Code and BeyondCode😋
Posts Tools Tags Categories BeyondCode😋
Code and BeyondCode😋

Useful Linux CLI Commands

uncond train, and eval

train

1
2
3
4
5
6
7
8

# vae
OMP_NUM_THREADS=3 python main.py -b configs/autoencoder/kitti/autoencoder_c2_p4.yaml -t --gpus 0,1, -r logs/kitti/2025-02-27T11-19-07_autoencoder_c2_p4

OMP_NUM_THREADS=3 python main.py -b configs/autoencoder/kitti/autoencoder_c2_p4.yaml -t --gpus 3,
# lidm both for cond and uncond
OMP_NUM_THREADS=3 python main.py -b configs/lidar_diffusion/kitti/uncond_vig_dec0.yaml -t --gpus 0,

OMP_NUM_THREADS=3 python main.py -b configs/lidar_diffusion/kitti/cond_t2l_vig_dec0.yaml -t --gpus 1,3 -r logs/kitti/2025-02-25T12-14-50_cond_t2l_vig_dec0

eval

1
2
3
4
5
6
7
8
9

# sample and eval
CUDA_VISIBLE_DEVICES=2 python scripts/sample.py -d kitti -r models/lidm/kitti/vig_dec0/epoch=000027.ckpt -n 2000 -s 250 --eval # for current best model, and specify the seed

CUDA_VISIBLE_DEVICES=1 python scripts/sample.py -d kitti -r models/lidm/kitti/vig_dec0/epoch=000027.ckpt -n 2000 -c 25 --eval # for inference steps

CUDA_VISIBLE_DEVICES=1 python scripts/sample.py -d kitti -r models/lidm/kitti/vig_dec0/last.ckpt -n 2000 --eval

# eval only
CUDA_VISIBLE_DEVICES=1 python scripts/sample.py -d kitti -f models/lidm/kitti/gcn_tr1d/samples/00452624/2025-02-03-15-44-42/numpy/samples.pcd --eval

conditional sample

1
2

# text2lidar
OMP_NUM_THREADS=3 CUDA_VISIBLE_DEVICES=1 python scripts/text2lidar.py -r models/lidm/kitti/t2l/epoch=000020.ckpt -d kitti -p "there is a crosswalk in front of the ego vehicle" # 有个十字路口的场景

docker setup

unix helpful commands

find and delete folders containing sth

在Unix系统中，你可以使用find命令结合rm命令来删除所有包含break的文件夹。这里有一个命令示例，它会查找当前路径下所有包含break的文件夹，并删除它们：

1

find . -type d -name '*2025-03-02-10-04-43*' -exec rm -rf {} +

1

find . -type f -name '*00002*' -exec rm {} +

下面是命令的详细解释： find .：从当前目录开始查找。 -type d：查找的类型是目录。 -name ‘*break*’：查找名称中包含break的目录。 -exec：对找到的每个项目执行后面的命令。 rm -rf {} +：删除找到的每个目录及其内容。{}是一个占位符，代表find命令找到的每个项目，+表示将所有找到的项目传递给rm命令，而不是一次传递一个。

zip

1

zip -r topo_1000.zip *.txt

peek the gpu usage

1

watch -n 1 nvidia-smi

peek the cpu usage

1

top

1

OMP_NUM_THREADS=3

kill a process

1

kill -9 <pid>

在你提供的截图中，通过 ps aux | grep python 命令列出了包含 “python” 关键词的进程。 如果你想要终止所有正在运行的 Python 进程，你可以使用 pkill 命令：

1

pkill -f python

peek the disk usage

1

df -h

ssh to a remote server

1

ssh -P <port> <username>@<remote_server_ip>

nohup

1

nohup python GLiDR_kitti.py --data /data/data_odometry_static_dynamic/scan/ --exp_name find_shapes_kitti_mlpdiff --beam 16 --dim 8 --batch_size 32 --mode kitti > training_mlp_diff_100.log 2>&1 &

cd

1
2

cd - # 回到上一级目录
cd "/path/to/directory with spaces" # 进入带有空格的目录

tmux后台使用逻辑

1
2
3

# install tmux
sudo apt-get update
sudo apt-get install tmux

Session

  • 新建会话
1

tmux new -s lidm

  • 离开会话 Ctrl+b d

  • 重连会话

1

tmux attach -t my_session
  • 列出会话
1

tmux ls

  • 杀死会话 Ctrl+b :kill-session

  • 上下滚动 Ctrl+b [ q

Window

  • 创建新窗口 Ctrl+b c

  • 切换窗口 Ctrl+b p （上一个窗口） Ctrl+b n （下一个窗口）Ctrl+b <number> （指定窗口）Ctrl+b w （窗口列表）

  • 为窗口命名
    Ctrl+b ,

Panes

  • 切分窗格 Ctrl+b % 垂直分割 Ctrl+b " 水平分割

  • 转换窗口 Ctrl+b <arrow key>

  • 关闭当前 pane Ctrl+b x

build essentials

1
2

sudo apt-get update
sudo apt-get install build-essential

ssh docker image transfer

docker run

1
2
3
4
5

nvidia-docker run -it --name 'hz_lidargen' --shm-size 16g -v /data0/dataset/:/data -v /home/liujiuming/hz:/home/hz fzhiheng/deep_envs:py38-torch110-cu113-pointnet2 /bin/bash

docker start hz_lidargen # lidardiff / lidargen / ultralidar

docker exec -it hz_lidargen /bin/bash

docker save and load

1

docker save -o my-image.tar my-image:tag

1
2
3
4
5
6

scp 'epoch=000037.ckpt' liujiuming@10.129.22.67:~/GLIDR/GLiDR-main/LiDAR-Diffusion-main/models/lidm/kitti/uncond_vig_dec1

# folder
scp -r my-folder user@remote-server:/path/to/destination

docker load -i /path/to/destination/my-image.tar

ifconfig

1

ifconfig

1
2
3

6012 10.129.22.57
6017 10.129.22.67 
6016 10.129.22.66

Python Lib

knn_cuda

1

pip install --upgrade https://github.com/unlimblue/KNN_CUDA/releases/download/0.2/KNN_CUDA-0.2-py3-none-any.whl -i http://pypi.doubanio.com/simple/ --trusted-host pypi.doubanio.com

pointnet2_ops_lib

1

pip install git+https://github.com/erikwijmans/Pointnet2_PyTorch.git#subdirectory=pointnet2_ops_lib

topologylayer

1
2

pip install git+https://github.com/Topologic/topologylayer.git
# topologylayer np.int --> np.int64

ninja

https://kimi.moonshot.cn/chat/ctjnjk8pe77ot9u80lvg

1

sudo apt-get install ninja-build

pip configs / 换源 / conda 换源

1

pip install open3d -i https://pypi.tuna.tsinghua.edu.cn/simple # 临时

1

pip config set global.index-url https://pypi.tuna.tsinghua.edu.cn/simple # 永久

1
2
3
4
5
6
7
8

conda config --show channels
conda config --set show_channel_urls yes
# 加进去
conda config --add channels https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/free/
conda config --add channels https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/main/
# conda-forge
conda config --add channels https://mirrors.tuna.tsinghua.edu.cn/anaconda/cloud/conda-forge/
# 注意windows prefers http instead of https

lidargen / ultralidar commands

1
2
3

export PYTHONPATH=/home/hz/mmdetection3d:$PYTHONPATH # in docker
CUDA_VISIBLE_DEVICES=3 ./tools/dist_test.sh configs/ultralidar_kitti360_static_blank_code.py ./epoch_200.pth 1 --eval "mIoU"
CUDA_VISIBLE_DEVICES=3 ./tools/dist_test.sh configs/ultralidar_kitti360_gene.py ./epoch_200.pth 1 --eval "mIoU"
for viz draft
1
2
3
4
5
6

export KITTI360_DATASET=/data/KITTI-360/
OMP_NUM_THREADS=3 CUDA_VISIBLE_DEVICES=3 python lidargen.py --sample --exp kitti_pretrained --config kitti.yml

OMP_NUM_THREADS=3 CUDA_VISIBLE_DEVICES=3 python try.py --loop_count 200 --mid 4

OMP_NUM_THREADS=3 CUDA_VISIBLE_DEVICES=3 python try.py --loop_count 5000 --mid 4