MetaTransformer
Meta-Transformer for Unified Multimodal Learning
Install / Use
/learn @invictus717/MetaTransformerREADME
Meta-Transformer with Large Language Models ✨✨✨
We're thrilled to present 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.
🔥🔥 Project Website is at OneLLM.
🌟 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, 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 | ckpt | Meta-Transformer-L14 | LAION-2B | Large | 302M | ckpt | ckpt
</div> <!-- </details> --> <!-- <details> -->- Demo of Use for Pretrained Encoder
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 yiyuanzhang.ai@gmail.com ,kaixionggong@gmail.com, zhangkaipeng@pjlab.org.cn, or xyyue@ie.cuhk.edu.hk!
<br></br>
