# Alibaba — OFA: Towards Building a One-For-All Model

- Company: Alibaba (alibaba.com)
- Announced: 2022-11-14T08:01:41+00:00
- Category: new-model
- Coverage: not counted
- Announcement: yes
- Group: models
- Source: https://qwenlm.github.io/blog/ofa/
- Record: https://forck.live/items/2322-ofa-towards-building-a-one-for-all-model
- Subject: Qwen
- Models affected: OFA

Alibaba proposes OFA, a unified multimodal pretrained model that handles both understanding and generation tasks across modalities, trained with instruction-based multitask pretraining.

## Evidence

Verbatim from https://qwenlm.github.io/blog/ofa/:

> we propose OFA1, namely One-For-All, a unified multimodal pretrained model that unifies understanding and generation tasks concerning modalities into a single framework

---

Record: https://forck.live/items/2322-ofa-towards-building-a-one-for-all-model
Catalogue: https://forck.live/llms.txt
Current issue: https://forck.live/feed.md