Code and pre-trained models for MultiMAE: Multi-modal Multi-task Masked Autoencoders

Visual Intelligence & Learning Lab, Swiss Federal Institute of Technology (EPFL)

Last update: Jan 6, 2023

Related tags

Deep Learning MultiMAE

Overview

MultiMAE: Multi-modal Multi-task Masked Autoencoders

Roman Bachmann*, David Mizrahi*, Andrei Atanov, Amir Zamir

Website | arXiv | BibTeX

Official PyTorch implementation and pre-trained models for MultiMAE: Multi-modal Multi-task Masked Autoencoders.

We introduce Multi-modal Multi-task Masked Autoencoders (MultiMAE), an efficient and effective pre-training strategy for Vision Transformers. Given a small random sample of visible patches from multiple modalities, the MultiMAE pre-training objective is to reconstruct the masked-out regions. Once pre-trained, a single MultiMAE encoder can then be used for both single-modal and multi-modal downstream transfer, yielding competitive to or significantly better results than the baselines.

Catalog

Pre-trained models
MultiMAE pre-training code
ImageNet-1K classification fine-tuning code
Semantic segmentation fine-tuning code (single-modal & multi-modal)
Depth estimation fine-tuning code
Taskonomy fine-tuning code
Colab & Hugging Face demos

Pre-trained models

We provide the weights of our pre-trained MultiMAE ViT-B model, in MultiViT (multi-modal) format and timm (RGB-only) format.

For comparison, we also provide the weights of a MAE ViT-B model that we pre-trained using the official MAE codebase following the recommended settings.

Method	Arch.	Pre-training modalities	Pre-training epochs	Weights (MultiViT)	Weights (timm)	Config
MAE	ViT-B	RGB	1600	download	download	See MAE
MultiMAE	ViT-B	RGB+D+S	1600	download	download	link

These pre-trained models can then be fine-tuned using this codebase to reach the following performance:

Method	Classif. (@1)	Semantic Segmentation (mIoU)							Depth (δ1)
	ImageNet-1K (RGB)	ADE20K (RGB)	Hypersim (RGB / D / RGB + D)			NYUv2 (RGB / D / RGB + D)			NYUv2 (RGB)
Sup. (DeiT)	81.8	45.8	33.9	-	-	50.1	-	-	80.7
MAE	83.3	46.2	36.5	-	-	50.8	-	-	85.1
MultiMAE	83.3	46.2	37.0	38.5	47.6	52.0	41.4	56.0	86.4

Model formats

We provide pre-trained weights in two different formats: the single-modal ViT / timm format, which is compatible with other popular ViT repositories (e.g., timm, DINO, MAE), and the multi-modal MultiMAE / MultiViT format, which is used throughout this codebase for multi-modal pre-training and fine-tuning. See multimae/multimae.py for the documentation and implementation of MultiMAE / MultiViT.

You can convert between these formats using the provided vit2multimae_converter.py and multimae2vit_converter.py scripts.

Usage

Set-up

See SETUP.md for set-up instructions.

Pre-training

See PRETRAINING.md for pre-training instructions.

Fine-tuning

See FINETUNING.md for fine-tuning instructions.

Demo & visualizations

For interactive demos, please see our website. Open our Colab notebook to play around with the visualization code, or simply upload an image to our Hugging Face Spaces demo.

Acknowledgement

This repository is built using the timm, DeiT, DINO, MoCo v3, BEiT, MAE-priv, and MAE repositories.

License

This project is under the CC-BY-NC 4.0 license. See LICENSE for details.

Citation

If you find this repository helpful, please consider citing our work:

@article{bachmann2022multimae,
  author    = {Roman Bachmann and David Mizrahi and Andrei Atanov and Amir Zamir},
  title     = {{MultiMAE}: Multi-modal Multi-task Masked Autoencoders},
  journal   = {arXiv preprint arXiv:2204.01678},
  year      = {2022},
}

Comments

ADE20K dataset structure for semantic segmentation

Hi,

First of all, thanks for your amazing work!

We're trying to reproduce the paper results and stumbled over how to set up semantic segmentation finetuning with ADE20K. To our surprise, the data loader seems to expect the same root/task_a/class_x/xxx.ext folder hierarchy known from classification. However, as the images naturally contain more than a single semantic class, we're not sure how the images are supposed to be arranged.

Could you hence give us a hint on how the data should be structured to work with the provided ft_ade_64e_multimae-b_rgb.yaml configuration?

Thank you, Paul

opened by pkwagner 6
The reason why depth map should be divided 2**16 ?

Thank you for your great Multi-MAE, We observed that in https://github.com/EPFL-VILAB/MultiMAE/blob/main/utils/datasets.py line 96, you use img = torch.Tensor(np.array(task_dict[task]) / 2 ** 16). Can you tell me the reason why depth map should be divided 2**16 ? Is there any problems without this operation?

opened by yingfei1016 6
add web demo/model to Huggingface

Hi, would you be interested in adding MultiMAE to Hugging Face? The Hub offers free hosting, and it would make your work more accessible and visible to the rest of the ML community. Models/datasets/spaces(web demos) can be added to a user account or organization similar to github.

Example from other organizations: Keras: https://huggingface.co/keras-io Microsoft: https://huggingface.co/microsoft Facebook: https://huggingface.co/facebook

Example spaces with repos: github: https://github.com/salesforce/BLIP Spaces: https://huggingface.co/spaces/salesforce/BLIP

github: https://github.com/facebookresearch/omnivore Spaces: https://huggingface.co/spaces/akhaliq/omnivore

and here are guides for adding spaces/models/datasets to your org

How to add a Space: https://huggingface.co/blog/gradio-spaces how to add models: https://huggingface.co/docs/hub/adding-a-model uploading a dataset: https://huggingface.co/docs/datasets/upload_dataset.html

Please let us know if you would be interested and if you have any questions, we can also help with the technical implementation.

opened by AK391 4
Linear probing results

Hey, Thank you for providing the code for the paper. The paper is really interesting and the project page is very well done!

I was wondering whether you've tested the performance of linear probing on the RGB image when trained with all 3 modalities. The results of the original MAE paper were not very good, it is interesting to understand if the additional supervision creates better representations that translate into better linear probing scores.

Thanks, Eliahu

opened by eliahuhorwitz 4

Query about semseg domain in pre-training

Hi, I have successful made the pesudo labels and trained ‘rgb’ in/out-domain multimae model.

But when I trained model with 'rgb-semseg' in/out-domain, I met an error in multimae/input_adapters.py line 232

# Create patches [B, C, H, W] -> [B, (H*W), C]
x_patch = rearrange(self.proj(x), 'b d nh nw -> b (nh nw) d')

The full log is log.txt. x.size() is [batchsize, 64, 56, 56] before line 232. I can't find out what's wrong.

What's more, I don't know why the pseudo semeg label image resize into 1/4 (that is 224*224->56*56)in utils/datasets.py line 105

# Convert to Tensor
for task in task_dict:
    if task in ['depth']:
        img = torch.Tensor(np.array(task_dict[task]) / 2 ** 16)
        img = img.unsqueeze(0)  # 1 x H x W
    elif task in ['rgb']:
        img = TF.to_tensor(task_dict[task])
        img = TF.normalize(img, mean=self.rgb_mean, std=self.rgb_std)
    elif task in ['semseg', 'semseg_coco']:
        # TODO: add this to a config instead
        # Rescale to 0.25x size (stride 4)
        scale_factor = 0.25
        img = task_dict[task].resize((int(self.input_size * scale_factor), int(self.input_size * scale_factor)))
        # Using pil_to_tensor keeps it in uint8, to_tensor converts it to float (rescaled to [0, 1])
        img = TF.pil_to_tensor(img).to(torch.long).squeeze(0)

and then use nn.Conv2d in multimae/input_adapters.py line 198

if self.interpolate_class_emb:
    self.proj = nn.Sequential(
        nn.Upsample(scale_factor=(1 / self.P_H, 1 / self.P_W),
                    mode='bilinear'),  # Actually a downsample operation
        nn.Conv2d(in_channels=self.dim_class_emb, out_channels=self.dim_tokens,
                    kernel_size=1, stride=1),
    )
else:
    self.proj = nn.Conv2d(
        in_channels=self.dim_class_emb, out_channels=self.dim_tokens,
        kernel_size=(self.P_H, self.P_W), stride=(self.P_H, self.P_W)
    )
)

Thank you for any help.

opened by Chianghui-Wong 3

Problem during evaluate the pretrained model

Problem

Hi! I encountered some problem while just try to evaluate this model with same config as Demo on Colab.

Environment

Ubuntu 22.04 CUDA Kernel 10.1 CUDA Runtime 11.3 Pytorch 1.12.0

Terminal

/opt/conda/conda-bld/pytorch_1656352645774/work/aten/src/ATen/native/cuda/ScatterGatherKernel.cu:144: operator(): block: [6825,0,0], thread: [0,0,0] Assertionidx_dim >= 0 && idx_dim < index_size && "index out of bounds"failed. /opt/conda/conda-bld/pytorch_1656352645774/work/aten/src/ATen/native/cuda/ScatterGatherKernel.cu:144: operator(): block: [6825,0,0], thread: [1,0,0] Assertionidx_dim >= 0 && idx_dim < index_size && "index out of bounds"failed. /opt/conda/conda-bld/pytorch_1656352645774/work/aten/src/ATen/native/cuda/ScatterGatherKernel.cu:144: operator(): block: [6825,0,0], thread: [2,0,0] Assertionidx_dim >= 0 && idx_dim < index_size && "index out of bounds"failed. /opt/conda/conda-bld/pytorch_1656352645774/work/aten/src/ATen/native/cuda/ScatterGatherKernel.cu:144: operator(): block: [6825,0,0], thread: [3,0,0] Assertionidx_dim >= 0 && idx_dim < index_size && "index out of bounds"failed. /opt/conda/conda-bld/pytorch_1656352645774/work/aten/src/ATen/native/cuda/ScatterGatherKernel.cu:144: operator(): block: [6825,0,0], thread: [4,0,0] Assertionidx_dim >= 0 && idx_dim < index_size && "index out of bounds"failed. /opt/conda/conda-bld/pytorch_1656352645774/work/aten/src/ATen/native/cuda/ScatterGatherKernel.cu:144: operator(): block: [6825,0,0], thread: [5,0,0] Assertionidx_dim >= 0 && idx_dim < index_size && "index out of bounds"failed. /opt/conda/conda-bld/pytorch_1656352645774/work/aten/src/ATen/native/cuda/ScatterGatherKernel.cu:144: operator(): block: [6825,0,0], thread: [6,0,0] Assertionidx_dim >= 0 && idx_dim < index_size && "index out of bounds"failed. /opt/conda/conda-bld/pytorch_1656352645774/work/aten/src/ATen/native/cuda/ScatterGatherKernel.cu:144: operator(): block: [6825,0,0], thread: [7,0,0] Assertionidx_dim >= 0 && idx_dim < index_size && "index out of bounds"failed. /opt/conda/conda-bld/pytorch_1656352645774/work/aten/src/ATen/native/cuda/ScatterGatherKernel.cu:144: operator(): block: [6825,0,0], thread: [8,0,0] Assertionidx_dim >= 0 && idx_dim < index_size && "index out of bounds"failed. /opt/conda/conda-bld/pytorch_1656352645774/work/aten/src/ATen/native/cuda/ScatterGatherKernel.cu:144: operator(): block: [6825,0,0], thread: [9,0,0] Assertionidx_dim >= 0 && idx_dim < index_size && "index out of bounds"failed. /opt/conda/conda-bld/pytorch_1656352645774/work/aten/src/ATen/native/cuda/ScatterGatherKernel.cu:144: operator(): block: [6825,0,0], thread: [10,0,0] Assertionidx_dim >= 0 && idx_dim < index_size && "index out of bounds"failed. /opt/conda/conda-bld/pytorch_1656352645774/work/aten/src/ATen/native/cuda/ScatterGatherKernel.cu:144: operator(): block: [6825,0,0], thread: [11,0,0] Assertionidx_dim >= 0 && idx_dim < index_size && "index out of bounds"failed. /opt/conda/conda-bld/pytorch_1656352645774/work/aten/src/ATen/native/cuda/ScatterGatherKernel.cu:144: operator(): block: [6825,0,0], thread: [12,0,0] Assertionidx_dim >= 0 && idx_dim < index_size && "index out of bounds"failed. /opt/conda/conda-bld/pytorch_1656352645774/work/aten/src/ATen/native/cuda/ScatterGatherKernel.cu:144: operator(): block: [6825,0,0], thread: [13,0,0] Assertionidx_dim >= 0 && idx_dim < index_size && "index out of bounds"failed. /opt/conda/conda-bld/pytorch_1656352645774/work/aten/src/ATen/native/cuda/ScatterGatherKernel.cu:144: operator(): block: [6825,0,0], thread: [14,0,0] Assertionidx_dim >= 0 && idx_dim < index_size && "index out of bounds"failed. /opt/conda/conda-bld/pytorch_1656352645774/work/aten/src/ATen/native/cuda/ScatterGatherKernel.cu:144: operator(): block: [6825,0,0], thread: [15,0,0] Assertionidx_dim >= 0 && idx_dim < index_size && "index out of bounds"failed. /opt/conda/conda-bld/pytorch_1656352645774/work/aten/src/ATen/native/cuda/ScatterGatherKernel.cu:144: operator(): block: [6825,0,0], thread: [16,0,0] Assertionidx_dim >= 0 && idx_dim < index_size && "index out of bounds"failed. /opt/conda/conda-bld/pytorch_1656352645774/work/aten/src/ATen/native/cuda/ScatterGatherKernel.cu:144: operator(): block: [6825,0,0], thread: [17,0,0] Assertionidx_dim >= 0 && idx_dim < index_size && "index out of bounds"failed. /opt/conda/conda-bld/pytorch_1656352645774/work/aten/src/ATen/native/cuda/ScatterGatherKernel.cu:144: operator(): block: [6825,0,0], thread: [18,0,0] Assertionidx_dim >= 0 && idx_dim < index_size && "index out of bounds"failed. /opt/conda/conda-bld/pytorch_1656352645774/work/aten/src/ATen/native/cuda/ScatterGatherKernel.cu:144: operator(): block: [6825,0,0], thread: [19,0,0] Assertionidx_dim >= 0 && idx_dim < index_size && "index out of bounds"failed. /opt/conda/conda-bld/pytorch_1656352645774/work/aten/src/ATen/native/cuda/ScatterGatherKernel.cu:144: operator(): block: [6825,0,0], thread: [20,0,0] Assertionidx_dim >= 0 && idx_dim < index_size && "index out of bounds"failed. /opt/conda/conda-bld/pytorch_1656352645774/work/aten/src/ATen/native/cuda/ScatterGatherKernel.cu:144: operator(): block: [6825,0,0], thread: [21,0,0] Assertionidx_dim >= 0 && idx_dim < index_size && "index out of bounds"failed. /opt/conda/conda-bld/pytorch_1656352645774/work/aten/src/ATen/native/cuda/ScatterGatherKernel.cu:144: operator(): block: [6825,0,0], thread: [22,0,0] Assertionidx_dim >= 0 && idx_dim < index_size && "index out of bounds"failed. /opt/conda/conda-bld/pytorch_1656352645774/work/aten/src/ATen/native/cuda/ScatterGatherKernel.cu:144: operator(): block: [6825,0,0], thread: [23,0,0] Assertionidx_dim >= 0 && idx_dim < index_size && "index out of bounds"failed. /opt/conda/conda-bld/pytorch_1656352645774/work/aten/src/ATen/native/cuda/ScatterGatherKernel.cu:144: operator(): block: [6825,0,0], thread: [24,0,0] Assertionidx_dim >= 0 && idx_dim < index_size && "index out of bounds"failed. /opt/conda/conda-bld/pytorch_1656352645774/work/aten/src/ATen/native/cuda/ScatterGatherKernel.cu:144: operator(): block: [6825,0,0], thread: [25,0,0] Assertionidx_dim >= 0 && idx_dim < index_size && "index out of bounds"failed. /opt/conda/conda-bld/pytorch_1656352645774/work/aten/src/ATen/native/cuda/ScatterGatherKernel.cu:144: operator(): block: [6825,0,0], thread: [26,0,0] Assertionidx_dim >= 0 && idx_dim < index_size && "index out of bounds"failed. /opt/conda/conda-bld/pytorch_1656352645774/work/aten/src/ATen/native/cuda/ScatterGatherKernel.cu:144: operator(): block: [6825,0,0], thread: [27,0,0] Assertionidx_dim >= 0 && idx_dim < index_size && "index out of bounds"failed. /opt/conda/conda-bld/pytorch_1656352645774/work/aten/src/ATen/native/cuda/ScatterGatherKernel.cu:144: operator(): block: [6825,0,0], thread: [28,0,0] Assertionidx_dim >= 0 && idx_dim < index_size && "index out of bounds"failed. /opt/conda/conda-bld/pytorch_1656352645774/work/aten/src/ATen/native/cuda/ScatterGatherKernel.cu:144: operator(): block: [6825,0,0], thread: [29,0,0] Assertionidx_dim >= 0 && idx_dim < index_size && "index out of bounds"failed. /opt/conda/conda-bld/pytorch_1656352645774/work/aten/src/ATen/native/cuda/ScatterGatherKernel.cu:144: operator(): block: [6825,0,0], thread: [30,0,0] Assertionidx_dim >= 0 && idx_dim < index_size && "index out of bounds"failed. /opt/conda/conda-bld/pytorch_1656352645774/work/aten/src/ATen/native/cuda/ScatterGatherKernel.cu:144: operator(): block: [6825,0,0], thread: [31,0,0] Assertionidx_dim >= 0 && idx_dim < index_size && "index out of bounds"failed. Traceback (most recent call last): File "/home/jxr/anaconda3/envs/python/lib/python3.10/runpy.py", line 196, in _run_module_as_main return _run_code(code, main_globals, None, File "/home/jxr/anaconda3/envs/python/lib/python3.10/runpy.py", line 86, in _run_code exec(code, run_globals) File "/home/jxr/.vscode-server/extensions/ms-python.python-2022.12.0/pythonFiles/lib/python/debugpy/adapter/../../debugpy/launcher/../../debugpy/__main__.py", line 39, in <module> cli.main() File "/home/jxr/.vscode-server/extensions/ms-python.python-2022.12.0/pythonFiles/lib/python/debugpy/adapter/../../debugpy/launcher/../../debugpy/../debugpy/server/cli.py", line 430, in main run() File "/home/jxr/.vscode-server/extensions/ms-python.python-2022.12.0/pythonFiles/lib/python/debugpy/adapter/../../debugpy/launcher/../../debugpy/../debugpy/server/cli.py", line 284, in run_file runpy.run_path(target, run_name="__main__") File "/home/jxr/.vscode-server/extensions/ms-python.python-2022.12.0/pythonFiles/lib/python/debugpy/_vendored/pydevd/_pydevd_bundle/pydevd_runpy.py", line 321, in run_path return _run_module_code(code, init_globals, run_name, File "/home/jxr/.vscode-server/extensions/ms-python.python-2022.12.0/pythonFiles/lib/python/debugpy/_vendored/pydevd/_pydevd_bundle/pydevd_runpy.py", line 135, in _run_module_code _run_code(code, mod_globals, init_globals, File "/home/jxr/.vscode-server/extensions/ms-python.python-2022.12.0/pythonFiles/lib/python/debugpy/_vendored/pydevd/_pydevd_bundle/pydevd_runpy.py", line 124, in _run_code exec(code, run_globals) File "/home/jxr/3D-MultiMAE/MultiMAE/try_model.py", line 118, in <module> preds, masks = multimae.forward( File "/home/jxr/3D-MultiMAE/MultiMAE/multimae/multimae.py", line 350, in forward encoder_tokens = self.encoder(input_tokens) File "/home/jxr/anaconda3/envs/python/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1130, in _call_impl return forward_call(*input, **kwargs) File "/home/jxr/anaconda3/envs/python/lib/python3.10/site-packages/torch/nn/modules/container.py", line 139, in forward input = module(input) File "/home/jxr/anaconda3/envs/python/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1130, in _call_impl return forward_call(*input, **kwargs) File "/home/jxr/3D-MultiMAE/MultiMAE/multimae/multimae_utils.py", line 230, in forward x = x + self.drop_path(self.attn(self.norm1(x))) File "/home/jxr/anaconda3/envs/python/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1130, in _call_impl return forward_call(*input, **kwargs) File "/home/jxr/3D-MultiMAE/MultiMAE/multimae/multimae_utils.py", line 175, in forward attn = (q @ k.transpose(-2, -1)) * self.scale RuntimeError: CUDA error: device-side assert triggered CUDA kernel errors might be asynchronously reported at some other API call,so the stacktrace below might be incorrect. For debugging consider passing CUDA_LAUNCH_BLOCKING=1.

opened by kamzero 2
Example usage of regular MAE Weights

Hey awesome work! I am trying to figure out how to modify the demo notebook to use the regular MAE instead of multiMAE. In particular i comment out all depth and semseg info but the resulting image infilling looks corrupted. Could you by chance share an example of proper usage of the regular MAE weights? Thanks so much for the help!

opened by mhamilton723 2
Some doubts about pseudo labels

Hi, I am pseudo-tagging the imagenet-1k, and encountering some difficulties.

Firstly, I wonder what would happen if the classes of semeg are more than 255? How to use one channel depth png image to represent them? (Although COCO datasets is only 80 classes, the imagenet is more than 255 classes when fine-tuning)

Secondly, on the example of Colab notebook, the rgb2depth model of DPT could not input any size of imagenet pictures. How could we save all the pseudo labels down before the data augmentation cutting it into 224*224? We need to align the original images with the pseudo labeled image should we?

Thank you for any help.

opened by Chianghui-Wong 2

is it normal to see this during finetuning?

_IncompatibleKeys(missing_keys=['output_adapters.cls.norm.weight', 'output_adapters.cls.norm.bias', 'output_adapters.cls.head.weight', 'output_adapters.cls.head.bias'], unexpected_keys=[])

Could be happening because of deleting the output adapter.

@dmizr thank you for replying to previous issues.

opened by forkbabu 1

how to evaluate/test mae or multimae on test dataset?

@dmizr @amir32002 @roman-bachmann thanks for the paper

is there a way in the code already to perform testing on a different dataset or is it required to code it up ourselves?

opened by forkbabu 1
colab error

Hello,

Thank you for the code. However, I am not able to run this: !wget https://drive.switch.ch/index.php/s/RFfTZwyKROKKx0l/download in the google colab. Can you please recheck?

Thanks

opened by AnukritiSinghh 1

Releases(pretrained-weights)

pretrained-weights(Mar 31, 2022)

Source code(tar.gz)
Source code(zip)
mae-b_dec512d8b_1600e_multivit-c477195b.pth(423.56 MB)
mae-b_dec512d8b_1600e_timm-f74f3a8d.pth(426.97 MB)
multimae-b_98_rgb+-depth-semseg_1600e_multivit-afff3f8c.pth(376.13 MB)
multimae-b_98_rgb+-depth-semseg_1600e_timm-bafa5499.pth(327.33 MB)

Code and pre-trained models for MultiMAE: Multi-modal Multi-task Masked Autoencoders

Related tags

Overview

MultiMAE: Multi-modal Multi-task Masked Autoencoders

Catalog

Pre-trained models

Model formats

Usage

Set-up

Pre-training

Fine-tuning

Demo & visualizations

Acknowledgement

License

Citation

Comments

Problem

Environment

Terminal

Releases(pretrained-weights)

pretrained-weights(Mar 31, 2022)

Owner

Visual Intelligence & Learning Lab, Swiss Federal Institute of Technology (EPFL)

VideoMAE: Masked Autoencoders are Data-Efficient Learners for Self-Supervised Video Pre-Training

PyTorch code for the paper "Complementarity is the King: Multi-modal and Multi-grained Hierarchical Semantic Enhancement Network for Cross-modal Retrieval".

Implementation of PyTorch-based multi-task pre-trained models

Unofficial PyTorch implementation of Masked Autoencoders Are Scalable Vision Learners

PyTorch implementation of Masked Autoencoders Are Scalable Vision Learners for self-supervised ViT.

An pytorch implementation of Masked Autoencoders Are Scalable Vision Learners

Re-implememtation of MAE (Masked Autoencoders Are Scalable Vision Learners) using PyTorch.

A collection of pre-trained StyleGAN2 models trained on different datasets at different resolution.

A Comprehensive Empirical Study of Vision-Language Pre-trained Model for Supervised Cross-Modal Retrieval

The official code for PRIMER: Pyramid-based Masked Sentence Pre-training for Multi-document Summarization

Codes to pre-train T5 (Text-to-Text Transfer Transformer) models pre-trained on Japanese web texts

FocusFace: Multi-task Contrastive Learning for Masked Face Recognition

Code of U2Fusion: a unified unsupervised image fusion network for multiple image fusion tasks, including multi-modal, multi-exposure and multi-focus image fusion.

Pre-trained BERT Models for Ancient and Medieval Greek, and associated code for LaTeCH 2021 paper titled - "A Pilot Study for BERT Language Modelling and Morphological Analysis for Ancient and Medieval Greek"

VIMPAC: Video Pre-Training via Masked Token Prediction and Contrastive Learning

[CVPR 2022 Oral] Versatile Multi-Modal Pre-Training for Human-Centric Perception

Source code and dataset for ACL2021 paper: "ERICA: Improving Entity and Relation Understanding for Pre-trained Language Models via Contrastive Learning".

Code, Data and Demo for Paper: Controllable Generation from Pre-trained Language Models via Inverse Prompting