Ecosyste.ms: Awesome

An open API service indexing awesome lists of open source software.

Awesome Lists | Featured Topics | Projects

https://github.com/invictus717/MetaTransformer

Meta-Transformer for Unified Multimodal Learning
https://github.com/invictus717/MetaTransformer

artificial-intelligence computer-vision foundationmodel machine-learning multimedia multimodal transformers

Last synced: about 2 months ago
JSON representation

Meta-Transformer for Unified Multimodal Learning

Host: GitHub
URL: https://github.com/invictus717/MetaTransformer
Owner: invictus717
License: apache-2.0
Created: 2023-07-08T12:40:54.000Z (about 1 year ago)
Default Branch: master
Last Pushed: 2023-12-05T07:36:11.000Z (10 months ago)
Last Synced: 2024-06-19T03:11:32.090Z (3 months ago)
Topics: artificial-intelligence, computer-vision, foundationmodel, machine-learning, multimedia, multimodal, transformers
Language: Python
Homepage: https://arxiv.org/abs/2307.10802
Size: 21.7 MB
Stars: 1,459
Watchers: 22
Forks: 113
Open Issues: 2
Metadata Files:
- Readme: README.md
- License: LICENSE

Awesome Lists containing this project

awesome-foundation-model - MetaTransformer
StarryDivineSky - invictus717/MetaTransformer - Transformer 框架与多模态大型语言模型相结合，该模型执行多模态联合训练，支持更多模态，包括 fMRI、深度图和法线图，并在 25 个基准测试中展示了非常令人印象深刻的性能。作为基础模型，Meta-Transformer 可以处理来自 12 种模态的数据，这决定了它可以支持广泛的应用程序。如图所示，Meta-Transformer可以为下游任务提供服务，包括股票分析📈、天气预报❄️ ⛄ ☁️ ☔ ☀️ ⚡、遥感📡、自动驾驶🚗、社交网络🌍、语音识别🔉等。表 1：Meta-Transformer 能够处理多达 12 种模态，包括自然语言、RGB 图像、点云、音频、视频、表格数据、图形、时间序列数据、高光谱图像、IMU 、医学图像和红外图像。此存储库旨在探索 transformer 在多模态学习中的潜力和可扩展性。我们利用 Transformer 的优势来处理长度变化序列。然后，我们按照元方案提出数据到序列的标记化，然后将其应用于 12 种模态，包括文本、图像、点云、音频、视频、红外、超光谱、X 射线、表格、图形、时间序列和惯性测量单元（IMU）数据。在获得令牌序列后，我们采用模态共享编码器来提取不同模态的表示。借助特定于任务的磁头，Meta-Transformer 可以处理不同模态的各种任务，例如：分类、检测和分割。 (多模态大模型 / 网络服务_其他)

README

        










    Yiyuan Zhang^1,2* 

    Kaixiong Gong^1,2* 

    Kaipeng Zhang^2,† 

    

    Hongsheng Li ^1,2 

    Yu Qiao ² 

    Wanli Ouyang² 

    Xiangyu Yue^1,†,‡







    ¹

    Multimedia Lab, The Chinese University of Hong Kong 

    

    ² OpenGVLab，Shanghai AI Laboratory 

    

    ^* Equal Contribution 

    ^† Corresponding Author 

    ^‡ Project Lead 



-----------------

[![arXiv](https://img.shields.io/badge/arxiv-2307.10802-b31b1b?style=plastic&color=b31b1b&link=https%3A%2F%2Farxiv.org%2Fabs%2F2307.10802)](https://arxiv.org/abs/2307.10802)

[![website](https://img.shields.io/badge/Project-Website-brightgreen)](https://kxgong.github.io/meta_transformer/)

[![blog-cn](https://img.shields.io/badge/%E6%9C%BA%E5%99%A8%E4%B9%8B%E5%BF%83-%E7%AE%80%E4%BB%8B-brightgreen)](https://mp.weixin.qq.com/s/r38bzqdJxDZUvtDI0c9CEw)

[![Hugging Face Spaces](https://img.shields.io/badge/%F0%9F%A4%97%20Hugging%20Face-Space-blue)](https://huggingface.co/papers/2307.10802)

[![OpenXLab](https://cdn-static.openxlab.org.cn/header/openxlab_models.svg)](https://openxlab.org.cn/models/detail/zhangyiyuan/MetaTransformer)

![](https://img.shields.io/github/stars/invictus717/MetaTransformer?style=social)



     

## Meta-Transformer with Large Language Models ✨✨✨

We're thrilled to present [OneLLM](https://github.com/csuhan/OneLLM), ensembling Meta-Transformer framework with Multimodal Large Language Models, which performs multimodal joint training🚀, supports more modalities including fMRI, Depth and Normal Maps 🚀, and demonstrates very impressive performances on **25** benchmarks🚀🚀🚀. 

🔥🔥 The code, pretrained models, and datasets are publicly available at [OneLLM](https://github.com/csuhan/OneLLM).

🔥🔥 Project Website is at [OneLLM](https://onellm.csuhan.com/).

### 🌟 Single Foundation Model Supports A Wide Range of Applications

As a foundation model, Meta-Transformer can handle data from 12 modalities, which determines that it can support a wide range of applications. As shown in this figure, Meta-Transformer can provide services for downstream tasks including stock analysis 📈, weather forecasting ☀️ ☔ ☁️ ❄️ ⛄ ⚡, remote sensing 📡, autonomous driving 🚗, social network 🌍, speech recognition 🔉, etc.







**Table 1**: Meta-Transformer is capable of handling up to 12 modalities, including natural language , RGB images , point clouds , audios , videos , tabular data , graph , time series data , hyper-spectral images , IMU , medical images , and infrared images .







## 🚩🚩🚩 Shared-Encoder, Unpaired Data, More Modalities 



  

  



This repository is built to explore the potential and extensibility of transformers for multimodal learning. We utilize the advantages of Transformers to deal with length-variant sequences. Then we propose the *Data-to-Sequence* tokenization following a meta-scheme, then we apply it to 12 modalities including text, image, point cloud, audio, video, infrared, hyper-spectral, X-Ray, tabular, graph, time-series, and Inertial Measurement Unit (IMU) data.







After obtaining the token sequence, we employ a modality-shared encoder to extract representation across different modalities. With task-specific heads, Meta-Transformer can handle various tasks on the different modalities, such as: classification, detection, and segmentation.







# 🌟 News

* **2023.8.17:** Release code to directly get embeddings from multiple modalities. We will further release code on utilizing Meta-Transformer for Human-Centric vision tasks.

* **2023.8.2:** 🎉🎉🎉 The implementation of Meta-Transformer for image, point cloud, graph, tabular, time-series, X-Ray, hyper-spectrum, LiDAR data has been released. We also release a very powerful foundation model for Autonomous Driving 🚀🚀🚀.  

* **2023.7.22:** Pretrained weights and a usage demo for our Meta-Transformer have been released. Comprehensive documentation and implementation of the image modality are underway and will be released soon. Stay tuned for more exciting updates!⌛⌛⌛

* **2023.7.21:** Paper is released at [arxiv](https://arxiv.org/abs/2307.10802), and code will be gradually released.

* **2023.7.8:** Github Repository Initialization.

# 🔓 Model Zoo

 Open-source Modality-Agnostic Models 






|      Model      |   Pretraining   | Scale | #Param |                                               Download | 国内下载源                                               |

| :------------: | :----------: | :----------------------: | :----: | :---------------------------------------------------------------------------------------------------: | :--------: | 

| Meta-Transformer-B16  | LAION-2B |         Base          |  85M  |   [ckpt](https://drive.google.com/file/d/19ahcN2QKknkir_bayhTW5rucuAiX0OXq/view?usp=sharing)    | [ckpt](https://download.openxlab.org.cn/models/zhangyiyuan/MetaTransformer/weight//Meta-Transformer_base_patch16_encoder)

| Meta-Transformer-L14  | LAION-2B |         Large          |  302M  |   [ckpt](https://drive.google.com/file/d/15EtzCBAQSqmelhdLz6k880A19_RpcX9B/view?usp=drive_link)   | [ckpt](https://download.openxlab.org.cn/models/zhangyiyuan/MetaTransformer/weight//Meta-Transformer_large_patch14_encoder)



* Demo of Use for Pretrained Encoder

```python

import torch 

import torch.nn as nn

from timm.models.vision_transformer import Block

from Data2Seq import Data2Seq

video_tokenier = Data2Seq(modality='video',dim=768)

audio_tokenier = Data2Seq(modality='audio',dim=768)

time_series_tokenier = Data2Seq(modality='time-series',dim=768)

features = torch.concat([video_tokenizer(video), audio_tokenizer(audio), time_series_tokenizer(time_data)],dim=1)

# For base-scale encoder:

ckpt = torch.load("Meta-Transformer_base_patch16_encoder.pth")

encoder = nn.Sequential(*[

            Block(

                dim=768,

                num_heads=12,

                mlp_ratio=4.,

                qkv_bias=True,

                norm_layer=nn.LayerNorm,

                act_layer=nn.GELU

            )

            for i in range(12)])

encoder.load_state_dict(ckpt,strict=True)

# For large-scale encoder:

ckpt = torch.load("Meta-Transformer_large_patch14_encoder.pth")

encoder = nn.Sequential(*[

            Block(

                dim=1024,

                num_heads=16,

                mlp_ratio=4.,

                qkv_bias=True,

                norm_layer=nn.LayerNorm,

                act_layer=nn.GELU

            )

            for i in range(24)])

encoder.load_state_dict(ckpt,strict=True)

encoded_features = encoder(features)

```

# 🕙 ToDo

- [ x ] Meta-Transformer with Large Language Models.

- [ x ] Multimodal Joint Training with Meta-Transformer.

- [ x ] Support More Modalities and More Tasks.

# Contact

🚀🚀🚀 We aspire to shape this repository into **a formidable foundation for mainstream AI perception tasks across diverse modalities**. Your contributions can play a significant role in this endeavor, and we warmly welcome your participation in our project!

To contact us, never hestitate to send an email to `[email protected]` ,`[email protected]`, `[email protected]`, or `[email protected]`!




 

# Citation

If the code and paper help your research, please kindly cite:

```

@article{zhang2023meta,

  title={Meta-transformer: A unified framework for multimodal learning},

  author={Zhang, Yiyuan and Gong, Kaixiong and Zhang, Kaipeng and Li, Hongsheng and Qiao, Yu and Ouyang, Wanli and Yue, Xiangyu},

  journal={arXiv preprint arXiv:2307.10802},

  year={2023}

}

```

# License

This project is released under the [Apache 2.0 license](LICENSE).

# Acknowledgement

This code is developed based on excellent open-sourced projects including [MMClassification](https://github.com/open-mmlab/mmpretrain/tree/mmcls-1.x), [MMDetection](https://github.com/open-mmlab/mmdetection), [MMsegmentation](https://github.com/open-mmlab/mmsegmentation), [OpenPoints](https://github.com/guochengqian/openpoints), [Time-Series-Library](https://github.com/thuml/Time-Series-Library), [Graphomer](https://github.com/microsoft/Graphormer), [SpectralFormer](https://github.com/danfenghong/IEEE_TGRS_SpectralFormer), and [ViT-Adapter](https://github.com/czczup/ViT-Adapter).