Reproduction of Vision Transformer in Tensorflow2. Train from scratch and Finetune.

Last update: Dec 27, 2022

Related tags

Overview

Vision Transformer(ViT) in Tensorflow2

Tensorflow2 implementation of the Vision Transformer(ViT).

This repository is for An image is worth 16x16 words: Transformers for image recognition at scale and How to train your ViT? Data, Augmentation, and Regularization in Vision Transformers.

Limitations.

Due to memory limitations, only the ti/16, s/16, and b/16 models were tested.
Due to memory limitations, batch_size 2048 in s16 and 1024 in b/16 (in paper, 4096).
Due to computational resource limitations, only reproduce using imagenet1k.

All experimental results and graphs are opend in Wandb.

https://docs.google.com/spreadsheets/d/1j0lFlaMuqccFiHj3eQVpZYIbSoXY6Pz6oEW76x7g25M/edit?usp=sharing
upstream: https://wandb.ai/justhungryman/vit
downstream: https://wandb.ai/justhungryman/vit-downstream/
In case of an experiment in which the tpu is stopped, it is resumed (duplicated experiment name but different start epoch).

Model weights

Since this is personal project, it is hard to train with large datasets like imagenet21k. For a pretrain model with good performance, see the official repo. But if you really need it, contact me.

Install dependencies

pip install -r requirements

All experiments were done on tpu_v3-8 with the support of TRC. But you can experiment on GPU. Check conf/config.yaml and conf/downstream.yaml

  # TPU options
  env:
    mode: tpu
    gcp_project: {your_project}
    tpu_name: node-1
    tpu_zone: europe-west4-a
    mixed_precision: True
  # GPU options
  # env:
  #   mode: gpu
  #   mixed_precision: True

Train from scratch

python run.py experiment=vit-s16-aug_light1-bs_2048-wd_0.1-do_0.1-dp_0.1-lr_1e-3 base.project_name=vit-s16-aug_light1-bs_2048-wd_0.1-do_0.1-dp_0.1-lr_1e-3 base.save_dir={your_save_dir} base.env.gcp_project={your_gcp_project} base.env.tpu_name={your_tpu_name} base.debug=False

Downstream

python run.py --config-name=downstream experiment=downstream-imagenet-ti16_384 base.pretrained={your_checkpoint} base.project_name={your_project_name} base.save_dir={your_save_dir} base.env.gcp_project={your_gcp_project} base.env.tpu_name={your_tpu_name} base.debug=False

Board

To track metics, you can use wandb or tensorboard (default: wandb). You can change in conf/callbacks/{filename.yaml}.

modules:
  - type: MonitorCallback
  - type: TerminateOnNaN
  - type: ProgbarLogger
    params:
      count_mode: steps
  - type: ModelCheckpoint
    params:
      filepath: ???
      save_weights_only: True
  - type: Wandb
    project: vit
    nested_dict: False
    hide_config: True
    params: 
      monitor: val_loss
      save_model: False
  # - type: TensorBoard
  #   params:
  #     log_dir: ???
  #     histogram_freq: 1

TFC

This open source was assisted by TPU Research Cloud (TRC) program

Thank you for providing the TPU.

Citations

@article{dosovitskiy2020image,
  title={An image is worth 16x16 words: Transformers for image recognition at scale},
  author={Dosovitskiy, Alexey and Beyer, Lucas and Kolesnikov, Alexander and Weissenborn, Dirk and Zhai, Xiaohua and Unterthiner, Thomas and Dehghani, Mostafa and Minderer, Matthias and Heigold, Georg and Gelly, Sylvain and others},
  journal={arXiv preprint arXiv:2010.11929},
  year={2020}
}

@article{steiner2021train,
  title={How to train your ViT? Data, Augmentation, and Regularization in Vision Transformers},
  author={Steiner, Andreas and Kolesnikov, Alexander and Zhai, Xiaohua and Wightman, Ross and Uszkoreit, Jakob and Beyer, Lucas},
  journal={arXiv preprint arXiv:2106.10270},
  year={2021}
}

Reproduction of Vision Transformer in Tensorflow2. Train from scratch and Finetune.

Related tags

Overview

Vision Transformer(ViT) in Tensorflow2

Limitations.

Model weights

Install dependencies

Train from scratch

Downstream

Board

TFC

Citations

Owner

sungjun lee

Visualizer for neural network, deep learning, and machine learning models

VLG-Net: Video-Language Graph Matching Networks for Video Grounding

In real-world applications of machine learning, reliable and safe systems must consider measures of performance beyond standard test set accuracy

NeuralForecast is a Python library for time series forecasting with deep learning models

[NeurIPS 2021] Low-Rank Subspaces in GANs

Pre-trained Deep Learning models and demos (high quality and extremely fast)

some classic model used to segment the medical images like CT、X-ray and so on

A lightweight face-recognition toolbox and pipeline based on tensorflow-lite

This is the official source code for SLATE. We provide the code for the model, the training code, and a dataset loader for the 3D Shapes dataset. This code is implemented in Pytorch.

This repository is a series of notebooks that show solutions for the projects at Dataquest.io.

OHLC Average Prediction of Apple Inc. Using LSTM Recurrent Neural Network

Dynamic Graph Event Detection

pytorch implementation of openpose including Hand and Body Pose Estimation.

InferPy: Deep Probabilistic Modeling with Tensorflow Made Easy

Self-Supervised Image Denoising via Iterative Data Refinement

Code for CVPR 2021 paper TransNAS-Bench-101: Improving Transferrability and Generalizability of Cross-Task Neural Architecture Search.

Enabling dynamic analysis of Legacy Embedded Systems in full emulated environment

sequitur is a library that lets you create and train an autoencoder for sequential data in just two lines of code

1st Solution For NeurIPS 2021 Competition on ML4CO Dual Task

novel deep learning research works with PaddlePaddle