ViT-22B

Name: Vision Transformers-22B
Author: Google Research

基础大模型

Vision Transformers-22B

Release date: 2023-02-10更新于: 2023-03-13 11:21:30.267612

Live demo

Vision Transformers-22B is an AI model published by Google Research, released on 2023-02-10, for 基础大模型, with 220.0B parameters, and 2K tokens context length.

Data sourced primarily from official releases (GitHub, Hugging Face, papers), then benchmark leaderboards, then third-party evaluators. Learn about our data methodology

ViT-22B

Model basics

Reasoning traces

Not supported

Thinking modes

Thinking modes not supported

Context length

2K tokens

Max output length

No data

Model type

基础大模型

Release date

2023-02-10

Model file size

No data

MoE architecture

Total params / Active params

220.0B / N/A

Knowledge cutoff

No data

ViT-22B

Open source & experience

Code license

No data

Weights license

No data

GitHub repo

GitHub link unavailable

Hugging Face

Hugging Face link unavailable

Live demo

No live demo

Vision Transformers-22B

Model Overview

ViT-22B是谷歌最新发布的220亿参数的视觉大模型。此前，虽然语言模型的参数一直增长，但是在视觉领域，模型的参数只有40亿。而往上的规模训练会出现一些不稳定情况。因此谷歌精心设计了ViT模型，使其有很高的并行训练效率。

ViT-22B

Model basics

Open source & experience

Official resources

API details

Benchmark Results

Publisher

Model Overview

DataLearner 官方微信