You can not select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.
ColossalAI/examples/images/vit
Baizhou Zhang 295b38fecf
[example] update vit example for hybrid parallel plugin (#4641)
1 year ago
..
README.md [example] update vit example for hybrid parallel plugin (#4641) 1 year ago
args.py [example] update vit example for hybrid parallel plugin (#4641) 1 year ago
data.py [example] update vit example for hybrid parallel plugin (#4641) 1 year ago
requirements.txt [example] update ViT example using booster api (#3940) 1 year ago
run_benchmark.sh [example] update vit example for hybrid parallel plugin (#4641) 1 year ago
run_demo.sh [example] update vit example for hybrid parallel plugin (#4641) 1 year ago
test_ci.sh [example] update vit example for hybrid parallel plugin (#4641) 1 year ago
vit_benchmark.py [example] update vit example for hybrid parallel plugin (#4641) 1 year ago
vit_train_demo.py [example] update vit example for hybrid parallel plugin (#4641) 1 year ago

README.md

Overview

Vision Transformer is a class of Transformer model tailored for computer vision tasks. It was first proposed in paper An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale and achieved SOTA results on various tasks at that time.

In our example, we are using pretrained weights of ViT loaded from HuggingFace. We adapt the ViT training code to ColossalAI by leveraging Boosting API loaded with a chosen plugin, where each plugin corresponds to a specific kind of training strategy. This example supports plugins including TorchDDPPlugin (DDP), LowLevelZeroPlugin (Zero1/Zero2), GeminiPlugin (Gemini) and HybridParallelPlugin (any combination of tensor/pipeline/data parallel).

Run Demo

By running the following script:

bash run_demo.sh

You will finetune a a ViT-base model on this dataset, with more than 8000 images of bean leaves. This dataset is for image classification task and there are 3 labels: ['angular_leaf_spot', 'bean_rust', 'healthy'].

The script can be modified if you want to try another set of hyperparameters or change to another ViT model with different size.

The demo code refers to this blog.

Run Benchmark

You can run benchmark for ViT model by running the following script:

bash run_benchmark.sh

The script will test performance (throughput & peak memory usage) for each combination of hyperparameters. You can also play with this script to configure your own set of hyperparameters for testing.