#83 · Primary category： Foundation Models
Aria
Codebase for Aria - an Open Multimodal Native MoE
Project last updated：01/22/25
GitHub Stars
1.1K
Forks
89
Contributors
11
License
Apache-2.0
Why we included this project
Aria is worth a look if you want one open model that actually handles text, images, documents, and video instead of stitching together several specialized systems. It is a mixture-of-experts model with 25.3B total parameters but only 3.9B active per token, which keeps inference and fine-tuning noticeably cheaper than a dense model of similar size. The repo ships a working codebase with clear install steps and Hugging Face integration, so you can pull the weights, ask a chat-style question about an image, or fine-tune on your own data with a single A100-class GPU. With a 64K context window and strong results on video and document understanding, it is a practical base for real vision assistants or document analysis pipelines, and the Apache-2.0 license makes it easy to build on.
Articles for this project
No articles for this project yet.
To suggest a topic or contribute an article, contact us.
Related projects in this category
transformers
🤗 Transformers: the model-definition framework for state-of-the-art machine learning models in text, vision, audio, and multimodal models, for both inference and training.
CLIP
CLIP (Contrastive Language-Image Pretraining), Predict the most relevant text snippet given an image
MiniCPM-V
A Pocket-Sized MLLM for Ultra-Efficient Image and Video Understanding on Your Phone
generative-models
Generative Models by Stability AI
unilm
Large-scale Self-supervised Pre-training Across Tasks, Languages, and Modalities