cc

.gitignore

@@ -0,0 +1,4 @@
+__pycache__
+*.zip
+wandb
+logs

Makefile

@@ -0,0 +1,30 @@
+
+help:  ## Show help
+	@grep -E '^[.a-zA-Z_-]+:.*?## .*$$' $(MAKEFILE_LIST) | awk 'BEGIN {FS = ":.*?## "}; {printf "\033[36m%-30s\033[0m %s\n", $$1, $$2}'
+
+clean: ## Clean autogenerated files
+	rm -rf dist
+	find . -type f -name "*.DS_Store" -ls -delete
+	find . | grep -E "(__pycache__|\.pyc|\.pyo)" | xargs rm -rf
+	find . | grep -E ".pytest_cache" | xargs rm -rf
+	find . | grep -E ".ipynb_checkpoints" | xargs rm -rf
+	rm -f .coverage
+
+clean-logs: ## Clean logs
+	rm -rf logs/**
+
+format: ## Run pre-commit hooks
+	pre-commit run -a
+
+sync: ## Merge changes from main branch to your current branch
+	git pull
+	git pull origin main
+
+test: ## Run not slow tests
+	pytest -k "not slow"
+
+test-full: ## Run all tests
+	pytest
+
+train: ## Train the model
+	python src/train.py

README.md

@@ -0,0 +1,1297 @@
+<div align="center">
+
+# Lightning-Hydra-Template
+
+[![python](https://img.shields.io/badge/-Python_3.8_%7C_3.9_%7C_3.10-blue?logo=python&logoColor=white)](https://github.com/pre-commit/pre-commit)
+[![pytorch](https://img.shields.io/badge/PyTorch_2.0+-ee4c2c?logo=pytorch&logoColor=white)](https://pytorch.org/get-started/locally/)
+[![lightning](https://img.shields.io/badge/-Lightning_2.0+-792ee5?logo=pytorchlightning&logoColor=white)](https://pytorchlightning.ai/)
+[![hydra](https://img.shields.io/badge/Config-Hydra_1.3-89b8cd)](https://hydra.cc/)
+[![black](https://img.shields.io/badge/Code%20Style-Black-black.svg?labelColor=gray)](https://black.readthedocs.io/en/stable/)
+[![isort](https://img.shields.io/badge/%20imports-isort-%231674b1?style=flat&labelColor=ef8336)](https://pycqa.github.io/isort/) <br>
+[![tests](https://github.com/ashleve/lightning-hydra-template/actions/workflows/test.yml/badge.svg)](https://github.com/ashleve/lightning-hydra-template/actions/workflows/test.yml)
+[![code-quality](https://github.com/ashleve/lightning-hydra-template/actions/workflows/code-quality-main.yaml/badge.svg)](https://github.com/ashleve/lightning-hydra-template/actions/workflows/code-quality-main.yaml)
+[![codecov](https://codecov.io/gh/ashleve/lightning-hydra-template/branch/main/graph/badge.svg)](https://codecov.io/gh/ashleve/lightning-hydra-template) <br>
+[![license](https://img.shields.io/badge/License-MIT-green.svg?labelColor=gray)](https://github.com/ashleve/lightning-hydra-template#license)
+[![PRs](https://img.shields.io/badge/PRs-welcome-brightgreen.svg)](https://github.com/ashleve/lightning-hydra-template/pulls)
+[![contributors](https://img.shields.io/github/contributors/ashleve/lightning-hydra-template.svg)](https://github.com/ashleve/lightning-hydra-template/graphs/contributors)
+
+A clean template to kickstart your deep learning project 🚀⚡🔥<br>
+Click on [<kbd>Use this template</kbd>](https://github.com/ashleve/lightning-hydra-template/generate) to initialize new repository.
+
+_Suggestions are always welcome!_
+
+</div>
+
+<br>
+
+## 📌  Introduction
+
+**Why you might want to use it:**
+
+✅ Save on boilerplate <br>
+Easily add new models, datasets, tasks, experiments, and train on different accelerators, like multi-GPU, TPU or SLURM clusters.
+
+✅ Education <br>
+Thoroughly commented. You can use this repo as a learning resource.
+
+✅ Reusability <br>
+Collection of useful MLOps tools, configs, and code snippets. You can use this repo as a reference for various utilities.
+
+**Why you might not want to use it:**
+
+❌ Things break from time to time <br>
+Lightning and Hydra are still evolving and integrate many libraries, which means sometimes things break. For the list of currently known problems visit [this page](https://github.com/ashleve/lightning-hydra-template/labels/bug).
+
+❌ Not adjusted for data engineering <br>
+Template is not really adjusted for building data pipelines that depend on each other. It's more efficient to use it for model prototyping on ready-to-use data.
+
+❌ Overfitted to simple use case <br>
+The configuration setup is built with simple lightning training in mind. You might need to put some effort to adjust it for different use cases, e.g. lightning fabric.
+
+❌ Might not support your workflow <br>
+For example, you can't resume hydra-based multirun or hyperparameter search.
+
+> **Note**: _Keep in mind this is unofficial community project._
+
+<br>
+
+## Main Technologies
+
+[PyTorch Lightning](https://github.com/PyTorchLightning/pytorch-lightning) - a lightweight PyTorch wrapper for high-performance AI research. Think of it as a framework for organizing your PyTorch code.
+
+[Hydra](https://github.com/facebookresearch/hydra) - a framework for elegantly configuring complex applications. The key feature is the ability to dynamically create a hierarchical configuration by composition and override it through config files and the command line.
+
+<br>
+
+## Main Ideas
+
+- [**Rapid Experimentation**](#your-superpowers): thanks to hydra command line superpowers
+- [**Minimal Boilerplate**](#how-it-works): thanks to automating pipelines with config instantiation
+- [**Main Configs**](#main-config): allow you to specify default training configuration
+- [**Experiment Configs**](#experiment-config): allow you to override chosen hyperparameters and version control experiments
+- [**Workflow**](#workflow): comes down to 4 simple steps
+- [**Experiment Tracking**](#experiment-tracking): Tensorboard, W&B, Neptune, Comet, MLFlow and CSVLogger
+- [**Logs**](#logs): all logs (checkpoints, configs, etc.) are stored in a dynamically generated folder structure
+- [**Hyperparameter Search**](#hyperparameter-search): simple search is effortless with Hydra plugins like Optuna Sweeper
+- [**Tests**](#tests): generic, easy-to-adapt smoke tests for speeding up the development
+- [**Continuous Integration**](#continuous-integration): automatically test and lint your repo with Github Actions
+- [**Best Practices**](#best-practices): a couple of recommended tools, practices and standards
+
+<br>
+
+## Project Structure
+
+The directory structure of new project looks like this:
+
+```
+├── .github                   <- Github Actions workflows
+│
+├── configs                   <- Hydra configs
+│   ├── callbacks                <- Callbacks configs
+│   ├── data                     <- Data configs
+│   ├── debug                    <- Debugging configs
+│   ├── experiment               <- Experiment configs
+│   ├── extras                   <- Extra utilities configs
+│   ├── hparams_search           <- Hyperparameter search configs
+│   ├── hydra                    <- Hydra configs
+│   ├── local                    <- Local configs
+│   ├── logger                   <- Logger configs
+│   ├── model                    <- Model configs
+│   ├── paths                    <- Project paths configs
+│   ├── trainer                  <- Trainer configs
+│   │
+│   ├── eval.yaml             <- Main config for evaluation
+│   └── train.yaml            <- Main config for training
+│
+├── data                   <- Project data
+│
+├── logs                   <- Logs generated by hydra and lightning loggers
+│
+├── notebooks              <- Jupyter notebooks. Naming convention is a number (for ordering),
+│                             the creator's initials, and a short `-` delimited description,
+│                             e.g. `1.0-jqp-initial-data-exploration.ipynb`.
+│
+├── scripts                <- Shell scripts
+│
+├── src                    <- Source code
+│   ├── data                     <- Data scripts
+│   ├── models                   <- Model scripts
+│   ├── utils                    <- Utility scripts
+│   │
+│   ├── eval.py                  <- Run evaluation
+│   └── train.py                 <- Run training
+│
+├── tests                  <- Tests of any kind
+│
+├── .env.example              <- Example of file for storing private environment variables
+├── .gitignore                <- List of files ignored by git
+├── .pre-commit-config.yaml   <- Configuration of pre-commit hooks for code formatting
+├── .project-root             <- File for inferring the position of project root directory
+├── environment.yaml          <- File for installing conda environment
+├── Makefile                  <- Makefile with commands like `make train` or `make test`
+├── pyproject.toml            <- Configuration options for testing and linting
+├── requirements.txt          <- File for installing python dependencies
+├── setup.py                  <- File for installing project as a package
+└── README.md
+```
+
+<br>
+
+## 🚀  Quickstart
+
+```bash
+# clone project
+git clone https://github.com/ashleve/lightning-hydra-template
+cd lightning-hydra-template
+
+# [OPTIONAL] create conda environment
+conda create -n myenv python=3.9
+conda activate myenv
+
+# install pytorch according to instructions
+# https://pytorch.org/get-started/
+
+# install requirements
+pip install -r requirements.txt
+```
+
+Template contains example with MNIST classification.<br>
+When running `python src/train.py` you should see something like this:
+
+<div align="center">
+
+![](https://github.com/ashleve/lightning-hydra-template/blob/resources/terminal.png)
+
+</div>
+
+## ⚡  Your Superpowers
+
+<details>
+<summary><b>Override any config parameter from command line</b></summary>
+
+```bash
+python train.py trainer.max_epochs=20 model.optimizer.lr=1e-4
+```
+
+> **Note**: You can also add new parameters with `+` sign.
+
+```bash
+python train.py +model.new_param="owo"
+```
+
+</details>
+
+<details>
+<summary><b>Train on CPU, GPU, multi-GPU and TPU</b></summary>
+
+```bash
+# train on CPU
+python train.py trainer=cpu
+
+# train on 1 GPU
+python train.py trainer=gpu
+
+# train on TPU
+python train.py +trainer.tpu_cores=8
+
+# train with DDP (Distributed Data Parallel) (4 GPUs)
+python train.py trainer=ddp trainer.devices=4
+
+# train with DDP (Distributed Data Parallel) (8 GPUs, 2 nodes)
+python train.py trainer=ddp trainer.devices=4 trainer.num_nodes=2
+
+# simulate DDP on CPU processes
+python train.py trainer=ddp_sim trainer.devices=2
+
+# accelerate training on mac
+python train.py trainer=mps
+```
+
+> **Warning**: Currently there are problems with DDP mode, read [this issue](https://github.com/ashleve/lightning-hydra-template/issues/393) to learn more.
+
+</details>
+
+<details>
+<summary><b>Train with mixed precision</b></summary>
+
+```bash
+# train with pytorch native automatic mixed precision (AMP)
+python train.py trainer=gpu +trainer.precision=16
+```
+
+</details>
+
+<!-- deepspeed support still in beta
+<details>
+<summary><b>Optimize large scale models on multiple GPUs with Deepspeed</b></summary>
+
+```bash
+python train.py +trainer.
+```
+
+</details>
+ -->
+
+<details>
+<summary><b>Train model with any logger available in PyTorch Lightning, like W&B or Tensorboard</b></summary>
+
+```yaml
+# set project and entity names in `configs/logger/wandb`
+wandb:
+  project: "your_project_name"
+  entity: "your_wandb_team_name"
+```
+
+```bash
+# train model with Weights&Biases (link to wandb dashboard should appear in the terminal)
+python train.py logger=wandb
+```
+
+> **Note**: Lightning provides convenient integrations with most popular logging frameworks. Learn more [here](#experiment-tracking).
+
+> **Note**: Using wandb requires you to [setup account](https://www.wandb.com/) first. After that just complete the config as below.
+
+> **Note**: Click [here](https://wandb.ai/hobglob/template-dashboard/) to see example wandb dashboard generated with this template.
+
+</details>
+
+<details>
+<summary><b>Train model with chosen experiment config</b></summary>
+
+```bash
+python train.py experiment=example
+```
+
+> **Note**: Experiment configs are placed in [configs/experiment/](configs/experiment/).
+
+</details>
+
+<details>
+<summary><b>Attach some callbacks to run</b></summary>
+
+```bash
+python train.py callbacks=default
+```
+
+> **Note**: Callbacks can be used for things such as as model checkpointing, early stopping and [many more](https://pytorch-lightning.readthedocs.io/en/latest/extensions/callbacks.html#built-in-callbacks).
+
+> **Note**: Callbacks configs are placed in [configs/callbacks/](configs/callbacks/).
+
+</details>
+
+<details>
+<summary><b>Use different tricks available in Pytorch Lightning</b></summary>
+
+```yaml
+# gradient clipping may be enabled to avoid exploding gradients
+python train.py +trainer.gradient_clip_val=0.5
+
+# run validation loop 4 times during a training epoch
+python train.py +trainer.val_check_interval=0.25
+
+# accumulate gradients
+python train.py +trainer.accumulate_grad_batches=10
+
+# terminate training after 12 hours
+python train.py +trainer.max_time="00:12:00:00"
+```
+
+> **Note**: PyTorch Lightning provides about [40+ useful trainer flags](https://pytorch-lightning.readthedocs.io/en/latest/common/trainer.html#trainer-flags).
+
+</details>
+
+<details>
+<summary><b>Easily debug</b></summary>
+
+```bash
+# runs 1 epoch in default debugging mode
+# changes logging directory to `logs/debugs/...`
+# sets level of all command line loggers to 'DEBUG'
+# enforces debug-friendly configuration
+python train.py debug=default
+
+# run 1 train, val and test loop, using only 1 batch
+python train.py debug=fdr
+
+# print execution time profiling
+python train.py debug=profiler
+
+# try overfitting to 1 batch
+python train.py debug=overfit
+
+# raise exception if there are any numerical anomalies in tensors, like NaN or +/-inf
+python train.py +trainer.detect_anomaly=true
+
+# use only 20% of the data
+python train.py +trainer.limit_train_batches=0.2 \
++trainer.limit_val_batches=0.2 +trainer.limit_test_batches=0.2
+```
+
+> **Note**: Visit [configs/debug/](configs/debug/) for different debugging configs.
+
+</details>
+
+<details>
+<summary><b>Resume training from checkpoint</b></summary>
+
+```yaml
+python train.py ckpt_path="/path/to/ckpt/name.ckpt"
+```
+
+> **Note**: Checkpoint can be either path or URL.
+
+> **Note**: Currently loading ckpt doesn't resume logger experiment, but it will be supported in future Lightning release.
+
+</details>
+
+<details>
+<summary><b>Evaluate checkpoint on test dataset</b></summary>
+
+```yaml
+python eval.py ckpt_path="/path/to/ckpt/name.ckpt"
+```
+
+> **Note**: Checkpoint can be either path or URL.
+
+</details>
+
+<details>
+<summary><b>Create a sweep over hyperparameters</b></summary>
+
+```bash
+# this will run 6 experiments one after the other,
+# each with different combination of batch_size and learning rate
+python train.py -m data.batch_size=32,64,128 model.lr=0.001,0.0005
+```
+
+> **Note**: Hydra composes configs lazily at job launch time. If you change code or configs after launching a job/sweep, the final composed configs might be impacted.
+
+</details>
+
+<details>
+<summary><b>Create a sweep over hyperparameters with Optuna</b></summary>
+
+```bash
+# this will run hyperparameter search defined in `configs/hparams_search/mnist_optuna.yaml`
+# over chosen experiment config
+python train.py -m hparams_search=mnist_optuna experiment=example
+```
+
+> **Note**: Using [Optuna Sweeper](https://hydra.cc/docs/next/plugins/optuna_sweeper) doesn't require you to add any boilerplate to your code, everything is defined in a [single config file](configs/hparams_search/mnist_optuna.yaml).
+
+> **Warning**: Optuna sweeps are not failure-resistant (if one job crashes then the whole sweep crashes).
+
+</details>
+
+<details>
+<summary><b>Execute all experiments from folder</b></summary>
+
+```bash
+python train.py -m 'experiment=glob(*)'
+```
+
+> **Note**: Hydra provides special syntax for controlling behavior of multiruns. Learn more [here](https://hydra.cc/docs/next/tutorials/basic/running_your_app/multi-run). The command above executes all experiments from [configs/experiment/](configs/experiment/).
+
+</details>
+
+<details>
+<summary><b>Execute run for multiple different seeds</b></summary>
+
+```bash
+python train.py -m seed=1,2,3,4,5 trainer.deterministic=True logger=csv tags=["benchmark"]
+```
+
+> **Note**: `trainer.deterministic=True` makes pytorch more deterministic but impacts the performance.
+
+</details>
+
+<details>
+<summary><b>Execute sweep on a remote AWS cluster</b></summary>
+
+> **Note**: This should be achievable with simple config using [Ray AWS launcher for Hydra](https://hydra.cc/docs/next/plugins/ray_launcher). Example is not implemented in this template.
+
+</details>
+
+<!-- <details>
+<summary><b>Execute sweep on a SLURM cluster</b></summary>
+
+> This should be achievable with either [the right lightning trainer flags](https://pytorch-lightning.readthedocs.io/en/latest/clouds/cluster.html?highlight=SLURM#slurm-managed-cluster) or simple config using [Submitit launcher for Hydra](https://hydra.cc/docs/plugins/submitit_launcher). Example is not yet implemented in this template.
+
+</details> -->
+
+<details>
+<summary><b>Use Hydra tab completion</b></summary>
+
+> **Note**: Hydra allows you to autocomplete config argument overrides in shell as you write them, by pressing `tab` key. Read the [docs](https://hydra.cc/docs/tutorials/basic/running_your_app/tab_completion).
+
+</details>
+
+<details>
+<summary><b>Apply pre-commit hooks</b></summary>
+
+```bash
+pre-commit run -a
+```
+
+> **Note**: Apply pre-commit hooks to do things like auto-formatting code and configs, performing code analysis or removing output from jupyter notebooks. See [# Best Practices](#best-practices) for more.
+
+Update pre-commit hook versions in `.pre-commit-config.yaml` with:
+
+```bash
+pre-commit autoupdate
+```
+
+</details>
+
+<details>
+<summary><b>Run tests</b></summary>
+
+```bash
+# run all tests
+pytest
+
+# run tests from specific file
+pytest tests/test_train.py
+
+# run all tests except the ones marked as slow
+pytest -k "not slow"
+```
+
+</details>
+
+<details>
+<summary><b>Use tags</b></summary>
+
+Each experiment should be tagged in order to easily filter them across files or in logger UI:
+
+```bash
+python train.py tags=["mnist","experiment_X"]
+```
+
+> **Note**: You might need to escape the bracket characters in your shell with `python train.py tags=\["mnist","experiment_X"\]`.
+
+If no tags are provided, you will be asked to input them from command line:
+
+```bash
+>>> python train.py tags=[]
+[2022-07-11 15:40:09,358][src.utils.utils][INFO] - Enforcing tags! <cfg.extras.enforce_tags=True>
+[2022-07-11 15:40:09,359][src.utils.rich_utils][WARNING] - No tags provided in config. Prompting user to input tags...
+Enter a list of comma separated tags (dev):
+```
+
+If no tags are provided for multirun, an error will be raised:
+
+```bash
+>>> python train.py -m +x=1,2,3 tags=[]
+ValueError: Specify tags before launching a multirun!
+```
+
+> **Note**: Appending lists from command line is currently not supported in hydra :(
+
+</details>
+
+<br>
+
+## ❤️  Contributions
+
+This project exists thanks to all the people who contribute.
+
+![Contributors](https://readme-contributors.now.sh/ashleve/lightning-hydra-template?extension=jpg&width=400&aspectRatio=1)
+
+Have a question? Found a bug? Missing a specific feature? Feel free to file a new issue, discussion or PR with respective title and description.
+
+Before making an issue, please verify that:
+
+- The problem still exists on the current `main` branch.
+- Your python dependencies are updated to recent versions.
+
+Suggestions for improvements are always welcome!
+
+<br>
+
+## How It Works
+
+All PyTorch Lightning modules are dynamically instantiated from module paths specified in config. Example model config:
+
+```yaml
+_target_: src.models.mnist_model.MNISTLitModule
+lr: 0.001
+net:
+  _target_: src.models.components.simple_dense_net.SimpleDenseNet
+  input_size: 784
+  lin1_size: 256
+  lin2_size: 256
+  lin3_size: 256
+  output_size: 10
+```
+
+Using this config we can instantiate the object with the following line:
+
+```python
+model = hydra.utils.instantiate(config.model)
+```
+
+This allows you to easily iterate over new models! Every time you create a new one, just specify its module path and parameters in appropriate config file. <br>
+
+Switch between models and datamodules with command line arguments:
+
+```bash
+python train.py model=mnist
+```
+
+Example pipeline managing the instantiation logic: [src/train.py](src/train.py).
+
+<br>
+
+## Main Config
+
+Location: [configs/train.yaml](configs/train.yaml) <br>
+Main project config contains default training configuration.<br>
+It determines how config is composed when simply executing command `python train.py`.<br>
+
+<details>
+<summary><b>Show main project config</b></summary>
+
+```yaml
+# order of defaults determines the order in which configs override each other
+defaults:
+  - _self_
+  - data: mnist.yaml
+  - model: mnist.yaml
+  - callbacks: default.yaml
+  - logger: null # set logger here or use command line (e.g. `python train.py logger=csv`)
+  - trainer: default.yaml
+  - paths: default.yaml
+  - extras: default.yaml
+  - hydra: default.yaml
+
+  # experiment configs allow for version control of specific hyperparameters
+  # e.g. best hyperparameters for given model and datamodule
+  - experiment: null
+
+  # config for hyperparameter optimization
+  - hparams_search: null
+
+  # optional local config for machine/user specific settings
+  # it's optional since it doesn't need to exist and is excluded from version control
+  - optional local: default.yaml
+
+  # debugging config (enable through command line, e.g. `python train.py debug=default)
+  - debug: null
+
+# task name, determines output directory path
+task_name: "train"
+
+# tags to help you identify your experiments
+# you can overwrite this in experiment configs
+# overwrite from command line with `python train.py tags="[first_tag, second_tag]"`
+# appending lists from command line is currently not supported :(
+# https://github.com/facebookresearch/hydra/issues/1547
+tags: ["dev"]
+
+# set False to skip model training
+train: True
+
+# evaluate on test set, using best model weights achieved during training
+# lightning chooses best weights based on the metric specified in checkpoint callback
+test: True
+
+# simply provide checkpoint path to resume training
+ckpt_path: null
+
+# seed for random number generators in pytorch, numpy and python.random
+seed: null
+```
+
+</details>
+
+<br>
+
+## Experiment Config
+
+Location: [configs/experiment](configs/experiment)<br>
+Experiment configs allow you to overwrite parameters from main config.<br>
+For example, you can use them to version control best hyperparameters for each combination of model and dataset.
+
+<details>
+<summary><b>Show example experiment config</b></summary>
+
+```yaml
+# @package _global_
+
+# to execute this experiment run:
+# python train.py experiment=example
+
+defaults:
+  - override /data: mnist.yaml
+  - override /model: mnist.yaml
+  - override /callbacks: default.yaml
+  - override /trainer: default.yaml
+
+# all parameters below will be merged with parameters from default configurations set above
+# this allows you to overwrite only specified parameters
+
+tags: ["mnist", "simple_dense_net"]
+
+seed: 12345
+
+trainer:
+  min_epochs: 10
+  max_epochs: 10
+  gradient_clip_val: 0.5
+
+model:
+  optimizer:
+    lr: 0.002
+  net:
+    lin1_size: 128
+    lin2_size: 256
+    lin3_size: 64
+
+data:
+  batch_size: 64
+
+logger:
+  wandb:
+    tags: ${tags}
+    group: "mnist"
+```
+
+</details>
+
+<br>
+
+## Workflow
+
+**Basic workflow**
+
+1. Write your PyTorch Lightning module (see [models/mnist_module.py](src/models/mnist_module.py) for example)
+2. Write your PyTorch Lightning datamodule (see [data/mnist_datamodule.py](src/data/mnist_datamodule.py) for example)
+3. Write your experiment config, containing paths to model and datamodule
+4. Run training with chosen experiment config:
+   ```bash
+   python src/train.py experiment=experiment_name.yaml
+   ```
+
+**Experiment design**
+
+_Say you want to execute many runs to plot how accuracy changes in respect to batch size._
+
+1. Execute the runs with some config parameter that allows you to identify them easily, like tags:
+
+   ```bash
+   python train.py -m logger=csv data.batch_size=16,32,64,128 tags=["batch_size_exp"]
+   ```
+
+2. Write a script or notebook that searches over the `logs/` folder and retrieves csv logs from runs containing given tags in config. Plot the results.
+
+<br>
+
+## Logs
+
+Hydra creates new output directory for every executed run.
+
+Default logging structure:
+
+```
+├── logs
+│   ├── task_name
+│   │   ├── runs                        # Logs generated by single runs
+│   │   │   ├── YYYY-MM-DD_HH-MM-SS       # Datetime of the run
+│   │   │   │   ├── .hydra                  # Hydra logs
+│   │   │   │   ├── csv                     # Csv logs
+│   │   │   │   ├── wandb                   # Weights&Biases logs
+│   │   │   │   ├── checkpoints             # Training checkpoints
+│   │   │   │   └── ...                     # Any other thing saved during training
+│   │   │   └── ...
+│   │   │
+│   │   └── multiruns                   # Logs generated by multiruns
+│   │       ├── YYYY-MM-DD_HH-MM-SS       # Datetime of the multirun
+│   │       │   ├──1                        # Multirun job number
+│   │       │   ├──2
+│   │       │   └── ...
+│   │       └── ...
+│   │
+│   └── debugs                          # Logs generated when debugging config is attached
+│       └── ...
+```
+
+</details>
+
+You can change this structure by modifying paths in [hydra configuration](configs/hydra).
+
+<br>
+
+## Experiment Tracking
+
+PyTorch Lightning supports many popular logging frameworks: [Weights&Biases](https://www.wandb.com/), [Neptune](https://neptune.ai/), [Comet](https://www.comet.ml/), [MLFlow](https://mlflow.org), [Tensorboard](https://www.tensorflow.org/tensorboard/).
+
+These tools help you keep track of hyperparameters and output metrics and allow you to compare and visualize results. To use one of them simply complete its configuration in [configs/logger](configs/logger) and run:
+
+```bash
+python train.py logger=logger_name
+```
+
+You can use many of them at once (see [configs/logger/many_loggers.yaml](configs/logger/many_loggers.yaml) for example).
+
+You can also write your own logger.
+
+Lightning provides convenient method for logging custom metrics from inside LightningModule. Read the [docs](https://pytorch-lightning.readthedocs.io/en/latest/extensions/logging.html#automatic-logging) or take a look at [MNIST example](src/models/mnist_module.py).
+
+<br>
+
+## Tests
+
+Template comes with generic tests implemented with `pytest`.
+
+```bash
+# run all tests
+pytest
+
+# run tests from specific file
+pytest tests/test_train.py
+
+# run all tests except the ones marked as slow
+pytest -k "not slow"
+```
+
+Most of the implemented tests don't check for any specific output - they exist to simply verify that executing some commands doesn't end up in throwing exceptions. You can execute them once in a while to speed up the development.
+
+Currently, the tests cover cases like:
+
+- running 1 train, val and test step
+- running 1 epoch on 1% of data, saving ckpt and resuming for the second epoch
+- running 2 epochs on 1% of data, with DDP simulated on CPU
+
+And many others. You should be able to modify them easily for your use case.
+
+There is also `@RunIf` decorator implemented, that allows you to run tests only if certain conditions are met, e.g. GPU is available or system is not windows. See the [examples](tests/test_train.py).
+
+<br>
+
+## Hyperparameter Search
+
+You can define hyperparameter search by adding new config file to [configs/hparams_search](configs/hparams_search).
+
+<details>
+<summary><b>Show example hyperparameter search config</b></summary>
+
+```yaml
+# @package _global_
+
+defaults:
+  - override /hydra/sweeper: optuna
+
+# choose metric which will be optimized by Optuna
+# make sure this is the correct name of some metric logged in lightning module!
+optimized_metric: "val/acc_best"
+
+# here we define Optuna hyperparameter search
+# it optimizes for value returned from function with @hydra.main decorator
+hydra:
+  sweeper:
+    _target_: hydra_plugins.hydra_optuna_sweeper.optuna_sweeper.OptunaSweeper
+
+    # 'minimize' or 'maximize' the objective
+    direction: maximize
+
+    # total number of runs that will be executed
+    n_trials: 20
+
+    # choose Optuna hyperparameter sampler
+    # docs: https://optuna.readthedocs.io/en/stable/reference/samplers.html
+    sampler:
+      _target_: optuna.samplers.TPESampler
+      seed: 1234
+      n_startup_trials: 10 # number of random sampling runs before optimization starts
+
+    # define hyperparameter search space
+    params:
+      model.optimizer.lr: interval(0.0001, 0.1)
+      data.batch_size: choice(32, 64, 128, 256)
+      model.net.lin1_size: choice(64, 128, 256)
+      model.net.lin2_size: choice(64, 128, 256)
+      model.net.lin3_size: choice(32, 64, 128, 256)
+```
+
+</details>
+
+Next, execute it with: `python train.py -m hparams_search=mnist_optuna`
+
+Using this approach doesn't require adding any boilerplate to code, everything is defined in a single config file. The only necessary thing is to return the optimized metric value from the launch file.
+
+You can use different optimization frameworks integrated with Hydra, like [Optuna, Ax or Nevergrad](https://hydra.cc/docs/plugins/optuna_sweeper/).
+
+The `optimization_results.yaml` will be available under `logs/task_name/multirun` folder.
+
+This approach doesn't support resuming interrupted search and advanced techniques like prunning - for more sophisticated search and workflows, you should probably write a dedicated optimization task (without multirun feature).
+
+<br>
+
+## Continuous Integration
+
+Template comes with CI workflows implemented in Github Actions:
+
+- `.github/workflows/test.yaml`: running all tests with pytest
+- `.github/workflows/code-quality-main.yaml`: running pre-commits on main branch for all files
+- `.github/workflows/code-quality-pr.yaml`: running pre-commits on pull requests for modified files only
+
+<br>
+
+## Distributed Training
+
+Lightning supports multiple ways of doing distributed training. The most common one is DDP, which spawns separate process for each GPU and averages gradients between them. To learn about other approaches read the [lightning docs](https://lightning.ai/docs/pytorch/latest/advanced/speed.html).
+
+You can run DDP on mnist example with 4 GPUs like this:
+
+```bash
+python train.py trainer=ddp
+```
+
+> **Note**: When using DDP you have to be careful how you write your models - read the [docs](https://lightning.ai/docs/pytorch/latest/advanced/speed.html).
+
+<br>
+
+## Accessing Datamodule Attributes In Model
+
+The simplest way is to pass datamodule attribute directly to model on initialization:
+
+```python
+# ./src/train.py
+datamodule = hydra.utils.instantiate(config.data)
+model = hydra.utils.instantiate(config.model, some_param=datamodule.some_param)
+```
+
+> **Note**: Not a very robust solution, since it assumes all your datamodules have `some_param` attribute available.
+
+Similarly, you can pass a whole datamodule config as an init parameter:
+
+```python
+# ./src/train.py
+model = hydra.utils.instantiate(config.model, dm_conf=config.data, _recursive_=False)
+```
+
+You can also pass a datamodule config parameter to your model through variable interpolation:
+
+```yaml
+# ./configs/model/my_model.yaml
+_target_: src.models.my_module.MyLitModule
+lr: 0.01
+some_param: ${data.some_param}
+```
+
+Another approach is to access datamodule in LightningModule directly through Trainer:
+
+```python
+# ./src/models/mnist_module.py
+def on_train_start(self):
+  self.some_param = self.trainer.datamodule.some_param
+```
+
+> **Note**: This only works after the training starts since otherwise trainer won't be yet available in LightningModule.
+
+<br>
+
+## Best Practices
+
+<details>
+<summary><b>Use Miniconda</b></summary>
+
+It's usually unnecessary to install full anaconda environment, miniconda should be enough (weights around 80MB).
+
+Big advantage of conda is that it allows for installing packages without requiring certain compilers or libraries to be available in the system (since it installs precompiled binaries), so it often makes it easier to install some dependencies e.g. cudatoolkit for GPU support.
+
+It also allows you to access your environments globally which might be more convenient than creating new local environment for every project.
+
+Example installation:
+
+```bash
+wget https://repo.anaconda.com/miniconda/Miniconda3-latest-Linux-x86_64.sh
+bash Miniconda3-latest-Linux-x86_64.sh
+```
+
+Update conda:
+
+```bash
+conda update -n base -c defaults conda
+```
+
+Create new conda environment:
+
+```bash
+conda create -n myenv python=3.10
+conda activate myenv
+```
+
+</details>
+
+<details>
+<summary><b>Use automatic code formatting</b></summary>
+
+Use pre-commit hooks to standardize code formatting of your project and save mental energy.<br>
+Simply install pre-commit package with:
+
+```bash
+pip install pre-commit
+```
+
+Next, install hooks from [.pre-commit-config.yaml](.pre-commit-config.yaml):
+
+```bash
+pre-commit install
+```
+
+After that your code will be automatically reformatted on every new commit.
+
+To reformat all files in the project use command:
+
+```bash
+pre-commit run -a
+```
+
+To update hook versions in [.pre-commit-config.yaml](.pre-commit-config.yaml) use:
+
+```bash
+pre-commit autoupdate
+```
+
+</details>
+
+<details>
+<summary><b>Set private environment variables in .env file</b></summary>
+
+System specific variables (e.g. absolute paths to datasets) should not be under version control or it will result in conflict between different users. Your private keys also shouldn't be versioned since you don't want them to be leaked.<br>
+
+Template contains `.env.example` file, which serves as an example. Create a new file called `.env` (this name is excluded from version control in .gitignore).
+You should use it for storing environment variables like this:
+
+```
+MY_VAR=/home/user/my_system_path
+```
+
+All variables from `.env` are loaded in `train.py` automatically.
+
+Hydra allows you to reference any env variable in `.yaml` configs like this:
+
+```yaml
+path_to_data: ${oc.env:MY_VAR}
+```
+
+</details>
+
+<details>
+<summary><b>Name metrics using '/' character</b></summary>
+
+Depending on which logger you're using, it's often useful to define metric name with `/` character:
+
+```python
+self.log("train/loss", loss)
+```
+
+This way loggers will treat your metrics as belonging to different sections, which helps to get them organised in UI.
+
+</details>
+
+<details>
+<summary><b>Use torchmetrics</b></summary>
+
+Use official [torchmetrics](https://github.com/PytorchLightning/metrics) library to ensure proper calculation of metrics. This is especially important for multi-GPU training!
+
+For example, instead of calculating accuracy by yourself, you should use the provided `Accuracy` class like this:
+
+```python
+from torchmetrics.classification.accuracy import Accuracy
+
+
+class LitModel(LightningModule):
+    def __init__(self)
+        self.train_acc = Accuracy()
+        self.val_acc = Accuracy()
+
+    def training_step(self, batch, batch_idx):
+        ...
+        acc = self.train_acc(predictions, targets)
+        self.log("train/acc", acc)
+        ...
+
+    def validation_step(self, batch, batch_idx):
+        ...
+        acc = self.val_acc(predictions, targets)
+        self.log("val/acc", acc)
+        ...
+```
+
+Make sure to use different metric instance for each step to ensure proper value reduction over all GPU processes.
+
+Torchmetrics provides metrics for most use cases, like F1 score or confusion matrix. Read [documentation](https://torchmetrics.readthedocs.io/en/latest/#more-reading) for more.
+
+</details>
+
+<details>
+<summary><b>Follow PyTorch Lightning style guide</b></summary>
+
+The style guide is available [here](https://pytorch-lightning.readthedocs.io/en/latest/starter/style_guide.html).<br>
+
+1. Be explicit in your init. Try to define all the relevant defaults so that the user doesn’t have to guess. Provide type hints. This way your module is reusable across projects!
+
+   ```python
+   class LitModel(LightningModule):
+       def __init__(self, layer_size: int = 256, lr: float = 0.001):
+   ```
+
+2. Preserve the recommended method order.
+
+   ```python
+   class LitModel(LightningModule):
+
+       def __init__():
+           ...
+
+       def forward():
+           ...
+
+       def training_step():
+           ...
+
+       def training_step_end():
+           ...
+
+       def on_train_epoch_end():
+           ...
+
+       def validation_step():
+           ...
+
+       def validation_step_end():
+           ...
+
+       def on_validation_epoch_end():
+           ...
+
+       def test_step():
+           ...
+
+       def test_step_end():
+           ...
+
+       def on_test_epoch_end():
+           ...
+
+       def configure_optimizers():
+           ...
+
+       def any_extra_hook():
+           ...
+   ```
+
+</details>
+
+<details>
+<summary><b>Version control your data and models with DVC</b></summary>
+
+Use [DVC](https://dvc.org) to version control big files, like your data or trained ML models.<br>
+To initialize the dvc repository:
+
+```bash
+dvc init
+```
+
+To start tracking a file or directory, use `dvc add`:
+
+```bash
+dvc add data/MNIST
+```
+
+DVC stores information about the added file (or a directory) in a special .dvc file named data/MNIST.dvc, a small text file with a human-readable format. This file can be easily versioned like source code with Git, as a placeholder for the original data:
+
+```bash
+git add data/MNIST.dvc data/.gitignore
+git commit -m "Add raw data"
+```
+
+</details>
+
+<details>
+<summary><b>Support installing project as a package</b></summary>
+
+It allows other people to easily use your modules in their own projects.
+Change name of the `src` folder to your project name and complete the `setup.py` file.
+
+Now your project can be installed from local files:
+
+```bash
+pip install -e .
+```
+
+Or directly from git repository:
+
+```bash
+pip install git+git://github.com/YourGithubName/your-repo-name.git --upgrade
+```
+
+So any file can be easily imported into any other file like so:
+
+```python
+from project_name.models.mnist_module import MNISTLitModule
+from project_name.data.mnist_datamodule import MNISTDataModule
+```
+
+</details>
+
+<details>
+<summary><b>Keep local configs out of code versioning</b></summary>
+
+Some configurations are user/machine/installation specific (e.g. configuration of local cluster, or harddrive paths on a specific machine). For such scenarios, a file [configs/local/default.yaml](configs/local/) can be created which is automatically loaded but not tracked by Git.
+
+For example, you can use it for a SLURM cluster config:
+
+```yaml
+# @package _global_
+
+defaults:
+  - override /hydra/launcher@_here_: submitit_slurm
+
+data_dir: /mnt/scratch/data/
+
+hydra:
+  launcher:
+    timeout_min: 1440
+    gpus_per_task: 1
+    gres: gpu:1
+  job:
+    env_set:
+      MY_VAR: /home/user/my/system/path
+      MY_KEY: asdgjhawi8y23ihsghsueity23ihwd
+```
+
+</details>
+
+<br>
+
+## Resources
+
+This template was inspired by:
+
+- [PyTorchLightning/deep-learning-project-template](https://github.com/PyTorchLightning/deep-learning-project-template)
+- [drivendata/cookiecutter-data-science](https://github.com/drivendata/cookiecutter-data-science)
+- [lucmos/nn-template](https://github.com/lucmos/nn-template)
+
+Other useful repositories:
+
+- [jxpress/lightning-hydra-template-vertex-ai](https://github.com/jxpress/lightning-hydra-template-vertex-ai) - lightning-hydra-template integration with Vertex AI hyperparameter tuning and custom training job
+
+</details>
+
+<br>
+
+## License
+
+Lightning-Hydra-Template is licensed under the MIT License.
+
+```
+MIT License
+
+Copyright (c) 2021 ashleve
+
+Permission is hereby granted, free of charge, to any person obtaining a copy
+of this software and associated documentation files (the "Software"), to deal
+in the Software without restriction, including without limitation the rights
+to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+copies of the Software, and to permit persons to whom the Software is
+furnished to do so, subject to the following conditions:
+
+The above copyright notice and this permission notice shall be included in all
+copies or substantial portions of the Software.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+SOFTWARE.
+```
+
+<br>
+<br>
+<br>
+<br>
+
+**DELETE EVERYTHING ABOVE FOR YOUR PROJECT**
+
+______________________________________________________________________
+
+<div align="center">
+
+# Your Project Name
+
+<a href="https://pytorch.org/get-started/locally/"><img alt="PyTorch" src="https://img.shields.io/badge/PyTorch-ee4c2c?logo=pytorch&logoColor=white"></a>
+<a href="https://pytorchlightning.ai/"><img alt="Lightning" src="https://img.shields.io/badge/-Lightning-792ee5?logo=pytorchlightning&logoColor=white"></a>
+<a href="https://hydra.cc/"><img alt="Config: Hydra" src="https://img.shields.io/badge/Config-Hydra-89b8cd"></a>
+<a href="https://github.com/ashleve/lightning-hydra-template"><img alt="Template" src="https://img.shields.io/badge/-Lightning--Hydra--Template-017F2F?style=flat&logo=github&labelColor=gray"></a><br>
+[![Paper](http://img.shields.io/badge/paper-arxiv.1001.2234-B31B1B.svg)](https://www.nature.com/articles/nature14539)
+[![Conference](http://img.shields.io/badge/AnyConference-year-4b44ce.svg)](https://papers.nips.cc/paper/2020)
+
+</div>
+
+## Description
+
+What it does
+
+## Installation
+
+#### Pip
+
+```bash
+# clone project
+git clone https://github.com/YourGithubName/your-repo-name
+cd your-repo-name
+
+# [OPTIONAL] create conda environment
+conda create -n myenv python=3.9
+conda activate myenv
+
+# install pytorch according to instructions
+# https://pytorch.org/get-started/
+
+# install requirements
+pip install -r requirements.txt
+```
+
+#### Conda
+
+```bash
+# clone project
+git clone https://github.com/YourGithubName/your-repo-name
+cd your-repo-name
+
+# create conda environment and install dependencies
+conda env create -f environment.yaml -n myenv
+
+# activate conda environment
+conda activate myenv
+```
+
+## How to run
+
+Train model with default configuration
+
+```bash
+# train on CPU
+python src/train.py trainer=cpu
+
+# train on GPU
+python src/train.py trainer=gpu
+```
+
+Train model with chosen experiment configuration from [configs/experiment/](configs/experiment/)
+
+```bash
+python src/train.py experiment=experiment_name.yaml
+```
+
+You can override any parameter from command line like this
+
+```bash
+python src/train.py trainer.max_epochs=20 data.batch_size=64
+```

configs/__init__.py

@@ -0,0 +1 @@
+# this file is needed here to include configs when building project as a package

configs/callbacks/default.yaml

@@ -0,0 +1,22 @@
+defaults:
+  - model_checkpoint
+  - early_stopping
+  - model_summary
+  - rich_progress_bar
+  - _self_
+
+model_checkpoint:
+  dirpath: ${paths.root_dir}/checkpoints
+  filename: "epoch_{epoch:03d}"
+  monitor: "val/psnr"
+  mode: "max"
+  save_last: True
+  auto_insert_metric_name: False
+
+early_stopping:
+  monitor: "val/psnr"
+  patience: 100
+  mode: "max"
+
+model_summary:
+  max_depth: -1

configs/callbacks/early_stopping.yaml

@@ -0,0 +1,15 @@
+# https://lightning.ai/docs/pytorch/stable/api/lightning.pytorch.callbacks.EarlyStopping.html
+
+early_stopping:
+  _target_: lightning.pytorch.callbacks.EarlyStopping
+  monitor: ??? # quantity to be monitored, must be specified !!!
+  min_delta: 0. # minimum change in the monitored quantity to qualify as an improvement
+  patience: 3 # number of checks with no improvement after which training will be stopped
+  verbose: False # verbosity mode
+  mode: "min" # "max" means higher metric value is better, can be also "min"
+  strict: True # whether to crash the training if monitor is not found in the validation metrics
+  check_finite: True # when set True, stops training when the monitor becomes NaN or infinite
+  stopping_threshold: null # stop training immediately once the monitored quantity reaches this threshold
+  divergence_threshold: null # stop training as soon as the monitored quantity becomes worse than this threshold
+  check_on_train_epoch_end: null # whether to run early stopping at the end of the training epoch
+  # log_rank_zero_only: False  # this keyword argument isn't available in stable version

configs/callbacks/model_checkpoint.yaml

@@ -0,0 +1,17 @@
+# https://lightning.ai/docs/pytorch/stable/api/lightning.pytorch.callbacks.ModelCheckpoint.html
+
+model_checkpoint:
+  _target_: lightning.pytorch.callbacks.ModelCheckpoint
+  dirpath: null # directory to save the model file
+  filename: null # checkpoint filename
+  monitor: null # name of the logged metric which determines when model is improving
+  verbose: False # verbosity mode
+  save_last: null # additionally always save an exact copy of the last checkpoint to a file last.ckpt
+  save_top_k: 1 # save k best models (determined by above metric)
+  mode: "min" # "max" means higher metric value is better, can be also "min"
+  auto_insert_metric_name: True # when True, the checkpoints filenames will contain the metric name
+  save_weights_only: False # if True, then only the model’s weights will be saved
+  every_n_train_steps: null # number of training steps between checkpoints
+  train_time_interval: null # checkpoints are monitored at the specified time interval
+  every_n_epochs: null # number of epochs between checkpoints
+  save_on_train_epoch_end: null # whether to run checkpointing at the end of the training epoch or the end of validation

configs/callbacks/model_summary.yaml

@@ -0,0 +1,5 @@
+# https://lightning.ai/docs/pytorch/stable/api/lightning.pytorch.callbacks.RichModelSummary.html
+
+model_summary:
+  _target_: lightning.pytorch.callbacks.RichModelSummary
+  max_depth: 1 # the maximum depth of layer nesting that the summary will include

configs/callbacks/rich_progress_bar.yaml

@@ -0,0 +1,4 @@
+# https://lightning.ai/docs/pytorch/latest/api/lightning.pytorch.callbacks.RichProgressBar.html
+
+rich_progress_bar:
+  _target_: lightning.pytorch.callbacks.RichProgressBar

configs/data/swim.yaml

@@ -0,0 +1,4 @@
+_target_: swim.data.swim_data.SwimDataModule
+root_dir: /home/qninh/projects/swim_/datasets/swim_data
+batch_size: 4
+img_size: 64

configs/debug/default.yaml

@@ -0,0 +1,35 @@
+# @package _global_
+
+# default debugging setup, runs 1 full epoch
+# other debugging configs can inherit from this one
+
+# overwrite task name so debugging logs are stored in separate folder
+task_name: "debug"
+
+# disable callbacks and loggers during debugging
+callbacks: null
+logger: null
+
+extras:
+  ignore_warnings: False
+  enforce_tags: False
+
+# sets level of all command line loggers to 'DEBUG'
+# https://hydra.cc/docs/tutorials/basic/running_your_app/logging/
+hydra:
+  job_logging:
+    root:
+      level: DEBUG
+
+  # use this to also set hydra loggers to 'DEBUG'
+  # verbose: True
+
+trainer:
+  max_epochs: 1
+  accelerator: cpu # debuggers don't like gpus
+  devices: 1 # debuggers don't like multiprocessing
+  detect_anomaly: true # raise exception if NaN or +/-inf is detected in any tensor
+
+data:
+  num_workers: 0 # debuggers don't like multiprocessing
+  pin_memory: False # disable gpu memory pin

configs/debug/fdr.yaml

@@ -0,0 +1,9 @@
+# @package _global_
+
+# runs 1 train, 1 validation and 1 test step
+
+defaults:
+  - default
+
+trainer:
+  fast_dev_run: true

configs/debug/limit.yaml

@@ -0,0 +1,12 @@
+# @package _global_
+
+# uses only 1% of the training data and 5% of validation/test data
+
+defaults:
+  - default
+
+trainer:
+  max_epochs: 3
+  limit_train_batches: 0.01
+  limit_val_batches: 0.05
+  limit_test_batches: 0.05

configs/debug/overfit.yaml

@@ -0,0 +1,13 @@
+# @package _global_
+
+# overfits to 3 batches
+
+defaults:
+  - default
+
+trainer:
+  max_epochs: 20
+  overfit_batches: 3
+
+# model ckpt and early stopping need to be disabled during overfitting
+callbacks: null

configs/debug/profiler.yaml

@@ -0,0 +1,12 @@
+# @package _global_
+
+# runs with execution time profiling
+
+defaults:
+  - default
+
+trainer:
+  max_epochs: 1
+  profiler: "simple"
+  # profiler: "advanced"
+  # profiler: "pytorch"

configs/eval.yaml

@@ -0,0 +1,18 @@
+# @package _global_
+
+defaults:
+  - _self_
+  - data: mnist # choose datamodule with `test_dataloader()` for evaluation
+  - model: mnist
+  - logger: null
+  - trainer: default
+  - paths: default
+  - extras: default
+  - hydra: default
+
+task_name: "eval"
+
+tags: ["dev"]
+
+# passing checkpoint path is necessary for evaluation
+ckpt_path: ???

configs/experiment/example.yaml

@@ -0,0 +1,41 @@
+# @package _global_
+
+# to execute this experiment run:
+# python train.py experiment=example
+
+defaults:
+  - override /data: mnist
+  - override /model: mnist
+  - override /callbacks: default
+  - override /trainer: default
+
+# all parameters below will be merged with parameters from default configurations set above
+# this allows you to overwrite only specified parameters
+
+tags: ["mnist", "simple_dense_net"]
+
+seed: 12345
+
+trainer:
+  min_epochs: 10
+  max_epochs: 10
+  gradient_clip_val: 0.5
+
+model:
+  optimizer:
+    lr: 0.002
+  net:
+    lin1_size: 128
+    lin2_size: 256
+    lin3_size: 64
+  compile: false
+
+data:
+  batch_size: 64
+
+logger:
+  wandb:
+    tags: ${tags}
+    group: "mnist"
+  aim:
+    experiment: "mnist"

configs/extras/default.yaml

@@ -0,0 +1,8 @@
+# disable python warnings if they annoy you
+ignore_warnings: False
+
+# ask user for tags if none are provided in the config
+enforce_tags: True
+
+# pretty print config tree at the start of the run using Rich library
+print_config: True

configs/hparams_search/mnist_optuna.yaml

@@ -0,0 +1,52 @@
+# @package _global_
+
+# example hyperparameter optimization of some experiment with Optuna:
+# python train.py -m hparams_search=mnist_optuna experiment=example
+
+defaults:
+  - override /hydra/sweeper: optuna
+
+# choose metric which will be optimized by Optuna
+# make sure this is the correct name of some metric logged in lightning module!
+optimized_metric: "val/acc_best"
+
+# here we define Optuna hyperparameter search
+# it optimizes for value returned from function with @hydra.main decorator
+# docs: https://hydra.cc/docs/next/plugins/optuna_sweeper
+hydra:
+  mode: "MULTIRUN" # set hydra to multirun by default if this config is attached
+
+  sweeper:
+    _target_: hydra_plugins.hydra_optuna_sweeper.optuna_sweeper.OptunaSweeper
+
+    # storage URL to persist optimization results
+    # for example, you can use SQLite if you set 'sqlite:///example.db'
+    storage: null
+
+    # name of the study to persist optimization results
+    study_name: null
+
+    # number of parallel workers
+    n_jobs: 1
+
+    # 'minimize' or 'maximize' the objective
+    direction: maximize
+
+    # total number of runs that will be executed
+    n_trials: 20
+
+    # choose Optuna hyperparameter sampler
+    # you can choose bayesian sampler (tpe), random search (without optimization), grid sampler, and others
+    # docs: https://optuna.readthedocs.io/en/stable/reference/samplers.html
+    sampler:
+      _target_: optuna.samplers.TPESampler
+      seed: 1234
+      n_startup_trials: 10 # number of random sampling runs before optimization starts
+
+    # define hyperparameter search space
+    params:
+      model.optimizer.lr: interval(0.0001, 0.1)
+      data.batch_size: choice(32, 64, 128, 256)
+      model.net.lin1_size: choice(64, 128, 256)
+      model.net.lin2_size: choice(64, 128, 256)
+      model.net.lin3_size: choice(32, 64, 128, 256)

configs/hydra/default.yaml

@@ -0,0 +1,19 @@
+# https://hydra.cc/docs/configure_hydra/intro/
+
+# enable color logging
+defaults:
+  - override hydra_logging: colorlog
+  - override job_logging: colorlog
+
+# output directory, generated dynamically on each run
+run:
+  dir: ${paths.log_dir}/${task_name}/runs/${now:%Y-%m-%d}_${now:%H-%M-%S}
+sweep:
+  dir: ${paths.log_dir}/${task_name}/multiruns/${now:%Y-%m-%d}_${now:%H-%M-%S}
+  subdir: ${hydra.job.num}
+
+job_logging:
+  handlers:
+    file:
+      # Incorporates fix from https://github.com/facebookresearch/hydra/pull/2242
+      filename: ${hydra.runtime.output_dir}/${task_name}.log

configs/logger/aim.yaml

@@ -0,0 +1,28 @@
+# https://aimstack.io/
+
+# example usage in lightning module:
+# https://github.com/aimhubio/aim/blob/main/examples/pytorch_lightning_track.py
+
+# open the Aim UI with the following command (run in the folder containing the `.aim` folder):
+# `aim up`
+
+aim:
+  _target_: aim.pytorch_lightning.AimLogger
+  repo: ${paths.root_dir} # .aim folder will be created here
+  # repo: "aim://ip_address:port" # can instead provide IP address pointing to Aim remote tracking server which manages the repo, see https://aimstack.readthedocs.io/en/latest/using/remote_tracking.html#
+
+  # aim allows to group runs under experiment name
+  experiment: null # any string, set to "default" if not specified
+
+  train_metric_prefix: "train/"
+  val_metric_prefix: "val/"
+  test_metric_prefix: "test/"
+
+  # sets the tracking interval in seconds for system usage metrics (CPU, GPU, memory, etc.)
+  system_tracking_interval: 10 # set to null to disable system metrics tracking
+
+  # enable/disable logging of system params such as installed packages, git info, env vars, etc.
+  log_system_params: true
+
+  # enable/disable tracking console logs (default value is true)
+  capture_terminal_logs: false # set to false to avoid infinite console log loop issue https://github.com/aimhubio/aim/issues/2550

configs/logger/comet.yaml

@@ -0,0 +1,12 @@
+# https://www.comet.ml
+
+comet:
+  _target_: lightning.pytorch.loggers.comet.CometLogger
+  api_key: ${oc.env:COMET_API_TOKEN} # api key is loaded from environment variable
+  save_dir: "${paths.output_dir}"
+  project_name: "lightning-hydra-template"
+  rest_api_key: null
+  # experiment_name: ""
+  experiment_key: null # set to resume experiment
+  offline: False
+  prefix: ""

configs/logger/csv.yaml

@@ -0,0 +1,7 @@
+# csv logger built in lightning
+
+csv:
+  _target_: lightning.pytorch.loggers.csv_logs.CSVLogger
+  save_dir: "${paths.output_dir}"
+  name: "csv/"
+  prefix: ""

configs/logger/many_loggers.yaml

@@ -0,0 +1,9 @@
+# train with many loggers at once
+
+defaults:
+  # - comet
+  - csv
+  # - mlflow
+  # - neptune
+  - tensorboard
+  - wandb

configs/logger/mlflow.yaml

@@ -0,0 +1,12 @@
+# https://mlflow.org
+
+mlflow:
+  _target_: lightning.pytorch.loggers.mlflow.MLFlowLogger
+  # experiment_name: ""
+  # run_name: ""
+  tracking_uri: ${paths.log_dir}/mlflow/mlruns # run `mlflow ui` command inside the `logs/mlflow/` dir to open the UI
+  tags: null
+  # save_dir: "./mlruns"
+  prefix: ""
+  artifact_location: null
+  # run_id: ""

configs/logger/neptune.yaml

@@ -0,0 +1,9 @@
+# https://neptune.ai
+
+neptune:
+  _target_: lightning.pytorch.loggers.neptune.NeptuneLogger
+  api_key: ${oc.env:NEPTUNE_API_TOKEN} # api key is loaded from environment variable
+  project: username/lightning-hydra-template
+  # name: ""
+  log_model_checkpoints: True
+  prefix: ""

configs/logger/tensorboard.yaml

@@ -0,0 +1,10 @@
+# https://www.tensorflow.org/tensorboard/
+
+tensorboard:
+  _target_: lightning.pytorch.loggers.tensorboard.TensorBoardLogger
+  save_dir: "${paths.output_dir}/tensorboard/"
+  name: null
+  log_graph: False
+  default_hp_metric: True
+  prefix: ""
+  # version: ""

configs/logger/wandb.yaml

@@ -0,0 +1,16 @@
+# https://wandb.ai
+
+wandb:
+  _target_: lightning.pytorch.loggers.wandb.WandbLogger
+  # name: "" # name of the run (normally generated by wandb)
+  save_dir: "${paths.output_dir}"
+  offline: False
+  id: null # pass correct id to resume experiment!
+  anonymous: null # enable anonymous logging
+  project: "swim"
+  log_model: False # upload lightning ckpts
+  prefix: "" # a string to put at the beginning of metric keys
+  # entity: "" # set to name of your wandb team
+  group: ""
+  tags: []
+  job_type: ""

configs/model/autoencoder.yaml

@@ -0,0 +1,11 @@
+_target_: swim.models.autoencoder.Autoencoder
+
+learning_rate: 1e-4
+
+channels: 128
+channel_multipliers: [1, 2, 4]
+n_resnet_blocks: 1
+in_channels: 3
+out_channels: 3
+z_channels: 4
+emb_channels: 4

configs/paths/default.yaml

@@ -0,0 +1,18 @@
+# path to root directory
+# this requires PROJECT_ROOT environment variable to exist
+# you can replace it with "." if you want the root to be the current working directory
+root_dir: ${oc.env:PROJECT_ROOT}
+
+# path to data directory
+data_dir: ${paths.root_dir}/data/
+
+# path to logging directory
+log_dir: ${paths.root_dir}/logs/
+
+# path to output directory, created dynamically by hydra
+# path generation pattern is specified in `configs/hydra/default.yaml`
+# use it to store all files generated during the run, like ckpts and metrics
+output_dir: ${hydra:runtime.output_dir}
+
+# path to working directory
+work_dir: ${hydra:runtime.cwd}

configs/train.yaml

@@ -0,0 +1,49 @@
+# @package _global_
+
+# specify here default configuration
+# order of defaults determines the order in which configs override each other
+defaults:
+  - _self_
+  - data: swim
+  - model: autoencoder
+  - callbacks: default
+  - logger: wandb # set logger here or use command line (e.g. `python train.py logger=tensorboard`)
+  - trainer: gpu
+  - paths: default
+  - extras: default
+  - hydra: default
+
+  # experiment configs allow for version control of specific hyperparameters
+  # e.g. best hyperparameters for given model and datamodule
+  - experiment: null
+
+  # config for hyperparameter optimization
+  - hparams_search: null
+
+  # optional local config for machine/user specific settings
+  # it's optional since it doesn't need to exist and is excluded from version control
+  - optional local: default
+
+  # debugging config (enable through command line, e.g. `python train.py debug=default)
+  - debug: null
+
+# task name, determines output directory path
+task_name: "train"
+
+# tags to help you identify your experiments
+# you can overwrite this in experiment configs
+# overwrite from command line with `python train.py tags="[first_tag, second_tag]"`
+tags: ["dev"]
+
+# set False to skip model training
+train: True
+
+# evaluate on test set, using best model weights achieved during training
+# lightning chooses best weights based on the metric specified in checkpoint callback
+test: True
+
+# simply provide checkpoint path to resume training
+ckpt_path: null
+
+# seed for random number generators in pytorch, numpy and python.random
+seed: 42

configs/trainer/cpu.yaml

@@ -0,0 +1,5 @@
+defaults:
+  - default
+
+accelerator: cpu
+devices: 1

configs/trainer/ddp.yaml

@@ -0,0 +1,9 @@
+defaults:
+  - default
+
+strategy: ddp
+
+accelerator: gpu
+devices: 4
+num_nodes: 1
+sync_batchnorm: True

configs/trainer/ddp_sim.yaml

@@ -0,0 +1,7 @@
+defaults:
+  - default
+
+# simulate DDP on CPU, useful for debugging
+accelerator: cpu
+devices: 2
+strategy: ddp_spawn

configs/trainer/default.yaml

@@ -0,0 +1,19 @@
+_target_: lightning.pytorch.trainer.Trainer
+
+default_root_dir: ${paths.output_dir}
+
+min_epochs: 1 # prevents early stopping
+max_epochs: 10
+
+accelerator: cpu
+devices: 1
+
+# mixed precision for extra speed-up
+# precision: 16
+
+# perform a validation loop every N training epochs
+check_val_every_n_epoch: 1
+
+# set True to to ensure deterministic results
+# makes training slower but gives more reproducibility than just setting seeds
+deterministic: False

configs/trainer/gpu.yaml

@@ -0,0 +1,5 @@
+defaults:
+  - default
+
+accelerator: gpu
+devices: 1

configs/trainer/mps.yaml

@@ -0,0 +1,5 @@
+defaults:
+  - default
+
+accelerator: mps
+devices: 1

environment.yaml

@@ -0,0 +1,45 @@
+# reasons you might want to use `environment.yaml` instead of `requirements.txt`:
+# - pip installs packages in a loop, without ensuring dependencies across all packages
+#   are fulfilled simultaneously, but conda achieves proper dependency control across
+#   all packages
+# - conda allows for installing packages without requiring certain compilers or
+#   libraries to be available in the system, since it installs precompiled binaries
+
+name: myenv
+
+channels:
+  - pytorch
+  - conda-forge
+  - defaults
+
+# it is strongly recommended to specify versions of packages installed through conda
+# to avoid situation when version-unspecified packages install their latest major
+# versions which can sometimes break things
+
+# current approach below keeps the dependencies in the same major versions across all
+# users, but allows for different minor and patch versions of packages where backwards
+# compatibility is usually guaranteed
+
+dependencies:
+  - python=3.10
+  - pytorch=2.*
+  - torchvision=0.*
+  - lightning=2.*
+  - torchmetrics=0.*
+  - hydra-core=1.*
+  - rich=13.*
+  - pre-commit=3.*
+  - pytest=7.*
+
+  # --------- loggers --------- #
+  # - wandb
+  # - neptune-client
+  # - mlflow
+  # - comet-ml
+  # - aim>=3.16.2 # no lower than 3.16.2, see https://github.com/aimhubio/aim/issues/2550
+
+  - pip>=23
+  - pip:
+      - hydra-optuna-sweeper
+      - hydra-colorlog
+      - rootutils

pyproject.toml

@@ -0,0 +1,25 @@
+[tool.pytest.ini_options]
+addopts = [
+  "--color=yes",
+  "--durations=0",
+  "--strict-markers",
+  "--doctest-modules",
+]
+filterwarnings = [
+  "ignore::DeprecationWarning",
+  "ignore::UserWarning",
+]
+log_cli = "True"
+markers = [
+  "slow: slow tests",
+]
+minversion = "6.0"
+testpaths = "tests/"
+
+[tool.coverage.report]
+exclude_lines = [
+    "pragma: nocover",
+    "raise NotImplementedError",
+    "raise NotImplementedError()",
+    "if __name__ == .__main__.:",
+]

requirements.txt

@@ -0,0 +1,24 @@
+# --------- pytorch --------- #
+torch>=2.0.0
+torchvision>=0.15.0
+lightning>=2.0.0
+torchmetrics>=0.11.4
+
+# --------- hydra --------- #
+hydra-core==1.3.2
+hydra-colorlog==1.2.0
+hydra-optuna-sweeper==1.2.0
+
+# --------- loggers --------- #
+wandb
+# neptune-client
+# mlflow
+# comet-ml
+# aim>=3.16.2  # no lower than 3.16.2, see https://github.com/aimhubio/aim/issues/2550
+
+# --------- others --------- #
+rootutils       # standardizing the project root setup
+pre-commit      # hooks for applying linters on commit
+rich            # beautiful text formatting in terminal
+pytest          # tests
+# sh            # for running bash commands in some tests (linux/macos only)

scripts/schedule.sh

@@ -0,0 +1,7 @@
+#!/bin/bash
+# Schedule execution of many runs
+# Run from root folder with: bash scripts/schedule.sh
+
+python src/train.py trainer.max_epochs=5 logger=csv
+
+python src/train.py trainer.max_epochs=10 logger=csv

setup.py

@@ -0,0 +1,21 @@
+#!/usr/bin/env python
+
+from setuptools import find_packages, setup
+
+setup(
+    name="src",
+    version="0.0.1",
+    description="Describe Your Cool Project",
+    author="",
+    author_email="",
+    url="https://github.com/user/project",
+    install_requires=["lightning", "hydra-core"],
+    packages=find_packages(),
+    # use this to customize global commands available in the terminal after installing the package
+    entry_points={
+        "console_scripts": [
+            "train_command = src.train:main",
+            "eval_command = src.eval:main",
+        ]
+    },
+)

swim/data/swim_data.py

@@ -0,0 +1,145 @@
+from typing import Literal, List
+
+import os
+import json
+import torch
+import torchvision.transforms as T
+from torch.utils.data import Dataset, DataLoader
+from PIL import Image
+from lightning import LightningDataModule
+
+
+class SwimDataset(Dataset):
+    def __init__(
+        self,
+        root_dir: str = "./datasets/swim_data",
+        split: Literal["train", "val"] = "train",
+        img_size: int = 512,
+    ):
+        super().__init__()
+        self.root_dir = root_dir
+        self.split_dir = os.path.join(root_dir, split)
+        self.img_size = img_size
+
+        if split == "train":
+            self.transform = T.Compose(
+                [
+                    T.Resize(img_size),  # smaller edge of image resized to img_size
+                    T.RandomCrop(img_size),  # get a random crop of img_size x img_size
+                    T.RandomHorizontalFlip(),
+                    T.ToTensor(),
+                    T.Normalize((0.5, 0.5, 0.5), (0.5, 0.5, 0.5)),
+                ]
+            )
+        elif split == "val":
+            self.transform = T.Compose(
+                [
+                    T.Resize(img_size),
+                    T.CenterCrop(img_size),
+                    T.ToTensor(),
+                    T.Normalize((0.5, 0.5, 0.5), (0.5, 0.5, 0.5)),
+                ]
+            )
+
+        with open(os.path.join(self.split_dir, "labels.json"), "r") as f:
+            self.data = json.load(f)
+
+        # filter out images that are both at night and have adverse weather conditions
+        self.data = [
+            img
+            for img in self.data
+            if not (img["timeofday"] == "night" and img["weather"] != "clear")
+        ]
+
+    def __len__(self):
+        return len(self.data) // 100
+
+    def __getitem__(self, idx):
+        data = self.data[idx]
+
+        # load image
+        img_path = os.path.join(self.split_dir, "images", data["name"])
+        img = Image.open(img_path).convert("RGB")
+        img = self.transform(img)
+
+        # load style
+        if data["weather"] != "clear":
+            style_name = data["weather"]
+        elif data["timeofday"] == "night":
+            style_name = "night"
+        else:
+            style_name = "clear"
+
+        # true if image has any styles
+        style_flag = style_name != "clear"
+
+        # one-hot encode style
+        style = torch.zeros(4)
+
+        if style_flag:
+            style[self.get_stylenames().index(style_name)] = 1
+
+        return {
+            "image": img,
+            "style": style,
+            "style_flag": style_flag,
+        }
+
+    def get_stylenames(self) -> List[str]:
+        return ["rain", "snow", "fog", "night"]
+
+
+class SwimDataModule(LightningDataModule):
+    def __init__(
+        self,
+        root_dir: str = "./datasets/swim_data",
+        batch_size: int = 1,
+        img_size: int = 512,
+    ):
+        super().__init__()
+        self.root_dir = root_dir
+        self.img_size = img_size
+        self.batch_size = batch_size
+
+    def setup(self, stage=None):
+        if stage == "fit" or stage is None:
+            self.train_dataset = SwimDataset(
+                root_dir=self.root_dir, split="train", img_size=self.img_size
+            )
+            self.val_dataset = SwimDataset(
+                root_dir=self.root_dir, split="val", img_size=self.img_size
+            )
+
+    def train_dataloader(self):
+        return DataLoader(
+            self.train_dataset,
+            batch_size=self.batch_size,
+            shuffle=True,
+            num_workers=4,
+            collate_fn=self.custom_collate_fn,
+        )
+
+    def val_dataloader(self):
+        return DataLoader(
+            self.val_dataset,
+            batch_size=self.batch_size,
+            shuffle=False,
+            num_workers=4,
+            collate_fn=self.custom_collate_fn,
+        )
+
+    def test_dataloader(self):
+        return DataLoader(
+            self.val_dataset,
+            batch_size=1,
+            shuffle=False,
+            num_workers=4,
+            collate_fn=self.custom_collate_fn,
+        )
+
+    @staticmethod
+    def custom_collate_fn(batch):
+        images = torch.stack([item["image"] for item in batch])
+        styles = torch.stack([item["style"] for item in batch])
+        style_flags = [item["style_flag"] for item in batch]
+        return {"images": images, "styles": styles, "style_flags": style_flags}

swim/eval.py

@@ -0,0 +1,99 @@
+from typing import Any, Dict, List, Tuple
+
+import hydra
+import rootutils
+from lightning import LightningDataModule, LightningModule, Trainer
+from lightning.pytorch.loggers import Logger
+from omegaconf import DictConfig
+
+rootutils.setup_root(__file__, indicator=".project-root", pythonpath=True)
+# ------------------------------------------------------------------------------------ #
+# the setup_root above is equivalent to:
+# - adding project root dir to PYTHONPATH
+#       (so you don't need to force user to install project as a package)
+#       (necessary before importing any local modules e.g. `from src import utils`)
+# - setting up PROJECT_ROOT environment variable
+#       (which is used as a base for paths in "configs/paths/default.yaml")
+#       (this way all filepaths are the same no matter where you run the code)
+# - loading environment variables from ".env" in root dir
+#
+# you can remove it if you:
+# 1. either install project as a package or move entry files to project root dir
+# 2. set `root_dir` to "." in "configs/paths/default.yaml"
+#
+# more info: https://github.com/ashleve/rootutils
+# ------------------------------------------------------------------------------------ #
+
+from swim.utils import (
+    RankedLogger,
+    extras,
+    instantiate_loggers,
+    log_hyperparameters,
+    task_wrapper,
+)
+
+log = RankedLogger(__name__, rank_zero_only=True)
+
+
+@task_wrapper
+def evaluate(cfg: DictConfig) -> Tuple[Dict[str, Any], Dict[str, Any]]:
+    """Evaluates given checkpoint on a datamodule testset.
+
+    This method is wrapped in optional @task_wrapper decorator, that controls the behavior during
+    failure. Useful for multiruns, saving info about the crash, etc.
+
+    :param cfg: DictConfig configuration composed by Hydra.
+    :return: Tuple[dict, dict] with metrics and dict with all instantiated objects.
+    """
+    assert cfg.ckpt_path
+
+    log.info(f"Instantiating datamodule <{cfg.data._target_}>")
+    datamodule: LightningDataModule = hydra.utils.instantiate(cfg.data)
+
+    log.info(f"Instantiating model <{cfg.model._target_}>")
+    model: LightningModule = hydra.utils.instantiate(cfg.model)
+
+    log.info("Instantiating loggers...")
+    logger: List[Logger] = instantiate_loggers(cfg.get("logger"))
+
+    log.info(f"Instantiating trainer <{cfg.trainer._target_}>")
+    trainer: Trainer = hydra.utils.instantiate(cfg.trainer, logger=logger)
+
+    object_dict = {
+        "cfg": cfg,
+        "datamodule": datamodule,
+        "model": model,
+        "logger": logger,
+        "trainer": trainer,
+    }
+
+    if logger:
+        log.info("Logging hyperparameters!")
+        log_hyperparameters(object_dict)
+
+    log.info("Starting testing!")
+    trainer.test(model=model, datamodule=datamodule, ckpt_path=cfg.ckpt_path)
+
+    # for predictions use trainer.predict(...)
+    # predictions = trainer.predict(model=model, dataloaders=dataloaders, ckpt_path=cfg.ckpt_path)
+
+    metric_dict = trainer.callback_metrics
+
+    return metric_dict, object_dict
+
+
+@hydra.main(version_base="1.3", config_path="../configs", config_name="eval.yaml")
+def main(cfg: DictConfig) -> None:
+    """Main entry point for evaluation.
+
+    :param cfg: DictConfig configuration composed by Hydra.
+    """
+    # apply extra utilities
+    # (e.g. ask for tags if none are provided in cfg, print cfg tree, etc.)
+    extras(cfg)
+
+    evaluate(cfg)
+
+
+if __name__ == "__main__":
+    main()