Everlyn-Labs/Everlyn-1
An open autoregressive model for generating images and videos with multimodal understanding capabilities.
Not currently ranked — collecting fresh signals.
star history
Everlyn-1 presents research on foundational video AI models with three main components: a distribution matching approach for video tokenization using Wasserstein distance, EfficientARV for efficient autoregressive image and video generation, and ANTRP for reducing hallucinations in multimodal large language models. The project explores image animation, inpainting, video prediction, and integration with MLLMs.
Frequently asked
- What is Everlyn-Labs/Everlyn-1?
- An open autoregressive model for generating images and videos with multimodal understanding capabilities.
- Is Everlyn-1 open source?
- Yes — Everlyn-Labs/Everlyn-1 is an open-source project tracked on heatdrop.
- How popular is Everlyn-1?
- Everlyn-Labs/Everlyn-1 has 2.9k stars on GitHub.
- Where can I find Everlyn-1?
- Everlyn-Labs/Everlyn-1 is on GitHub at https://github.com/Everlyn-Labs/Everlyn-1.