric multimodal large language models (MLLMs) from NYU's VISIONx group, with training and evaluation toolkits. It includes released model weights, instruction-tuning datasets (Cambrian 10M), and a 26-benchmark MLLM evaluation suite.

Use cases

• train a multimodal LLM with vision-centric encoders
• fine-tune an MLLM on image instruction-tuning data
• evaluate multimodal LLMs across 26 benchmarks
• run image question answering with an open vision-language model
• study how vision representations like CLIP and DINO affect MLLM performance
• download open multimodal instruction-tuning datasets

When to choose

• you need fully open weights, data, and code for multimodal LLM research
• you want to experiment with vision encoder combinations for MLLMs
• you need a reproducible MLLM benchmarking suite on HPC clusters

When to avoid

• you need a production-ready multimodal API or hosted service
• you only need text-only LLMs
• you need video-native multimodal models rather than image-based ones

Facets

library · maturity active

machine-learning deep-learning llm-training chatbot image-processing large-language-models computer-vision machine-learning artificial-intelligence python multimodal-llm vision-language-model instruction-tuning clip dino mllm research model-training evaluation-suite gpu linux

Member repositories

For agents

markdown · JSON · MCP: product_card(name="cambrian-mllm/cambrian")