MetaTransformer
<p align="center" width="100%"> <img src="assets\Meta-Transformer_banner.png" width="80%" height="80%"> </p> <div> <div align="center"> <a href='https://scholar.google.com/citations?user=KuYlJCIAAAAJ&hl=en' target='_blank'>Yiyuan Zhang<sup>1,2*</sup></a>  <a href='https://kxgong.github.io/' target='_blank'>Kaixiong Gong<sup>1,2*</sup></a>  <a href='http://kpzhang93.github.io/' target='_blank'>Kaipeng Zhang<sup>2,†</sup></a>  </br> <a href='http://www.ee.cuhk.edu.hk/~hsli/' target='_blank'>Hongsheng Li <sup>1,2</sup></a>  <a href='https://mmlab.siat.ac.cn/yuqiao/index.html' target='_blank'>Yu Qiao <sup>2</sup></a>  <a href='https://wlouyang.github.io/' target='_blank'>Wanli Ouyang<sup>2</sup></a>  <a href='http://people.eecs.berkeley.edu/~xyyue/' target='_blank'>Xiangyu Yue<sup>1,†,‡</sup></a> </div> <div> <div align="center"> <sup>1</sup> <a href='http://mmlab.ie.cuhk.edu.hk/' target='_blank'>Multimedia Lab, The Chinese University of Hong Kong</a>  </br> <sup>2</sup> <a href='https://github.com/OpenGVLab' target='_blank'>OpenGVLab，Shanghai AI Laboratory </a></br> <sup>*</sup> Equal Contribution  <sup>†</sup> Corresponding Author  <sup>‡</sup> Project Lead  </div> ----------------- [](https://arxiv.org/abs/2307.10802) [](https://kxgong.github.io/meta_transformer/) [](https://mp.weixin.qq.com/s/r38bzqdJxDZUvtDI0c9CEw) [](https://huggingface.co/papers/2307.10802) [](https://openxlab.org.cn/models/detail/zhangyiyuan/MetaTransformer)  <a href="https://twitter.com/_akhaliq/status/1682248055637041152"><img src="https://img.icons8.com/color/48/000000/twitter.png" width="25" height="25"></a> <a href="https://www.youtube.com/watch?v=V8L8xbsTyls&ab_channel=CSBoard"><img src="https://img.icons8.com/color/48/000000/youtube-play.png" width="25" height="25"></a> <a href='https://huggingface.co/kxgong/Meta-Transformer'> <img src="assets\icons\huggingface.png" width="25" height="25"> </a> <a href='https://open.spotify.com/episode/6JJxcy2zMtTwr4jXPQEXjh'> <img src="https://upload.wikimedia.org/wikipedia/commons/1/19/Spotify_logo_without_text.svg" width="20" height="20"></a> ## 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. <p align="center" width="100%"> <img src="assets\Meta-Transformer_application.png" width="100%" height="100%"> </p> **Table 1**: Meta-Transformer is capable of handling up to 12 modalities, including natural language <img src="assets\icons\text.jpg" width="15" height="15">, RGB images <img src="assets\icons\img.jpg" width="15" height="15">, point clouds <img src="assets\icons\pcd.jpg" width="15" height="15">, audios <img src="assets\icons\audio.jpg" width="15" height="15">, videos <img src="assets\icons\video.jpg" width="15" height="15">, tabular data <img src="assets\icons\table.jpg" width="15" height="15">, graph <img src="assets\icons\graph.jpg" width="15" height="15">, time series data <img src="assets\icons\time.jpg" width="15" height="15">, hyper-spectral images <img src="assets\icons\hyper.jpg" width="15" height="15">, IMU <img src="assets\icons\imu.jpg" width="15" height="15">, medical images <img src="assets\icons\xray.jpg" width="15" height="15">, and infrared images <img src="assets\icons\infrared.jpg" width="15" height="15">. <p align="left"> <img src="assets\Meta-Transformer_cmp.png" width=100%> </p> ## 🚩🚩🚩 Shared-Encoder, Unpaired Data, More Modalities <div> <img class="image" src="assets\Meta-Transformer_teaser.png" width="52%" height="100%"> <img class="image" src="assets\Meta-Transformer_exp.png" width="45.2%" height="100%"> </div> 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. <p align="left"> <img src="assets\Meta-Transformer_data2seq.png" width=100%> </p> 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. <p align="left"> <img src="assets\Meta-Transformer_framework.png" width=100%> </p> # 🌟 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 <!-- <details> --> <summary> Open-source Modality-Agnostic Models </summary> <br> <div> | 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) </div> <!-- </details> --> <!-- <details> --> * 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) ``` <!-- </details> --> # 🕙 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]`! <br></br>   # 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).