GETTING_STARTED.md

Getting Started

This document provides basic instructions for training and evaluation using pycls.

• For general information about pycls, please see README.md
• For installation instructions, please see INSTALL.md

Training Models

Training on CIFAR with 1 GPU:

python tools/train_net.py \
    --cfg configs/cifar/resnet/R-56_nds_1gpu.yaml \
    OUT_DIR /tmp

Training on ImageNet with 1 GPU:

python tools/train_net.py \
    --cfg configs/imagenet/resnet/R-50-1x64d_step_1gpu.yaml \
    OUT_DIR /tmp

Training on ImageNet with 2 GPUs:

python tools/train_net.py \
    --cfg configs/imagenet/resnet/R-50-1x64d_step_2gpu.yaml \
    OUT_DIR /tmp

Finetuning Models

Finetuning on ImageNet with 1 GPU:

python tools/train_net.py \
    --cfg configs/imagenet/resnet/R-50-1x64d_step_1gpu.yaml \
    TRAIN.WEIGHTS /path/to/weights/file \
    OUT_DIR /tmp

Evaluating Models

Evaluation on ImageNet with 1 GPU:

python tools/test_net.py \
    --cfg configs/imagenet/resnet/R-50-1x64d_step_1gpu.yaml \
    TEST.WEIGHTS /path/to/weights/file \
    OUT_DIR /tmp