$ cd docker && docker compose up -d --build
$ docker compose down && docker compose up -d --build && docker exec -it [container_name] bash$ uv python pin 3.11
$ uv venv --python 3.11
$ source .venv/bin/activate
## uv init
$ uv sync
## uv add ...
$ uv add ruff
## run python program
$ uv run python main.py
$ uv lock
## uv sync # installs everything into .venv
$ uv export --format=requirements.txt > requirements.txt
$ deactivateref:
ref: https://lambda.ai/blog/set-up-a-tensorflow-gpu-docker-container-using-lambda-stack-dockerfile
ssh ubuntu@IP_ADDRESS -i ~/.ssh/lambda_cloudcurl -fsSL https://raw.githubusercontent.com/DenDen047/dotfiles/refs/heads/master/setup_scripts/lambda_cloud1.sh | bash
# if failed in the last step
sudo apt-get update && sudo apt-get install -y lambda-stack-cuda && sudo reboot
# after reboot, run the following command
curl -fsSL https://raw.githubusercontent.com/DenDen047/dotfiles/refs/heads/master/setup_scripts/lambda_cloud2.sh | bashref: https://modal.com/docs/guide
modal setup
modal run src/modal_sample.py.
├── README.md # プロジェクト概要と使い方を記述
├── conf/ # 実験設定ファイル (例: parameters.yml, secrets.yml)
├── data/ # データや中間成果物の一時保存場所
├── notebooks/ # JupyterLabでの実験ノート
├── pyproject.toml # Pythonプロジェクトの主要設定ファイル (PEP 518準拠)
├── setup.cfg # pyproject.toml未対応の設定を補完
└── src/ # Pythonパッケージコード (共通処理の関数やクラスなど)Please see the details here.
data/
├── 01_raw/ # Original, immutable data from source systems
├── 02_intermediate/ # Partially processed (cleaned/transformed) data
├── 03_primary/ # Canonical datasets for feature engineering
├── 04_feature/ # Engineered features ready for modeling
├── 05_model_input/ # Data prepared specifically for model training
├── 06_models/ # Trained models (e.g., .pkl, .h5 files)
├── 07_model_output/ # Model outputs like predictions or embeddings
└── 08_reporting/ # Reports, visualizations, dashboards, final outputsThis project follows git-flow.
- Starting Work: Begin new development or experiments by creating a branch off of the
developbranch. - Merging to
develop: Once work intended fordevelopis complete, merge it into thedevelopbranch (creating a Pull Request on GitHub is recommended if applicable) and delete the original branch. - Keeping History (Archiving): If you wish to keep the code and history of a branch without merging it into
develop(e.g., failed experiments, pure explorations), rename the branch toarchive/<original-branch-name>.- Example: To archive a branch named
exp/try-hyperparams-v1, rename it toarchive/exp/try-hyperparams-v1.
- Example: To archive a branch named
- Archived Branches: Branches prefixed with
archive/are kept for reference only and must not be merged into active branches likedevelopormain.
Following commitlint rule:
- https://github.com/conventional-changelog/commitlint
- https://www.conventionalcommits.org/en/v1.0.0/
- colormap:
turbo
ref: https://en.wikipedia.org/wiki/Active_and_passive_transformation
To avoid confusion, we distinguish between active and passive interpretations:
-
Active rotation / transformation
- Variables:
R_active,R_apply,R_obj,T_active - Meaning: Actively rotating points or vectors (e.g., applying to a point cloud).
- Variables:
-
Passive rotation / transformation
- Variables:
R_world_to_cam,R_frame,R_pose,T_world_to_cam - Meaning: Changing the coordinate frame (e.g., camera extrinsics).
- Variables:
-
Other common conventions
R_ext,T_ext: Extrinsic parameters (world → camera transformation).R_int,K: Intrinsic parameters (camera matrix).R_wc,R_cw: Shorthand forR_world_to_cam,R_cam_to_world.
-
Rotation matrices
R_passive = R_active.TR_active = R_passive.T
-
Transformation matrices (SE(3))
T_passive = T_active^-1T_active = T_passive^-1
This ensures consistent handling of both interpretations.
- 【Pythonの仮想環境を比較】〜オススメを紹介 〜
- モダンなPyTorchのテンプレ
- timm – PyTorch Image Models
- Best Practices for Python Coding