# UniVL Projects at AI Tinkerers

> Canonical HTML: https://aitinkerers.org/technologies/univl
> Markdown URL: https://aitinkerers.org/technologies/univl.md
> Technology record last updated: 2026-05-03T02:02:45Z
> Generated: 2026-09-23T07:37:50Z

UniVL is a unified video-and-language pre-training model designed to handle both multimodal understanding and generation tasks within a single framework.

Developed by researchers at Microsoft, UniVL addresses the historical gap between video-text understanding (like retrieval) and generation (like captioning). The model employs a Transformer-based backbone with four core components: two single-modal encoders, a cross-encoder, and a decoder. By pre-training on the massive HowTo100M dataset using five distinct objectives—including video-text alignment and language reconstruction—UniVL achieves state-of-the-art results across five major downstream tasks. Its flexible architecture allows it to adapt to diverse multimodal requirements, making it a highly efficient tool for developers working on complex video-to-text applications.

- Official technology site: https://github.com/microsoft/UniVL
- Public AI Tinkerers demos and talks: 1
- Result page: 1 of 1

## Recent Public Talks and Demos

### [Chatting with your videos and much more!](https://nyc.aitinkerers.org/talks/rsvp_NTIE8_HYjOU)

We recently launched our Generate API at a hackathon over the weekend which allows you to generate text about your videos by prompting. Ask the model to write reports, get summaries, and come up with chapters - whatever you need. Here is a link to our blog post about the hackathon: https://app.twelvelabs.io/blog/introducing-the-multimodal-ai-23labs-hackathon

- Event context: AI Tinkerers NYC - October Meetup — 2023-10-26 — New York City
- Public talk page: https://nyc.aitinkerers.org/talks/rsvp_NTIE8_HYjOU

## Related Technologies

- [Generate API](https://aitinkerers.org/technologies/generate-api) ([Markdown](https://aitinkerers.org/technologies/generate-api.md)) — 1 public demo
- [Imagen Video](https://aitinkerers.org/technologies/imagen-video) ([Markdown](https://aitinkerers.org/technologies/imagen-video.md)) — 2 public demos
- [Make-A-Video](https://aitinkerers.org/technologies/make-a-video) ([Markdown](https://aitinkerers.org/technologies/make-a-video.md)) — 2 public demos
- [Oscar](https://aitinkerers.org/technologies/oscar) ([Markdown](https://aitinkerers.org/technologies/oscar.md)) — 1 public demo
- [Runway Gen-2](https://aitinkerers.org/technologies/runway-gen-2) ([Markdown](https://aitinkerers.org/technologies/runway-gen-2.md)) — 2 public demos
- [Search API](https://aitinkerers.org/technologies/search-api) ([Markdown](https://aitinkerers.org/technologies/search-api.md)) — 2 public demos
- [VideoBERT](https://aitinkerers.org/technologies/videobert) ([Markdown](https://aitinkerers.org/technologies/videobert.md)) — 1 public demo
- [VideoCLIP](https://aitinkerers.org/technologies/videoclip) ([Markdown](https://aitinkerers.org/technologies/videoclip.md)) — 1 public demo
- [Video embeddings](https://aitinkerers.org/technologies/video-embeddings) ([Markdown](https://aitinkerers.org/technologies/video-embeddings.md)) — 2 public demos