Skip to content

Install.md

Latest commit

 

History

History
72 lines (49 loc) · 1.41 KB

Install.md

File metadata and controls

72 lines (49 loc) · 1.41 KB

Step-by-step installation instructions

Environment configuration

Create a conda virtual environment and activate it。

conda create -n tiofusion python=3.9
conda activate tiofusion
conda install cuda -c nvidia/label/cuda-12.1.1
conda install pytorch==2.1.0 torchvision==0.16.0 torchaudio==2.1.0 pytorch-cuda=12.1 -c pytorch -c nvidia
pip install openmim
mim install mmengine==0.9.0 mmcv==2.1.0 mmdet==3.2.0 mmdet3d==1.3.0

Configure CUDA_HOME

export CUDA_HOME={conda_path}/envs/mmlab

（optional）Install ninja to accelerate compilation

pip install ninja

Using the gcc compiler for linux and MSVC for windows, execute in the project root directory

python setup.py develop

Install GKT

cd TIOFUSION\model\pv2bev_encoder\ops\geometric_kernel_attn
python setup.py develop

train

single gpu

python train.py --config {} --work-dir {}

Use a specific gpu 0

python train.py --config {} --work-dir {} --gpu-ids 0

distributed training

python -m torch.distributed.launch --nproc_per_node=8 train.py --config {} --work-dir {} --launcher pytorch --gpus 8

Use specific gpu

CUDA_VISIBLE_DEVICES=0,3 python -m torch.distributed.launch --nproc_per_node=2 train.py --config {} --work-dir {} --launcher pytorch --gpus 2

test

 python test.py --config {} --load-from {checkpoinkfile}