Install all required dependencies in requirements.txt and replace all ..path/..to in the code to the absolute path
to corresponding resources.
*Only slurm-based distributed training support is implemented in this repo.
All experiment files are located in ssl_exp/fast_moco folder.
To perform self-supervised training on ImageNet, run:
sh run.sh <gpu-partition> To perform linear evaluation, make sure the pretrained weight is located in ./checkpoints folder or specified in config_finetune.yaml (saver.pretrain.path), run:
sh run_finetune.sh <gpu-partition> | Backbone | Model | Pretrained Epochs |
Top-1 Linear Evaluation Accuracy |
Pretrained Weight | md5 |
|---|---|---|---|---|---|
| Resnet-50 | Fast-MoCo | 100 | 73.5% | checkpoint | d6ea9023372c14db94b0dc285f216f99 |
| Resnet-50 | Fast-MoCo | 200 | 75.1% | checkpoint | 9f1c29ea305d9214f265fa460856db28 |
| Resnet-50 | Fast-MoCo | 400 | 75.5% | checkpoint | 79ae2aff26c6cb762feaf9155b137d4a |