README.md 42.2 KB
Newer Older
Mike Klingbiel's avatar
Mike Klingbiel committed
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755 756 757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 774 775 776 777 778 779 780 781 782 783 784 785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804 805 806 807 808 809 810 811 812 813 814 815 816 817 818 819 820 821 822 823 824 825 826 827 828 829 830 831 832 833 834 835 836 837 838 839 840 841 842 843 844 845 846 847 848 849 850 851 852 853 854 855 856 857 858 859 860 861 862 863 864 865 866 867 868 869 870 871 872 873 874 875 876 877 878 879 880 881 882 883 884 885 886 887 888 889 890 891 892 893 894 895 896 897 898 899 900 901 902 903 904 905 906 907 908 909 910 911 912 913 914 915 916 917 918 919 920 921 922 923 924 925 926 927 928 929 930 931 932 933 934 935 936 937 938 939 940 941 942 943 944 945 946 947 948 949 950 951 952 953 954 955 956 957 958 959 960 961 962 963 964 965 966 967 968 969 970 971 972 973 974 975 976 977 978 979 980 981 982 983 984 985 986 987 988 989 990 991 992 993 994 995 996 997 998 999 1000 1001 1002 1003 1004 1005 1006 1007 1008 1009 1010 1011 1012 1013 1014 1015 1016 1017 1018 1019 1020 1021 1022 1023 1024 1025 1026 1027 1028 1029 1030 1031 1032 1033 1034 1035 1036 1037 1038 1039 1040 1041 1042 1043 1044 1045 1046 1047 1048 1049 1050 1051 1052 1053 1054 1055 1056 1057 1058 1059 1060 1061 1062 1063 1064 1065 1066 1067 1068 1069 1070 1071 1072 1073 1074 1075 1076 1077 1078 1079 1080 1081 1082 1083 1084 1085 1086 1087 1088 1089 1090 1091 1092 1093 1094 1095 1096 1097 1098 1099 1100 1101 1102 1103 1104 1105 1106 1107 1108 1109 1110 1111 1112 1113 1114 1115 1116 1117 1118 1119 1120 1121 1122 1123 1124 1125 1126 1127 1128 1129 1130 1131 1132 1133 1134 1135 1136 1137 1138 1139 1140 1141 1142 1143 1144 1145 1146 1147 1148 1149 1150 1151 1152 1153 1154 1155 1156 1157 1158 1159 1160 1161 1162 1163 1164 1165 1166 1167 1168 1169 1170 1171 1172 1173 1174 1175 1176 1177 1178 1179 1180 1181 1182 1183 1184 1185 1186 1187 1188 1189 1190 1191 1192 1193 1194 1195 1196 1197 1198 1199 1200 1201 1202 1203 1204 1205 1206 1207 1208 1209 1210 1211 1212 1213 1214 1215 1216 1217 1218 1219 1220 1221 1222 1223 1224 1225 1226 1227 1228 1229 1230 1231 1232 1233 1234 1235 1236 1237 1238 1239 1240 1241 1242 1243 1244 1245 1246 1247 1248 1249 1250 1251 1252 1253 1254 1255 1256 1257 1258 1259 1260 1261 1262 1263 1264 1265 1266 1267 1268 1269 1270 1271 1272 1273 1274 1275 1276 1277 1278 1279 1280 1281 1282 1283 1284 1285 1286 1287 1288 1289 1290 1291 1292 1293
<div align="center">

# Lightning-Hydra-Template

[![python](https://img.shields.io/badge/-Python_3.7_%7C_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_1.8+-ee4c2c?logo=pytorch&logoColor=white)](https://pytorch.org/get-started/locally/)
[![lightning](https://img.shields.io/badge/-Lightning_1.6+-792ee5?logo=pytorchlightning&logoColor=white)](https://pytorchlightning.ai/)
[![hydra](https://img.shields.io/badge/Config-Hydra_1.2-89b8cd)](https://hydra.cc/)
[![black](https://img.shields.io/badge/Code%20Style-Black-black.svg?labelColor=gray)](https://black.readthedocs.io/en/stable/)
[![pre-commit](https://img.shields.io/badge/Pre--commit-enabled-brightgreen?logo=pre-commit&logoColor=white)](https://github.com/pre-commit/pre-commit)
[![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)
[![codecov](https://codecov.io/gh/ashleve/lightning-hydra-template/branch/main/graph/badge.svg)](https://codecov.io/gh/ashleve/lightning-hydra-template)
[![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)
[![license](https://img.shields.io/badge/License-MIT-green.svg?labelColor=gray)](https://github.com/ashleve/lightning-hydra-template#license)
[![contributors](https://img.shields.io/github/contributors/ashleve/lightning-hydra-template.svg)](https://github.com/ashleve/lightning-hydra-template/graphs/contributors)

<!-- <a href="https://www.python.org/"><img alt="Python" src="https://img.shields.io/badge/-Python 3.7+-blue?style=for-the-badge&logo=python&logoColor=white"></a> -->

<!-- <a href="https://pytorch.org/get-started/locally/"><img alt="PyTorch" src="https://img.shields.io/badge/-PyTorch 1.8+-ee4c2c?style=for-the-badge&logo=pytorch&logoColor=white"></a>
<a href="https://pytorchlightning.ai/"><img alt="Lightning" src="https://img.shields.io/badge/-Lightning 1.6+-792ee5?style=for-the-badge&logo=pytorchlightning&logoColor=white"></a>
<a href="https://hydra.cc/"><img alt="Config: hydra" src="https://img.shields.io/badge/config-hydra 1.2-89b8cd?style=for-the-badge&labelColor=gray"></a>
<a href="https://black.readthedocs.io/en/stable/"><img alt="Code style: black" src="https://img.shields.io/badge/code%20style-black-black.svg?style=for-the-badge&labelColor=gray"></a> -->

A clean and scalable 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 should use it:**

- Convenient all-in-one technology stack for deep learning prototyping - allows you to rapidly iterate over new models, datasets and tasks on different hardware accelerators like CPUs, multi-GPUs or TPUs.
- A collection of best practices for efficient workflow and reproducibility.
- Thoroughly commented - you can use this repo as a reference and educational resource.

**Why you shouldn't use it:**

- Not fitted for data engineering - the template configuration setup is not designed for building data processing pipelines that depend on each other.
- Limits you as much as pytorch lightning limits you.
- 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).

<br>

<!-- ## Table Of Contents

- [📌&nbsp;&nbsp;Introduction](#introduction)
- [Main Technologies](#main-technologies)
- [Main Ideas Of This Template](#main-ideas-of-this-template)
- [Project Structure](#project-structure)
- [🚀&nbsp;&nbsp;Quickstart](#quickstart)
- [⚡&nbsp;&nbsp;Your Superpowers](#your-superpowers)
- [❤️&nbsp;&nbsp;Contributions](#️contributions)
- [How It Works](#how-it-works)
- [Main Config](#main-config)
- [Experiment Config](#experiment-config)
- [Workflow](#workflow)
- [Logs](#logs)
- [Experiment Tracking](#experiment-tracking)
- [Tests](#tests)
- [Hyperparameter Search](#hyperparameter-search)
- [Continuous Integration](#continuous-integration)
- [Distributed Training](#distributed-training)
- [Best Practices](#best-practices)
- [Resources](#resources)

<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 Of This Template

- **Predefined Structure**: clean and scalable so that work can easily be extended [# Project Structure](#project-structure)
- **Rapid Experimentation**: thanks to hydra command line superpowers | [# Your Superpowers](#your-superpowers)
- **Little Boilerplate**: thanks to automating pipelines with config instantiation | [# How It Works](#how-it-works)
- **Main Configs**: specify default training configuration | [# Main Config](#main-config)
- **Experiment Configs**: override chosen hyperparameters | [# Experiment Config](#experiment-config)
- **Workflow**: comes down to 4 simple steps | [# Workflow](#workflow)
- **Experiment Tracking**: Tensorboard, W&B, Neptune, Comet, MLFlow and CSVLogger | [# Experiment Tracking](#experiment-tracking)
- **Logs**: all logs (checkpoints, configs, etc.) are stored in a dynamically generated folder structure | [# Logs](#logs)
- **Hyperparameter Search**: made easier with Hydra plugins like Optuna Sweeper | [# Hyperparameter Search](#hyperparameter-search)
- **Tests**: generic, easy-to-adapt tests for speeding up the development | [# Tests](#tests)
- **Continuous Integration**: automatically test your repo with Github Actions | [# Continuous Integration](#continuous-integration)
- **Best Practices**: a couple of recommended tools, practices and standards | [# Best Practices](#best-practices)

<br>

## Project Structure

The directory structure of new project looks like this:

```
├── configs                   <- Hydra configuration files
│   ├── callbacks                <- Callbacks configs
│   ├── datamodule               <- Datamodule 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
│   ├── datamodules              <- Lightning datamodules
│   ├── models                   <- Lightning models
│   ├── tasks                    <- Different scenarios, like training, evaluation, etc.
│   ├── 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
├── 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

# log second gradient norm of the model
python train.py +trainer.track_grad_norm=2

# 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 datamodule.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.

</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"]
```

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

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/tasks/train_task.py](src/tasks/train_task.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_
  - datamodule: 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 /datamodule: 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

datamodule:
  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 [datamodules/mnist_datamodule.py](src/datamodules/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 datamodule.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/shell/test_basic_commands.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 hyperparemter 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)
      datamodule.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 advanced techniques like prunning - for more sophisticated search, 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

> **Note**: You need to enable the GitHub Actions from the settings in your repository.

<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://pytorch-lightning.readthedocs.io/en/latest/advanced/multi_gpu.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://pytorch-lightning.readthedocs.io/en/latest/advanced/multi_gpu.html).

<br>

## Accessing Datamodule Attributes In Model

The simplest way is to pass datamodule attribute directly to model on initialization:

```python
# ./src/tasks/train_task.py
datamodule = hydra.utils.instantiate(config.datamodule)
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/tasks/train_task.py
model = hydra.utils.instantiate(config.model, dm_conf=config.datamodule, _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: ${datamodule.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 for GPU environments</b></summary>

Use miniconda for your python environments (it's usually unnecessary to install full anaconda environment, miniconda should be enough).
It makes it easier to install some dependencies, like cudatoolkit for GPU support. It also allows you to access your environments globally.

Example installation:

```bash
wget https://repo.anaconda.com/miniconda/Miniconda3-latest-Linux-x86_64.sh
bash Miniconda3-latest-Linux-x86_64.sh
```

Create new conda environment:

```bash
conda create -n myenv python=3.8
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.

Currently template contains configurations of:

- **black** (python code formatting)
- **isort** (python import sorting)
- **pyupgrade** (upgrading python syntax to newer version)
- **docformatter** (python docstring formatting)
- **flake8** (python pep8 code analysis)
- **prettier** (yaml formatting)
- **nbstripout** (clearing output from jupyter notebooks)
- **bandit** (python security linter)
- **mdformat** (markdown formatting)
- **codespell** (word spellling linter)

To reformat all files in the project use command:

```bash
pre-commit run -a
```

</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 training_epoch_end():
           ...

       def validation_step():
           ...

       def validation_step_end():
           ...

       def validation_epoch_end():
           ...

       def test_step():
           ...

       def test_step_end():
           ...

       def 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.datamodules.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.

Example 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-learninig-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)
- [kedro-org/kedro](https://github.com/kedro-org/kedro)

Useful repositories:

- [pytorch/hydra-torch](https://github.com/pytorch/hydra-torch) - safely configuring PyTorch classes with Hydra
- [romesco/hydra-lightning](https://github.com/romesco/hydra-lightning) - safely configuring PyTorch Lightning classes with Hydra
- [PyTorchLightning/lightning-transformers](https://github.com/PyTorchLightning/lightning-transformers) - official Lightning Transformers repo built with Hydra

Other resources:

- [Cookiecutter Data Science Project Structure Opinions](http://drivendata.github.io/cookiecutter-data-science/#opinions)
- [The Machine Learning Reproducibility Checklist](https://www.cs.mcgill.ca/~jpineau/ReproducibilityChecklist.pdf)

</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

## How to run

Install dependencies

```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
```

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 datamodule.batch_size=64
```