├── .github ├── CODE-OF-CONDUCT.md ├── CONTRIBUTING.md ├── ISSUE_TEMPLATE │ ├── bug-report.md │ └── question.md └── workflows │ ├── test-basic.yaml │ ├── test-misc.yaml │ └── test-super-layer_model.yaml ├── .gitignore ├── .gitmodules ├── .latent-data ├── .gitignore ├── init-configs │ ├── README.md │ ├── bashrc │ ├── setup.sh │ ├── tmux.conf │ ├── vimrc │ └── zshrc └── splits │ └── .gitignore ├── CHANGE-LOG.md ├── LICENSE.md ├── README.md ├── configs ├── NeurIPS-2019 │ ├── C010-ResNet110.config │ ├── C010-ResNet164.config │ ├── C010-ResNet20.config │ ├── C010-ResNet32.config │ ├── C010-ResNet56.config │ ├── C100-ResNet110.config │ ├── C100-ResNet164.config │ ├── C100-ResNet20.config │ ├── C100-ResNet32.config │ ├── C100-ResNet56.config │ ├── ImageNet-ResNet18V1.config │ └── ImageNet-ResNet50V1.config ├── archs │ ├── CIFAR-ResNet08.config │ ├── CIFAR-ResNet1001.config │ ├── CIFAR-ResNet110.config │ ├── CIFAR-ResNet164.config │ ├── CIFAR-ResNet20.config │ ├── CIFAR-ResNet32.config │ ├── CIFAR-ResNet56.config │ ├── CIFAR-SIM05.config │ ├── CIFAR-WRN28-10.config │ ├── ImageNet-ResNet101V1.config │ ├── ImageNet-ResNet101V2.config │ ├── ImageNet-ResNet152V1.config │ ├── ImageNet-ResNet152V2.config │ ├── ImageNet-ResNet18V1.config │ ├── ImageNet-ResNet18V2.config │ ├── ImageNet-ResNet200V1.config │ ├── ImageNet-ResNet200V2.config │ ├── ImageNet-ResNet34V1.config │ ├── ImageNet-ResNet34V2.config │ ├── ImageNet-ResNet50V1.config │ ├── ImageNet-ResNet50V2.config │ ├── ImageNet-ResNext50-32x4dV2.config │ ├── NAS-CIFAR-DARTS.config │ ├── NAS-CIFAR-GDAS_V1.config │ ├── NAS-CIFAR-NAS.config │ ├── NAS-CIFAR-SETN.config │ ├── NAS-CIFAR-none.config │ ├── NAS-IMAGENET-DARTS_V2.config │ ├── NAS-IMAGENET-GDAS_V1.config │ ├── NAS-IMAGENET-SETN.config │ ├── NAS-IMAGENET-SETN1.config │ ├── NAS-IMAGENET-SETN2.config │ ├── NAS-IMAGENET-SETN3.config │ ├── NAS-IMAGENET-SETN4.config │ └── NAS-IMAGENET-none.config ├── compares │ ├── CIFAR010-TAS-R110.config │ ├── CIFAR010-TAS-R56.config │ ├── CIFAR100-FIX-R32.config │ ├── CIFAR100-RAND-R32.config │ ├── CIFAR100-TAS-R110.config │ ├── CIFAR100-TAS-R32.config │ └── CIFAR100-TAS-R56.config ├── nas-benchmark │ ├── CIFAR.config │ ├── ImageNet-16.config │ ├── ImageNet16-120-split.txt │ ├── LESS.config │ ├── algos │ │ ├── DARTS.config │ │ ├── ENAS.config │ │ ├── GDAS.config │ │ ├── R-EA.config │ │ ├── RANDOM.config │ │ ├── SETN.config │ │ └── weight-sharing.config │ ├── cifar-split.txt │ ├── cifar100-test-split.txt │ ├── hyper-opts │ │ ├── 01E.config │ │ ├── 12E.config │ │ ├── 200E.config │ │ └── 90E.config │ └── imagenet-16-120-test-split.txt ├── opts │ ├── CIFAR-E300-W5-L1-COS.config │ ├── CIFAR-E300-W5-L4-COS.config │ ├── CIFAR-E600-W5-L1-COS.config │ ├── CIFAR-Fast-Random.config │ ├── CIFAR-Slow-Random.config │ ├── Com-Paddle-NAS.config │ ├── Com-Paddle-RES.config │ ├── ImageNet-E120-Cos-Smooth.config │ ├── ImageNet-E120-Cos-Soft.config │ ├── ImageNet-E120-Step-Soft.config │ ├── NAS-CIFAR-V2.config │ ├── NAS-CIFAR.config │ └── NAS-IMAGENET.config ├── qlib │ ├── workflow_config_TabNet_Alpha360.yaml │ ├── workflow_config_alstm_Alpha360.yaml │ ├── workflow_config_doubleensemble_Alpha360.yaml │ ├── workflow_config_gru_Alpha360.yaml │ ├── workflow_config_lightgbm_Alpha360.yaml │ ├── workflow_config_lstm_Alpha360.yaml │ ├── workflow_config_mlp_Alpha360.yaml │ ├── workflow_config_naive_v1_Alpha360.yaml │ ├── workflow_config_naive_v2_Alpha360.yaml │ ├── workflow_config_sfm_Alpha360.yaml │ ├── workflow_config_transformer_Alpha360.yaml │ ├── workflow_config_transformer_basic_Alpha360.yaml │ └── workflow_config_xgboost_Alpha360.yaml ├── search-KD-opts │ ├── ImageNet-E120-Cos-Soft.config │ └── ImageNet-E150-MobileFast.config ├── search-archs │ ├── DARTS-NASNet-CIFAR.config │ ├── GDAS-NASNet-CIFAR.config │ ├── GDASFRC-NASNet-CIFAR.config │ ├── ImageNet-MobileNetV2-X.config │ ├── ImageNet-ResNet18V1.config │ ├── ImageNet-ResNet34V1.config │ └── ImageNet-ResNet50V1.config ├── search-opts │ ├── CIFAR.config │ ├── CIFARX.config │ ├── CIFARXX.config │ ├── DARTS-NASNet-CIFAR.config │ ├── GDAS-NASNet-CIFAR.config │ ├── ImageNet-MobileFast.config │ └── ImageNet-ResNet.config ├── temps │ ├── T-MobileNetV2-X.config │ ├── T-ResNet18.config │ ├── T-ResNet50.config │ ├── X-ResNet18.config │ └── X-ResNet50.config ├── yaml.data │ ├── cifar10.test │ └── cifar10.train ├── yaml.loss │ ├── cross-entropy │ ├── top1-ce │ └── top1.acc ├── yaml.model │ └── vit-cifar10.s0 └── yaml.opt │ └── vit.cifar ├── docs ├── BASELINE.md ├── CVPR-2019-GDAS.md ├── ICCV-2019-SETN.md ├── ICLR-2019-DARTS.md ├── NAS-Bench-201-PURE.md ├── NAS-Bench-201.md ├── NeurIPS-2019-TAS.md ├── README_CN.md └── requirements.txt ├── exps ├── NAS-Bench-201-algos │ ├── BOHB.py │ ├── DARTS-V1.py │ ├── DARTS-V2.py │ ├── ENAS.py │ ├── GDAS.py │ ├── RANDOM-NAS.py │ ├── RANDOM.py │ ├── README.md │ ├── R_EA.py │ ├── SETN.py │ └── reinforce.py ├── NAS-Bench-201 │ ├── check.py │ ├── dist-setup.py │ ├── functions.py │ ├── main.py │ ├── show-best.py │ ├── statistics-v2.py │ ├── statistics.py │ ├── test-correlation.py │ └── visualize.py ├── NATS-Bench │ ├── Analyze-time.py │ ├── draw-correlations.py │ ├── draw-fig2_5.py │ ├── draw-fig6.py │ ├── draw-fig7.py │ ├── draw-fig8.py │ ├── draw-ranks.py │ ├── draw-table.py │ ├── main-sss.py │ ├── main-tss.py │ ├── show-dataset.py │ ├── sss-collect.py │ ├── sss-file-manager.py │ ├── test-nats-api.py │ ├── tss-collect-patcher.py │ ├── tss-collect.py │ └── tss-file-manager.py ├── NATS-algos │ ├── README.md │ ├── bohb.py │ ├── random_wo_share.py │ ├── regularized_ea.py │ ├── reinforce.py │ ├── run-all.sh │ ├── search-cell.py │ └── search-size.py ├── TAS │ ├── prepare.py │ ├── search-shape.py │ └── search-transformable.py ├── basic │ ├── KD-main.py │ ├── basic-eval.py │ ├── basic-main.py │ └── xmain.py ├── experimental │ ├── GeMOSA │ │ ├── baselines │ │ │ ├── maml-ft.py │ │ │ ├── maml-nof.py │ │ │ ├── slbm-ft.py │ │ │ └── slbm-nof.py │ │ ├── basic-his.py │ │ ├── basic-prev.py │ │ ├── basic-same.py │ │ ├── main.py │ │ ├── meta_model.py │ │ ├── meta_model_ablation.py │ │ └── vis-synthetic.py │ ├── example-nas-bench.py │ ├── test-dks.py │ ├── test-dynamic.py │ ├── test-flops.py │ ├── test-nas-plot.py │ ├── test-resnest.py │ ├── test-ww-bench.py │ ├── test-ww.py │ ├── vis-nats-bench-algos.py │ ├── vis-nats-bench-ws.py │ └── visualize-nas-bench-x.py └── trading │ ├── baselines.py │ ├── organize_results.py │ └── workflow_tt.py ├── notebooks ├── NATS-Bench │ ├── BayesOpt.ipynb │ ├── find-largest.ipynb │ ├── issue-96.ipynb │ └── issue-97.ipynb ├── Q │ ├── qlib-data-play.ipynb │ └── workflow-test.ipynb ├── TOT │ ├── ES-Model-DC.ipynb │ ├── ES-Model-Drop.ipynb │ ├── Time-Curve.ipynb │ └── time-curve.py └── spaces-xmisc │ ├── random-search-transformer.ipynb │ ├── scheduler.ipynb │ ├── synthetic-data.ipynb │ ├── synthetic-env.ipynb │ ├── synthetic-visualize-env.ipynb │ └── test-transformer-encoder.ipynb ├── scripts-search ├── NAS-Bench-201-algos │ ├── BOHB.sh │ ├── DARTS-V1.sh │ ├── DARTS-V2.sh │ ├── ENAS.sh │ ├── GDAS.sh │ ├── R-EA.sh │ ├── RANDOM-NAS.sh │ ├── README.md │ ├── REINFORCE.sh │ ├── Random.sh │ ├── SETN.sh │ └── hps │ │ ├── DARTS-test-Gradient.sh │ │ └── GRID-RL.sh ├── NAS-Bench-201 │ ├── build.sh │ ├── meta-gen.sh │ ├── train-a-net.sh │ └── train-models.sh ├── NASNet-space-search-by-DARTS1V.sh ├── NASNet-space-search-by-GDAS-FRC.sh ├── NASNet-space-search-by-GDAS.sh ├── NASNet-space-search-by-SETN.sh ├── NATS │ └── search-size.sh ├── search-depth-cifar.sh ├── search-depth-gumbel.sh ├── search-shape-cifar.sh ├── search-width-cifar.sh └── search-width-gumbel.sh ├── scripts ├── NATS-Bench │ ├── train-shapes.sh │ └── train-topology.sh ├── TAS │ └── prepare.sh ├── base-train.sh ├── black.sh ├── experimental │ └── train-vit.sh ├── nas-infer-train.sh ├── retrain-searched-net.sh ├── tas-infer-train.sh └── trade │ ├── baseline.sh │ ├── tsf-all.sh │ ├── tsf-time.sh │ └── tsf.sh ├── setup.py ├── tests ├── test_basic_space.py ├── test_import.py ├── test_loader.py ├── test_math_static.py ├── test_misc_scheduler.py ├── test_super_att.py ├── test_super_container.py ├── test_super_mlp.py ├── test_super_norm.py ├── test_super_rearrange.py ├── test_super_vit.py ├── test_synthetic_env.py ├── test_synthetic_utils.py ├── test_tas.py ├── test_torch.sh └── test_torch_gpu_bugs.py └── xautodl ├── __init__.py ├── config_utils ├── __init__.py ├── args_utils.py ├── attention_args.py ├── basic_args.py ├── cls_init_args.py ├── cls_kd_args.py ├── config_utils.py ├── pruning_args.py ├── random_baseline.py ├── search_args.py ├── search_single_args.py └── share_args.py ├── datasets ├── DownsampledImageNet.py ├── LandmarkDataset.py ├── SearchDatasetWrap.py ├── __init__.py ├── get_dataset_with_transform.py ├── landmark_utils │ ├── __init__.py │ └── point_meta.py ├── math_base_funcs.py ├── math_core.py ├── math_dynamic_funcs.py ├── math_dynamic_generator.py ├── math_static_funcs.py ├── synthetic_core.py ├── synthetic_env.py ├── synthetic_utils.py └── test_utils.py ├── log_utils ├── __init__.py ├── logger.py ├── meter.py ├── pickle_wrap.py └── time_utils.py ├── models ├── CifarDenseNet.py ├── CifarResNet.py ├── CifarWideResNet.py ├── ImageNet_MobileNetV2.py ├── ImageNet_ResNet.py ├── SharedUtils.py ├── __init__.py ├── cell_infers │ ├── __init__.py │ ├── cells.py │ ├── nasnet_cifar.py │ └── tiny_network.py ├── cell_operations.py ├── cell_searchs │ ├── __init__.py │ ├── _test_module.py │ ├── generic_model.py │ ├── genotypes.py │ ├── search_cells.py │ ├── search_model_darts.py │ ├── search_model_darts_nasnet.py │ ├── search_model_enas.py │ ├── search_model_enas_utils.py │ ├── search_model_gdas.py │ ├── search_model_gdas_frc_nasnet.py │ ├── search_model_gdas_nasnet.py │ ├── search_model_random.py │ ├── search_model_setn.py │ └── search_model_setn_nasnet.py ├── clone_weights.py ├── initialization.py ├── shape_infers │ ├── InferCifarResNet.py │ ├── InferCifarResNet_depth.py │ ├── InferCifarResNet_width.py │ ├── InferImagenetResNet.py │ ├── InferMobileNetV2.py │ ├── InferTinyCellNet.py │ ├── __init__.py │ └── shared_utils.py └── shape_searchs │ ├── SearchCifarResNet.py │ ├── SearchCifarResNet_depth.py │ ├── SearchCifarResNet_width.py │ ├── SearchImagenetResNet.py │ ├── SearchSimResNet_width.py │ ├── SoftSelect.py │ ├── __init__.py │ └── generic_size_tiny_cell_model.py ├── nas_infer_model ├── DXYs │ ├── CifarNet.py │ ├── ImageNet.py │ ├── __init__.py │ ├── base_cells.py │ ├── construct_utils.py │ ├── genotypes.py │ └── head_utils.py ├── __init__.py └── operations.py ├── procedures ├── __init__.py ├── advanced_main.py ├── basic_main.py ├── eval_funcs.py ├── funcs_nasbench.py ├── metric_utils.py ├── optimizers.py ├── q_exps.py ├── search_main.py ├── search_main_v2.py ├── simple_KD_main.py └── starts.py ├── spaces ├── __init__.py ├── basic_op.py └── basic_space.py ├── trade_models ├── __init__.py ├── naive_v1_model.py ├── naive_v2_model.py ├── quant_transformer.py └── transformers.py ├── utils ├── __init__.py ├── affine_utils.py ├── evaluation_utils.py ├── flop_benchmark.py ├── gpu_manager.py ├── hash_utils.py ├── nas_utils.py ├── qlib_utils.py ├── str_utils.py ├── temp_sync.py └── weight_watcher.py ├── xlayers ├── __init__.py ├── misc_utils.py ├── super_activations.py ├── super_attention.py ├── super_attention_v2.py ├── super_container.py ├── super_core.py ├── super_dropout.py ├── super_linear.py ├── super_module.py ├── super_norm.py ├── super_positional_embedding.py ├── super_rearrange.py ├── super_trade_stem.py ├── super_transformer.py ├── super_utils.py └── weight_init.py ├── xmisc ├── __init__.py ├── logger_utils.py ├── meter_utils.py ├── module_utils.py ├── sampler_utils.py ├── scheduler_utils.py ├── time_utils.py ├── torch_utils.py └── yaml_utils.py └── xmodels ├── __init__.py ├── core.py ├── transformers.py └── transformers_quantum.py /.github/CONTRIBUTING.md: -------------------------------------------------------------------------------- 1 | # Contributing to AutoDL-Projects 2 | 3 | :+1::tada: First off, thanks for taking the time to contribute! :tada::+1: 4 | 5 | The following is a set of guidelines for contributing to AutoDL-Projects. 6 | 7 | ## Table Of Contents 8 | 9 | [How Can I Contribute?](#how-can-i-contribute) 10 | * [Reporting Bugs](#reporting-bugs) 11 | * [Suggesting Enhancements](#suggesting-enhancements) 12 | * [Your First Code Contribution](#your-first-code-contribution) 13 | 14 | ## How Can I Contribute? 15 | 16 | ### Reporting Bugs 17 | 18 | This section guides you through submitting a bug report for AutoDL-Projects. 19 | Following these guidelines helps maintainers and the community understand your report :pencil:, reproduce the behavior :computer: :computer:, and find related reports :mag_right:. 20 | 21 | When you are creating a bug report, please include as many details as possible. 22 | Fill out [the required template](https://github.com/D-X-Y/AutoDL-Projects/blob/main/.github/ISSUE_TEMPLATE/bug-report.md). The information it asks for helps us resolve issues faster. 23 | 24 | > **Note:** If you find a **Closed** issue that seems like it is the same thing that you're experiencing, open a new issue and include a link to the original issue in the body of your new one. 25 | 26 | ### Suggesting Enhancements 27 | 28 | Please feel free to email me (dongxuanyi888@gmail.com). 29 | 30 | ### Your First Code Contribution 31 | 32 | Please feel free to open an Pull Requests. 33 | 34 | -------------------------------------------------------------------------------- /.github/ISSUE_TEMPLATE/bug-report.md: -------------------------------------------------------------------------------- 1 | --- 2 | name: Bug Report 3 | about: Create a report to help us improve 4 | title: '' 5 | labels: '' 6 | assignees: '' 7 | 8 | --- 9 | 10 | **Describe the bug** 11 | A clear and concise description of what the bug is. 12 | 13 | **To Reproduce** 14 | Please provide a small script to reproduce the behavior: 15 | ``` 16 | codes to reproduce the bug 17 | ``` 18 | Please let me know your OS, Python version, PyTorch version. 19 | 20 | **Expected behavior** 21 | A clear and concise description of what you expected to happen. 22 | 23 | **Screenshots** 24 | If applicable, add screenshots to help explain your problem. 25 | -------------------------------------------------------------------------------- /.github/ISSUE_TEMPLATE/question.md: -------------------------------------------------------------------------------- 1 | --- 2 | name: Questions w.r.t. an AutoDL algorithm 3 | about: Ask questions about or discuss on some algorithms 4 | title: '' 5 | labels: '' 6 | assignees: '' 7 | 8 | --- 9 | 10 | **Which Algorithm** 11 | Please put the title of the AutoDL algorithm that you want to ask here. 12 | 13 | **Describe the Question** 14 | A clear and concise description of what the bug is. 15 | 16 | -------------------------------------------------------------------------------- /.github/workflows/test-basic.yaml: -------------------------------------------------------------------------------- 1 | name: Test Spaces 2 | on: 3 | push: 4 | branches: 5 | - main 6 | pull_request: 7 | branches: 8 | - main 9 | 10 | 11 | jobs: 12 | build: 13 | strategy: 14 | matrix: 15 | os: [ubuntu-18.04, ubuntu-20.04, macos-latest] 16 | python-version: [3.6, 3.7, 3.8, 3.9] 17 | 18 | runs-on: ${{ matrix.os }} 19 | steps: 20 | - uses: actions/checkout@v2 21 | 22 | - name: Set up Python ${{ matrix.python-version }} 23 | uses: actions/setup-python@v2 24 | with: 25 | python-version: ${{ matrix.python-version }} 26 | 27 | - name: Lint with Black 28 | run: | 29 | python -m pip install black 30 | python --version 31 | python -m black --version 32 | echo $PWD ; ls 33 | python -m black ./exps -l 88 --check --diff --verbose 34 | python -m black ./tests -l 88 --check --diff --verbose 35 | python -m black ./xautodl/x* -l 88 --check --diff --verbose 36 | python -m black ./xautodl/spaces -l 88 --check --diff --verbose 37 | python -m black ./xautodl/trade_models -l 88 --check --diff --verbose 38 | python -m black ./xautodl/procedures -l 88 --check --diff --verbose 39 | python -m black ./xautodl/config_utils -l 88 --check --diff --verbose 40 | python -m black ./xautodl/log_utils -l 88 --check --diff --verbose 41 | 42 | - name: Install XAutoDL from source 43 | run: | 44 | pip install . 45 | 46 | - name: Test Search Space 47 | run: | 48 | python -m pip install pytest 49 | python -m pip install torch torchvision 50 | python -m pip install parameterized 51 | echo $PWD 52 | echo "Show what we have here:" 53 | ls 54 | python --version 55 | python -m pytest ./tests/test_import.py -s 56 | python -m pytest ./tests/test_basic_space.py -s 57 | shell: bash 58 | 59 | - name: Test Math 60 | run: | 61 | python -m pytest ./tests/test_math*.py -s 62 | shell: bash 63 | 64 | - name: Test Synthetic Data 65 | run: | 66 | python -m pytest ./tests/test_synthetic*.py -s 67 | shell: bash 68 | -------------------------------------------------------------------------------- /.github/workflows/test-misc.yaml: -------------------------------------------------------------------------------- 1 | name: Test Xmisc 2 | on: 3 | push: 4 | branches: 5 | - main 6 | pull_request: 7 | branches: 8 | - main 9 | 10 | 11 | jobs: 12 | build: 13 | strategy: 14 | matrix: 15 | os: [ubuntu-18.04, ubuntu-20.04, macos-latest] 16 | python-version: [3.6, 3.7, 3.8, 3.9] 17 | 18 | runs-on: ${{ matrix.os }} 19 | steps: 20 | - uses: actions/checkout@v2 21 | 22 | - name: Set up Python ${{ matrix.python-version }} 23 | uses: actions/setup-python@v2 24 | with: 25 | python-version: ${{ matrix.python-version }} 26 | 27 | - name: Install XAutoDL from source 28 | run: | 29 | pip install . 30 | 31 | - name: Test Xmisc 32 | run: | 33 | python -m pip install pytest 34 | python -m pip install torch torchvision 35 | python -m pip install parameterized 36 | echo $PWD 37 | echo "Show what we have here:" 38 | ls 39 | python --version 40 | python -m pytest ./tests/test_misc* -s 41 | shell: bash 42 | -------------------------------------------------------------------------------- /.github/workflows/test-super-layer_model.yaml: -------------------------------------------------------------------------------- 1 | name: Test Super Model 2 | on: 3 | push: 4 | branches: 5 | - main 6 | pull_request: 7 | branches: 8 | - main 9 | 10 | 11 | jobs: 12 | build: 13 | strategy: 14 | matrix: 15 | os: [ubuntu-18.04, ubuntu-20.04, macos-latest] 16 | python-version: [3.6, 3.7, 3.8, 3.9] 17 | 18 | runs-on: ${{ matrix.os }} 19 | steps: 20 | - uses: actions/checkout@v2 21 | 22 | - name: Set up Python ${{ matrix.python-version }} 23 | uses: actions/setup-python@v2 24 | with: 25 | python-version: ${{ matrix.python-version }} 26 | 27 | - name: Install XAutoDL from source 28 | run: | 29 | pip install . 30 | 31 | - name: Test Super Model 32 | run: | 33 | python -m pip install pytest 34 | python -m pip install parameterized 35 | python -m pip install torch torchvision 36 | python -m pytest ./tests/test_super_*.py 37 | shell: bash 38 | 39 | - name: Test TAS (NeurIPS 2019) 40 | run: | 41 | python -m pytest ./tests/test_tas.py 42 | shell: bash 43 | -------------------------------------------------------------------------------- /.gitignore: -------------------------------------------------------------------------------- 1 | .DS_Store 2 | # Byte-compiled / optimized / DLL files 3 | __pycache__/ 4 | *.py[cod] 5 | *$py.class 6 | 7 | # C extensions 8 | *.so 9 | 10 | # Distribution / packaging 11 | .Python 12 | env/ 13 | build/ 14 | develop-eggs/ 15 | dist/ 16 | downloads/ 17 | eggs/ 18 | .eggs/ 19 | lib64/ 20 | parts/ 21 | sdist/ 22 | var/ 23 | *.egg-info/ 24 | .installed.cfg 25 | *.egg 26 | 27 | # PyInstaller 28 | # Usually these files are written by a python script from a template 29 | # before PyInstaller builds the exe, so as to inject date/other infos into it. 30 | *.manifest 31 | *.spec 32 | 33 | # Installer logs 34 | pip-log.txt 35 | pip-delete-this-directory.txt 36 | 37 | # Unit test / coverage reports 38 | htmlcov/ 39 | .tox/ 40 | .coverage 41 | .coverage.* 42 | .cache 43 | nosetests.xml 44 | coverage.xml 45 | *,cover 46 | .hypothesis/ 47 | 48 | # Translations 49 | *.mo 50 | *.pot 51 | 52 | # Django stuff: 53 | *.log 54 | local_settings.py 55 | 56 | # Flask stuff: 57 | instance/ 58 | .webassets-cache 59 | 60 | # Scrapy stuff: 61 | .scrapy 62 | 63 | # Sphinx documentation 64 | docs/_build/ 65 | 66 | # PyBuilder 67 | target/ 68 | 69 | # IPython Notebook 70 | .ipynb_checkpoints 71 | 72 | # pyenv 73 | .python-version 74 | 75 | # celery beat schedule file 76 | celerybeat-schedule 77 | 78 | # dotenv 79 | .env 80 | 81 | # virtualenv 82 | venv/ 83 | ENV/ 84 | 85 | # Spyder project settings 86 | .spyderproject 87 | 88 | # Rope project settings 89 | .ropeproject 90 | 91 | # Pycharm project 92 | .idea 93 | snapshots 94 | *.pytorch 95 | *.tar.bz 96 | data 97 | .*.swp 98 | main_main.py 99 | *.pdf 100 | */*.pdf 101 | 102 | # Device 103 | scripts-nas/.nfs00* 104 | */.nfs00* 105 | *.DS_Store 106 | 107 | # logs and snapshots 108 | output 109 | logs 110 | 111 | # snapshot 112 | a.pth 113 | cal-merge*.sh 114 | GPU-*.sh 115 | cal.sh 116 | aaa 117 | cx.sh 118 | 119 | NAS-Bench-*-v1_0.pth 120 | lib/NAS-Bench-*-v1_0.pth 121 | others/TF 122 | scripts-search/l2s-algos 123 | TEMP-L.sh 124 | 125 | .nfs00* 126 | *.swo 127 | */*.swo 128 | 129 | # Visual Studio Code 130 | .vscode 131 | mlruns* 132 | outputs 133 | 134 | pytest_cache 135 | *.pkl 136 | *.pth 137 | 138 | *.tgz 139 | -------------------------------------------------------------------------------- /.gitmodules: -------------------------------------------------------------------------------- 1 | [submodule ".latent-data/qlib"] 2 | path = .latent-data/qlib 3 | url = https://github.com/microsoft/qlib.git 4 | [submodule ".latent-data/NATS-Bench"] 5 | path = .latent-data/NATS-Bench 6 | url = https://github.com/D-X-Y/NATS-Bench.git 7 | [submodule ".latent-data/NAS-Bench-201"] 8 | path = .latent-data/NAS-Bench-201 9 | url = https://github.com/D-X-Y/NAS-Bench-201.git 10 | -------------------------------------------------------------------------------- /.latent-data/.gitignore: -------------------------------------------------------------------------------- 1 | * 2 | -------------------------------------------------------------------------------- /.latent-data/init-configs/README.md: -------------------------------------------------------------------------------- 1 | # Git Commands 2 | 3 | ``` 4 | git clone --recurse-submodules git@github.com:D-X-Y/AutoDL-Projects.git 5 | 6 | git submodule init 7 | git submodule update 8 | git pull orign main 9 | 10 | git submodule update --remote --recursive 11 | ``` 12 | 13 | Pylint check for Q-lib: 14 | ``` 15 | python -m black __init__.py -l 120 16 | 17 | pytest -W ignore::DeprecationWarning qlib/tests/test_all_pipeline.py 18 | ``` 19 | 20 | 21 | ``` 22 | conda update --all 23 | ``` 24 | 25 | ## [phillytools](https://phillytools.azurewebsites.net/master/get_started/2_installation.html) 26 | 27 | ``` 28 | conda create -n pt6 python=3.7 29 | 30 | conda activate pt6 31 | 32 | pip install -U phillytools --extra-index-url https://msrpypi.azurewebsites.net/stable/7e404de797f4e1eeca406c1739b00867 --extra-index-url https://azuremlsdktestpypi.azureedge.net/K8s-Compute/D58E86006C65 33 | ``` 34 | -------------------------------------------------------------------------------- /.latent-data/init-configs/setup.sh: -------------------------------------------------------------------------------- 1 | #!/bin/bash 2 | 3 | script=$(readlink -f "$0") 4 | script_dir=$(dirname "$script") 5 | echo "script-directory: $script_dir" 6 | 7 | cp ${script_dir}/tmux.conf ~/.tmux.conf 8 | cp ${script_dir}/vimrc ~/.vimrc 9 | cp ${script_dir}/bashrc ~/.bashrc 10 | cp ${script_dir}/condarc ~/.condarc 11 | 12 | wget https://repo.anaconda.com/miniconda/Miniconda3-4.7.12.1-Linux-x86_64.sh 13 | wget https://repo.anaconda.com/miniconda/Miniconda3-latest-Linux-x86_64.sh 14 | wget https://repo.anaconda.com/archive/Anaconda3-2021.05-Linux-x86_64.sh 15 | -------------------------------------------------------------------------------- /.latent-data/init-configs/vimrc: -------------------------------------------------------------------------------- 1 | color desert 2 | "设置背景色，每种配色有两种方案，一个light、一个dark 3 | "set background=light 4 | ""打开语法高亮 5 | syntax enable 6 | set background=dark 7 | "colorscheme solarized 8 | colorscheme desert 9 | let g:solarized_termcolors=256 10 | 11 | if has('gui_running') 12 | set background=light 13 | else 14 | set background=dark 15 | endif 16 | "显示行号 17 | "set number 18 | ""设置缩进有三个取值cindent(c风格)、smartindent(智能模式，其实不觉得有什么智能)、autoindent(简单的与上一行保持一致) 19 | set autoindent 20 | "在windows版本中vim的退格键模式默认与vi兼容，与我们的使用习惯不太符合，下边这条可以改过来 21 | "set backspace=indent,eol,start 22 | ""用空格键替换制表符 23 | "set expandtab 24 | "制表符占4个空格 25 | "set tabstop=4 26 | ""默认缩进4个空格大小 27 | "set shiftwidth=4 28 | "增量式搜索 29 | "set incsearch 30 | ""高亮搜索 31 | set hlsearch 32 | set statusline=[%F]%y%r%m%*%=[Line:%l/%L,Column:%c][%p%%] 33 | set laststatus=2 34 | set ruler 35 | "有时中文会显示乱码，用一下几条命令解决 36 | "let &termencoding=&encoding 37 | "set fileencodings=utf-8,gbk 38 | ""很多插件都会要求的配置检测文件类型 39 | :filetype on 40 | :filetype plugin on 41 | :filetype indent on 42 | "下边这个很有用可以根据不同的文件类型执行不同的命令 43 | ""例如：如果是c/c++类型 44 | :autocmd FileType c,cpp,cuda :set foldmethod=syntax 45 | :autocmd FileType c,cpp,cuda :set number 46 | :autocmd FileType c,cpp,cuda :set cindent 47 | :autocmd FileType c,cpp,cuda :set expandtab 48 | :autocmd FileType c,cpp,cuda :set tabstop=4 49 | :autocmd FileType c,cpp,cuda :set shiftwidth=4 50 | "例如：如果是python类型 51 | :autocmd FileType python :set number 52 | :autocmd FileType python :set foldmethod=syntax 53 | :autocmd FileType python :set paste 54 | :autocmd FileType python :set expandtab 55 | :autocmd FileType python :set tabstop=2 56 | :autocmd FileType python :set shiftwidth=2 57 | "例如：如果是python类型 58 | :autocmd FileType matlab :set number 59 | :autocmd FileType matlab :set foldmethod=syntax 60 | :autocmd FileType matlab :set paste 61 | :autocmd FileType matlab :set expandtab 62 | :autocmd FileType matlab :set tabstop=2 63 | :autocmd FileType matlab :set shiftwidth=2 64 | -------------------------------------------------------------------------------- /.latent-data/splits/.gitignore: -------------------------------------------------------------------------------- 1 | * 2 | -------------------------------------------------------------------------------- /CHANGE-LOG.md: -------------------------------------------------------------------------------- 1 | # This file shows the major updates of this repo. 2 | 3 | - [2020.04.11] [4ef9531](https://github.com/D-X-Y/AutoDL-Projects/tree/4ef9531) Add change log as `CHANGE-LOG.md`. 4 | - [2019.12.20] [69ca086](https://github.com/D-X-Y/AutoDL-Projects/tree/69ca086) Release NAS-Bench-201. 5 | - [2019.09.28] [f8f3f38](https://github.com/D-X-Y/AutoDL-Projects/tree/f8f3f38) TAS and SETN codes were publicly released. 6 | - [2019.01.31] [13e908f](https://github.com/D-X-Y/AutoDL-Projects/tree/13e908f) GDAS codes were publicly released. 7 | - [2020.07.01] [a45808b](https://github.com/D-X-Y/AutoDL-Projects/tree/a45808b) Upgrade NAS-API to the 2.0 version. 8 | - [2020.09.16] [7052265](https://github.com/D-X-Y/AutoDL-Projects/tree/7052265) Create NATS-BENCH. 9 | - [2020.10.15] [446262a](https://github.com/D-X-Y/AutoDL-Projects/tree/446262a) Update NATS-BENCH to version 1.0 10 | - [2020.12.20] [dae387a](https://github.com/D-X-Y/AutoDL-Projects/tree/dae387a) Update NATS-BENCH to version 1.1 11 | - [2021.05.18] [98fadf8](https://github.com/D-X-Y/AutoDL-Projects/tree/98fadf8) Before moving to `xautodl` 12 | - [2021.05.21] [df99173](https://github.com/D-X-Y/AutoDL-Projects/tree/df99173) `xautodl` is close to ready 13 | -------------------------------------------------------------------------------- /LICENSE.md: -------------------------------------------------------------------------------- 1 | MIT License 2 | 3 | Copyright (c) since 2019.01.01, author: Xuanyi Dong (GitHub: https://github.com/D-X-Y) 4 | 5 | Permission is hereby granted, free of charge, to any person obtaining a copy 6 | of this software and associated documentation files (the "Software"), to deal 7 | in the Software without restriction, including without limitation the rights 8 | to use, copy, modify, merge, publish, distribute, sublicense, and/or sell 9 | copies of the Software, and to permit persons to whom the Software is 10 | furnished to do so, subject to the following conditions: 11 | 12 | The above copyright notice and this permission notice shall be included in all 13 | copies or substantial portions of the Software. 14 | 15 | THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR 16 | IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, 17 | FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE 18 | AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER 19 | LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, 20 | OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE 21 | SOFTWARE. 22 | -------------------------------------------------------------------------------- /configs/NeurIPS-2019/C010-ResNet110.config: -------------------------------------------------------------------------------- 1 | { 2 | "dataset" : ["str" , "cifar"], 3 | "arch" : ["str" , "resnet"], 4 | "depth" : ["int" , "110"], 5 | "module" : ["str" , "ResNetBasicblock"], 6 | "super_type" : ["str" , "infer-shape"], 7 | "zero_init_residual" : ["bool" , "0"], 8 | "class_num" : ["int" , "10"], 9 | "search_mode" : ["str" , "shape"], 10 | "xchannels" : ["int" , ["3", "16", "14", "16", "14", "16", "16", "16", "16", "16", "14", "16", "16", "16", "12", "16", "16", "16", "9", "16", "8", "16", "4", "16", "4", "4", "4", "16", "4", "4", "4", "4", "6", "6", "4", "6", "11", "4", "32", "32", "32", "32", "32", "32", "32", "32", "28", "32", "32", "28", "22", "22", "22", "32", "32", "25", "28", "9", "9", "28", "12", "9", "12", "32", "9", "9", "22", "12", "16", "9", "12", "9", "9", "9", "64", "64", "64", "64", "64", "64", "64", "64", "64", "64", "64", "64", "64", "64", "64", "38", "64", "25", "19", "19", "19", "19", "19", "25", "32", "19", "19", "25", "25", "19", "19", "38", "38", "19", "19", "51"]], 11 | "xblocks" : ["int" , ["11", "11", "9"]], 12 | "estimated_FLOP" : ["float" , "117.498238"] 13 | } -------------------------------------------------------------------------------- /configs/NeurIPS-2019/C010-ResNet164.config: -------------------------------------------------------------------------------- 1 | { 2 | "dataset" : ["str" , "cifar"], 3 | "arch" : ["str" , "resnet"], 4 | "depth" : ["int" , "164"], 5 | "module" : ["str" , "ResNetBottleneck"], 6 | "super_type" : ["str" , "infer-shape"], 7 | "zero_init_residual" : ["bool" , "0"], 8 | "class_num" : ["int" , "10"], 9 | "search_mode" : ["str" , "shape"], 10 | "xchannels" : ["int" , ["3", "16", "8", "16", "64", "6", "16", "64", "14", "16", "64", "8", "16", "64", "4", "16", "64", "6", "16", "64", "6", "16", "64", "11", "11", "64", "4", "14", "64", "4", "4", "57", "4", "16", "64", "9", "12", "57", "4", "16", "64", "4", "8", "57", "6", "6", "51", "6", "4", "44", "6", "4", "57", "6", "6", "19", "32", "32", "128", "32", "32", "128", "32", "32", "128", "9", "32", "128", "32", "32", "128", "25", "28", "115", "12", "32", "128", "32", "32", "128", "32", "32", "128", "32", "32", "102", "28", "32", "128", "16", "32", "128", "28", "19", "128", "32", "9", "51", "16", "12", "102", "12", "22", "115", "9", "12", "51", "12", "16", "38", "64", "64", "256", "64", "64", "256", "64", "64", "256", "64", "64", "256", "64", "64", "256", "64", "64", "256", "64", "64", "256", "64", "64", "256", "64", "64", "256", "64", "64", "256", "64", "57", "204", "64", "25", "179", "19", "25", "204", "44", "19", "153", "38", "25", "76", "19", "32", "128", "19", "51", "76", "57", "32", "76"]], 11 | "xblocks" : ["int" , ["13", "15", "13"]], 12 | "estimated_FLOP" : ["float" , "173.023672"] 13 | } -------------------------------------------------------------------------------- /configs/NeurIPS-2019/C010-ResNet20.config: -------------------------------------------------------------------------------- 1 | { 2 | "dataset" : ["str" , "cifar"], 3 | "arch" : ["str" , "resnet"], 4 | "depth" : ["int" , "20"], 5 | "module" : ["str" , "ResNetBasicblock"], 6 | "super_type" : ["str" , "infer-shape"], 7 | "zero_init_residual" : ["bool" , "0"], 8 | "class_num" : ["int" , "10"], 9 | "search_mode" : ["str" , "shape"], 10 | "xchannels" : ["int" , ["3", "16", "6", "4", "4", "4", "4", "4", "32", "32", "12", "19", "32", "28", "64", "64", "64", "64", "64", "44"]], 11 | "xblocks" : ["int" , ["3", "3", "3"]], 12 | "estimated_FLOP" : ["float" , "22.444472"] 13 | } -------------------------------------------------------------------------------- /configs/NeurIPS-2019/C010-ResNet32.config: -------------------------------------------------------------------------------- 1 | { 2 | "dataset" : ["str" , "cifar"], 3 | "arch" : ["str" , "resnet"], 4 | "depth" : ["int" , "32"], 5 | "module" : ["str" , "ResNetBasicblock"], 6 | "super_type" : ["str" , "infer-shape"], 7 | "zero_init_residual" : ["bool" , "0"], 8 | "class_num" : ["int" , "10"], 9 | "search_mode" : ["str" , "shape"], 10 | "xchannels" : ["int" , ["3", "16", "6", "4", "12", "4", "4", "16", "9", "9", "6", "14", "32", "32", "9", "19", "28", "9", "32", "19", "32", "9", "64", "64", "64", "64", "64", "64", "64", "32", "38", "32"]], 11 | "xblocks" : ["int" , ["5", "5", "5"]], 12 | "estimated_FLOP" : ["float" , "34.945344"] 13 | } -------------------------------------------------------------------------------- /configs/NeurIPS-2019/C010-ResNet56.config: -------------------------------------------------------------------------------- 1 | { 2 | "dataset" : ["str" , "cifar"], 3 | "arch" : ["str" , "resnet"], 4 | "depth" : ["int" , "56"], 5 | "module" : ["str" , "ResNetBasicblock"], 6 | "super_type" : ["str" , "infer-shape"], 7 | "zero_init_residual" : ["bool" , "0"], 8 | "class_num" : ["int" , "10"], 9 | "search_mode" : ["str" , "shape"], 10 | "xchannels" : ["int" , ["3", "16", "16", "16", "14", "11", "9", "16", "12", "16", "6", "16", "4", "8", "4", "14", "6", "4", "4", "4", "32", "32", "32", "32", "32", "32", "22", "28", "32", "32", "19", "9", "19", "16", "9", "25", "16", "9", "64", "64", "64", "64", "64", "64", "64", "64", "64", "51", "19", "19", "32", "19", "19", "32", "19", "25"]], 11 | "xblocks" : ["int" , ["5", "5", "5"]], 12 | "estimated_FLOP" : ["float" , "57.93305"] 13 | } -------------------------------------------------------------------------------- /configs/NeurIPS-2019/C100-ResNet110.config: -------------------------------------------------------------------------------- 1 | { 2 | "dataset" : ["str" , "cifar"], 3 | "arch" : ["str" , "resnet"], 4 | "depth" : ["int" , "110"], 5 | "module" : ["str" , "ResNetBasicblock"], 6 | "super_type" : ["str" , "infer-shape"], 7 | "zero_init_residual" : ["bool" , "0"], 8 | "class_num" : ["int" , "100"], 9 | "search_mode" : ["str" , "shape"], 10 | "xchannels" : ["int" , ["3", "16", "14", "16", "11", "14", "16", "16", "11", "16", "9", "14", "12", "16", "16", "16", "8", "16", "14", "16", "12", "4", "11", "16", "4", "4", "4", "16", "12", "4", "8", "4", "9", "4", "6", "14", "4", "4", "32", "32", "32", "32", "28", "28", "32", "32", "32", "32", "32", "28", "32", "28", "25", "32", "32", "32", "9", "9", "32", "32", "9", "25", "28", "32", "28", "9", "9", "32", "12", "12", "9", "22", "12", "9", "64", "64", "64", "64", "64", "64", "64", "64", "64", "64", "64", "64", "64", "64", "64", "44", "64", "57", "19", "19", "19", "19", "25", "19", "25", "19", "25", "19", "19", "25", "19", "19", "19", "25", "25", "19"]], 11 | "xblocks" : ["int" , ["13", "9", "11"]], 12 | "estimated_FLOP" : ["float" , "117.653164"] 13 | } -------------------------------------------------------------------------------- /configs/NeurIPS-2019/C100-ResNet164.config: -------------------------------------------------------------------------------- 1 | { 2 | "dataset" : ["str" , "cifar"], 3 | "arch" : ["str" , "resnet"], 4 | "depth" : ["int" , "164"], 5 | "module" : ["str" , "ResNetBottleneck"], 6 | "super_type" : ["str" , "infer-shape"], 7 | "zero_init_residual" : ["bool" , "0"], 8 | "class_num" : ["int" , "100"], 9 | "search_mode" : ["str" , "shape"], 10 | "xchannels" : ["int" , ["3", "16", "16", "16", "57", "6", "11", "64", "4", "6", "51", "6", "9", "64", "4", "8", "64", "4", "14", "64", "4", "8", "64", "4", "8", "64", "6", "12", "64", "6", "16", "64", "8", "16", "64", "14", "12", "64", "4", "16", "64", "4", "14", "64", "11", "16", "64", "4", "14", "64", "11", "4", "64", "4", "4", "19", "25", "32", "128", "28", "32", "115", "28", "32", "128", "25", "32", "128", "32", "32", "128", "25", "32", "128", "12", "32", "128", "25", "32", "128", "28", "32", "128", "25", "28", "128", "32", "32", "128", "28", "19", "128", "32", "32", "128", "19", "28", "128", "9", "19", "128", "28", "9", "89", "28", "19", "128", "9", "16", "38", "64", "64", "256", "64", "64", "256", "64", "64", "256", "64", "64", "256", "64", "64", "256", "64", "64", "256", "64", "64", "256", "64", "64", "256", "64", "64", "256", "64", "64", "204", "64", "64", "179", "64", "64", "102", "64", "64", "102", "44", "19", "76", "19", "19", "76", "19", "38", "76", "25", "38", "153", "44", "25", "230"]], 11 | "xblocks" : ["int" , ["15", "15", "15"]], 12 | "estimated_FLOP" : ["float" , "165.583512"] 13 | } -------------------------------------------------------------------------------- /configs/NeurIPS-2019/C100-ResNet20.config: -------------------------------------------------------------------------------- 1 | { 2 | "dataset" : ["str" , "cifar"], 3 | "arch" : ["str" , "resnet"], 4 | "depth" : ["int" , "20"], 5 | "module" : ["str" , "ResNetBasicblock"], 6 | "super_type" : ["str" , "infer-shape"], 7 | "zero_init_residual" : ["bool" , "0"], 8 | "class_num" : ["int" , "100"], 9 | "search_mode" : ["str" , "shape"], 10 | "xchannels" : ["int" , ["3", "16", "4", "4", "4", "4", "6", "4", "32", "32", "9", "19", "32", "28", "64", "64", "64", "64", "64", "64"]], 11 | "xblocks" : ["int" , ["3", "3", "3"]], 12 | "estimated_FLOP" : ["float" , "22.433792"] 13 | } -------------------------------------------------------------------------------- /configs/NeurIPS-2019/C100-ResNet32.config: -------------------------------------------------------------------------------- 1 | { 2 | "dataset" : ["str" , "cifar"], 3 | "arch" : ["str" , "resnet"], 4 | "depth" : ["int" , "32"], 5 | "module" : ["str" , "ResNetBasicblock"], 6 | "super_type" : ["str" , "infer-shape"], 7 | "zero_init_residual" : ["bool" , "0"], 8 | "class_num" : ["int" , "100"], 9 | "xchannels" : ["int" , ["3", "16", "4", "4", "6", "11", "6", "4", "8", "4", "4", "4", "32", "32", "9", "28", "28", "28", "28", "28", "32", "32", "64", "64", "64", "64", "64", "64", "64", "64", "64", "64"]], 10 | "xblocks" : ["int" , ["5", "5", "5"]], 11 | "estimated_FLOP" : ["float" , "42.47"] 12 | } 13 | -------------------------------------------------------------------------------- /configs/NeurIPS-2019/C100-ResNet56.config: -------------------------------------------------------------------------------- 1 | { 2 | "dataset" : ["str" , "cifar"], 3 | "arch" : ["str" , "resnet"], 4 | "depth" : ["int" , "56"], 5 | "module" : ["str" , "ResNetBasicblock"], 6 | "super_type" : ["str" , "infer-shape"], 7 | "zero_init_residual" : ["bool" , "0"], 8 | "class_num" : ["int" , "100"], 9 | "search_mode" : ["str" , "shape"], 10 | "xchannels" : ["int" , ["3", "16", "16", "9", "14", "16", "14", "16", "8", "16", "8", "14", "4", "4", "4", "8", "4", "6", "4", "4", "32", "32", "32", "28", "32", "32", "32", "22", "32", "32", "32", "9", "25", "19", "25", "12", "9", "9", "64", "64", "64", "64", "64", "64", "64", "64", "64", "51", "19", "19", "19", "19", "25", "38", "19", "19"]], 11 | "xblocks" : ["int" , ["5", "5", "7"]], 12 | "estimated_FLOP" : ["float" , "59.472556"] 13 | } -------------------------------------------------------------------------------- /configs/NeurIPS-2019/ImageNet-ResNet18V1.config: -------------------------------------------------------------------------------- 1 | { 2 | "dataset" : ["str" , "imagenet"], 3 | "arch" : ["str" , "resnet"], 4 | "block_name" : ["str" , "BasicBlock"], 5 | "layers" : ["int" , ["2", "2", "2", "2"]], 6 | "deep_stem" : ["bool" , "0"], 7 | "zero_init_residual" : ["bool" , "1"], 8 | "class_num" : ["int" , "1000"], 9 | "search_mode" : ["str" , "shape"], 10 | "xchannels" : ["int" , ["3", "64", "25", "64", "38", "19", "128", "128", "38", "38", "256", "256", "256", "256", "512", "512", "512", "512"]], 11 | "xblocks" : ["int" , ["1", "1", "2", "2"]], 12 | "super_type" : ["str" , "infer-shape"], 13 | "estimated_FLOP" : ["float" , "1120.44032"] 14 | } -------------------------------------------------------------------------------- /configs/NeurIPS-2019/ImageNet-ResNet50V1.config: -------------------------------------------------------------------------------- 1 | { 2 | "dataset" : ["str" , "imagenet"], 3 | "arch" : ["str" , "resnet"], 4 | "block_name" : ["str" , "Bottleneck"], 5 | "layers" : ["int" , ["3", "4", "6", "3"]], 6 | "deep_stem" : ["bool" , "0"], 7 | "zero_init_residual" : ["bool" , "1"], 8 | "class_num" : ["int" , "1000"], 9 | "search_mode" : ["str" , "shape"], 10 | "xchannels" : ["int" , ["3", "45", "45", "30", "102", "33", "60", "154", "68", "70", "180", "38", "38", "307", "38", "38", "410", "64", "128", "358", "38", "51", "256", "76", "76", "512", "76", "76", "512", "179", "256", "614", "100", "102", "307", "179", "230", "614", "204", "102", "307", "153", "153", "1228", "512", "512", "1434", "512", "512", "1844"]], 11 | "xblocks" : ["int" , ["3", "4", "5", "3"]], 12 | "super_type" : ["str" , "infer-shape"], 13 | "estimated_FLOP" : ["float" , "2291.316289"] 14 | } 15 | -------------------------------------------------------------------------------- /configs/archs/CIFAR-ResNet08.config: -------------------------------------------------------------------------------- 1 | { 2 | "dataset" : ["str", "cifar"], 3 | "arch" : ["str", "resnet"], 4 | "depth" : ["int", 8], 5 | "module" : ["str", "ResNetBasicblock"], 6 | "super_type": ["str" , "basic"], 7 | "zero_init_residual" : ["bool", "0"] 8 | } 9 | -------------------------------------------------------------------------------- /configs/archs/CIFAR-ResNet1001.config: -------------------------------------------------------------------------------- 1 | { 2 | "dataset" : ["str", "cifar"], 3 | "arch" : ["str", "resnet"], 4 | "depth" : ["int", 1001], 5 | "module" : ["str", "ResNetBottleneck"], 6 | "super_type": ["str" , "basic"], 7 | "zero_init_residual" : ["bool", "0"] 8 | } 9 | -------------------------------------------------------------------------------- /configs/archs/CIFAR-ResNet110.config: -------------------------------------------------------------------------------- 1 | { 2 | "dataset" : ["str", "cifar"], 3 | "arch" : ["str", "resnet"], 4 | "depth" : ["int", 110], 5 | "module" : ["str", "ResNetBasicblock"], 6 | "super_type": ["str" , "basic"], 7 | "zero_init_residual" : ["bool", "0"] 8 | } 9 | -------------------------------------------------------------------------------- /configs/archs/CIFAR-ResNet164.config: -------------------------------------------------------------------------------- 1 | { 2 | "dataset" : ["str", "cifar"], 3 | "arch" : ["str", "resnet"], 4 | "depth" : ["int", 164], 5 | "module" : ["str", "ResNetBottleneck"], 6 | "super_type": ["str" , "basic"], 7 | "zero_init_residual" : ["bool", "0"] 8 | } 9 | -------------------------------------------------------------------------------- /configs/archs/CIFAR-ResNet20.config: -------------------------------------------------------------------------------- 1 | { 2 | "dataset" : ["str", "cifar"], 3 | "arch" : ["str", "resnet"], 4 | "depth" : ["int", 20], 5 | "module" : ["str", "ResNetBasicblock"], 6 | "super_type": ["str" , "basic"], 7 | "zero_init_residual" : ["bool", "0"] 8 | } 9 | -------------------------------------------------------------------------------- /configs/archs/CIFAR-ResNet32.config: -------------------------------------------------------------------------------- 1 | { 2 | "dataset" : ["str", "cifar"], 3 | "arch" : ["str", "resnet"], 4 | "depth" : ["int", 32], 5 | "module" : ["str", "ResNetBasicblock"], 6 | "super_type": ["str" , "basic"], 7 | "zero_init_residual" : ["bool", "0"] 8 | } 9 | -------------------------------------------------------------------------------- /configs/archs/CIFAR-ResNet56.config: -------------------------------------------------------------------------------- 1 | { 2 | "dataset" : ["str", "cifar"], 3 | "arch" : ["str", "resnet"], 4 | "depth" : ["int", 56], 5 | "module" : ["str", "ResNetBasicblock"], 6 | "super_type": ["str" , "basic"], 7 | "zero_init_residual" : ["bool", "0"] 8 | } 9 | -------------------------------------------------------------------------------- /configs/archs/CIFAR-SIM05.config: -------------------------------------------------------------------------------- 1 | { 2 | "dataset" : ["str", "cifar"], 3 | "arch" : ["str", "simres"], 4 | "depth" : ["int", 5], 5 | "super_type": ["str" , "basic"], 6 | "zero_init_residual" : ["bool", "0"] 7 | } 8 | -------------------------------------------------------------------------------- /configs/archs/CIFAR-WRN28-10.config: -------------------------------------------------------------------------------- 1 | { 2 | "dataset" : ["str", "cifar"], 3 | "arch" : ["str", "wideresnet"], 4 | "depth" : ["int", 28], 5 | "wide_factor":["int", 10], 6 | "dropout" : ["bool", 0], 7 | "super_type": ["str" , "basic"] 8 | } 9 | -------------------------------------------------------------------------------- /configs/archs/ImageNet-ResNet101V1.config: -------------------------------------------------------------------------------- 1 | { 2 | "dataset" : ["str", "imagenet"], 3 | "arch" : ["str", "resnet"], 4 | "block_name" : ["str", "Bottleneck"], 5 | "layers" : ["int", [3,4,23,3]], 6 | "deep_stem" : ["bool", 0], 7 | "zero_init_residual" : ["bool", "1"], 8 | "groups" : ["int", 1], 9 | "width_per_group" : ["int", 64], 10 | "norm_layer" : ["none", "None"] 11 | 12 | } 13 | -------------------------------------------------------------------------------- /configs/archs/ImageNet-ResNet101V2.config: -------------------------------------------------------------------------------- 1 | { 2 | "dataset" : ["str", "imagenet"], 3 | "arch" : ["str", "resnet"], 4 | "block_name" : ["str", "Bottleneck"], 5 | "layers" : ["int", [3,4,23,3]], 6 | "deep_stem" : ["bool", 1], 7 | "zero_init_residual" : ["bool", "1"], 8 | "groups" : ["int", 1], 9 | "width_per_group" : ["int", 64], 10 | "norm_layer" : ["none", "None"] 11 | 12 | } 13 | -------------------------------------------------------------------------------- /configs/archs/ImageNet-ResNet152V1.config: -------------------------------------------------------------------------------- 1 | { 2 | "dataset" : ["str", "imagenet"], 3 | "arch" : ["str", "resnet"], 4 | "block_name" : ["str", "Bottleneck"], 5 | "layers" : ["int", [3,8,36,3]], 6 | "deep_stem" : ["bool", 0], 7 | "zero_init_residual" : ["bool", "1"], 8 | "groups" : ["int", 1], 9 | "width_per_group" : ["int", 64], 10 | "norm_layer" : ["none", "None"] 11 | } 12 | -------------------------------------------------------------------------------- /configs/archs/ImageNet-ResNet152V2.config: -------------------------------------------------------------------------------- 1 | { 2 | "dataset" : ["str", "imagenet"], 3 | "arch" : ["str", "resnet"], 4 | "block_name" : ["str", "Bottleneck"], 5 | "layers" : ["int", [3,8,36,3]], 6 | "deep_stem" : ["bool", 1], 7 | "zero_init_residual" : ["bool", "1"], 8 | "groups" : ["int", 1], 9 | "width_per_group" : ["int", 64], 10 | "norm_layer" : ["none", "None"] 11 | } 12 | -------------------------------------------------------------------------------- /configs/archs/ImageNet-ResNet18V1.config: -------------------------------------------------------------------------------- 1 | { 2 | "dataset" : ["str", "imagenet"], 3 | "arch" : ["str", "resnet"], 4 | "block_name" : ["str", "BasicBlock"], 5 | "layers" : ["int", [2,2,2,2]], 6 | "deep_stem" : ["bool", 0], 7 | "zero_init_residual" : ["bool", "1"], 8 | "groups" : ["int", 1], 9 | "width_per_group" : ["int", 64], 10 | "norm_layer" : ["none", "None"] 11 | } 12 | -------------------------------------------------------------------------------- /configs/archs/ImageNet-ResNet18V2.config: -------------------------------------------------------------------------------- 1 | { 2 | "dataset" : ["str", "imagenet"], 3 | "arch" : ["str", "resnet"], 4 | "block_name" : ["str", "BasicBlock"], 5 | "layers" : ["int", [2,2,2,2]], 6 | "deep_stem" : ["bool", 1], 7 | "zero_init_residual" : ["bool", "1"], 8 | "groups" : ["int", 1], 9 | "width_per_group" : ["int", 64], 10 | "norm_layer" : ["none", "None"] 11 | 12 | } 13 | -------------------------------------------------------------------------------- /configs/archs/ImageNet-ResNet200V1.config: -------------------------------------------------------------------------------- 1 | { 2 | "dataset" : ["str", "imagenet"], 3 | "arch" : ["str", "resnet"], 4 | "block_name" : ["str", "Bottleneck"], 5 | "layers" : ["int", [3,24,36,3]], 6 | "deep_stem" : ["bool", 0], 7 | "zero_init_residual" : ["bool", "1"], 8 | "groups" : ["int", 1], 9 | "width_per_group" : ["int", 64], 10 | "norm_layer" : ["none", "None"] 11 | } 12 | -------------------------------------------------------------------------------- /configs/archs/ImageNet-ResNet200V2.config: -------------------------------------------------------------------------------- 1 | { 2 | "dataset" : ["str", "imagenet"], 3 | "arch" : ["str", "resnet"], 4 | "block_name" : ["str", "Bottleneck"], 5 | "layers" : ["int", [3,24,36,3]], 6 | "deep_stem" : ["bool", 1], 7 | "zero_init_residual" : ["bool", "1"], 8 | "groups" : ["int", 1], 9 | "width_per_group" : ["int", 64], 10 | "norm_layer" : ["none", "None"] 11 | } 12 | -------------------------------------------------------------------------------- /configs/archs/ImageNet-ResNet34V1.config: -------------------------------------------------------------------------------- 1 | { 2 | "dataset" : ["str", "imagenet"], 3 | "arch" : ["str", "resnet"], 4 | "block_name" : ["str", "BasicBlock"], 5 | "layers" : ["int", [3,4,6,3]], 6 | "deep_stem" : ["bool", 0], 7 | "zero_init_residual" : ["bool", "1"], 8 | "groups" : ["int", 1], 9 | "width_per_group" : ["int", 64], 10 | "norm_layer" : ["none", "None"] 11 | 12 | } 13 | -------------------------------------------------------------------------------- /configs/archs/ImageNet-ResNet34V2.config: -------------------------------------------------------------------------------- 1 | { 2 | "dataset" : ["str", "imagenet"], 3 | "arch" : ["str", "resnet"], 4 | "block_name" : ["str", "BasicBlock"], 5 | "layers" : ["int", [3,4,6,3]], 6 | "deep_stem" : ["bool", 1], 7 | "zero_init_residual" : ["bool", "1"], 8 | "groups" : ["int", 1], 9 | "width_per_group" : ["int", 64], 10 | "norm_layer" : ["none", "None"] 11 | 12 | } 13 | -------------------------------------------------------------------------------- /configs/archs/ImageNet-ResNet50V1.config: -------------------------------------------------------------------------------- 1 | { 2 | "dataset" : ["str", "imagenet"], 3 | "arch" : ["str", "resnet"], 4 | "block_name" : ["str", "Bottleneck"], 5 | "layers" : ["int", [3,4,6,3]], 6 | "deep_stem" : ["bool", 0], 7 | "zero_init_residual" : ["bool", "1"], 8 | "groups" : ["int", 1], 9 | "width_per_group" : ["int", 64], 10 | "norm_layer" : ["none", "None"] 11 | 12 | } 13 | -------------------------------------------------------------------------------- /configs/archs/ImageNet-ResNet50V2.config: -------------------------------------------------------------------------------- 1 | { 2 | "dataset" : ["str", "imagenet"], 3 | "arch" : ["str", "resnet"], 4 | "block_name" : ["str", "Bottleneck"], 5 | "layers" : ["int", [3,4,6,3]], 6 | "deep_stem" : ["bool", 1], 7 | "zero_init_residual" : ["bool", "1"], 8 | "groups" : ["int", 1], 9 | "width_per_group" : ["int", 64], 10 | "norm_layer" : ["none", "None"] 11 | 12 | } 13 | -------------------------------------------------------------------------------- /configs/archs/ImageNet-ResNext50-32x4dV2.config: -------------------------------------------------------------------------------- 1 | { 2 | "dataset" : ["str", "imagenet"], 3 | "arch" : ["str", "resnet"], 4 | "block_name" : ["str", "Bottleneck"], 5 | "layers" : ["int", [3,4,6,3]], 6 | "deep_stem" : ["bool", 1], 7 | "zero_init_residual" : ["bool", "1"], 8 | "groups" : ["int", 32], 9 | "width_per_group" : ["int", 4], 10 | "norm_layer" : ["none", "None"] 11 | 12 | } 13 | -------------------------------------------------------------------------------- /configs/archs/NAS-CIFAR-DARTS.config: -------------------------------------------------------------------------------- 1 | { 2 | "arch" : ["str", "dxys"], 3 | "genotype" : ["str", "DARTS"], 4 | "dataset" : ["str", "cifar"], 5 | "ichannel" : ["int", 36], 6 | "layers" : ["int", 6], 7 | "stem_multi": ["int", 3], 8 | "auxiliary" : ["bool", 1], 9 | "drop_path_prob": ["float", 0.2] 10 | } 11 | -------------------------------------------------------------------------------- /configs/archs/NAS-CIFAR-GDAS_V1.config: -------------------------------------------------------------------------------- 1 | { 2 | "arch" : ["str", "dxys"], 3 | "genotype" : ["str", "GDAS_V1"], 4 | "dataset" : ["str", "cifar"], 5 | "ichannel" : ["int", 36], 6 | "layers" : ["int", 6], 7 | "stem_multi": ["int", 3], 8 | "auxiliary" : ["bool", 1], 9 | "drop_path_prob": ["float", 0.2] 10 | } 11 | -------------------------------------------------------------------------------- /configs/archs/NAS-CIFAR-NAS.config: -------------------------------------------------------------------------------- 1 | { 2 | "arch" : ["str", "dxys"], 3 | "genotype" : ["str", "NASNet"], 4 | "dataset" : ["str", "cifar"], 5 | "ichannel" : ["int", 33], 6 | "layers" : ["int", 6], 7 | "stem_multi": ["int", 3], 8 | "auxiliary" : ["bool", 1], 9 | "drop_path_prob": ["float", 0.2] 10 | } 11 | -------------------------------------------------------------------------------- /configs/archs/NAS-CIFAR-SETN.config: -------------------------------------------------------------------------------- 1 | { 2 | "arch" : ["str", "dxys"], 3 | "genotype" : ["str", "SETN"], 4 | "dataset" : ["str", "cifar"], 5 | "ichannel" : ["int", 36], 6 | "layers" : ["int", 6], 7 | "stem_multi": ["int", 3], 8 | "auxiliary" : ["bool", 1], 9 | "drop_path_prob": ["float", 0.2] 10 | } 11 | -------------------------------------------------------------------------------- /configs/archs/NAS-CIFAR-none.config: -------------------------------------------------------------------------------- 1 | { 2 | "super_type": ["str", "infer-nasnet.cifar"], 3 | "genotype" : ["none", "none"], 4 | "dataset" : ["str", "cifar"], 5 | "ichannel" : ["int", 33], 6 | "layers" : ["int", 6], 7 | "stem_multi": ["int", 3], 8 | "auxiliary" : ["bool", 1], 9 | "drop_path_prob": ["float", 0.2] 10 | } 11 | -------------------------------------------------------------------------------- /configs/archs/NAS-IMAGENET-DARTS_V2.config: -------------------------------------------------------------------------------- 1 | { 2 | "arch" : ["str", "dxys"], 3 | "genotype" : ["str", "DARTS_V2"], 4 | "dataset" : ["str", "imagenet"], 5 | "ichannel" : ["int", 48], 6 | "layers" : ["int", 4], 7 | "auxiliary" : ["bool", 1], 8 | "drop_path_prob": ["float", 0] 9 | } 10 | -------------------------------------------------------------------------------- /configs/archs/NAS-IMAGENET-GDAS_V1.config: -------------------------------------------------------------------------------- 1 | { 2 | "arch" : ["str", "dxys"], 3 | "genotype" : ["str", "GDAS_V1"], 4 | "dataset" : ["str", "imagenet"], 5 | "ichannel" : ["int", 50], 6 | "layers" : ["int", 4], 7 | "auxiliary" : ["bool", 1], 8 | "drop_path_prob": ["float", 0] 9 | } 10 | -------------------------------------------------------------------------------- /configs/archs/NAS-IMAGENET-SETN.config: -------------------------------------------------------------------------------- 1 | { 2 | "arch" : ["str", "dxys"], 3 | "genotype" : ["str", "SETN"], 4 | "dataset" : ["str", "imagenet"], 5 | "ichannel" : ["int", 58], 6 | "layers" : ["int", 2], 7 | "auxiliary" : ["bool", 1], 8 | "drop_path_prob": ["float", 0] 9 | } 10 | -------------------------------------------------------------------------------- /configs/archs/NAS-IMAGENET-SETN1.config: -------------------------------------------------------------------------------- 1 | { 2 | "arch" : ["str", "dxys"], 3 | "genotype" : ["str", "SETN"], 4 | "dataset" : ["str", "imagenet"], 5 | "ichannel" : ["int", 73], 6 | "layers" : ["int", 1], 7 | "auxiliary" : ["bool", 1], 8 | "drop_path_prob": ["float", 0] 9 | } 10 | -------------------------------------------------------------------------------- /configs/archs/NAS-IMAGENET-SETN2.config: -------------------------------------------------------------------------------- 1 | { 2 | "arch" : ["str", "dxys"], 3 | "genotype" : ["str", "SETN"], 4 | "dataset" : ["str", "imagenet"], 5 | "ichannel" : ["int", 58], 6 | "layers" : ["int", 2], 7 | "auxiliary" : ["bool", 1], 8 | "drop_path_prob": ["float", 0] 9 | } 10 | -------------------------------------------------------------------------------- /configs/archs/NAS-IMAGENET-SETN3.config: -------------------------------------------------------------------------------- 1 | { 2 | "arch" : ["str", "dxys"], 3 | "genotype" : ["str", "SETN"], 4 | "dataset" : ["str", "imagenet"], 5 | "ichannel" : ["int", 49], 6 | "layers" : ["int", 3], 7 | "auxiliary" : ["bool", 1], 8 | "drop_path_prob": ["float", 0] 9 | } 10 | -------------------------------------------------------------------------------- /configs/archs/NAS-IMAGENET-SETN4.config: -------------------------------------------------------------------------------- 1 | { 2 | "arch" : ["str", "dxys"], 3 | "genotype" : ["str", "SETN"], 4 | "dataset" : ["str", "imagenet"], 5 | "ichannel" : ["int", 44], 6 | "layers" : ["int", 4], 7 | "auxiliary" : ["bool", 1], 8 | "drop_path_prob": ["float", 0] 9 | } 10 | -------------------------------------------------------------------------------- /configs/archs/NAS-IMAGENET-none.config: -------------------------------------------------------------------------------- 1 | { 2 | "super_type": ["str", "infer-nasnet.imagenet"], 3 | "genotype" : ["none", "none"], 4 | "dataset" : ["str", "imagenet"], 5 | "ichannel" : ["int", 50], 6 | "layers" : ["int", 4], 7 | "auxiliary" : ["bool", 1], 8 | "drop_path_prob": ["float", 0] 9 | } 10 | -------------------------------------------------------------------------------- /configs/compares/CIFAR010-TAS-R110.config: -------------------------------------------------------------------------------- 1 | { 2 | "dataset" : ["str" , "cifar"], 3 | "arch" : ["str" , "resnet"], 4 | "depth" : ["int" , "110"], 5 | "module" : ["str" , "ResNetBasicblock"], 6 | "super_type" : ["str" , "infer-shape"], 7 | "zero_init_residual" : ["bool" , "0"], 8 | "class_num" : ["int" , "10"], 9 | "search_mode" : ["str" , "shape"], 10 | "xchannels" : ["int" , ["3", "16", "14", "16", "14", "16", "16", "16", "16", "16", "14", "16", "16", "16", "12", "16", "16", "16", "9", "16", "8", "16", "4", "16", "4", "4", "4", "16", "4", "4", "4", "4", "6", "6", "4", "6", "11", "4", "32", "32", "32", "32", "32", "32", "32", "32", "28", "32", "32", "28", "22", "22", "22", "32", "32", "25", "28", "9", "9", "28", "12", "9", "12", "32", "9", "9", "22", "12", "16", "9", "12", "9", "9", "9", "64", "64", "64", "64", "64", "64", "64", "64", "64", "64", "64", "64", "64", "64", "64", "38", "64", "25", "19", "19", "19", "19", "19", "25", "32", "19", "19", "25", "25", "19", "19", "38", "38", "19", "19", "51"]], 11 | "xblocks" : ["int" , ["11", "11", "9"]], 12 | "estimated_FLOP" : ["float" , "117.498238"] 13 | } -------------------------------------------------------------------------------- /configs/compares/CIFAR010-TAS-R56.config: -------------------------------------------------------------------------------- 1 | { 2 | "dataset" : ["str" , "cifar"], 3 | "arch" : ["str" , "resnet"], 4 | "depth" : ["int" , "56"], 5 | "module" : ["str" , "ResNetBasicblock"], 6 | "super_type" : ["str" , "infer-shape"], 7 | "zero_init_residual" : ["bool" , "0"], 8 | "class_num" : ["int" , "10"], 9 | "search_mode" : ["str" , "shape"], 10 | "xchannels" : ["int" , ["3", "16", "16", "16", "14", "14", "14", "16", "12", "16", "6", "16", "4", "4", "4", "14", "4", "4", "6", "4", "32", "32", "32", "32", "32", "32", "22", "25", "16", "32", "19", "9", "9", "16", "25", "12", "16", "9", "64", "64", "64", "64", "64", "64", "64", "64", "64", "51", "19", "19", "32", "51", "25", "32", "19", "19"]], 11 | "xblocks" : ["int" , ["5", "5", "5"]], 12 | "estimated_FLOP" : ["float" , "57.52595"] 13 | } -------------------------------------------------------------------------------- /configs/compares/CIFAR100-FIX-R32.config: -------------------------------------------------------------------------------- 1 | { 2 | "dataset" : ["str" , "cifar"], 3 | "arch" : ["str" , "resnet"], 4 | "depth" : ["int" , "32"], 5 | "module" : ["str" , "ResNetBasicblock"], 6 | "super_type" : ["str" , "infer-width"], 7 | "zero_init_residual" : ["bool" , "0"], 8 | "class_num" : ["int" , "100"], 9 | "xchannels" : ["int" , ["3", "12", "12", "12", "12", "12", "12", "12", "12", "12", "12", "12", "25", "25", "25", "25", "25", "25", "25", "25", "25", "25", "50", "50", "50", "50", "50", "50", "50", "50", "50", "50"]], 10 | "estimated_FLOP" : ["float" , "41.095816"] 11 | } 12 | -------------------------------------------------------------------------------- /configs/compares/CIFAR100-RAND-R32.config: -------------------------------------------------------------------------------- 1 | { 2 | "dataset" : ["str" , "cifar"], 3 | "arch" : ["str" , "resnet"], 4 | "depth" : ["int" , "32"], 5 | "module" : ["str" , "ResNetBasicblock"], 6 | "super_type" : ["str" , "infer-width"], 7 | "zero_init_residual" : ["bool" , "0"], 8 | "class_num" : ["int" , "100"], 9 | "xchannels" : ["int" , ["3", "4", "11", "11", "11", "12", "14", "16", "8", "9", "6", "12", "28", "32", "28", "32", "12", "25", "28", "22", "28", "25", "57", "19", "38", "64", "64", "51", "57", "64", "64", "57"]], 10 | "estimated_FLOP" : ["float" , "42.908996"] 11 | } 12 | -------------------------------------------------------------------------------- /configs/compares/CIFAR100-TAS-R110.config: -------------------------------------------------------------------------------- 1 | { 2 | "dataset" : ["str" , "cifar"], 3 | "arch" : ["str" , "resnet"], 4 | "depth" : ["int" , "110"], 5 | "module" : ["str" , "ResNetBasicblock"], 6 | "super_type" : ["str" , "infer-shape"], 7 | "zero_init_residual" : ["bool" , "0"], 8 | "class_num" : ["int" , "100"], 9 | "search_mode" : ["str" , "shape"], 10 | "xchannels" : ["int" , ["3", "16", "14", "16", "11", "14", "16", "16", "11", "16", "9", "14", "12", "16", "16", "16", "8", "16", "14", "16", "12", "4", "11", "16", "4", "4", "4", "16", "12", "4", "8", "4", "9", "4", "6", "14", "4", "4", "32", "32", "32", "32", "28", "28", "32", "32", "32", "32", "32", "28", "32", "28", "25", "32", "32", "32", "9", "9", "32", "32", "9", "25", "28", "32", "28", "9", "9", "32", "12", "12", "9", "22", "12", "9", "64", "64", "64", "64", "64", "64", "64", "64", "64", "64", "64", "64", "64", "64", "64", "44", "64", "57", "19", "19", "19", "19", "25", "19", "25", "19", "25", "19", "19", "25", "19", "19", "19", "25", "25", "19"]], 11 | "xblocks" : ["int" , ["13", "9", "11"]], 12 | "estimated_FLOP" : ["float" , "117.653164"] 13 | } -------------------------------------------------------------------------------- /configs/compares/CIFAR100-TAS-R32.config: -------------------------------------------------------------------------------- 1 | { 2 | "dataset" : ["str" , "cifar"], 3 | "arch" : ["str" , "resnet"], 4 | "depth" : ["int" , "32"], 5 | "module" : ["str" , "ResNetBasicblock"], 6 | "super_type" : ["str" , "infer-width"], 7 | "zero_init_residual" : ["bool" , "0"], 8 | "class_num" : ["int" , "100"], 9 | "xchannels" : ["int" , ["3", "16", "4", "4", "4", "14", "6", "4", "8", "4", "4", "4", "32", "32", "9", "28", "28", "28", "28", "28", "32", "32", "64", "64", "64", "64", "64", "64", "64", "64", "64", "64"]], 10 | "estimated_FLOP" : ["float" , "42.493184"] 11 | } 12 | -------------------------------------------------------------------------------- /configs/compares/CIFAR100-TAS-R56.config: -------------------------------------------------------------------------------- 1 | { 2 | "dataset" : ["str" , "cifar"], 3 | "arch" : ["str" , "resnet"], 4 | "depth" : ["int" , "56"], 5 | "module" : ["str" , "ResNetBasicblock"], 6 | "super_type" : ["str" , "infer-shape"], 7 | "zero_init_residual" : ["bool" , "0"], 8 | "class_num" : ["int" , "100"], 9 | "search_mode" : ["str" , "shape"], 10 | "xchannels" : ["int" , ["3", "16", "16", "9", "14", "16", "14", "16", "8", "16", "8", "14", "4", "4", "4", "8", "4", "6", "4", "4", "32", "32", "32", "28", "32", "32", "32", "22", "32", "32", "32", "9", "25", "19", "25", "12", "9", "9", "64", "64", "64", "64", "64", "64", "64", "64", "64", "51", "19", "19", "19", "19", "25", "38", "19", "19"]], 11 | "xblocks" : ["int" , ["5", "5", "7"]], 12 | "estimated_FLOP" : ["float" , "59.472556"] 13 | } -------------------------------------------------------------------------------- /configs/nas-benchmark/CIFAR.config: -------------------------------------------------------------------------------- 1 | { 2 | "scheduler": ["str", "cos"], 3 | "eta_min" : ["float", "0.0"], 4 | "epochs" : ["int", "200"], 5 | "warmup" : ["int", "0"], 6 | "optim" : ["str", "SGD"], 7 | "LR" : ["float", "0.1"], 8 | "decay" : ["float", "0.0005"], 9 | "momentum" : ["float", "0.9"], 10 | "nesterov" : ["bool", "1"], 11 | "criterion": ["str", "Softmax"], 12 | "batch_size": ["int", "256"] 13 | } 14 | -------------------------------------------------------------------------------- /configs/nas-benchmark/ImageNet-16.config: -------------------------------------------------------------------------------- 1 | { 2 | "scheduler": ["str", "cos"], 3 | "eta_min" : ["float", "0.0"], 4 | "epochs" : ["int", "200"], 5 | "warmup" : ["int", "0"], 6 | "optim" : ["str", "SGD"], 7 | "LR" : ["float", "0.1"], 8 | "decay" : ["float", "0.0005"], 9 | "momentum" : ["float", "0.9"], 10 | "nesterov" : ["bool", "1"], 11 | "criterion": ["str", "Softmax"], 12 | "batch_size": ["int", "256"] 13 | } 14 | -------------------------------------------------------------------------------- /configs/nas-benchmark/LESS.config: -------------------------------------------------------------------------------- 1 | { 2 | "scheduler": ["str", "cos"], 3 | "eta_min" : ["float", "0.0"], 4 | "epochs" : ["int", "12"], 5 | "warmup" : ["int", "0"], 6 | "optim" : ["str", "SGD"], 7 | "LR" : ["float", "0.1"], 8 | "decay" : ["float", "0.0005"], 9 | "momentum" : ["float", "0.9"], 10 | "nesterov" : ["bool", "1"], 11 | "criterion": ["str", "Softmax"], 12 | "batch_size": ["int", "256"] 13 | } 14 | -------------------------------------------------------------------------------- /configs/nas-benchmark/algos/DARTS.config: -------------------------------------------------------------------------------- 1 | { 2 | "scheduler": ["str", "cos"], 3 | "LR" : ["float", "0.025"], 4 | "eta_min" : ["float", "0.001"], 5 | "epochs" : ["int", "50"], 6 | "warmup" : ["int", "0"], 7 | "optim" : ["str", "SGD"], 8 | "decay" : ["float", "0.0005"], 9 | "momentum" : ["float", "0.9"], 10 | "nesterov" : ["bool", "1"], 11 | "criterion": ["str", "Softmax"], 12 | "batch_size": ["int", "64"] 13 | } 14 | -------------------------------------------------------------------------------- /configs/nas-benchmark/algos/ENAS.config: -------------------------------------------------------------------------------- 1 | { 2 | "scheduler": ["str", "cos"], 3 | "LR" : ["float", "0.05"], 4 | "eta_min" : ["float", "0.0005"], 5 | "epochs" : ["int", "250"], 6 | "T_max" : ["int", "10"], 7 | "warmup" : ["int", "0"], 8 | "optim" : ["str", "SGD"], 9 | "decay" : ["float", "0.00025"], 10 | "momentum" : ["float", "0.9"], 11 | "nesterov" : ["bool", "1"], 12 | "controller_lr" : ["float", "0.001"], 13 | "controller_betas": ["float", [0, 0.999]], 14 | "controller_eps" : ["float", 0.001], 15 | "criterion": ["str", "Softmax"], 16 | "batch_size": ["int", "128"] 17 | } 18 | -------------------------------------------------------------------------------- /configs/nas-benchmark/algos/GDAS.config: -------------------------------------------------------------------------------- 1 | { 2 | "scheduler": ["str", "cos"], 3 | "LR" : ["float", "0.025"], 4 | "eta_min" : ["float", "0.001"], 5 | "epochs" : ["int", "250"], 6 | "warmup" : ["int", "0"], 7 | "optim" : ["str", "SGD"], 8 | "decay" : ["float", "0.0005"], 9 | "momentum" : ["float", "0.9"], 10 | "nesterov" : ["bool", "1"], 11 | "criterion": ["str", "Softmax"], 12 | "batch_size": ["int", "64"] 13 | } 14 | -------------------------------------------------------------------------------- /configs/nas-benchmark/algos/R-EA.config: -------------------------------------------------------------------------------- 1 | { 2 | "scheduler": ["str", "cos"], 3 | "eta_min" : ["float", "0.001"], 4 | "epochs" : ["int", "25"], 5 | "warmup" : ["int", "0"], 6 | "optim" : ["str", "SGD"], 7 | "LR" : ["float", "0.025"], 8 | "decay" : ["float", "0.0005"], 9 | "momentum" : ["float", "0.9"], 10 | "nesterov" : ["bool", "1"], 11 | "criterion": ["str", "Softmax"], 12 | "batch_size": ["int", "64"] 13 | } 14 | -------------------------------------------------------------------------------- /configs/nas-benchmark/algos/RANDOM.config: -------------------------------------------------------------------------------- 1 | { 2 | "scheduler": ["str", "cos"], 3 | "LR" : ["float", "0.025"], 4 | "eta_min" : ["float", "0.001"], 5 | "epochs" : ["int", "250"], 6 | "warmup" : ["int", "0"], 7 | "optim" : ["str", "SGD"], 8 | "decay" : ["float", "0.0005"], 9 | "momentum" : ["float", "0.9"], 10 | "nesterov" : ["bool", "1"], 11 | "criterion": ["str", "Softmax"], 12 | "batch_size": ["int", "64"], 13 | "test_batch_size": ["int", "512"] 14 | } 15 | -------------------------------------------------------------------------------- /configs/nas-benchmark/algos/SETN.config: -------------------------------------------------------------------------------- 1 | { 2 | "scheduler": ["str", "cos"], 3 | "LR" : ["float", "0.025"], 4 | "eta_min" : ["float", "0.001"], 5 | "epochs" : ["int", "250"], 6 | "warmup" : ["int", "0"], 7 | "optim" : ["str", "SGD"], 8 | "decay" : ["float", "0.0005"], 9 | "momentum" : ["float", "0.9"], 10 | "nesterov" : ["bool", "1"], 11 | "criterion": ["str", "Softmax"], 12 | "batch_size": ["int", "64"], 13 | "test_batch_size": ["int", "512"] 14 | } 15 | -------------------------------------------------------------------------------- /configs/nas-benchmark/algos/weight-sharing.config: -------------------------------------------------------------------------------- 1 | { 2 | "scheduler": ["str", "cos"], 3 | "LR" : ["float", "0.025"], 4 | "eta_min" : ["float", "0.001"], 5 | "epochs" : ["int", "100"], 6 | "warmup" : ["int", "0"], 7 | "optim" : ["str", "SGD"], 8 | "decay" : ["float", "0.0005"], 9 | "momentum" : ["float", "0.9"], 10 | "nesterov" : ["bool", "1"], 11 | "criterion": ["str", "Softmax"], 12 | "batch_size": ["int", "64"], 13 | "test_batch_size": ["int", "512"] 14 | } 15 | -------------------------------------------------------------------------------- /configs/nas-benchmark/hyper-opts/01E.config: -------------------------------------------------------------------------------- 1 | { 2 | "scheduler": ["str", "cos"], 3 | "eta_min" : ["float", "0.0"], 4 | "epochs" : ["int", "1"], 5 | "warmup" : ["int", "0"], 6 | "optim" : ["str", "SGD"], 7 | "LR" : ["float", "0.1"], 8 | "decay" : ["float", "0.0005"], 9 | "momentum" : ["float", "0.9"], 10 | "nesterov" : ["bool", "1"], 11 | "criterion": ["str", "Softmax"], 12 | "batch_size": ["int", "256"] 13 | } 14 | -------------------------------------------------------------------------------- /configs/nas-benchmark/hyper-opts/12E.config: -------------------------------------------------------------------------------- 1 | { 2 | "scheduler": ["str", "cos"], 3 | "eta_min" : ["float", "0.0"], 4 | "epochs" : ["int", "12"], 5 | "warmup" : ["int", "0"], 6 | "optim" : ["str", "SGD"], 7 | "LR" : ["float", "0.1"], 8 | "decay" : ["float", "0.0005"], 9 | "momentum" : ["float", "0.9"], 10 | "nesterov" : ["bool", "1"], 11 | "criterion": ["str", "Softmax"], 12 | "batch_size": ["int", "256"] 13 | } -------------------------------------------------------------------------------- /configs/nas-benchmark/hyper-opts/200E.config: -------------------------------------------------------------------------------- 1 | { 2 | "scheduler": ["str", "cos"], 3 | "eta_min" : ["float", "0.0"], 4 | "epochs" : ["int", "200"], 5 | "warmup" : ["int", "0"], 6 | "optim" : ["str", "SGD"], 7 | "LR" : ["float", "0.1"], 8 | "decay" : ["float", "0.0005"], 9 | "momentum" : ["float", "0.9"], 10 | "nesterov" : ["bool", "1"], 11 | "criterion": ["str", "Softmax"], 12 | "batch_size": ["int", "256"] 13 | } 14 | -------------------------------------------------------------------------------- /configs/nas-benchmark/hyper-opts/90E.config: -------------------------------------------------------------------------------- 1 | { 2 | "scheduler": ["str", "cos"], 3 | "eta_min" : ["float", "0.0"], 4 | "epochs" : ["int", "90"], 5 | "warmup" : ["int", "0"], 6 | "optim" : ["str", "SGD"], 7 | "LR" : ["float", "0.1"], 8 | "decay" : ["float", "0.0005"], 9 | "momentum" : ["float", "0.9"], 10 | "nesterov" : ["bool", "1"], 11 | "criterion": ["str", "Softmax"], 12 | "batch_size": ["int", "256"] 13 | } 14 | -------------------------------------------------------------------------------- /configs/opts/CIFAR-E300-W5-L1-COS.config: -------------------------------------------------------------------------------- 1 | { 2 | "scheduler": ["str", "cos"], 3 | "eta_min" : ["float", "0.0"], 4 | "epochs" : ["int", "300"], 5 | "warmup" : ["int", "5"], 6 | "optim" : ["str", "SGD"], 7 | "LR" : ["float", "0.1"], 8 | "decay" : ["float", "0.0005"], 9 | "momentum" : ["float", "0.9"], 10 | "nesterov" : ["bool", "1"], 11 | "criterion": ["str", "Softmax"], 12 | "auxiliary": ["float", "-1"] 13 | } 14 | -------------------------------------------------------------------------------- /configs/opts/CIFAR-E300-W5-L4-COS.config: -------------------------------------------------------------------------------- 1 | { 2 | "scheduler": ["str", "cos"], 3 | "eta_min" : ["float", "0.0"], 4 | "epochs" : ["int", "300"], 5 | "warmup" : ["int", "5"], 6 | "optim" : ["str", "SGD"], 7 | "LR" : ["float", "0.4"], 8 | "decay" : ["float", "0.0005"], 9 | "momentum" : ["float", "0.9"], 10 | "nesterov" : ["bool", "1"], 11 | "criterion": ["str", "Softmax"] 12 | } 13 | -------------------------------------------------------------------------------- /configs/opts/CIFAR-E600-W5-L1-COS.config: -------------------------------------------------------------------------------- 1 | { 2 | "scheduler": ["str", "cos"], 3 | "eta_min" : ["float", "0.0"], 4 | "epochs" : ["int", "600"], 5 | "warmup" : ["int", "5"], 6 | "optim" : ["str", "SGD"], 7 | "LR" : ["float", "0.1"], 8 | "decay" : ["float", "0.0005"], 9 | "momentum" : ["float", "0.9"], 10 | "nesterov" : ["bool", "1"], 11 | "criterion": ["str", "Softmax"] 12 | } 13 | -------------------------------------------------------------------------------- /configs/opts/CIFAR-Fast-Random.config: -------------------------------------------------------------------------------- 1 | { 2 | "scheduler": ["str", "cos"], 3 | "eta_min" : ["float", "0.0"], 4 | "epochs" : ["int", "100"], 5 | "warmup" : ["int", "5"], 6 | "optim" : ["str", "SGD"], 7 | "LR" : ["float", "0.1"], 8 | "decay" : ["float", "0.0005"], 9 | "momentum" : ["float", "0.9"], 10 | "nesterov" : ["bool", "1"], 11 | "criterion": ["str", "Softmax"], 12 | "auxiliary": ["float", "-1"] 13 | } 14 | -------------------------------------------------------------------------------- /configs/opts/CIFAR-Slow-Random.config: -------------------------------------------------------------------------------- 1 | { 2 | "scheduler": ["str", "cos"], 3 | "eta_min" : ["float", "0.0"], 4 | "epochs" : ["int", "300"], 5 | "warmup" : ["int", "5"], 6 | "optim" : ["str", "SGD"], 7 | "LR" : ["float", "0.1"], 8 | "decay" : ["float", "0.0005"], 9 | "momentum" : ["float", "0.9"], 10 | "nesterov" : ["bool", "1"], 11 | "criterion": ["str", "Softmax"], 12 | "auxiliary": ["float", "-1"] 13 | } 14 | -------------------------------------------------------------------------------- /configs/opts/Com-Paddle-NAS.config: -------------------------------------------------------------------------------- 1 | { 2 | "scheduler": ["str", "cos"], 3 | "eta_min" : ["float", "0.0"], 4 | "epochs" : ["int", "595"], 5 | "warmup" : ["int", "5"], 6 | "optim" : ["str", "SGD"], 7 | "LR" : ["float", "0.025"], 8 | "decay" : ["float", "0.0005"], 9 | "momentum" : ["float", "0.9"], 10 | "nesterov" : ["bool", "1"], 11 | "criterion": ["str", "Softmax"], 12 | "auxiliary": ["float", "0.4"] 13 | } 14 | -------------------------------------------------------------------------------- /configs/opts/Com-Paddle-RES.config: -------------------------------------------------------------------------------- 1 | { 2 | "scheduler": ["str", "cos"], 3 | "eta_min" : ["float", "0.0"], 4 | "epochs" : ["int", "300"], 5 | "warmup" : ["int", "5"] , 6 | "optim" : ["str", "SGD"], 7 | "LR" : ["float", "0.1"], 8 | "decay" : ["float", "0.0005"], 9 | "momentum" : ["float", "0.9"], 10 | "nesterov" : ["bool" , "1"] , 11 | "criterion": ["str" , "Softmax"], 12 | "auxiliary": ["float", "-1"] 13 | } 14 | -------------------------------------------------------------------------------- /configs/opts/ImageNet-E120-Cos-Smooth.config: -------------------------------------------------------------------------------- 1 | { 2 | "scheduler": ["str", "cos"], 3 | "eta_min" : ["float", "0.0"], 4 | "epochs" : ["int", "120"], 5 | "warmup" : ["int", "5"], 6 | "optim" : ["str", "SGD"], 7 | "LR" : ["float", "0.1"], 8 | "decay" : ["float", "0.0001"], 9 | "momentum" : ["float", "0.9"], 10 | "nesterov" : ["bool", "1"], 11 | "criterion": ["str", "SmoothSoftmax"], 12 | "label_smooth": ["float", 0.1], 13 | "auxiliary": ["float", "-1"] 14 | } 15 | -------------------------------------------------------------------------------- /configs/opts/ImageNet-E120-Cos-Soft.config: -------------------------------------------------------------------------------- 1 | { 2 | "scheduler": ["str", "cos"], 3 | "eta_min" : ["float", "0.0"], 4 | "epochs" : ["int", "120"], 5 | "warmup" : ["int", "5"], 6 | "optim" : ["str", "SGD"], 7 | "LR" : ["float", "0.1"], 8 | "decay" : ["float", "0.0001"], 9 | "momentum" : ["float", "0.9"], 10 | "nesterov" : ["bool", "1"], 11 | "criterion": ["str", "Softmax"], 12 | "auxiliary": ["float", "-1"] 13 | } 14 | -------------------------------------------------------------------------------- /configs/opts/ImageNet-E120-Step-Soft.config: -------------------------------------------------------------------------------- 1 | { 2 | "scheduler": ["str", "multistep"], 3 | "epochs" : ["int", "120"], 4 | "warmup" : ["int", "5"], 5 | "optim" : ["str", "SGD"], 6 | "LR" : ["float", "0.1"], 7 | "milestones":["int", [30, 60, 90]], 8 | "gammas" : ["float", [0.1, 0.1, 0.1]], 9 | "decay" : ["float", "0.0001"], 10 | "momentum" : ["float", "0.9"], 11 | "nesterov" : ["bool", "1"], 12 | "criterion": ["str", "Softmax"], 13 | "auxiliary": ["float", "-1"] 14 | } 15 | -------------------------------------------------------------------------------- /configs/opts/NAS-CIFAR-V2.config: -------------------------------------------------------------------------------- 1 | { 2 | "scheduler": ["str", "cos"], 3 | "eta_min" : ["float", "0.0"], 4 | "epochs" : ["int", "595"], 5 | "warmup" : ["int", "5"], 6 | "optim" : ["str", "SGD"], 7 | "LR" : ["float", "0.025"], 8 | "decay" : ["float", "0.0003"], 9 | "momentum" : ["float", "0.9"], 10 | "nesterov" : ["bool", "1"], 11 | "criterion": ["str", "Softmax"], 12 | "auxiliary": ["float", "0.4"] 13 | } 14 | -------------------------------------------------------------------------------- /configs/opts/NAS-CIFAR.config: -------------------------------------------------------------------------------- 1 | { 2 | "scheduler": ["str", "cos"], 3 | "eta_min" : ["float", "0.0"], 4 | "epochs" : ["int", "295"], 5 | "warmup" : ["int", "5"], 6 | "optim" : ["str", "SGD"], 7 | "LR" : ["float", "0.025"], 8 | "decay" : ["float", "0.0005"], 9 | "momentum" : ["float", "0.9"], 10 | "nesterov" : ["bool", "1"], 11 | "criterion": ["str", "Softmax"], 12 | "auxiliary": ["float", "0.4"] 13 | } 14 | -------------------------------------------------------------------------------- /configs/opts/NAS-IMAGENET.config: -------------------------------------------------------------------------------- 1 | { 2 | "scheduler": ["str", "cos"], 3 | "eta_min" : ["float", "0.0"], 4 | "epochs" : ["int", "245"], 5 | "warmup" : ["int", "5"], 6 | "optim" : ["str", "SGD"], 7 | "LR" : ["float", "0.1"], 8 | "decay" : ["float", "0.00003"], 9 | "momentum" : ["float", "0.9"], 10 | "nesterov" : ["bool", "1"], 11 | "criterion": ["str", "SmoothSoftmax"], 12 | "label_smooth": ["float", 0.1], 13 | "auxiliary" : ["float", "0.4"] 14 | } 15 | -------------------------------------------------------------------------------- /configs/qlib/workflow_config_naive_v1_Alpha360.yaml: -------------------------------------------------------------------------------- 1 | qlib_init: 2 | provider_uri: "~/.qlib/qlib_data/cn_data" 3 | region: cn 4 | market: &market all 5 | benchmark: &benchmark SH000300 6 | data_handler_config: &data_handler_config 7 | start_time: 2008-01-01 8 | end_time: 2020-08-01 9 | fit_start_time: 2008-01-01 10 | fit_end_time: 2014-12-31 11 | instruments: *market 12 | infer_processors: [] 13 | learn_processors: [] 14 | label: ["Ref($close, -2) / Ref($close, -1) - 1"] 15 | port_analysis_config: &port_analysis_config 16 | strategy: 17 | class: TopkDropoutStrategy 18 | module_path: qlib.contrib.strategy.strategy 19 | kwargs: 20 | topk: 50 21 | n_drop: 5 22 | backtest: 23 | verbose: False 24 | limit_threshold: 0.095 25 | account: 100000000 26 | benchmark: *benchmark 27 | deal_price: close 28 | open_cost: 0.0005 29 | close_cost: 0.0015 30 | min_cost: 5 31 | task: 32 | model: 33 | class: NAIVE_V1 34 | module_path: trade_models.naive_v1_model 35 | kwargs: 36 | d_feat: 6 37 | dataset: 38 | class: DatasetH 39 | module_path: qlib.data.dataset 40 | kwargs: 41 | handler: 42 | class: Alpha360 43 | module_path: qlib.contrib.data.handler 44 | kwargs: *data_handler_config 45 | segments: 46 | train: [2008-01-01, 2014-12-31] 47 | valid: [2015-01-01, 2016-12-31] 48 | test: [2017-01-01, 2020-08-01] 49 | record: 50 | - class: SignalRecord 51 | module_path: qlib.workflow.record_temp 52 | kwargs: {} 53 | - class: SignalMseRecord 54 | module_path: qlib.contrib.workflow.record_temp 55 | kwargs: {} 56 | - class: SigAnaRecord 57 | module_path: qlib.workflow.record_temp 58 | kwargs: 59 | ana_long_short: False 60 | ann_scaler: 252 61 | - class: PortAnaRecord 62 | module_path: qlib.workflow.record_temp 63 | kwargs: 64 | config: *port_analysis_config 65 | -------------------------------------------------------------------------------- /configs/qlib/workflow_config_naive_v2_Alpha360.yaml: -------------------------------------------------------------------------------- 1 | qlib_init: 2 | provider_uri: "~/.qlib/qlib_data/cn_data" 3 | region: cn 4 | market: &market all 5 | benchmark: &benchmark SH000300 6 | data_handler_config: &data_handler_config 7 | start_time: 2008-01-01 8 | end_time: 2020-08-01 9 | fit_start_time: 2008-01-01 10 | fit_end_time: 2014-12-31 11 | instruments: *market 12 | infer_processors: [] 13 | learn_processors: [] 14 | label: ["Ref($close, -2) / Ref($close, -1) - 1"] 15 | port_analysis_config: &port_analysis_config 16 | strategy: 17 | class: TopkDropoutStrategy 18 | module_path: qlib.contrib.strategy.strategy 19 | kwargs: 20 | topk: 50 21 | n_drop: 5 22 | backtest: 23 | verbose: False 24 | limit_threshold: 0.095 25 | account: 100000000 26 | benchmark: *benchmark 27 | deal_price: close 28 | open_cost: 0.0005 29 | close_cost: 0.0015 30 | min_cost: 5 31 | task: 32 | model: 33 | class: NAIVE_V2 34 | module_path: trade_models.naive_v2_model 35 | kwargs: 36 | d_feat: 6 37 | dataset: 38 | class: DatasetH 39 | module_path: qlib.data.dataset 40 | kwargs: 41 | handler: 42 | class: Alpha360 43 | module_path: qlib.contrib.data.handler 44 | kwargs: *data_handler_config 45 | segments: 46 | train: [2008-01-01, 2014-12-31] 47 | valid: [2015-01-01, 2016-12-31] 48 | test: [2017-01-01, 2020-08-01] 49 | record: 50 | - class: SignalRecord 51 | module_path: qlib.workflow.record_temp 52 | kwargs: {} 53 | - class: SignalMseRecord 54 | module_path: qlib.contrib.workflow.record_temp 55 | kwargs: {} 56 | - class: SigAnaRecord 57 | module_path: qlib.workflow.record_temp 58 | kwargs: 59 | ana_long_short: False 60 | ann_scaler: 252 61 | - class: PortAnaRecord 62 | module_path: qlib.workflow.record_temp 63 | kwargs: 64 | config: *port_analysis_config 65 | -------------------------------------------------------------------------------- /configs/search-KD-opts/ImageNet-E120-Cos-Soft.config: -------------------------------------------------------------------------------- 1 | { 2 | "scheduler": ["str", "cos"], 3 | "eta_min" : ["float", "0.0"], 4 | "epochs" : ["int", "125"], 5 | "T_max" : ["int", "120"], 6 | "warmup" : ["int", "5"], 7 | "optim" : ["str", "SGD"], 8 | "LR" : ["float", "0.1"], 9 | "decay" : ["float", "0.0001"], 10 | "momentum" : ["float", "0.9"], 11 | "nesterov" : ["bool", "1"], 12 | "criterion": ["str", "Softmax"], 13 | "auxiliary": ["float", "-1"] 14 | } 15 | -------------------------------------------------------------------------------- /configs/search-KD-opts/ImageNet-E150-MobileFast.config: -------------------------------------------------------------------------------- 1 | { 2 | "scheduler": ["str", "cos"], 3 | "eta_min" : ["float", "0.0"], 4 | "epochs" : ["int", "155"], 5 | "T_max" : ["int", "150"], 6 | "warmup" : ["int", "0"], 7 | "gamma" : ["float", "0.98"], 8 | "optim" : ["str", "SGD"], 9 | "LR" : ["float", "0.05"], 10 | "decay" : ["float", "0.00004"], 11 | "momentum" : ["float", "0.9"], 12 | "nesterov" : ["bool", "0"], 13 | "criterion": ["str", "Softmax"], 14 | "auxiliary": ["float", "-1"] 15 | } 16 | -------------------------------------------------------------------------------- /configs/search-archs/DARTS-NASNet-CIFAR.config: -------------------------------------------------------------------------------- 1 | { 2 | "super_type" : ["str", "nasnet-super"], 3 | "name" : ["str", "DARTS"], 4 | "C" : ["int", "16" ], 5 | "N" : ["int", "2" ], 6 | "steps" : ["int", "4" ], 7 | "multiplier" : ["int", "4" ], 8 | "stem_multiplier" : ["int", "3" ] 9 | } 10 | -------------------------------------------------------------------------------- /configs/search-archs/GDAS-NASNet-CIFAR.config: -------------------------------------------------------------------------------- 1 | { 2 | "super_type" : ["str", "nasnet-super"], 3 | "name" : ["str", "GDAS"], 4 | "C" : ["int", "16" ], 5 | "N" : ["int", "2" ], 6 | "steps" : ["int", "4" ], 7 | "multiplier" : ["int", "4" ], 8 | "stem_multiplier" : ["int", "3" ] 9 | } 10 | -------------------------------------------------------------------------------- /configs/search-archs/GDASFRC-NASNet-CIFAR.config: -------------------------------------------------------------------------------- 1 | { 2 | "super_type" : ["str", "nasnet-super"], 3 | "name" : ["str", "GDAS_FRC"], 4 | "C" : ["int", "16" ], 5 | "N" : ["int", "2" ], 6 | "steps" : ["int", "4" ], 7 | "multiplier" : ["int", "4" ], 8 | "stem_multiplier" : ["int", "3" ] 9 | } 10 | -------------------------------------------------------------------------------- /configs/search-archs/ImageNet-MobileNetV2-X.config: -------------------------------------------------------------------------------- 1 | { 2 | "dataset" : ["str", "imagenet"], 3 | "arch" : ["str", "MobileNetV2"], 4 | "width_mult" : ["float", 1.0], 5 | "dropout" : ["float", 0.0], 6 | "input_channel" : ["int", 32], 7 | "last_channel" : ["int", 1280], 8 | "block_name" : ["str", "InvertedResidual"] 9 | } 10 | -------------------------------------------------------------------------------- /configs/search-archs/ImageNet-ResNet18V1.config: -------------------------------------------------------------------------------- 1 | { 2 | "dataset" : ["str", "imagenet"], 3 | "arch" : ["str", "resnet"], 4 | "block_name" : ["str", "BasicBlock"], 5 | "layers" : ["int", [2,2,2,2]], 6 | "deep_stem" : ["bool", 0], 7 | "zero_init_residual" : ["bool", "1"] 8 | } 9 | -------------------------------------------------------------------------------- /configs/search-archs/ImageNet-ResNet34V1.config: -------------------------------------------------------------------------------- 1 | { 2 | "dataset" : ["str", "imagenet"], 3 | "arch" : ["str", "resnet"], 4 | "block_name" : ["str", "BasicBlock"], 5 | "layers" : ["int", [3,4,6,3]], 6 | "deep_stem" : ["bool", 0], 7 | "zero_init_residual" : ["bool", "1"] 8 | } 9 | -------------------------------------------------------------------------------- /configs/search-archs/ImageNet-ResNet50V1.config: -------------------------------------------------------------------------------- 1 | { 2 | "dataset" : ["str", "imagenet"], 3 | "arch" : ["str", "resnet"], 4 | "block_name" : ["str", "Bottleneck"], 5 | "layers" : ["int", [3,4,6,3]], 6 | "deep_stem" : ["bool", 0], 7 | "zero_init_residual" : ["bool", "1"] 8 | } 9 | -------------------------------------------------------------------------------- /configs/search-opts/CIFAR.config: -------------------------------------------------------------------------------- 1 | { 2 | "scheduler": ["str", "cos"], 3 | "eta_min" : ["float", "0.0"], 4 | "epochs" : ["int", "300"], 5 | "warmup" : ["int", "5"], 6 | "optim" : ["str", "SGD"], 7 | "LR" : ["float", "0.1"], 8 | "decay" : ["float", "0.0005"], 9 | "momentum" : ["float", "0.9"], 10 | "nesterov" : ["bool", "1"], 11 | "criterion": ["str", "Softmax"], 12 | "arch_LR" : ["float", "0.001"], 13 | "arch_decay" : ["float", "0.001"] 14 | } 15 | -------------------------------------------------------------------------------- /configs/search-opts/CIFARX.config: -------------------------------------------------------------------------------- 1 | { 2 | "scheduler": ["str", "cos"], 3 | "eta_min" : ["float", "0.0"], 4 | "epochs" : ["int", "600"], 5 | "warmup" : ["int", "0"], 6 | "optim" : ["str", "SGD"], 7 | "LR" : ["float", "0.1"], 8 | "decay" : ["float", "0.0005"], 9 | "momentum" : ["float", "0.9"], 10 | "nesterov" : ["bool", "1"], 11 | "criterion": ["str", "Softmax"], 12 | "arch_LR" : ["float", "0.001"], 13 | "arch_decay" : ["float", "0.001"] 14 | } 15 | -------------------------------------------------------------------------------- /configs/search-opts/CIFARXX.config: -------------------------------------------------------------------------------- 1 | { 2 | "scheduler": ["str", "cos"], 3 | "eta_min" : ["float", "0.0"], 4 | "epochs" : ["int", "900"], 5 | "warmup" : ["int", "5"], 6 | "optim" : ["str", "SGD"], 7 | "LR" : ["float", "0.1"], 8 | "decay" : ["float", "0.0005"], 9 | "momentum" : ["float", "0.9"], 10 | "nesterov" : ["bool", "1"], 11 | "criterion": ["str", "Softmax"], 12 | "arch_LR" : ["float", "0.001"], 13 | "arch_decay" : ["float", "0.001"] 14 | } 15 | -------------------------------------------------------------------------------- /configs/search-opts/DARTS-NASNet-CIFAR.config: -------------------------------------------------------------------------------- 1 | { 2 | "scheduler": ["str", "cos"], 3 | "LR" : ["float", "0.025"], 4 | "eta_min" : ["float", "0.001"], 5 | "epochs" : ["int", "50"], 6 | "warmup" : ["int", "0"], 7 | "optim" : ["str", "SGD"], 8 | "decay" : ["float", "0.0003"], 9 | "momentum" : ["float", "0.9"], 10 | "nesterov" : ["bool", "1"], 11 | "criterion": ["str", "Softmax"], 12 | "batch_size": ["int", "64"] 13 | } 14 | -------------------------------------------------------------------------------- /configs/search-opts/GDAS-NASNet-CIFAR.config: -------------------------------------------------------------------------------- 1 | { 2 | "scheduler": ["str", "cos"], 3 | "LR" : ["float", "0.025"], 4 | "eta_min" : ["float", "0.001"], 5 | "epochs" : ["int", "250"], 6 | "warmup" : ["int", "0"], 7 | "optim" : ["str", "SGD"], 8 | "decay" : ["float", "0.0005"], 9 | "momentum" : ["float", "0.9"], 10 | "nesterov" : ["bool", "1"], 11 | "criterion": ["str", "Softmax"], 12 | "batch_size": ["int", "256"] 13 | } 14 | -------------------------------------------------------------------------------- /configs/search-opts/ImageNet-MobileFast.config: -------------------------------------------------------------------------------- 1 | { 2 | "scheduler": ["str", "cos"], 3 | "eta_min" : ["float", "0.0"], 4 | "epochs" : ["int", "150"], 5 | "warmup" : ["int", "0"], 6 | "gamma" : ["float", "0.98"], 7 | "optim" : ["str", "SGD"], 8 | "LR" : ["float", "0.05"], 9 | "decay" : ["float", "0.00004"], 10 | "momentum" : ["float", "0.9"], 11 | "nesterov" : ["bool", "0"], 12 | "criterion": ["str", "Softmax"], 13 | "arch_LR" : ["float", "0.001"], 14 | "arch_decay" : ["float", "0.001"] 15 | } 16 | -------------------------------------------------------------------------------- /configs/search-opts/ImageNet-ResNet.config: -------------------------------------------------------------------------------- 1 | { 2 | "scheduler": ["str", "cos"], 3 | "eta_min" : ["float", "0.0"], 4 | "epochs" : ["int", "115"], 5 | "warmup" : ["int", "5"], 6 | "optim" : ["str", "SGD"], 7 | "LR" : ["float", "0.1"], 8 | "decay" : ["float", "0.0001"], 9 | "momentum" : ["float", "0.9"], 10 | "nesterov" : ["bool", "1"], 11 | "criterion": ["str", "Softmax"], 12 | "arch_LR" : ["float", "0.001"], 13 | "arch_decay" : ["float", "0.001"] 14 | } 15 | -------------------------------------------------------------------------------- /configs/temps/T-MobileNetV2-X.config: -------------------------------------------------------------------------------- 1 | { 2 | "dataset" : ["str", "imagenet"], 3 | "arch" : ["str", "MobileNetV2"], 4 | "dropout" : ["float", 0.0], 5 | "super_type" : ["str", "infer-shape"], 6 | "xchannels" : ["str", "3-32 32-16 16-96-24 24-86-24 24-86-32 32-96-32 32-77-32 32-77-32 32-231-48 48-269-48 48-269-64 64-269-64 64-192-67 67-246-77 77-288-96 96-346-96 96-576-160 160-480-128 128-576-320 320-768"], 7 | "xblocks" : ["int", [1, 2, 3, 3, 3, 3, 1]] 8 | } 9 | -------------------------------------------------------------------------------- /configs/temps/T-ResNet18.config: -------------------------------------------------------------------------------- 1 | { 2 | "dataset" : ["str" , "imagenet"], 3 | "arch" : ["str" , "resnet"], 4 | "block_name" : ["str" , "BasicBlock"], 5 | "layers" : ["int" , ["2", "2", "2", "2"]], 6 | "deep_stem" : ["bool" , "0"], 7 | "class_num" : ["int" , "1000"], 8 | "search_mode" : ["str" , "shape"], 9 | "xchannels" : ["int" , ["3", "48", "64", "38", "25", "57", "102", "89", "102", "115", "128", "230", "230", "256", "153", "358", "358", "460"]], 10 | "xblocks" : ["int" , ["2", "2", "2", "2"]], 11 | "super_type" : ["str" , "infer-shape"], 12 | "estimated_FLOP" : ["float" , "953.381522"], 13 | "zero_init_residual" : ["bool" , "1"] 14 | } 15 | -------------------------------------------------------------------------------- /configs/temps/T-ResNet50.config: -------------------------------------------------------------------------------- 1 | { 2 | "dataset" : ["str" , "imagenet"], 3 | "arch" : ["str" , "resnet"], 4 | "block_name" : ["str" , "Bottleneck"], 5 | "layers" : ["int" , ["3", "4", "6", "3"]], 6 | "deep_stem" : ["bool" , "0"], 7 | "zero_init_residual" : ["bool" , "1"], 8 | "class_num" : ["int" , "1000"], 9 | "search_mode" : ["str" , "shape"], 10 | "xchannels" : ["int" , ["3", "45", "45", "30", "102", "33", "60", "154", "68", "70", "180", "38", "38", "307", "38", "38", "410", "64", "128", "358", "38", "51", "256", "76", "76", "512", "76", "76", "512", "179", "256", "614", "100", "102", "307", "179", "230", "614", "204", "102", "307", "153", "153", "1228", "512", "512", "1434", "512", "512", "1844"]], 11 | "xblocks" : ["int" , ["3", "4", "5", "3"]], 12 | "super_type" : ["str" , "infer-shape"], 13 | "estimated_FLOP" : ["float" , "2291.316289"] 14 | } 15 | -------------------------------------------------------------------------------- /configs/temps/X-ResNet18.config: -------------------------------------------------------------------------------- 1 | { 2 | "dataset" : ["str" , "imagenet"], 3 | "arch" : ["str" , "resnet"], 4 | "block_name" : ["str" , "BasicBlock"], 5 | "layers" : ["int" , ["2", "2", "2", "2"]], 6 | "deep_stem" : ["bool" , "0"], 7 | "zero_init_residual" : ["bool" , "1"], 8 | "class_num" : ["int" , "1000"], 9 | "search_mode" : ["str" , "shape"], 10 | "xchannels" : ["int" , ["3", "64", "25", "64", "38", "19", "128", "128", "38", "38", "256", "256", "256", "256", "512", "512", "512", "512"]], 11 | "xblocks" : ["int" , ["1", "1", "2", "2"]], 12 | "super_type" : ["str" , "infer-shape"], 13 | "estimated_FLOP" : ["float" , "1120.44032"] 14 | } -------------------------------------------------------------------------------- /configs/temps/X-ResNet50.config: -------------------------------------------------------------------------------- 1 | { 2 | "dataset" : ["str" , "imagenet"], 3 | "arch" : ["str" , "resnet"], 4 | "block_name" : ["str" , "Bottleneck"], 5 | "layers" : ["int" , ["3", "4", "6", "3"]], 6 | "deep_stem" : ["bool" , "0"], 7 | "zero_init_residual" : ["bool" , "1"], 8 | "class_num" : ["int" , "1000"], 9 | "search_mode" : ["str" , "shape"], 10 | "xchannels" : ["int" , ["3", "64", "64", "64", "230", "25", "19", "128", "19", "38", "76", "128", "128", "512", "38", "38", "204", "38", "51", "256", "51", "51", "153", "256", "256", "1024", "256", "256", "1024", "256", "256", "921", "256", "256", "1024", "76", "76", "614", "102", "76", "307", "512", "512", "2048", "512", "512", "2048", "512", "512", "2048"]], 11 | "xblocks" : ["int" , ["1", "1", "2", "3"]], 12 | "super_type" : ["str" , "infer-shape"], 13 | "estimated_FLOP" : ["float" , "1680.93696"] 14 | } -------------------------------------------------------------------------------- /configs/yaml.data/cifar10.test: -------------------------------------------------------------------------------- 1 | class_or_func: CIFAR10 2 | module_path: torchvision.datasets 3 | args: [] 4 | kwargs: 5 | train: False 6 | download: True 7 | transform: 8 | class_or_func: Compose 9 | module_path: torchvision.transforms 10 | args: 11 | - 12 | - class_or_func: ToTensor 13 | module_path: torchvision.transforms 14 | args: [] 15 | kwargs: {} 16 | - class_or_func: Normalize 17 | module_path: torchvision.transforms 18 | args: [] 19 | kwargs: 20 | mean: [0.491, 0.482, 0.447] 21 | std: [0.247, 0.244, 0.262] 22 | kwargs: {} 23 | -------------------------------------------------------------------------------- /configs/yaml.data/cifar10.train: -------------------------------------------------------------------------------- 1 | class_or_func: CIFAR10 2 | module_path: torchvision.datasets 3 | args: [] 4 | kwargs: 5 | train: True 6 | download: True 7 | transform: 8 | class_or_func: Compose 9 | module_path: torchvision.transforms 10 | args: 11 | - 12 | - class_or_func: RandomHorizontalFlip 13 | module_path: torchvision.transforms 14 | args: [] 15 | kwargs: {} 16 | - class_or_func: RandomCrop 17 | module_path: torchvision.transforms 18 | args: [32] 19 | kwargs: {padding: 4} 20 | - class_or_func: ToTensor 21 | module_path: torchvision.transforms 22 | args: [] 23 | kwargs: {} 24 | - class_or_func: Normalize 25 | module_path: torchvision.transforms 26 | args: [] 27 | kwargs: 28 | mean: [0.491, 0.482, 0.447] 29 | std: [0.247, 0.244, 0.262] 30 | kwargs: {} 31 | -------------------------------------------------------------------------------- /configs/yaml.loss/cross-entropy: -------------------------------------------------------------------------------- 1 | class_or_func: CrossEntropyLoss 2 | module_path: torch.nn 3 | args: [] 4 | kwargs: {} 5 | -------------------------------------------------------------------------------- /configs/yaml.loss/top1-ce: -------------------------------------------------------------------------------- 1 | class_or_func: ComposeMetric 2 | module_path: xautodl.xmisc.meter_utils 3 | args: 4 | - class_or_func: Top1AccMetric 5 | module_path: xautodl.xmisc.meter_utils 6 | args: [False] 7 | kwargs: {} 8 | - class_or_func: CrossEntropyMetric 9 | module_path: xautodl.xmisc.meter_utils 10 | args: [False] 11 | kwargs: {} 12 | kwargs: {} 13 | -------------------------------------------------------------------------------- /configs/yaml.loss/top1.acc: -------------------------------------------------------------------------------- 1 | class_or_func: Top1AccMetric 2 | module_path: xautodl.xmisc.meter_utils 3 | args: [False] 4 | kwargs: {} 5 | -------------------------------------------------------------------------------- /configs/yaml.model/vit-cifar10.s0: -------------------------------------------------------------------------------- 1 | class_or_func: get_transformer 2 | module_path: xautodl.xmodels.transformers 3 | args: [vit-cifar10-p4-d4-h4-c32] 4 | kwargs: {} 5 | -------------------------------------------------------------------------------- /configs/yaml.opt/vit.cifar: -------------------------------------------------------------------------------- 1 | class_or_func: Adam 2 | module_path: torch.optim 3 | args: [] 4 | kwargs: 5 | betas: [0.9, 0.999] 6 | weight_decay: 0.1 7 | amsgrad: False 8 | -------------------------------------------------------------------------------- /docs/ICCV-2019-SETN.md: -------------------------------------------------------------------------------- 1 | # [One-Shot Neural Architecture Search via Self-Evaluated Template Network](https://arxiv.org/abs/1910.05733) 2 | 3 | 4 | 5 | Highlight: we equip one-shot NAS with an architecture sampler and train network weights using uniformly sampling. 6 | 7 | One-Shot Neural Architecture Search via Self-Evaluated Template Network is accepted by ICCV 2019. 8 | 9 | 10 | ## Requirements and Preparation 11 | 12 | Please install `Python>=3.6` and `PyTorch>=1.2.0`. 13 | 14 | ### Usefull tools 15 | 1. Compute the number of parameters and FLOPs of a model: 16 | ``` 17 | from utils import get_model_infos 18 | flop, param = get_model_infos(net, (1,3,32,32)) 19 | ``` 20 | 21 | 2. Different NAS-searched architectures are defined [here](https://github.com/D-X-Y/AutoDL-Projects/blob/main/lib/nas_infer_model/DXYs/genotypes.py). 22 | 23 | 24 | ## Usage 25 | 26 | Please use the following scripts to train the searched SETN-searched CNN on CIFAR-10, CIFAR-100, and ImageNet. 27 | ``` 28 | CUDA_VISIBLE_DEVICES=0 bash ./scripts/nas-infer-train.sh cifar10 SETN 96 -1 29 | CUDA_VISIBLE_DEVICES=0 bash ./scripts/nas-infer-train.sh cifar100 SETN 96 -1 30 | CUDA_VISIBLE_DEVICES=0,1,2,3 bash ./scripts/nas-infer-train.sh imagenet-1k SETN 256 -1 31 | ``` 32 | 33 | ### Searching on the NAS-Bench-201 search space 34 | The searching codes of SETN on a small search space (NAS-Bench-201). 35 | ``` 36 | CUDA_VISIBLE_DEVICES=0 bash ./scripts-search/algos/SETN.sh cifar10 1 -1 37 | ``` 38 | 39 | **Searching on the NASNet search space** is not ready yet. 40 | 41 | 42 | # Citation 43 | 44 | If you find that this project helps your research, please consider citing the following paper: 45 | ``` 46 | @inproceedings{dong2019one, 47 | title = {One-Shot Neural Architecture Search via Self-Evaluated Template Network}, 48 | author = {Dong, Xuanyi and Yang, Yi}, 49 | booktitle = {Proceedings of the IEEE International Conference on Computer Vision (ICCV)}, 50 | pages = {3681--3690}, 51 | year = {2019} 52 | } 53 | ``` 54 | -------------------------------------------------------------------------------- /docs/ICLR-2019-DARTS.md: -------------------------------------------------------------------------------- 1 | # DARTS: Differentiable Architecture Search 2 | 3 | DARTS: Differentiable Architecture Search is accepted by ICLR 2019. 4 | In this paper, Hanxiao proposed a differentiable neural architecture search method, named as DARTS. 5 | Recently, DARTS becomes very popular due to its simplicity and performance. 6 | 7 | ## Run DARTS on the NAS-Bench-201 search space 8 | ``` 9 | CUDA_VISIBLE_DEVICES=0 bash ./scripts-search/algos/DARTS-V2.sh cifar10 1 -1 10 | CUDA_VISIBLE_DEVICES=0 bash ./scripts-search/algos/GDAS.sh cifar10 1 -1 11 | ``` 12 | 13 | ## Run the first-order DARTS on the NASNet/DARTS search space 14 | This command will start to use the first-order DARTS to search architectures on the DARTS search space. 15 | ``` 16 | CUDA_VISIBLE_DEVICES=0 bash ./scripts-search/DARTS1V-search-NASNet-space.sh cifar10 -1 17 | ``` 18 | 19 | After searching, if you want to train the searched architecture found by the above scripts, you need to add the config of that architecture (will be printed in log) in [genotypes.py](https://github.com/D-X-Y/AutoDL-Projects/blob/main/lib/nas_infer_model/DXYs/genotypes.py). 20 | In future, I will add a more eligent way to train the searched architecture from the DARTS search space. 21 | 22 | 23 | # Citation 24 | 25 | ``` 26 | @inproceedings{liu2019darts, 27 | title = {{DARTS}: Differentiable architecture search}, 28 | author = {Liu, Hanxiao and Simonyan, Karen and Yang, Yiming}, 29 | booktitle = {International Conference on Learning Representations (ICLR)}, 30 | year = {2019} 31 | } 32 | ``` 33 | -------------------------------------------------------------------------------- /docs/requirements.txt: -------------------------------------------------------------------------------- 1 | torch 2 | torchvision 3 | 4 | nats_bench>=1.4 5 | nas_201_api 6 | -------------------------------------------------------------------------------- /exps/NAS-Bench-201-algos/README.md: -------------------------------------------------------------------------------- 1 | # NAS Algorithms evaluated in NAS-Bench-201 2 | 3 | The Python files in this folder are used to re-produce the results in our NAS-Bench-201 paper. 4 | 5 | We have upgraded the codes to be more general and extendable at [NATS-algos](https://github.com/D-X-Y/AutoDL-Projects/tree/main/exps/NATS-algos). 6 | 7 | **Notice** On 24 May 2021, the codes in `AutoDL` repo have been re-organized. If you find `module not found` error, please let me know. I will fix them ASAP. 8 | -------------------------------------------------------------------------------- /exps/NAS-Bench-201/dist-setup.py: -------------------------------------------------------------------------------- 1 | ##################################################### 2 | # Copyright (c) Xuanyi Dong [GitHub D-X-Y], 2019.08 # 3 | ##################################################### 4 | # [2020.02.25] Initialize the API as v1.1 5 | # [2020.03.09] Upgrade the API to v1.2 6 | # [2020.03.16] Upgrade the API to v1.3 7 | # [2020.06.30] Upgrade the API to v2.0 8 | import os 9 | from setuptools import setup 10 | 11 | 12 | def read(fname="README.md"): 13 | with open( 14 | os.path.join(os.path.dirname(__file__), fname), encoding="utf-8" 15 | ) as cfile: 16 | return cfile.read() 17 | 18 | 19 | setup( 20 | name="nas_bench_201", 21 | version="2.0", 22 | author="Xuanyi Dong", 23 | author_email="dongxuanyi888@gmail.com", 24 | description="API for NAS-Bench-201 (a benchmark for neural architecture search).", 25 | license="MIT", 26 | keywords="NAS Dataset API DeepLearning", 27 | url="https://github.com/D-X-Y/NAS-Bench-201", 28 | packages=["nas_201_api"], 29 | long_description=read("README.md"), 30 | long_description_content_type="text/markdown", 31 | classifiers=[ 32 | "Programming Language :: Python", 33 | "Topic :: Database", 34 | "Topic :: Scientific/Engineering :: Artificial Intelligence", 35 | "License :: OSI Approved :: MIT License", 36 | ], 37 | ) 38 | -------------------------------------------------------------------------------- /exps/NAS-Bench-201/show-best.py: -------------------------------------------------------------------------------- 1 | ##################################################### 2 | # Copyright (c) Xuanyi Dong [GitHub D-X-Y], 2020.01 # 3 | ################################################################################################ 4 | # python exps/NAS-Bench-201/show-best.py --api_path $HOME/.torch/NAS-Bench-201-v1_0-e61699.pth # 5 | ################################################################################################ 6 | import argparse 7 | from pathlib import Path 8 | 9 | from nas_201_api import NASBench201API as API 10 | 11 | if __name__ == "__main__": 12 | parser = argparse.ArgumentParser("Analysis of NAS-Bench-201") 13 | parser.add_argument( 14 | "--api_path", 15 | type=str, 16 | default=None, 17 | help="The path to the NAS-Bench-201 benchmark file.", 18 | ) 19 | args = parser.parse_args() 20 | 21 | meta_file = Path(args.api_path) 22 | assert meta_file.exists(), "invalid path for api : {:}".format(meta_file) 23 | 24 | api = API(str(meta_file)) 25 | 26 | # This will show the results of the best architecture based on the validation set of each dataset. 27 | arch_index, accuracy = api.find_best("cifar10-valid", "x-valid", None, None, False) 28 | print("FOR CIFAR-010, using the hyper-parameters with 200 training epochs :::") 29 | print("arch-index={:5d}, arch={:}".format(arch_index, api.arch(arch_index))) 30 | api.show(arch_index) 31 | print("") 32 | 33 | arch_index, accuracy = api.find_best("cifar100", "x-valid", None, None, False) 34 | print("FOR CIFAR-100, using the hyper-parameters with 200 training epochs :::") 35 | print("arch-index={:5d}, arch={:}".format(arch_index, api.arch(arch_index))) 36 | api.show(arch_index) 37 | print("") 38 | 39 | arch_index, accuracy = api.find_best("ImageNet16-120", "x-valid", None, None, False) 40 | print("FOR ImageNet16-120, using the hyper-parameters with 200 training epochs :::") 41 | print("arch-index={:5d}, arch={:}".format(arch_index, api.arch(arch_index))) 42 | api.show(arch_index) 43 | print("") 44 | -------------------------------------------------------------------------------- /exps/NATS-algos/README.md: -------------------------------------------------------------------------------- 1 | # NAS Algorithms evaluated in [NATS-Bench](https://arxiv.org/abs/2009.00437) 2 | 3 | The Python files in this folder are used to re-produce the results in ``NATS-Bench: Benchmarking NAS Algorithms for Architecture Topology and Size''. 4 | 5 | - [`search-size.py`](https://github.com/D-X-Y/AutoDL-Projects/blob/main/exps/NATS-algos/search-size.py) contains codes for weight-sharing-based search on the size search space. 6 | - [`search-cell.py`](https://github.com/D-X-Y/AutoDL-Projects/blob/main/exps/NATS-algos/search-cell.py) contains codes for weight-sharing-based search on the topology search space. 7 | - [`bohb.py`](https://github.com/D-X-Y/AutoDL-Projects/blob/main/exps/NATS-algos/bohb.py) contains the BOHB algorithm for both size and topology search spaces. 8 | - [`random_wo_share.py`](https://github.com/D-X-Y/AutoDL-Projects/blob/main/exps/NATS-algos/random_wo_share.py) contains the random search algorithm for both search spaces. 9 | - [`regularized_ea.py`](https://github.com/D-X-Y/AutoDL-Projects/blob/main/exps/NATS-algos/regularized_ea.py) contains the REA algorithm for both search spaces. 10 | - [`reinforce.py`](https://github.com/D-X-Y/AutoDL-Projects/blob/main/exps/NATS-algos/reinforce.py) contains the REINFORCE algorithm for both search spaces. 11 | 12 | ## Requirements 13 | 14 | - `nats_bench`>=v1.2 : you can use `pip install nats_bench` to install or from [sources](https://github.com/D-X-Y/NATS-Bench) 15 | - `hpbandster` : if you want to run BOHB 16 | 17 | ## Citation 18 | 19 | If you find that this project helps your research, please consider citing the related paper: 20 | ``` 21 | @article{dong2021nats, 22 | title = {{NATS-Bench}: Benchmarking NAS Algorithms for Architecture Topology and Size}, 23 | author = {Dong, Xuanyi and Liu, Lu and Musial, Katarzyna and Gabrys, Bogdan}, 24 | doi = {10.1109/TPAMI.2021.3054824}, 25 | journal = {IEEE Transactions on Pattern Analysis and Machine Intelligence (TPAMI)}, 26 | year = {2021}, 27 | note = {\mbox{doi}:\url{10.1109/TPAMI.2021.3054824}} 28 | } 29 | ``` 30 | -------------------------------------------------------------------------------- /exps/experimental/test-dks.py: -------------------------------------------------------------------------------- 1 | from dks.base.activation_getter import ( 2 | get_activation_function as _get_numpy_activation_function, 3 | ) 4 | from dks.base.activation_transform import _get_activations_params 5 | 6 | 7 | def subnet_max_func(x, r_fn): 8 | depth = 7 9 | res_x = r_fn(x) 10 | x = r_fn(x) 11 | for _ in range(depth): 12 | x = r_fn(r_fn(x)) + x 13 | return max(x, res_x) 14 | 15 | 16 | def subnet_max_func_v2(x, r_fn): 17 | depth = 2 18 | res_x = r_fn(x) 19 | 20 | x = r_fn(x) 21 | for _ in range(depth): 22 | x = 0.8 * r_fn(r_fn(x)) + 0.2 * x 23 | 24 | return max(x, res_x) 25 | 26 | 27 | def get_transformed_activations( 28 | activation_names, 29 | method="TAT", 30 | dks_params=None, 31 | tat_params=None, 32 | max_slope_func=None, 33 | max_curv_func=None, 34 | subnet_max_func=None, 35 | activation_getter=_get_numpy_activation_function, 36 | ): 37 | params = _get_activations_params( 38 | activation_names, 39 | method=method, 40 | dks_params=dks_params, 41 | tat_params=tat_params, 42 | max_slope_func=max_slope_func, 43 | max_curv_func=max_curv_func, 44 | subnet_max_func=subnet_max_func, 45 | ) 46 | return params 47 | 48 | 49 | params = get_transformed_activations( 50 | ["swish"], method="TAT", subnet_max_func=subnet_max_func 51 | ) 52 | print(params) 53 | 54 | params = get_transformed_activations( 55 | ["leaky_relu"], method="TAT", subnet_max_func=subnet_max_func_v2 56 | ) 57 | print(params) 58 | -------------------------------------------------------------------------------- /exps/experimental/test-dynamic.py: -------------------------------------------------------------------------------- 1 | ##################################################### 2 | # Copyright (c) Xuanyi Dong [GitHub D-X-Y], 2019.08 # 3 | ##################################################### 4 | # python test-dynamic.py 5 | ##################################################### 6 | import sys 7 | from pathlib import Path 8 | 9 | lib_dir = (Path(__file__).parent / ".." / "..").resolve() 10 | print("LIB-DIR: {:}".format(lib_dir)) 11 | if str(lib_dir) not in sys.path: 12 | sys.path.insert(0, str(lib_dir)) 13 | 14 | from xautodl.datasets.math_core import ConstantFunc 15 | from xautodl.datasets.math_core import GaussianDGenerator 16 | 17 | mean_generator = ConstantFunc(0) 18 | cov_generator = ConstantFunc(1) 19 | 20 | generator = GaussianDGenerator([mean_generator], [[cov_generator]], (-1, 1)) 21 | generator(0, 10) 22 | -------------------------------------------------------------------------------- /exps/experimental/test-flops.py: -------------------------------------------------------------------------------- 1 | import sys, time, random, argparse 2 | from copy import deepcopy 3 | import torchvision.models as models 4 | from pathlib import Path 5 | 6 | lib_dir = (Path(__file__).parent / ".." / ".." / "lib").resolve() 7 | if str(lib_dir) not in sys.path: 8 | sys.path.insert(0, str(lib_dir)) 9 | 10 | from utils import get_model_infos 11 | 12 | # from models.ImageNet_MobileNetV2 import MobileNetV2 13 | from torchvision.models.mobilenet import MobileNetV2 14 | 15 | 16 | def main(width_mult): 17 | # model = MobileNetV2(1001, width_mult, 32, 1280, 'InvertedResidual', 0.2) 18 | model = MobileNetV2(width_mult=width_mult) 19 | print(model) 20 | flops, params = get_model_infos(model, (2, 3, 224, 224)) 21 | print("FLOPs : {:}".format(flops)) 22 | print("Params : {:}".format(params)) 23 | print("-" * 50) 24 | 25 | 26 | if __name__ == "__main__": 27 | main(1.0) 28 | main(1.4) 29 | -------------------------------------------------------------------------------- /exps/experimental/test-resnest.py: -------------------------------------------------------------------------------- 1 | ##################################################### 2 | # Copyright (c) Xuanyi Dong [GitHub D-X-Y], 2020.06 # 3 | ##################################################### 4 | # python exps/experimental/test-resnest.py 5 | ##################################################### 6 | import sys, time, torch, random, argparse 7 | from PIL import ImageFile 8 | 9 | ImageFile.LOAD_TRUNCATED_IMAGES = True 10 | from copy import deepcopy 11 | from pathlib import Path 12 | 13 | lib_dir = (Path(__file__).parent / ".." / ".." / "lib").resolve() 14 | if str(lib_dir) not in sys.path: 15 | sys.path.insert(0, str(lib_dir)) 16 | from utils import get_model_infos 17 | 18 | torch.hub.list("zhanghang1989/ResNeSt", force_reload=True) 19 | 20 | for model_name, xshape in [ 21 | ("resnest50", (1, 3, 224, 224)), 22 | ("resnest101", (1, 3, 256, 256)), 23 | ("resnest200", (1, 3, 320, 320)), 24 | ("resnest269", (1, 3, 416, 416)), 25 | ]: 26 | # net = torch.hub.load('zhanghang1989/ResNeSt', model_name, pretrained=True) 27 | net = torch.hub.load("zhanghang1989/ResNeSt", model_name, pretrained=False) 28 | print("Model : {:}, input shape : {:}".format(model_name, xshape)) 29 | flops, param = get_model_infos(net, xshape) 30 | print("flops : {:.3f}M".format(flops)) 31 | print("params : {:.3f}M".format(param)) 32 | -------------------------------------------------------------------------------- /exps/experimental/test-ww.py: -------------------------------------------------------------------------------- 1 | import sys, time, random, argparse 2 | from copy import deepcopy 3 | import torchvision.models as models 4 | from pathlib import Path 5 | 6 | from xautodl.utils import weight_watcher 7 | 8 | 9 | def main(): 10 | # model = models.vgg19_bn(pretrained=True) 11 | # _, summary = weight_watcher.analyze(model, alphas=False) 12 | # for key, value in summary.items(): 13 | # print('{:10s} : {:}'.format(key, value)) 14 | 15 | _, summary = weight_watcher.analyze(models.vgg13(pretrained=True), alphas=False) 16 | print("vgg-13 : {:}".format(summary["lognorm"])) 17 | _, summary = weight_watcher.analyze(models.vgg13_bn(pretrained=True), alphas=False) 18 | print("vgg-13-BN : {:}".format(summary["lognorm"])) 19 | _, summary = weight_watcher.analyze(models.vgg16(pretrained=True), alphas=False) 20 | print("vgg-16 : {:}".format(summary["lognorm"])) 21 | _, summary = weight_watcher.analyze(models.vgg16_bn(pretrained=True), alphas=False) 22 | print("vgg-16-BN : {:}".format(summary["lognorm"])) 23 | _, summary = weight_watcher.analyze(models.vgg19(pretrained=True), alphas=False) 24 | print("vgg-19 : {:}".format(summary["lognorm"])) 25 | _, summary = weight_watcher.analyze(models.vgg19_bn(pretrained=True), alphas=False) 26 | print("vgg-19-BN : {:}".format(summary["lognorm"])) 27 | 28 | 29 | if __name__ == "__main__": 30 | main() 31 | -------------------------------------------------------------------------------- /notebooks/NATS-Bench/issue-97.ipynb: -------------------------------------------------------------------------------- 1 | { 2 | "cells": [ 3 | { 4 | "cell_type": "code", 5 | "execution_count": 1, 6 | "metadata": {}, 7 | "outputs": [ 8 | { 9 | "name": "stdout", 10 | "output_type": "stream", 11 | "text": [ 12 | "[2021-03-09 08:44:19] Try to use the default NATS-Bench (size) path from fast_mode=True and path=None.\n" 13 | ] 14 | } 15 | ], 16 | "source": [ 17 | "from nats_bench import create\n", 18 | "import numpy as np\n", 19 | "\n", 20 | "# Create the API for size search space\n", 21 | "api = create(None, 'sss', fast_mode=True, verbose=False)" 22 | ] 23 | }, 24 | { 25 | "cell_type": "code", 26 | "execution_count": 2, 27 | "metadata": {}, 28 | "outputs": [ 29 | { 30 | "name": "stdout", 31 | "output_type": "stream", 32 | "text": [ 33 | "There are 32768 architectures on the size search space\n" 34 | ] 35 | } 36 | ], 37 | "source": [ 38 | "print('There are {:} architectures on the size search space'.format(len(api)))\n", 39 | "\n", 40 | "c2acc = dict()\n", 41 | "for index in range(len(api)):\n", 42 | " info = api.get_more_info(index, 'cifar10', hp='90')\n", 43 | " config = api.get_net_config(index, 'cifar10')\n", 44 | " c2acc[config['channels']] = info['test-accuracy']" 45 | ] 46 | }, 47 | { 48 | "cell_type": "code", 49 | "execution_count": 4, 50 | "metadata": {}, 51 | "outputs": [ 52 | { 53 | "name": "stdout", 54 | "output_type": "stream", 55 | "text": [ 56 | "91.08546417236329\n" 57 | ] 58 | } 59 | ], 60 | "source": [ 61 | "print(np.mean(list(c2acc.values())))" 62 | ] 63 | } 64 | ], 65 | "metadata": { 66 | "kernelspec": { 67 | "display_name": "Python 3", 68 | "language": "python", 69 | "name": "python3" 70 | }, 71 | "language_info": { 72 | "codemirror_mode": { 73 | "name": "ipython", 74 | "version": 3 75 | }, 76 | "file_extension": ".py", 77 | "mimetype": "text/x-python", 78 | "name": "python", 79 | "nbconvert_exporter": "python", 80 | "pygments_lexer": "ipython3", 81 | "version": "3.8.3" 82 | } 83 | }, 84 | "nbformat": 4, 85 | "nbformat_minor": 4 86 | } 87 | -------------------------------------------------------------------------------- /scripts-search/NAS-Bench-201-algos/BOHB.sh: -------------------------------------------------------------------------------- 1 | #!/bin/bash 2 | # bash ./scripts-search/NAS-Bench-201-algos/BOHB.sh -1 3 | echo script name: $0 4 | echo $# arguments 5 | if [ "$#" -ne 2 ] ;then 6 | echo "Input illegal number of parameters " $# 7 | echo "Need 2 parameters for dataset and seed" 8 | exit 1 9 | fi 10 | if [ "$TORCH_HOME" = "" ]; then 11 | echo "Must set TORCH_HOME envoriment variable for data dir saving" 12 | exit 1 13 | else 14 | echo "TORCH_HOME : $TORCH_HOME" 15 | fi 16 | 17 | dataset=$1 18 | seed=$2 19 | channel=16 20 | num_cells=5 21 | max_nodes=4 22 | space=nas-bench-201 23 | #benchmark_file=${TORCH_HOME}/NAS-Bench-201-v1_0-e61699.pth 24 | benchmark_file=${TORCH_HOME}/NAS-Bench-201-v1_1-096897.pth 25 | 26 | save_dir=./output/search-cell-${space}/BOHB-${dataset} 27 | 28 | OMP_NUM_THREADS=4 python ./exps/NAS-Bench-201-algos/BOHB.py \ 29 | --save_dir ${save_dir} --max_nodes ${max_nodes} --channel ${channel} --num_cells ${num_cells} \ 30 | --dataset ${dataset} \ 31 | --search_space_name ${space} \ 32 | --arch_nas_dataset ${benchmark_file} \ 33 | --time_budget 12000 \ 34 | --n_iters 50 --num_samples 4 --random_fraction 0.0 --bandwidth_factor 3 \ 35 | --workers 4 --print_freq 200 --rand_seed ${seed} 36 | -------------------------------------------------------------------------------- /scripts-search/NAS-Bench-201-algos/DARTS-V1.sh: -------------------------------------------------------------------------------- 1 | #!/bin/bash 2 | # bash ./scripts-search/NAS-Bench-201-algos/DARTS-V1.sh cifar10 0 -1 3 | echo script name: $0 4 | echo $# arguments 5 | if [ "$#" -ne 3 ] ;then 6 | echo "Input illegal number of parameters " $# 7 | echo "Need 3 parameters for dataset, tracking_status, and seed" 8 | exit 1 9 | fi 10 | if [ "$TORCH_HOME" = "" ]; then 11 | echo "Must set TORCH_HOME envoriment variable for data dir saving" 12 | exit 1 13 | else 14 | echo "TORCH_HOME : $TORCH_HOME" 15 | fi 16 | 17 | dataset=$1 18 | BN=$2 19 | seed=$3 20 | channel=16 21 | num_cells=5 22 | max_nodes=4 23 | space=nas-bench-201 24 | 25 | if [ "$dataset" == "cifar10" ] || [ "$dataset" == "cifar100" ]; then 26 | data_path="$TORCH_HOME/cifar.python" 27 | else 28 | data_path="$TORCH_HOME/cifar.python/ImageNet16" 29 | fi 30 | #benchmark_file=${TORCH_HOME}/NAS-Bench-201-v1_0-e61699.pth 31 | benchmark_file=${TORCH_HOME}/NAS-Bench-201-v1_1-096897.pth 32 | 33 | save_dir=./output/search-cell-${space}/DARTS-V1-${dataset}-BN${BN} 34 | 35 | OMP_NUM_THREADS=4 python ./exps/NAS-Bench-201-algos/DARTS-V1.py \ 36 | --save_dir ${save_dir} --max_nodes ${max_nodes} --channel ${channel} --num_cells ${num_cells} \ 37 | --dataset ${dataset} --data_path ${data_path} \ 38 | --search_space_name ${space} \ 39 | --config_path configs/nas-benchmark/algos/DARTS.config \ 40 | --arch_nas_dataset ${benchmark_file} \ 41 | --track_running_stats ${BN} \ 42 | --arch_learning_rate 0.0003 --arch_weight_decay 0.001 \ 43 | --workers 4 --print_freq 200 --rand_seed ${seed} 44 | -------------------------------------------------------------------------------- /scripts-search/NAS-Bench-201-algos/DARTS-V2.sh: -------------------------------------------------------------------------------- 1 | #!/bin/bash 2 | # bash ./scripts-search/NAS-Bench-201-algos/DARTS-V2.sh cifar10 0 -1 3 | echo script name: $0 4 | echo $# arguments 5 | if [ "$#" -ne 3 ] ;then 6 | echo "Input illegal number of parameters " $# 7 | echo "Need 3 parameters for dataset, tracking_status, and seed" 8 | exit 1 9 | fi 10 | if [ "$TORCH_HOME" = "" ]; then 11 | echo "Must set TORCH_HOME envoriment variable for data dir saving" 12 | exit 1 13 | else 14 | echo "TORCH_HOME : $TORCH_HOME" 15 | fi 16 | 17 | dataset=$1 18 | BN=$2 19 | seed=$3 20 | channel=16 21 | num_cells=5 22 | max_nodes=4 23 | space=nas-bench-201 24 | 25 | if [ "$dataset" == "cifar10" ] || [ "$dataset" == "cifar100" ]; then 26 | data_path="$TORCH_HOME/cifar.python" 27 | else 28 | data_path="$TORCH_HOME/cifar.python/ImageNet16" 29 | fi 30 | #benchmark_file=${TORCH_HOME}/NAS-Bench-201-v1_0-e61699.pth 31 | benchmark_file=${TORCH_HOME}/NAS-Bench-201-v1_1-096897.pth 32 | 33 | save_dir=./output/search-cell-${space}/DARTS-V2-${dataset}-BN${BN} 34 | 35 | OMP_NUM_THREADS=4 python ./exps/NAS-Bench-201-algos/DARTS-V2.py \ 36 | --save_dir ${save_dir} --max_nodes ${max_nodes} --channel ${channel} --num_cells ${num_cells} \ 37 | --dataset ${dataset} --data_path ${data_path} \ 38 | --search_space_name ${space} \ 39 | --config_path configs/nas-benchmark/algos/DARTS.config \ 40 | --arch_nas_dataset ${benchmark_file} \ 41 | --track_running_stats ${BN} \ 42 | --arch_learning_rate 0.0003 --arch_weight_decay 0.001 \ 43 | --workers 4 --print_freq 200 --rand_seed ${seed} 44 | -------------------------------------------------------------------------------- /scripts-search/NAS-Bench-201-algos/ENAS.sh: -------------------------------------------------------------------------------- 1 | #!/bin/bash 2 | # Efficient Neural Architecture Search via Parameter Sharing, ICML 2018 3 | # bash ./scripts-search/NAS-Bench-201-algos/ENAS.sh cifar10 0 -1 4 | echo script name: $0 5 | echo $# arguments 6 | if [ "$#" -ne 3 ] ;then 7 | echo "Input illegal number of parameters " $# 8 | echo "Need 3 parameters for dataset, BN-tracking-status, and seed" 9 | exit 1 10 | fi 11 | if [ "$TORCH_HOME" = "" ]; then 12 | echo "Must set TORCH_HOME envoriment variable for data dir saving" 13 | exit 1 14 | else 15 | echo "TORCH_HOME : $TORCH_HOME" 16 | fi 17 | 18 | dataset=$1 19 | BN=$2 20 | seed=$3 21 | channel=16 22 | num_cells=5 23 | max_nodes=4 24 | space=nas-bench-201 25 | 26 | if [ "$dataset" == "cifar10" ] || [ "$dataset" == "cifar100" ]; then 27 | data_path="$TORCH_HOME/cifar.python" 28 | else 29 | data_path="$TORCH_HOME/cifar.python/ImageNet16" 30 | fi 31 | #benchmark_file=${TORCH_HOME}/NAS-Bench-201-v1_0-e61699.pth 32 | benchmark_file=${TORCH_HOME}/NAS-Bench-201-v1_1-096897.pth 33 | 34 | save_dir=./output/search-cell-${space}/ENAS-${dataset}-BN${BN} 35 | 36 | OMP_NUM_THREADS=4 python ./exps/NAS-Bench-201-algos/ENAS.py \ 37 | --save_dir ${save_dir} --max_nodes ${max_nodes} --channel ${channel} --num_cells ${num_cells} \ 38 | --dataset ${dataset} --data_path ${data_path} \ 39 | --search_space_name ${space} \ 40 | --arch_nas_dataset ${benchmark_file} \ 41 | --track_running_stats ${BN} \ 42 | --config_path ./configs/nas-benchmark/algos/ENAS.config \ 43 | --controller_entropy_weight 0.0001 \ 44 | --controller_bl_dec 0.99 \ 45 | --controller_train_steps 50 \ 46 | --controller_num_aggregate 20 \ 47 | --controller_num_samples 100 \ 48 | --workers 4 --print_freq 200 --rand_seed ${seed} 49 | -------------------------------------------------------------------------------- /scripts-search/NAS-Bench-201-algos/GDAS.sh: -------------------------------------------------------------------------------- 1 | #!/bin/bash 2 | # bash ./scripts-search/NAS-Bench-201-algos/GDAS.sh cifar10 0 -1 3 | echo script name: $0 4 | echo $# arguments 5 | if [ "$#" -ne 3 ] ;then 6 | echo "Input illegal number of parameters " $# 7 | echo "Need 3 parameters for dataset, BN-tracking, and seed" 8 | exit 1 9 | fi 10 | if [ "$TORCH_HOME" = "" ]; then 11 | echo "Must set TORCH_HOME envoriment variable for data dir saving" 12 | exit 1 13 | else 14 | echo "TORCH_HOME : $TORCH_HOME" 15 | fi 16 | 17 | dataset=$1 18 | BN=$2 19 | seed=$3 20 | channel=16 21 | num_cells=5 22 | max_nodes=4 23 | space=nas-bench-201 24 | 25 | if [ "$dataset" == "cifar10" ] || [ "$dataset" == "cifar100" ]; then 26 | data_path="$TORCH_HOME/cifar.python" 27 | else 28 | data_path="$TORCH_HOME/cifar.python/ImageNet16" 29 | fi 30 | #benchmark_file=${TORCH_HOME}/NAS-Bench-201-v1_0-e61699.pth 31 | benchmark_file=${TORCH_HOME}/NAS-Bench-201-v1_1-096897.pth 32 | 33 | save_dir=./output/search-cell-${space}/GDAS-${dataset}-BN${BN} 34 | 35 | OMP_NUM_THREADS=4 python ./exps/NAS-Bench-201-algos/GDAS.py \ 36 | --save_dir ${save_dir} --max_nodes ${max_nodes} --channel ${channel} --num_cells ${num_cells} \ 37 | --dataset ${dataset} --data_path ${data_path} \ 38 | --search_space_name ${space} \ 39 | --arch_nas_dataset ${benchmark_file} \ 40 | --config_path configs/nas-benchmark/algos/GDAS.config \ 41 | --tau_max 10 --tau_min 0.1 --track_running_stats ${BN} \ 42 | --arch_learning_rate 0.0003 --arch_weight_decay 0.001 \ 43 | --workers 4 --print_freq 200 --rand_seed ${seed} 44 | -------------------------------------------------------------------------------- /scripts-search/NAS-Bench-201-algos/R-EA.sh: -------------------------------------------------------------------------------- 1 | #!/bin/bash 2 | # Regularized Evolution for Image Classifier Architecture Search, AAAI 2019 3 | # bash ./scripts-search/NAS-Bench-201-algos/R-EA.sh cifar10 3 -1 4 | echo script name: $0 5 | echo $# arguments 6 | if [ "$#" -ne 3 ] ;then 7 | echo "Input illegal number of parameters " $# 8 | echo "Need 3 parameters for the-dataset-name, the-ea-sample-size and the-seed" 9 | exit 1 10 | fi 11 | if [ "$TORCH_HOME" = "" ]; then 12 | echo "Must set TORCH_HOME envoriment variable for data dir saving" 13 | exit 1 14 | else 15 | echo "TORCH_HOME : $TORCH_HOME" 16 | fi 17 | 18 | #dataset=cifar10 19 | dataset=$1 20 | sample_size=$2 21 | seed=$3 22 | channel=16 23 | num_cells=5 24 | max_nodes=4 25 | space=nas-bench-201 26 | #benchmark_file=${TORCH_HOME}/NAS-Bench-201-v1_0-e61699.pth 27 | benchmark_file=${TORCH_HOME}/NAS-Bench-201-v1_1-096897.pth 28 | 29 | save_dir=./output/search-cell-${space}/R-EA-${dataset}-SS${sample_size} 30 | 31 | OMP_NUM_THREADS=4 python ./exps/NAS-Bench-201-algos/R_EA.py \ 32 | --save_dir ${save_dir} --max_nodes ${max_nodes} --channel ${channel} --num_cells ${num_cells} \ 33 | --dataset ${dataset} \ 34 | --search_space_name ${space} \ 35 | --arch_nas_dataset ${benchmark_file} \ 36 | --time_budget 12000 \ 37 | --ea_cycles 200 --ea_population 10 --ea_sample_size ${sample_size} --ea_fast_by_api 1 \ 38 | --workers 4 --print_freq 200 --rand_seed ${seed} 39 | -------------------------------------------------------------------------------- /scripts-search/NAS-Bench-201-algos/RANDOM-NAS.sh: -------------------------------------------------------------------------------- 1 | #!/bin/bash 2 | # Random Search and Reproducibility for Neural Architecture Search, UAI 2019 3 | # bash ./scripts-search/NAS-Bench-201-algos/RANDOM-NAS.sh cifar10 0 -1 4 | echo script name: $0 5 | echo $# arguments 6 | if [ "$#" -ne 3 ] ;then 7 | echo "Input illegal number of parameters " $# 8 | echo "Need 3 parameters for dataset, BN-tracking-status, and seed" 9 | exit 1 10 | fi 11 | if [ "$TORCH_HOME" = "" ]; then 12 | echo "Must set TORCH_HOME envoriment variable for data dir saving" 13 | exit 1 14 | else 15 | echo "TORCH_HOME : $TORCH_HOME" 16 | fi 17 | 18 | dataset=$1 19 | BN=$2 20 | seed=$3 21 | channel=16 22 | num_cells=5 23 | max_nodes=4 24 | space=nas-bench-201 25 | 26 | if [ "$dataset" == "cifar10" ] || [ "$dataset" == "cifar100" ]; then 27 | data_path="$TORCH_HOME/cifar.python" 28 | else 29 | data_path="$TORCH_HOME/cifar.python/ImageNet16" 30 | fi 31 | #benchmark_file=${TORCH_HOME}/NAS-Bench-201-v1_0-e61699.pth 32 | benchmark_file=${TORCH_HOME}/NAS-Bench-201-v1_1-096897.pth 33 | 34 | save_dir=./output/search-cell-${space}/RANDOM-NAS-${dataset}-BN${BN} 35 | 36 | OMP_NUM_THREADS=4 python ./exps/NAS-Bench-201-algos/RANDOM-NAS.py \ 37 | --save_dir ${save_dir} --max_nodes ${max_nodes} --channel ${channel} --num_cells ${num_cells} \ 38 | --dataset ${dataset} --data_path ${data_path} \ 39 | --search_space_name ${space} \ 40 | --track_running_stats ${BN} \ 41 | --arch_nas_dataset ${benchmark_file} \ 42 | --config_path ./configs/nas-benchmark/algos/RANDOM.config \ 43 | --select_num 100 \ 44 | --workers 4 --print_freq 200 --rand_seed ${seed} 45 | -------------------------------------------------------------------------------- /scripts-search/NAS-Bench-201-algos/README.md: -------------------------------------------------------------------------------- 1 | # 10 NAS algorithms in NAS-Bench-201 2 | 3 | Each script in this folder corresponds to one NAS algorithm, you can simple run it by one command. 4 | -------------------------------------------------------------------------------- /scripts-search/NAS-Bench-201-algos/REINFORCE.sh: -------------------------------------------------------------------------------- 1 | #!/bin/bash 2 | # bash ./scripts-search/NAS-Bench-201-algos/REINFORCE.sh 0.001 -1 3 | echo script name: $0 4 | echo $# arguments 5 | if [ "$#" -ne 3 ] ;then 6 | echo "Input illegal number of parameters " $# 7 | echo "Need 3 parameters for dataset, LR, and seed" 8 | exit 1 9 | fi 10 | if [ "$TORCH_HOME" = "" ]; then 11 | echo "Must set TORCH_HOME envoriment variable for data dir saving" 12 | exit 1 13 | else 14 | echo "TORCH_HOME : $TORCH_HOME" 15 | fi 16 | 17 | dataset=$1 18 | LR=$2 19 | seed=$3 20 | channel=16 21 | num_cells=5 22 | max_nodes=4 23 | space=nas-bench-201 24 | #benchmark_file=${TORCH_HOME}/NAS-Bench-201-v1_0-e61699.pth 25 | benchmark_file=${TORCH_HOME}/NAS-Bench-201-v1_1-096897.pth 26 | 27 | save_dir=./output/search-cell-${space}/REINFORCE-${dataset}-${LR} 28 | 29 | OMP_NUM_THREADS=4 python ./exps/NAS-Bench-201-algos/reinforce.py \ 30 | --save_dir ${save_dir} --max_nodes ${max_nodes} --channel ${channel} --num_cells ${num_cells} \ 31 | --dataset ${dataset} \ 32 | --search_space_name ${space} \ 33 | --arch_nas_dataset ${benchmark_file} \ 34 | --time_budget 12000 \ 35 | --learning_rate ${LR} --EMA_momentum 0.9 \ 36 | --workers 4 --print_freq 200 --rand_seed ${seed} 37 | -------------------------------------------------------------------------------- /scripts-search/NAS-Bench-201-algos/Random.sh: -------------------------------------------------------------------------------- 1 | #!/bin/bash 2 | # bash ./scripts-search/NAS-Bench-201-algos/Random.sh -1 3 | echo script name: $0 4 | echo $# arguments 5 | if [ "$#" -ne 2 ] ;then 6 | echo "Input illegal number of parameters " $# 7 | echo "Need 2 parameters for dataset and seed" 8 | exit 1 9 | fi 10 | if [ "$TORCH_HOME" = "" ]; then 11 | echo "Must set TORCH_HOME envoriment variable for data dir saving" 12 | exit 1 13 | else 14 | echo "TORCH_HOME : $TORCH_HOME" 15 | fi 16 | 17 | dataset=$1 18 | seed=$2 19 | channel=16 20 | num_cells=5 21 | max_nodes=4 22 | space=nas-bench-201 23 | #benchmark_file=${TORCH_HOME}/NAS-Bench-201-v1_0-e61699.pth 24 | benchmark_file=${TORCH_HOME}/NAS-Bench-201-v1_1-096897.pth 25 | 26 | save_dir=./output/search-cell-${space}/RAND-${dataset} 27 | 28 | OMP_NUM_THREADS=4 python ./exps/NAS-Bench-201-algos/RANDOM.py \ 29 | --save_dir ${save_dir} --max_nodes ${max_nodes} --channel ${channel} --num_cells ${num_cells} \ 30 | --dataset ${dataset} \ 31 | --search_space_name ${space} \ 32 | --arch_nas_dataset ${benchmark_file} \ 33 | --time_budget 12000 \ 34 | --workers 4 --print_freq 200 --rand_seed ${seed} 35 | -------------------------------------------------------------------------------- /scripts-search/NAS-Bench-201-algos/SETN.sh: -------------------------------------------------------------------------------- 1 | #!/bin/bash 2 | # One-Shot Neural Architecture Search via Self-Evaluated Template Network, ICCV 2019 3 | # bash ./scripts-search/NAS-Bench-201-algos/SETN.sh cifar10 0 -1 4 | echo script name: $0 5 | echo $# arguments 6 | if [ "$#" -ne 3 ] ;then 7 | echo "Input illegal number of parameters " $# 8 | echo "Need 3 parameters for dataset, BN-tracking-status, and seed" 9 | exit 1 10 | fi 11 | if [ "$TORCH_HOME" = "" ]; then 12 | echo "Must set TORCH_HOME envoriment variable for data dir saving" 13 | exit 1 14 | else 15 | echo "TORCH_HOME : $TORCH_HOME" 16 | fi 17 | 18 | dataset=$1 19 | BN=$2 20 | seed=$3 21 | channel=16 22 | num_cells=5 23 | max_nodes=4 24 | space=nas-bench-201 25 | 26 | if [ "$dataset" == "cifar10" ] || [ "$dataset" == "cifar100" ]; then 27 | data_path="$TORCH_HOME/cifar.python" 28 | else 29 | data_path="$TORCH_HOME/cifar.python/ImageNet16" 30 | fi 31 | #benchmark_file=${TORCH_HOME}/NAS-Bench-201-v1_0-e61699.pth 32 | benchmark_file=${TORCH_HOME}/NAS-Bench-201-v1_1-096897.pth 33 | 34 | save_dir=./output/search-cell-${space}/SETN-${dataset}-BN${BN} 35 | 36 | OMP_NUM_THREADS=4 python ./exps/NAS-Bench-201-algos/SETN.py \ 37 | --save_dir ${save_dir} --max_nodes ${max_nodes} --channel ${channel} --num_cells ${num_cells} \ 38 | --dataset ${dataset} --data_path ${data_path} \ 39 | --search_space_name ${space} \ 40 | --arch_nas_dataset ${benchmark_file} \ 41 | --config_path configs/nas-benchmark/algos/SETN.config \ 42 | --track_running_stats ${BN} \ 43 | --arch_learning_rate 0.0003 --arch_weight_decay 0.001 \ 44 | --select_num 100 \ 45 | --workers 4 --print_freq 200 --rand_seed ${seed} 46 | -------------------------------------------------------------------------------- /scripts-search/NAS-Bench-201-algos/hps/DARTS-test-Gradient.sh: -------------------------------------------------------------------------------- 1 | #!/bin/bash 2 | # bash ./scripts-search/NAS-Bench-201-algos/DARTS-test-Gradient.sh cifar10 0 5 3 | echo script name: $0 4 | echo $# arguments 5 | if [ "$#" -ne 3 ] ;then 6 | echo "Input illegal number of parameters " $# 7 | echo "Need 3 parameters for dataset, tracking_status, and gradient_clip" 8 | exit 1 9 | fi 10 | if [ "$TORCH_HOME" = "" ]; then 11 | echo "Must set TORCH_HOME envoriment variable for data dir saving" 12 | exit 1 13 | else 14 | echo "TORCH_HOME : $TORCH_HOME" 15 | fi 16 | 17 | dataset=$1 18 | BN=$2 19 | gradient_clip=$3 20 | seed=-1 21 | channel=16 22 | num_cells=5 23 | max_nodes=4 24 | space=nas-bench-201 25 | 26 | if [ "$dataset" == "cifar10" ] || [ "$dataset" == "cifar100" ]; then 27 | data_path="$TORCH_HOME/cifar.python" 28 | else 29 | data_path="$TORCH_HOME/cifar.python/ImageNet16" 30 | fi 31 | #benchmark_file=${TORCH_HOME}/NAS-Bench-201-v1_0-e61699.pth 32 | benchmark_file=${TORCH_HOME}/NAS-Bench-201-v1_1-096897.pth 33 | 34 | save_dir=./output/search-cell-${space}/DARTS-V1-${dataset}-BN${BN}-Gradient${gradient_clip} 35 | 36 | OMP_NUM_THREADS=4 python ./exps/NAS-Bench-201-algos/DARTS-V1.py \ 37 | --save_dir ${save_dir} --max_nodes ${max_nodes} --channel ${channel} --num_cells ${num_cells} \ 38 | --dataset ${dataset} --data_path ${data_path} \ 39 | --search_space_name ${space} \ 40 | --config_path configs/nas-benchmark/algos/DARTS.config \ 41 | --arch_nas_dataset ${benchmark_file} \ 42 | --track_running_stats ${BN} --gradient_clip ${gradient_clip} \ 43 | --arch_learning_rate 0.0003 --arch_weight_decay 0.001 \ 44 | --workers 4 --print_freq 200 --rand_seed ${seed} 45 | -------------------------------------------------------------------------------- /scripts-search/NAS-Bench-201-algos/hps/GRID-RL.sh: -------------------------------------------------------------------------------- 1 | #!/bin/bash 2 | echo script name: $0 3 | 4 | #lrs="0.01 0.02 0.1 0.2 0.5 1.0 1.5 2.0 2.5 3.0" 5 | lrs="0.01 0.02 0.1 0.2 0.5" 6 | 7 | for lr in ${lrs} 8 | do 9 | bash ./scripts-search/NAS-Bench-201-algos/REINFORCE.sh ${lr} -1 10 | done 11 | -------------------------------------------------------------------------------- /scripts-search/NAS-Bench-201/build.sh: -------------------------------------------------------------------------------- 1 | #!/bin/bash 2 | # bash scripts-search/NAS-Bench-201/build.sh 3 | echo script name: $0 4 | echo $# arguments 5 | if [ "$#" -ne 0 ] ;then 6 | echo "Input illegal number of parameters " $# 7 | echo "Need 0 parameters" 8 | exit 1 9 | fi 10 | 11 | save_dir=./output/nas_bench_201_package 12 | echo "Prepare to build the package in ${save_dir}" 13 | rm -rf ${save_dir} 14 | mkdir -p ${save_dir} 15 | 16 | #cp NAS-Bench-201.md ${save_dir}/README.md 17 | sed '125,187d' NAS-Bench-201.md > ${save_dir}/README.md 18 | cp LICENSE.md ${save_dir}/LICENSE.md 19 | cp -r lib/nas_201_api ${save_dir}/ 20 | rm -rf ${save_dir}/nas_201_api/__pycache__ 21 | cp exps/NAS-Bench-201/dist-setup.py ${save_dir}/setup.py 22 | 23 | cd ${save_dir} 24 | # python setup.py sdist bdist_wheel 25 | # twine upload --repository-url https://test.pypi.org/legacy/ dist/* 26 | # twine upload dist/* 27 | -------------------------------------------------------------------------------- /scripts-search/NAS-Bench-201/meta-gen.sh: -------------------------------------------------------------------------------- 1 | #!/bin/bash 2 | # bash scripts-search/NAS-Bench-201/meta-gen.sh NAS-BENCH-201 4 3 | echo script name: $0 4 | echo $# arguments 5 | if [ "$#" -ne 2 ] ;then 6 | echo "Input illegal number of parameters " $# 7 | echo "Need 2 parameters for save-dir-name and maximum-node-in-cell" 8 | exit 1 9 | fi 10 | 11 | name=$1 12 | node=$2 13 | 14 | save_dir=./output/${name}-${node} 15 | 16 | python ./exps/NAS-Bench-201/main.py --mode meta --save_dir ${save_dir} --max_node ${node} 17 | -------------------------------------------------------------------------------- /scripts-search/NAS-Bench-201/train-a-net.sh: -------------------------------------------------------------------------------- 1 | #!/bin/bash 2 | # bash ./scripts-search/NAS-Bench-201/train-a-net.sh resnet 16 5 3 | echo script name: $0 4 | echo $# arguments 5 | if [ "$#" -ne 3 ] ;then 6 | echo "Input illegal number of parameters " $# 7 | echo "Need 3 parameters for network, channel, num-of-cells" 8 | exit 1 9 | fi 10 | if [ "$TORCH_HOME" = "" ]; then 11 | echo "Must set TORCH_HOME envoriment variable for data dir saving" 12 | exit 1 13 | else 14 | echo "TORCH_HOME : $TORCH_HOME" 15 | fi 16 | 17 | model=$1 18 | channel=$2 19 | num_cells=$3 20 | 21 | save_dir=./output/NAS-BENCH-201-4/ 22 | 23 | OMP_NUM_THREADS=4 python ./exps/NAS-Bench-201/main.py \ 24 | --mode specific-${model} --save_dir ${save_dir} --max_node 4 \ 25 | --datasets cifar10 cifar10 cifar100 ImageNet16-120 \ 26 | --use_less 0 \ 27 | --splits 1 0 0 0 \ 28 | --xpaths $TORCH_HOME/cifar.python \ 29 | $TORCH_HOME/cifar.python \ 30 | $TORCH_HOME/cifar.python \ 31 | $TORCH_HOME/cifar.python/ImageNet16 \ 32 | --channel ${channel} --num_cells ${num_cells} \ 33 | --workers 4 \ 34 | --seeds 777 888 999 35 | -------------------------------------------------------------------------------- /scripts-search/NAS-Bench-201/train-models.sh: -------------------------------------------------------------------------------- 1 | #!/bin/bash 2 | # bash ./scripts-search/train-models.sh 0/1 0 100 -1 '777 888 999' 3 | echo script name: $0 4 | echo $# arguments 5 | if [ "$#" -ne 5 ] ;then 6 | echo "Input illegal number of parameters " $# 7 | echo "Need 5 parameters for use-less-or-not, start-and-end, arch-index, and seeds" 8 | exit 1 9 | fi 10 | if [ "$TORCH_HOME" = "" ]; then 11 | echo "Must set TORCH_HOME envoriment variable for data dir saving" 12 | exit 1 13 | else 14 | echo "TORCH_HOME : $TORCH_HOME" 15 | fi 16 | 17 | use_less=$1 18 | xstart=$2 19 | xend=$3 20 | arch_index=$4 21 | all_seeds=$5 22 | 23 | save_dir=./output/NAS-BENCH-201-4/ 24 | 25 | if [ ${arch_index} == "-1" ]; then 26 | mode=new 27 | else 28 | mode=cover 29 | fi 30 | 31 | OMP_NUM_THREADS=4 python ./exps/NAS-Bench-201/main.py \ 32 | --mode ${mode} --save_dir ${save_dir} --max_node 4 \ 33 | --use_less ${use_less} \ 34 | --datasets cifar10 cifar10 cifar100 ImageNet16-120 \ 35 | --splits 1 0 0 0 \ 36 | --xpaths $TORCH_HOME/cifar.python \ 37 | $TORCH_HOME/cifar.python \ 38 | $TORCH_HOME/cifar.python \ 39 | $TORCH_HOME/cifar.python/ImageNet16 \ 40 | --channel 16 --num_cells 5 \ 41 | --workers 4 \ 42 | --srange ${xstart} ${xend} --arch_index ${arch_index} \ 43 | --seeds ${all_seeds} 44 | -------------------------------------------------------------------------------- /scripts-search/NASNet-space-search-by-DARTS1V.sh: -------------------------------------------------------------------------------- 1 | #!/bin/bash 2 | # bash ./scripts-search/NASNet-space-search-by-DARTS1V.sh cifar10 -1 3 | echo script name: $0 4 | echo $# arguments 5 | if [ "$#" -ne 2 ] ;then 6 | echo "Input illegal number of parameters " $# 7 | echo "Need 2 parameters for dataset, and seed" 8 | exit 1 9 | fi 10 | if [ "$TORCH_HOME" = "" ]; then 11 | echo "Must set TORCH_HOME envoriment variable for data dir saving" 12 | exit 1 13 | else 14 | echo "TORCH_HOME : $TORCH_HOME" 15 | fi 16 | 17 | dataset=$1 18 | BN=1 19 | seed=$2 20 | channel=16 21 | num_cells=5 22 | max_nodes=4 23 | space=darts 24 | 25 | if [ "$dataset" == "cifar10" ] || [ "$dataset" == "cifar100" ]; then 26 | data_path="$TORCH_HOME/cifar.python" 27 | else 28 | data_path="$TORCH_HOME/cifar.python/ImageNet16" 29 | fi 30 | 31 | save_dir=./output/search-cell-${space}/DARTS-V1-${dataset}-BN${BN} 32 | 33 | OMP_NUM_THREADS=4 python ./exps/NAS-Bench-201-algos/DARTS-V1.py \ 34 | --save_dir ${save_dir} --max_nodes ${max_nodes} --channel ${channel} --num_cells ${num_cells} \ 35 | --dataset ${dataset} --data_path ${data_path} \ 36 | --search_space_name ${space} \ 37 | --config_path configs/search-opts/DARTS-NASNet-CIFAR.config \ 38 | --model_config configs/search-archs/DARTS-NASNet-CIFAR.config \ 39 | --track_running_stats ${BN} \ 40 | --arch_learning_rate 0.0003 --arch_weight_decay 0.001 \ 41 | --workers 4 --print_freq 200 --rand_seed ${seed} 42 | -------------------------------------------------------------------------------- /scripts-search/NASNet-space-search-by-GDAS-FRC.sh: -------------------------------------------------------------------------------- 1 | #!/bin/bash 2 | # bash ./scripts-search/NASNet-space-search-by-GDAS-FRC.sh cifar10 0 -1 3 | echo script name: $0 4 | echo $# arguments 5 | if [ "$#" -ne 3 ] ;then 6 | echo "Input illegal number of parameters " $# 7 | echo "Need 3 parameters for dataset, track_running_stats, and seed" 8 | exit 1 9 | fi 10 | if [ "$TORCH_HOME" = "" ]; then 11 | echo "Must set TORCH_HOME envoriment variable for data dir saving" 12 | exit 1 13 | else 14 | echo "TORCH_HOME : $TORCH_HOME" 15 | fi 16 | 17 | dataset=$1 18 | track_running_stats=$2 19 | seed=$3 20 | space=darts 21 | 22 | if [ "$dataset" == "cifar10" ] || [ "$dataset" == "cifar100" ]; then 23 | data_path="$TORCH_HOME/cifar.python" 24 | else 25 | data_path="$TORCH_HOME/cifar.python/ImageNet16" 26 | fi 27 | 28 | save_dir=./output/search-cell-${space}/GDAS-FRC-${dataset}-BN${track_running_stats} 29 | 30 | OMP_NUM_THREADS=4 python ./exps/algos/GDAS.py \ 31 | --save_dir ${save_dir} \ 32 | --dataset ${dataset} --data_path ${data_path} \ 33 | --search_space_name ${space} \ 34 | --config_path configs/search-opts/GDAS-NASNet-CIFAR.config \ 35 | --model_config configs/search-archs/GDASFRC-NASNet-CIFAR.config \ 36 | --tau_max 10 --tau_min 0.1 --track_running_stats ${track_running_stats} \ 37 | --arch_learning_rate 0.0003 --arch_weight_decay 0.001 \ 38 | --workers 4 --print_freq 200 --rand_seed ${seed} 39 | -------------------------------------------------------------------------------- /scripts-search/NASNet-space-search-by-GDAS.sh: -------------------------------------------------------------------------------- 1 | #!/bin/bash 2 | # bash ./scripts-search/NASNet-space-search-by-GDAS.sh cifar10 1 -1 3 | echo script name: $0 4 | echo $# arguments 5 | if [ "$#" -ne 3 ] ;then 6 | echo "Input illegal number of parameters " $# 7 | echo "Need 3 parameters for dataset, track_running_stats, and seed" 8 | exit 1 9 | fi 10 | if [ "$TORCH_HOME" = "" ]; then 11 | echo "Must set TORCH_HOME envoriment variable for data dir saving" 12 | exit 1 13 | else 14 | echo "TORCH_HOME : $TORCH_HOME" 15 | fi 16 | 17 | dataset=$1 18 | track_running_stats=$2 19 | seed=$3 20 | space=darts 21 | 22 | if [ "$dataset" == "cifar10" ] || [ "$dataset" == "cifar100" ]; then 23 | data_path="$TORCH_HOME/cifar.python" 24 | else 25 | data_path="$TORCH_HOME/cifar.python/ImageNet16" 26 | fi 27 | 28 | save_dir=./output/search-cell-${space}/GDAS-${dataset}-BN${track_running_stats} 29 | 30 | OMP_NUM_THREADS=4 python ./exps/NAS-Bench-201-algos/GDAS.py \ 31 | --save_dir ${save_dir} \ 32 | --dataset ${dataset} --data_path ${data_path} \ 33 | --search_space_name ${space} \ 34 | --config_path configs/search-opts/GDAS-NASNet-CIFAR.config \ 35 | --model_config configs/search-archs/GDAS-NASNet-CIFAR.config \ 36 | --tau_max 10 --tau_min 0.1 --track_running_stats ${track_running_stats} \ 37 | --arch_learning_rate 0.0003 --arch_weight_decay 0.001 \ 38 | --workers 4 --print_freq 200 --rand_seed ${seed} 39 | -------------------------------------------------------------------------------- /scripts-search/NASNet-space-search-by-SETN.sh: -------------------------------------------------------------------------------- 1 | #!/bin/bash 2 | # bash ./scripts-search/NASNet-space-search-by-SETN.sh cifar10 1 -1 3 | # [TO BE DONE] 4 | echo script name: $0 5 | echo $# arguments 6 | if [ "$#" -ne 3 ] ;then 7 | echo "Input illegal number of parameters " $# 8 | echo "Need 3 parameters for dataset, track_running_stats, and seed" 9 | exit 1 10 | fi 11 | if [ "$TORCH_HOME" = "" ]; then 12 | echo "Must set TORCH_HOME envoriment variable for data dir saving" 13 | exit 1 14 | else 15 | echo "TORCH_HOME : $TORCH_HOME" 16 | fi 17 | 18 | dataset=$1 19 | track_running_stats=$2 20 | seed=$3 21 | space=darts 22 | 23 | if [ "$dataset" == "cifar10" ] || [ "$dataset" == "cifar100" ]; then 24 | data_path="$TORCH_HOME/cifar.python" 25 | else 26 | data_path="$TORCH_HOME/cifar.python/ImageNet16" 27 | fi 28 | 29 | save_dir=./output/search-cell-${space}/SETN-${dataset}-BN${track_running_stats} 30 | 31 | OMP_NUM_THREADS=4 python ./exps/NAS-Bench-201-algos/SETN.py \ 32 | --save_dir ${save_dir} \ 33 | --dataset ${dataset} --data_path ${data_path} \ 34 | --search_space_name ${space} \ 35 | --config_path configs/search-opts/SETN-NASNet-CIFAR.config \ 36 | --model_config configs/search-archs/SETN-NASNet-CIFAR.config \ 37 | --track_running_stats ${track_running_stats} \ 38 | --select_num 1000 \ 39 | --arch_learning_rate 0.0003 --arch_weight_decay 0.001 \ 40 | --workers 4 --print_freq 200 --rand_seed ${seed} 41 | -------------------------------------------------------------------------------- /scripts-search/NATS/search-size.sh: -------------------------------------------------------------------------------- 1 | #!/bin/bash 2 | # bash scripts-search/NATS/search-size.sh 0 0.3 777 3 | echo script name: $0 4 | echo $# arguments 5 | if [ "$#" -ne 3 ] ;then 6 | echo "Input illegal number of parameters " $# 7 | echo "Need 3 parameters for GPU-device, warmup-ratio, and seed" 8 | exit 1 9 | fi 10 | if [ "$TORCH_HOME" = "" ]; then 11 | echo "Must set TORCH_HOME envoriment variable for data dir saving" 12 | exit 1 13 | else 14 | echo "TORCH_HOME : $TORCH_HOME" 15 | fi 16 | 17 | device=$1 18 | ratio=$2 19 | seed=$3 20 | 21 | CUDA_VISIBLE_DEVICES=${device} python ./exps/NATS-algos/search-size.py --dataset cifar10 --data_path $TORCH_HOME/cifar.python --algo tas --warmup_ratio ${ratio} --rand_seed ${seed} 22 | CUDA_VISIBLE_DEVICES=${device} python ./exps/NATS-algos/search-size.py --dataset cifar100 --data_path $TORCH_HOME/cifar.python --algo tas --warmup_ratio ${ratio} --rand_seed ${seed} 23 | CUDA_VISIBLE_DEVICES=${device} python ./exps/NATS-algos/search-size.py --dataset ImageNet16-120 --data_path $TORCH_HOME/cifar.python/ImageNet16 --algo tas --warmup_ratio ${ratio} --rand_seed ${seed} 24 | 25 | # 26 | CUDA_VISIBLE_DEVICES=${device} python ./exps/NATS-algos/search-size.py --dataset cifar10 --data_path $TORCH_HOME/cifar.python --algo mask_gumbel --warmup_ratio ${ratio} --rand_seed ${seed} 27 | CUDA_VISIBLE_DEVICES=${device} python ./exps/NATS-algos/search-size.py --dataset cifar100 --data_path $TORCH_HOME/cifar.python --algo mask_gumbel --warmup_ratio ${ratio} --rand_seed ${seed} 28 | CUDA_VISIBLE_DEVICES=${device} python ./exps/NATS-algos/search-size.py --dataset ImageNet16-120 --data_path $TORCH_HOME/cifar.python/ImageNet16 --algo mask_gumbel --warmup_ratio ${ratio} --rand_seed ${seed} 29 | 30 | # 31 | CUDA_VISIBLE_DEVICES=${device} python ./exps/NATS-algos/search-size.py --dataset cifar10 --data_path $TORCH_HOME/cifar.python --algo mask_rl --arch_weight_decay 0 --warmup_ratio ${ratio} --rand_seed ${seed} 32 | CUDA_VISIBLE_DEVICES=${device} python ./exps/NATS-algos/search-size.py --dataset cifar100 --data_path $TORCH_HOME/cifar.python --algo mask_rl --arch_weight_decay 0 --warmup_ratio ${ratio} --rand_seed ${seed} 33 | CUDA_VISIBLE_DEVICES=${device} python ./exps/NATS-algos/search-size.py --dataset ImageNet16-120 --data_path $TORCH_HOME/cifar.python/ImageNet16 --algo mask_rl --arch_weight_decay 0 --warmup_ratio ${ratio} --rand_seed ${seed} 34 | -------------------------------------------------------------------------------- /scripts-search/search-depth-gumbel.sh: -------------------------------------------------------------------------------- 1 | #!/bin/bash 2 | # bash ./scripts-search/search-depth-gumbel.sh cifar10 ResNet110 CIFARX 0.57 777 3 | set -e 4 | echo script name: $0 5 | echo $# arguments 6 | if [ "$#" -ne 5 ] ;then 7 | echo "Input illegal number of parameters " $# 8 | echo "Need 5 parameters for the dataset and the-model-name and the-optimizer and FLOP-ratio and the-random-seed" 9 | exit 1 10 | fi 11 | if [ "$TORCH_HOME" = "" ]; then 12 | echo "Must set TORCH_HOME envoriment variable for data dir saving" 13 | exit 1 14 | else 15 | echo "TORCH_HOME : $TORCH_HOME" 16 | fi 17 | 18 | dataset=$1 19 | model=$2 20 | optim=$3 21 | expected_FLOP_ratio=$4 22 | rseed=$5 23 | 24 | bash ./scripts-search/search-depth-cifar.sh ${dataset} ${model} ${optim} 0.1 5 ${expected_FLOP_ratio} ${rseed} 25 | -------------------------------------------------------------------------------- /scripts-search/search-width-gumbel.sh: -------------------------------------------------------------------------------- 1 | #!/bin/bash 2 | # bash ./scripts-search/search-width-gumbel.sh cifar10 ResNet110 CIFARX 0.57 777 3 | set -e 4 | echo script name: $0 5 | echo $# arguments 6 | if [ "$#" -ne 5 ] ;then 7 | echo "Input illegal number of parameters " $# 8 | echo "Need 5 parameters for the dataset and the-model-name and the-optimizer and FLOP-ratio and the-random-seed" 9 | exit 1 10 | fi 11 | if [ "$TORCH_HOME" = "" ]; then 12 | echo "Must set TORCH_HOME envoriment variable for data dir saving" 13 | exit 1 14 | else 15 | echo "TORCH_HOME : $TORCH_HOME" 16 | fi 17 | 18 | dataset=$1 19 | model=$2 20 | optim=$3 21 | expected_FLOP_ratio=$4 22 | rseed=$5 23 | 24 | bash ./scripts-search/search-width-cifar.sh ${dataset} ${model} ${optim} 0.1 5 ${expected_FLOP_ratio} ${rseed} 25 | -------------------------------------------------------------------------------- /scripts/NATS-Bench/train-topology.sh: -------------------------------------------------------------------------------- 1 | #!/bin/bash 2 | ############################################################################## 3 | # NATS-Bench: Benchmarking NAS algorithms for Architecture Topology and Size # 4 | ############################################################################## 5 | # Copyright (c) Xuanyi Dong [GitHub D-X-Y], 2020.01 # 6 | ############################################################################## 7 | # [saturn1] CUDA_VISIBLE_DEVICES=0 bash scripts/NATS-Bench/train-topology.sh 00000-02000 200 "777 888 999" 8 | # [saturn1] CUDA_VISIBLE_DEVICES=0 bash scripts/NATS-Bench/train-topology.sh 02000-04000 200 "777 888 999" 9 | # [saturn1] CUDA_VISIBLE_DEVICES=1 bash scripts/NATS-Bench/train-topology.sh 04000-06000 200 "777 888 999" 10 | # [saturn1] CUDA_VISIBLE_DEVICES=1 bash scripts/NATS-Bench/train-topology.sh 06000-08000 200 "777 888 999" 11 | # 12 | # CUDA_VISIBLE_DEVICES=0 bash scripts/NATS-Bench/train-topology.sh 00000-05000 12 777 13 | # bash ./scripts/NATS-Bench/train-topology.sh 05001-10000 12 777 14 | # bash ./scripts/NATS-Bench/train-topology.sh 10001-14500 12 777 15 | # bash ./scripts/NATS-Bench/train-topology.sh 14501-15624 12 777 16 | # 17 | ############################################################################## 18 | echo script name: $0 19 | echo $# arguments 20 | if [ "$#" -ne 3 ] ;then 21 | echo "Input illegal number of parameters " $# 22 | echo "Need 3 parameters for start-and-end, hyper-parameters-opt-file, and seeds" 23 | exit 1 24 | fi 25 | if [ "$TORCH_HOME" = "" ]; then 26 | echo "Must set TORCH_HOME envoriment variable for data dir saving" 27 | exit 1 28 | else 29 | echo "TORCH_HOME : $TORCH_HOME" 30 | fi 31 | 32 | srange=$1 33 | opt=$2 34 | all_seeds=$3 35 | cpus=4 36 | 37 | save_dir=./output/NATS-Bench-topology/ 38 | 39 | OMP_NUM_THREADS=${cpus} python exps/NATS-Bench/main-tss.py \ 40 | --mode new --srange ${srange} --hyper ${opt} --save_dir ${save_dir} \ 41 | --datasets cifar10 cifar10 cifar100 ImageNet16-120 \ 42 | --splits 1 0 0 0 \ 43 | --xpaths $TORCH_HOME/cifar.python \ 44 | $TORCH_HOME/cifar.python \ 45 | $TORCH_HOME/cifar.python \ 46 | $TORCH_HOME/cifar.python/ImageNet16 \ 47 | --workers ${cpus} \ 48 | --seeds ${all_seeds} 49 | -------------------------------------------------------------------------------- /scripts/TAS/prepare.sh: -------------------------------------------------------------------------------- 1 | #!/bin/bash 2 | # bash ./scripts/TAS/prepare.sh 3 | #datasets='cifar10 cifar100 imagenet-1k' 4 | #ratios='0.5 0.8 0.9' 5 | ratios='0.5' 6 | save_dir=./.latent-data/splits 7 | 8 | for ratio in ${ratios} 9 | do 10 | python ./exps/TAS/prepare.py --name cifar10 --root $TORCH_HOME/cifar.python --save ${save_dir}/cifar10-${ratio}.pth --ratio ${ratio} 11 | python ./exps/TAS/prepare.py --name cifar100 --root $TORCH_HOME/cifar.python --save ${save_dir}/cifar100-${ratio}.pth --ratio ${ratio} 12 | python ./exps/TAS/prepare.py --name imagenet-1k --root $TORCH_HOME/ILSVRC2012 --save ${save_dir}/imagenet-1k-${ratio}.pth --ratio ${ratio} 13 | done 14 | -------------------------------------------------------------------------------- /scripts/base-train.sh: -------------------------------------------------------------------------------- 1 | #!/bin/bash 2 | # bash ./scripts/base-train.sh cifar10 ResNet110 E300 L1 256 -1 3 | echo script name: $0 4 | echo $# arguments 5 | if [ "$#" -ne 6 ] ;then 6 | echo "Input illegal number of parameters " $# 7 | echo "Need 6 parameters for the dataset and the-model-name and epochs and LR and the-batch-size and the-random-seed" 8 | exit 1 9 | fi 10 | if [ "$TORCH_HOME" = "" ]; then 11 | echo "Must set TORCH_HOME envoriment variable for data dir saving" 12 | exit 1 13 | else 14 | echo "TORCH_HOME : $TORCH_HOME" 15 | fi 16 | 17 | dataset=$1 18 | model=$2 19 | epoch=$3 20 | LR=$4 21 | batch=$5 22 | rseed=$6 23 | 24 | 25 | save_dir=./output/basic/${dataset}/${model}-${epoch}-${LR}-${batch} 26 | 27 | python --version 28 | 29 | OMP_NUM_THREADS=4 python ./exps/basic/basic-main.py --dataset ${dataset} \ 30 | --data_path $TORCH_HOME/cifar.python \ 31 | --model_config ./configs/archs/CIFAR-${model}.config \ 32 | --optim_config ./configs/opts/CIFAR-${epoch}-W5-${LR}-COS.config \ 33 | --procedure basic \ 34 | --save_dir ${save_dir} \ 35 | --cutout_length -1 \ 36 | --batch_size ${batch} --rand_seed ${rseed} --workers 4 \ 37 | --eval_frequency 1 --print_freq 100 --print_freq_eval 200 38 | -------------------------------------------------------------------------------- /scripts/black.sh: -------------------------------------------------------------------------------- 1 | #!/bin/bash 2 | # bash ./scripts/black.sh 3 | 4 | # script=$(readlink -f "$0") 5 | # scriptpath=$(dirname "$script") 6 | # echo $scriptpath 7 | 8 | # delete Python cache files 9 | find . | grep -E "(__pycache__|\.pyc|\.DS_Store|\.pyo$)" | xargs rm -rf 10 | 11 | black ./tests/ 12 | black ./xautodl/procedures 13 | black ./xautodl/datasets 14 | black ./xautodl/xlayers 15 | black ./exps/trading 16 | rm -rf ./xautodl.egg-info 17 | rm -rf ./build 18 | rm -rf ./dist 19 | rm -rf ./.pytest_cache 20 | -------------------------------------------------------------------------------- /scripts/experimental/train-vit.sh: -------------------------------------------------------------------------------- 1 | #!/bin/bash 2 | # bash ./scripts/experimental/train-vit.sh cifar10 -1 3 | echo script name: $0 4 | echo $# arguments 5 | if [ "$#" -ne 2 ] ;then 6 | echo "Input illegal number of parameters " $# 7 | echo "Need 2 parameters for dataset and random-seed" 8 | exit 1 9 | fi 10 | if [ "$TORCH_HOME" = "" ]; then 11 | echo "Must set TORCH_HOME envoriment variable for data dir saving" 12 | exit 1 13 | else 14 | echo "TORCH_HOME : $TORCH_HOME" 15 | fi 16 | 17 | dataset=$1 18 | rseed=$2 19 | 20 | save_dir=./outputs/${dataset}/vit-experimental 21 | 22 | python --version 23 | 24 | python ./exps/basic/xmain.py --save_dir ${save_dir} --rand_seed ${rseed} \ 25 | --train_data_config ./configs/yaml.data/${dataset}.train \ 26 | --valid_data_config ./configs/yaml.data/${dataset}.test \ 27 | --data_path $TORCH_HOME/cifar.python \ 28 | --model_config ./configs/yaml.model/vit-cifar10.s0 \ 29 | --optim_config ./configs/yaml.opt/vit.cifar \ 30 | --loss_config ./configs/yaml.loss/cross-entropy \ 31 | --metric_config ./configs/yaml.loss/top-ce \ 32 | --batch_size 256 \ 33 | --lr 0.003 --weight_decay 0.3 --scheduler warm-cos --steps 10000 34 | -------------------------------------------------------------------------------- /scripts/nas-infer-train.sh: -------------------------------------------------------------------------------- 1 | #!/bin/bash 2 | # bash ./scripts/nas-infer-train.sh cifar10 SETN 256 -1 3 | echo script name: $0 4 | echo $# arguments 5 | if [ "$#" -ne 4 ] ;then 6 | echo "Input illegal number of parameters " $# 7 | echo "Need 4 parameters for dataset, the-model-name, the-batch-size and the-random-seed" 8 | exit 1 9 | fi 10 | if [ "$TORCH_HOME" = "" ]; then 11 | echo "Must set TORCH_HOME envoriment variable for data dir saving" 12 | exit 1 13 | else 14 | echo "TORCH_HOME : $TORCH_HOME" 15 | fi 16 | 17 | dataset=$1 18 | model=$2 19 | batch=$3 20 | rseed=$4 21 | 22 | if [ ${dataset} == 'cifar10' ] || [ ${dataset} == 'cifar100' ]; then 23 | xpath=$TORCH_HOME/cifar.python 24 | base=CIFAR 25 | workers=4 26 | cutout_length=16 27 | elif [ ${dataset} == 'imagenet-1k' ]; then 28 | xpath=$TORCH_HOME/ILSVRC2012 29 | base=IMAGENET 30 | workers=28 31 | cutout_length=-1 32 | else 33 | exit 1 34 | echo 'Unknown dataset: '${dataset} 35 | fi 36 | 37 | SAVE_ROOT="./output" 38 | 39 | save_dir=${SAVE_ROOT}/nas-infer/${dataset}-${model}-${batch} 40 | 41 | python --version 42 | 43 | python ./exps/basic/basic-main.py --dataset ${dataset} \ 44 | --data_path ${xpath} --model_source nas \ 45 | --model_config ./configs/archs/NAS-${base}-${model}.config \ 46 | --optim_config ./configs/opts/NAS-${base}.config \ 47 | --procedure basic \ 48 | --save_dir ${save_dir} \ 49 | --cutout_length ${cutout_length} \ 50 | --batch_size ${batch} --rand_seed ${rseed} --workers ${workers} \ 51 | --eval_frequency 1 --print_freq 500 --print_freq_eval 1000 52 | -------------------------------------------------------------------------------- /scripts/retrain-searched-net.sh: -------------------------------------------------------------------------------- 1 | #!/bin/bash 2 | # bash ./scripts/retrain-searched-net.sh cifar10 ${NAME} ${PATH} 256 -1 3 | echo script name: $0 4 | echo $# arguments 5 | if [ "$#" -ne 5 ] ;then 6 | echo "Input illegal number of parameters " $# 7 | echo "Need 5 parameters for dataset, the save dir base name, the model path, the batch size, the random seed" 8 | exit 1 9 | fi 10 | if [ "$TORCH_HOME" = "" ]; then 11 | echo "Must set TORCH_HOME envoriment variable for data dir saving" 12 | exit 1 13 | else 14 | echo "TORCH_HOME : $TORCH_HOME" 15 | fi 16 | 17 | dataset=$1 18 | save_name=$2 19 | model_path=$3 20 | batch=$4 21 | rseed=$5 22 | 23 | if [ ${dataset} == 'cifar10' ] || [ ${dataset} == 'cifar100' ]; then 24 | xpath=$TORCH_HOME/cifar.python 25 | base=CIFAR 26 | workers=4 27 | cutout_length=16 28 | elif [ ${dataset} == 'imagenet-1k' ]; then 29 | xpath=$TORCH_HOME/ILSVRC2012 30 | base=IMAGENET 31 | workers=28 32 | cutout_length=-1 33 | else 34 | exit 1 35 | echo 'Unknown dataset: '${dataset} 36 | fi 37 | 38 | SAVE_ROOT="./output" 39 | 40 | save_dir=${SAVE_ROOT}/nas-infer/${dataset}-BS${batch}-${save_name} 41 | 42 | python --version 43 | 44 | python ./exps/basic/basic-main.py --dataset ${dataset} \ 45 | --data_path ${xpath} --model_source autodl-searched \ 46 | --model_config ./configs/archs/NAS-${base}-none.config \ 47 | --optim_config ./configs/opts/NAS-${base}.config \ 48 | --extra_model_path ${model_path} \ 49 | --procedure basic \ 50 | --save_dir ${save_dir} \ 51 | --cutout_length ${cutout_length} \ 52 | --batch_size ${batch} --rand_seed ${rseed} --workers ${workers} \ 53 | --eval_frequency 1 --print_freq 500 --print_freq_eval 1000 54 | -------------------------------------------------------------------------------- /scripts/tas-infer-train.sh: -------------------------------------------------------------------------------- 1 | #!/bin/bash 2 | # bash ./scripts/tas-infer-train.sh cifar10 C100-ResNet32 -1 3 | set -e 4 | echo script name: $0 5 | echo $# arguments 6 | if [ "$#" -ne 3 ] ;then 7 | echo "Input illegal number of parameters " $# 8 | echo "Need 3 parameters for the dataset and the-config-name and the-random-seed" 9 | exit 1 10 | fi 11 | if [ "$TORCH_HOME" = "" ]; then 12 | echo "Must set TORCH_HOME envoriment variable for data dir saving" 13 | exit 1 14 | else 15 | echo "TORCH_HOME : $TORCH_HOME" 16 | fi 17 | 18 | dataset=$1 19 | model=$2 20 | rseed=$3 21 | batch=256 22 | 23 | save_dir=./output/search-shape/TAS-INFER-${dataset}-${model} 24 | 25 | if [ ${dataset} == 'cifar10' ] || [ ${dataset} == 'cifar100' ]; then 26 | xpath=$TORCH_HOME/cifar.python 27 | opt_config=./configs/opts/CIFAR-E300-W5-L1-COS.config 28 | workers=4 29 | elif [ ${dataset} == 'imagenet-1k' ]; then 30 | xpath=$TORCH_HOME/ILSVRC2012 31 | #opt_config=./configs/opts/ImageNet-E120-Cos-Smooth.config 32 | opt_config=./configs/opts/RImageNet-E120-Cos-Soft.config 33 | workers=28 34 | else 35 | echo 'Unknown dataset: '${dataset} 36 | exit 1 37 | fi 38 | 39 | python --version 40 | 41 | # normal training 42 | xsave_dir=${save_dir}-NMT 43 | OMP_NUM_THREADS=4 python ./exps/basic/basic-main.py --dataset ${dataset} \ 44 | --data_path ${xpath} \ 45 | --model_config ./configs/NeurIPS-2019/${model}.config \ 46 | --optim_config ${opt_config} \ 47 | --procedure basic \ 48 | --save_dir ${xsave_dir} \ 49 | --cutout_length -1 \ 50 | --batch_size ${batch} --rand_seed ${rseed} --workers ${workers} \ 51 | --eval_frequency 1 --print_freq 100 --print_freq_eval 200 52 | 53 | # KD training 54 | xsave_dir=${save_dir}-KDT 55 | OMP_NUM_THREADS=4 python ./exps/basic/KD-main.py --dataset ${dataset} \ 56 | --data_path ${xpath} \ 57 | --model_config ./configs/NeurIPS-2019/${model}.config \ 58 | --optim_config ${opt_config} \ 59 | --KD_checkpoint ./.latent-data/basemodels/${dataset}/${model}.pth \ 60 | --procedure Simple-KD \ 61 | --save_dir ${xsave_dir} \ 62 | --KD_alpha 0.9 --KD_temperature 4 \ 63 | --cutout_length -1 \ 64 | --batch_size ${batch} --rand_seed ${rseed} --workers ${workers} \ 65 | --eval_frequency 1 --print_freq 100 --print_freq_eval 200 66 | -------------------------------------------------------------------------------- /scripts/trade/baseline.sh: -------------------------------------------------------------------------------- 1 | #!/bin/bash 2 | # 3 | # bash scripts/trade/baseline.sh 0 csi300 4 | # bash scripts/trade/baseline.sh 1 csi100 5 | # bash scripts/trade/baseline.sh 1 all 6 | # 7 | set -e 8 | echo script name: $0 9 | echo $# arguments 10 | 11 | if [ "$#" -ne 2 ] ;then 12 | echo "Input illegal number of parameters " $# 13 | exit 1 14 | fi 15 | 16 | gpu=$1 17 | market=$2 18 | 19 | # algorithms="NAIVE-V1 NAIVE-V2 MLP GRU LSTM ALSTM XGBoost LightGBM SFM TabNet DoubleE" 20 | algorithms="XGBoost LightGBM SFM TabNet DoubleE" 21 | 22 | for alg in ${algorithms} 23 | do 24 | python exps/trading/baselines.py --alg ${alg} --gpu ${gpu} --market ${market} 25 | done 26 | -------------------------------------------------------------------------------- /scripts/trade/tsf-all.sh: -------------------------------------------------------------------------------- 1 | #!/bin/bash 2 | # 3 | # bash scripts/trade/tsf-all.sh 0 csi300 0_0 4 | # bash scripts/trade/tsf-all.sh 0 csi300 0.1_0 5 | # bash scripts/trade/tsf-all.sh 1 all 6 | # 7 | set -e 8 | echo script name: $0 9 | echo $# arguments 10 | 11 | if [ "$#" -ne 3 ] ;then 12 | echo "Input illegal number of parameters " $# 13 | exit 1 14 | fi 15 | 16 | gpu=$1 17 | market=$2 18 | drop=$3 19 | 20 | channels="6 12 24 32 48 64" 21 | #depths="1 2 3 4 5 6 7 8" 22 | 23 | for channel in ${channels} 24 | do 25 | python exps/trading/baselines.py --alg TSF-1x${channel}-drop${drop} \ 26 | TSF-2x${channel}-drop${drop} \ 27 | TSF-3x${channel}-drop${drop} \ 28 | TSF-4x${channel}-drop${drop} \ 29 | TSF-5x${channel}-drop${drop} \ 30 | TSF-6x${channel}-drop${drop} \ 31 | TSF-7x${channel}-drop${drop} \ 32 | TSF-8x${channel}-drop${drop} \ 33 | --gpu ${gpu} --market ${market} --shared_dataset True 34 | done 35 | -------------------------------------------------------------------------------- /scripts/trade/tsf-time.sh: -------------------------------------------------------------------------------- 1 | #!/bin/bash 2 | # 3 | # bash scripts/trade/tsf-time.sh 0 csi300 TSF-2x24-drop0_0 4 | # bash scripts/trade/tsf-time.sh 1 csi100 5 | # bash scripts/trade/tsf-time.sh 1 all 6 | # 7 | set -e 8 | echo script name: $0 9 | echo $# arguments 10 | 11 | if [ "$#" -ne 3 ] ;then 12 | echo "Input illegal number of parameters " $# 13 | exit 1 14 | fi 15 | 16 | gpu=$1 17 | market=$2 18 | base=$3 19 | xtimes="2008-01-01 2008-07-01 2009-01-01 2009-07-01 2010-01-01 2011-01-01 2012-01-01 2013-01-01" 20 | 21 | for xtime in ${xtimes} 22 | do 23 | 24 | python exps/trading/baselines.py --alg ${base}s${xtime} --gpu ${gpu} --market ${market} --shared_dataset False 25 | 26 | done 27 | -------------------------------------------------------------------------------- /scripts/trade/tsf.sh: -------------------------------------------------------------------------------- 1 | #!/bin/bash 2 | # 3 | # bash scripts/trade/tsf.sh 0 csi300 3 0_0 4 | # bash scripts/trade/tsf.sh 0 csi300 3 0.1_0 5 | # bash scripts/trade/tsf.sh 1 csi100 3 0.2_0 6 | # bash scripts/trade/tsf.sh 1 all 3 0.1_0 7 | # 8 | set -e 9 | echo script name: $0 10 | echo $# arguments 11 | 12 | if [ "$#" -ne 4 ] ;then 13 | echo "Input illegal number of parameters " $# 14 | exit 1 15 | fi 16 | 17 | gpu=$1 18 | market=$2 19 | depth=$3 20 | drop=$4 21 | 22 | channels="6 12 24 32 48 64" 23 | 24 | for channel in ${channels} 25 | do 26 | 27 | python exps/trading/baselines.py --alg TSF-${depth}x${channel}-drop${drop} --gpu ${gpu} --market ${market} 28 | 29 | done 30 | -------------------------------------------------------------------------------- /setup.py: -------------------------------------------------------------------------------- 1 | ##################################################### 2 | # Copyright (c) Xuanyi Dong [GitHub D-X-Y], 2021.05 # 3 | ##################################################### 4 | """The setup function for pypi.""" 5 | # The following is to make nats_bench avaliable on Python Package Index (PyPI) 6 | # 7 | # conda install -c conda-forge twine # Use twine to upload nats_bench to pypi 8 | # 9 | # python setup.py sdist bdist_wheel 10 | # python setup.py --help-commands 11 | # twine check dist/* 12 | # 13 | # twine upload --repository-url https://test.pypi.org/legacy/ dist/* 14 | # twine upload dist/* 15 | # https://pypi.org/project/xautodl 16 | # 17 | # TODO(xuanyidong): upload it to conda 18 | # 19 | # [2021.06.01] v0.9.9 20 | # [2021.08.14] v1.0.0 21 | # 22 | import os 23 | from setuptools import setup, find_packages 24 | 25 | NAME = "xautodl" 26 | REQUIRES_PYTHON = ">=3.6" 27 | DESCRIPTION = "Automated Deep Learning Package" 28 | 29 | VERSION = "1.0.0" 30 | 31 | 32 | def read(fname="README.md"): 33 | with open( 34 | os.path.join(os.path.dirname(__file__), fname), encoding="utf-8" 35 | ) as cfile: 36 | return cfile.read() 37 | 38 | 39 | # What packages are required for this module to be executed? 40 | REQUIRED = ["numpy>=1.16.5", "pyyaml>=5.0.0", "fvcore"] 41 | 42 | packages = find_packages( 43 | exclude=("tests", "scripts", "scripts-search", "lib*", "exps*") 44 | ) 45 | print("packages: {:}".format(packages)) 46 | 47 | setup( 48 | name=NAME, 49 | version=VERSION, 50 | author="Xuanyi Dong", 51 | author_email="dongxuanyi888@gmail.com", 52 | description=DESCRIPTION, 53 | license="MIT Licence", 54 | keywords="NAS Dataset API DeepLearning", 55 | url="https://github.com/D-X-Y/AutoDL-Projects", 56 | packages=packages, 57 | install_requires=REQUIRED, 58 | python_requires=REQUIRES_PYTHON, 59 | long_description=read("README.md"), 60 | long_description_content_type="text/markdown", 61 | classifiers=[ 62 | "Programming Language :: Python", 63 | "Programming Language :: Python :: 3", 64 | "Topic :: Database", 65 | "Topic :: Scientific/Engineering :: Artificial Intelligence", 66 | "License :: OSI Approved :: MIT License", 67 | ], 68 | ) 69 | -------------------------------------------------------------------------------- /tests/test_import.py: -------------------------------------------------------------------------------- 1 | ##################################################### 2 | # Copyright (c) Xuanyi Dong [GitHub D-X-Y], 2021.03 # 3 | ##################################################### 4 | # pytest ./tests/test_import.py # 5 | ##################################################### 6 | def test_import(): 7 | from xautodl import config_utils 8 | from xautodl import datasets 9 | from xautodl import log_utils 10 | from xautodl import models 11 | from xautodl import nas_infer_model 12 | from xautodl import procedures 13 | from xautodl import trade_models 14 | from xautodl import utils 15 | 16 | from xautodl import xlayers 17 | from xautodl import xmisc 18 | from xautodl import xmodels 19 | from xautodl import spaces 20 | 21 | print("Check all imports done") 22 | -------------------------------------------------------------------------------- /tests/test_loader.py: -------------------------------------------------------------------------------- 1 | ##################################################### 2 | # Copyright (c) Xuanyi Dong [GitHub D-X-Y], 2021.03 # 3 | ##################################################### 4 | # pytest tests/test_loader.py -s # 5 | ##################################################### 6 | import unittest 7 | import tempfile 8 | import torch 9 | 10 | from xautodl.datasets import get_datasets 11 | 12 | 13 | def test_simple(): 14 | xdir = tempfile.mkdtemp() 15 | train_data, valid_data, xshape, class_num = get_datasets("cifar10", xdir, -1) 16 | print(train_data) 17 | print(valid_data) 18 | 19 | xloader = torch.utils.data.DataLoader( 20 | train_data, batch_size=256, shuffle=True, num_workers=4, pin_memory=True 21 | ) 22 | print(xloader) 23 | print(next(iter(xloader))) 24 | 25 | for i, data in enumerate(xloader): 26 | print(i) 27 | 28 | 29 | test_simple() 30 | -------------------------------------------------------------------------------- /tests/test_math_static.py: -------------------------------------------------------------------------------- 1 | ##################################################### 2 | # Copyright (c) Xuanyi Dong [GitHub D-X-Y], 2021.03 # 3 | ##################################################### 4 | # pytest tests/test_math_static.py -s # 5 | ##################################################### 6 | import unittest 7 | 8 | from xautodl.datasets.math_core import QuadraticSFunc 9 | from xautodl.datasets.math_core import ConstantFunc 10 | 11 | 12 | class TestConstantFunc(unittest.TestCase): 13 | """Test the constant function.""" 14 | 15 | def test_simple(self): 16 | function = ConstantFunc(0.1) 17 | for i in range(100): 18 | assert function(i) == 0.1 19 | 20 | 21 | class TestQuadraticSFunc(unittest.TestCase): 22 | """Test the quadratic function.""" 23 | 24 | def test_simple(self): 25 | function = QuadraticSFunc({0: 1, 1: 2, 2: 1}) 26 | print(function) 27 | for x in (0, 0.5, 1): 28 | print("f({:})={:}".format(x, function(x))) 29 | thresh = 1e-7 30 | self.assertTrue(abs(function(0) - 1) < thresh) 31 | self.assertTrue(abs(function(0.5) - 0.5 * 0.5 - 2 * 0.5 - 1) < thresh) 32 | self.assertTrue(abs(function(1) - 1 - 2 - 1) < thresh) 33 | -------------------------------------------------------------------------------- /tests/test_super_rearrange.py: -------------------------------------------------------------------------------- 1 | ##################################################### 2 | # Copyright (c) Xuanyi Dong [GitHub D-X-Y], 2021.03 # 3 | ##################################################### 4 | # pytest ./tests/test_super_rearrange.py -s # 5 | ##################################################### 6 | import unittest 7 | 8 | import torch 9 | from xautodl import xlayers 10 | 11 | 12 | class TestSuperReArrange(unittest.TestCase): 13 | """Test the super re-arrange layer.""" 14 | 15 | def test_super_re_arrange(self): 16 | layer = xlayers.SuperReArrange( 17 | "b c (h p1) (w p2) -> b (h w) (c p1 p2)", p1=4, p2=4 18 | ) 19 | tensor = torch.rand((8, 4, 32, 32)) 20 | print("The tensor shape: {:}".format(tensor.shape)) 21 | print(layer) 22 | outs = layer(tensor) 23 | print("The output tensor shape: {:}".format(outs.shape)) 24 | assert tuple(outs.shape) == (8, 32 * 32 // 16, 4 * 4 * 4) 25 | -------------------------------------------------------------------------------- /tests/test_super_vit.py: -------------------------------------------------------------------------------- 1 | ##################################################### 2 | # Copyright (c) Xuanyi Dong [GitHub D-X-Y], 2021.03 # 3 | ##################################################### 4 | # pytest ./tests/test_super_vit.py -s # 5 | ##################################################### 6 | import unittest 7 | from parameterized import parameterized 8 | 9 | import torch 10 | from xautodl.xmodels import transformers 11 | from xautodl.utils.flop_benchmark import count_parameters 12 | 13 | 14 | class TestSuperViT(unittest.TestCase): 15 | """Test the super re-arrange layer.""" 16 | 17 | def test_super_vit(self): 18 | model = transformers.get_transformer("vit-base-16") 19 | tensor = torch.rand((2, 3, 224, 224)) 20 | print("The tensor shape: {:}".format(tensor.shape)) 21 | # print(model) 22 | outs = model(tensor) 23 | print("The output tensor shape: {:}".format(outs.shape)) 24 | 25 | @parameterized.expand( 26 | [ 27 | ["vit-cifar10-p4-d4-h4-c32", 32], 28 | ["vit-base-16", 224], 29 | ["vit-large-16", 224], 30 | ["vit-huge-14", 224], 31 | ] 32 | ) 33 | def test_imagenet(self, name, resolution): 34 | tensor = torch.rand((2, 3, resolution, resolution)) 35 | config = transformers.name2config[name] 36 | model = transformers.get_transformer(config) 37 | outs = model(tensor) 38 | size = count_parameters(model, "mb", True) 39 | print( 40 | "{:10s} : size={:.2f}MB, out-shape: {:}".format( 41 | name, size, tuple(outs.shape) 42 | ) 43 | ) 44 | -------------------------------------------------------------------------------- /tests/test_synthetic_env.py: -------------------------------------------------------------------------------- 1 | ##################################################### 2 | # Copyright (c) Xuanyi Dong [GitHub D-X-Y], 2021.04 # 3 | ##################################################### 4 | # pytest tests/test_synthetic_env.py -s # 5 | ##################################################### 6 | import unittest 7 | 8 | from xautodl.datasets.synthetic_core import get_synthetic_env 9 | 10 | 11 | class TestSynethicEnv(unittest.TestCase): 12 | """Test the synethtic environment.""" 13 | 14 | def test_simple(self): 15 | versions = ["v1", "v2", "v3", "v4"] 16 | for version in versions: 17 | env = get_synthetic_env(version=version) 18 | print(env) 19 | for timestamp, (x, y) in env: 20 | self.assertEqual(x.shape, (1000, env._data_generator.ndim)) 21 | -------------------------------------------------------------------------------- /tests/test_synthetic_utils.py: -------------------------------------------------------------------------------- 1 | ##################################################### 2 | # Copyright (c) Xuanyi Dong [GitHub D-X-Y], 2021.03 # 3 | ##################################################### 4 | # pytest tests/test_synthetic_utils.py -s # 5 | ##################################################### 6 | import unittest 7 | 8 | from xautodl.datasets.synthetic_core import TimeStamp 9 | 10 | 11 | class TestTimeStamp(unittest.TestCase): 12 | """Test the timestamp generator.""" 13 | 14 | def test_simple(self): 15 | for mode in (None, "train", "valid", "test"): 16 | generator = TimeStamp(0, 1) 17 | print(generator) 18 | for idx, (i, xtime) in enumerate(generator): 19 | self.assertTrue(i == idx) 20 | if idx == 0: 21 | self.assertTrue(xtime == 0) 22 | if idx + 1 == len(generator): 23 | self.assertTrue(abs(xtime - 1) < 1e-8) 24 | -------------------------------------------------------------------------------- /tests/test_tas.py: -------------------------------------------------------------------------------- 1 | ################################################## 2 | # Copyright (c) Xuanyi Dong [GitHub D-X-Y], 2019 # 3 | ################################################## 4 | import torch 5 | import torch.nn as nn 6 | import unittest 7 | 8 | from xautodl.models.shape_searchs.SoftSelect import ChannelWiseInter 9 | 10 | 11 | class TestTASFunc(unittest.TestCase): 12 | """Test the TAS function.""" 13 | 14 | def test_channel_interplation(self): 15 | tensors = torch.rand((16, 128, 7, 7)) 16 | 17 | for oc in range(200, 210): 18 | out_v1 = ChannelWiseInter(tensors, oc, "v1") 19 | out_v2 = ChannelWiseInter(tensors, oc, "v2") 20 | assert (out_v1 == out_v2).any().item() == 1 21 | for oc in range(48, 160): 22 | out_v1 = ChannelWiseInter(tensors, oc, "v1") 23 | out_v2 = ChannelWiseInter(tensors, oc, "v2") 24 | assert (out_v1 == out_v2).any().item() == 1 25 | -------------------------------------------------------------------------------- /tests/test_torch.sh: -------------------------------------------------------------------------------- 1 | # bash ./tests/test_torch.sh 2 | 3 | pytest ./tests/test_torch_gpu_bugs.py::test_create -s 4 | CUDA_VISIBLE_DEVICES="" pytest ./tests/test_torch_gpu_bugs.py::test_load -s 5 | -------------------------------------------------------------------------------- /tests/test_torch_gpu_bugs.py: -------------------------------------------------------------------------------- 1 | ##################################################### 2 | # Copyright (c) Xuanyi Dong [GitHub D-X-Y], 2021.03 # 3 | ##################################################### 4 | # pytest ./tests/test_torch_gpu_bugs.py::test_create 5 | # 6 | # CUDA_VISIBLE_DEVICES="" pytest ./tests/test_torch_gpu_bugs.py::test_load 7 | ##################################################### 8 | import os, sys, time, torch 9 | import pickle 10 | import tempfile 11 | from pathlib import Path 12 | 13 | root_dir = (Path(__file__).parent / ".." / "..").resolve() 14 | 15 | from xautodl.trade_models.quant_transformer import QuantTransformer 16 | 17 | 18 | def test_create(): 19 | """Test the basic quant-model.""" 20 | if not torch.cuda.is_available(): 21 | return 22 | quant_model = QuantTransformer(GPU=0) 23 | temp_dir = root_dir / "tests" / ".pytest_cache" 24 | temp_dir.mkdir(parents=True, exist_ok=True) 25 | temp_file = temp_dir / "quant-model.pkl" 26 | with temp_file.open("wb") as f: 27 | # quant_model.to(None) 28 | quant_model.to("cpu") 29 | # del quant_model.model 30 | # del quant_model.train_optimizer 31 | pickle.dump(quant_model, f) 32 | print("save into {:}".format(temp_file)) 33 | 34 | 35 | def test_load(): 36 | temp_file = root_dir / "tests" / ".pytest_cache" / "quant-model.pkl" 37 | with temp_file.open("rb") as f: 38 | model = pickle.load(f) 39 | print(model.model) 40 | print(model.train_optimizer) 41 | -------------------------------------------------------------------------------- /xautodl/__init__.py: -------------------------------------------------------------------------------- 1 | ##################################################### 2 | # Copyright (c) Xuanyi Dong [GitHub D-X-Y], 2021.05 # 3 | ##################################################### 4 | # An Automated Deep Learning Package to support # 5 | # research activities. # 6 | ##################################################### 7 | 8 | 9 | def version(): 10 | versions = ["0.9.9"] # 2021.06.01 11 | versions = ["1.0.0"] # 2021.08.14 12 | return versions[-1] 13 | -------------------------------------------------------------------------------- /xautodl/config_utils/__init__.py: -------------------------------------------------------------------------------- 1 | ################################################## 2 | # Copyright (c) Xuanyi Dong [GitHub D-X-Y], 2019 # 3 | ################################################## 4 | # general config related functions 5 | from .config_utils import load_config, dict2config, configure2str 6 | 7 | # the args setting for different experiments 8 | from .basic_args import obtain_basic_args 9 | from .attention_args import obtain_attention_args 10 | from .random_baseline import obtain_RandomSearch_args 11 | from .cls_kd_args import obtain_cls_kd_args 12 | from .cls_init_args import obtain_cls_init_args 13 | from .search_single_args import obtain_search_single_args 14 | from .search_args import obtain_search_args 15 | 16 | # for network pruning 17 | from .pruning_args import obtain_pruning_args 18 | 19 | # utils for args 20 | from .args_utils import arg_str2bool 21 | -------------------------------------------------------------------------------- /xautodl/config_utils/args_utils.py: -------------------------------------------------------------------------------- 1 | import argparse 2 | 3 | 4 | def arg_str2bool(v): 5 | if isinstance(v, bool): 6 | return v 7 | elif v.lower() in ("yes", "true", "t", "y", "1"): 8 | return True 9 | elif v.lower() in ("no", "false", "f", "n", "0"): 10 | return False 11 | else: 12 | raise argparse.ArgumentTypeError("Boolean value expected.") 13 | -------------------------------------------------------------------------------- /xautodl/config_utils/attention_args.py: -------------------------------------------------------------------------------- 1 | import random, argparse 2 | from .share_args import add_shared_args 3 | 4 | 5 | def obtain_attention_args(): 6 | parser = argparse.ArgumentParser( 7 | description="Train a classification model on typical image classification datasets.", 8 | formatter_class=argparse.ArgumentDefaultsHelpFormatter, 9 | ) 10 | parser.add_argument("--resume", type=str, help="Resume path.") 11 | parser.add_argument("--init_model", type=str, help="The initialization model path.") 12 | parser.add_argument( 13 | "--model_config", type=str, help="The path to the model configuration" 14 | ) 15 | parser.add_argument( 16 | "--optim_config", type=str, help="The path to the optimizer configuration" 17 | ) 18 | parser.add_argument("--procedure", type=str, help="The procedure basic prefix.") 19 | parser.add_argument("--att_channel", type=int, help=".") 20 | parser.add_argument("--att_spatial", type=str, help=".") 21 | parser.add_argument("--att_active", type=str, help=".") 22 | add_shared_args(parser) 23 | # Optimization options 24 | parser.add_argument( 25 | "--batch_size", type=int, default=2, help="Batch size for training." 26 | ) 27 | args = parser.parse_args() 28 | 29 | if args.rand_seed is None or args.rand_seed < 0: 30 | args.rand_seed = random.randint(1, 100000) 31 | assert args.save_dir is not None, "save-path argument can not be None" 32 | return args 33 | -------------------------------------------------------------------------------- /xautodl/config_utils/basic_args.py: -------------------------------------------------------------------------------- 1 | ################################################## 2 | # Copyright (c) Xuanyi Dong [GitHub D-X-Y], 2020 # 3 | ################################################## 4 | import random, argparse 5 | from .share_args import add_shared_args 6 | 7 | 8 | def obtain_basic_args(): 9 | parser = argparse.ArgumentParser( 10 | description="Train a classification model on typical image classification datasets.", 11 | formatter_class=argparse.ArgumentDefaultsHelpFormatter, 12 | ) 13 | parser.add_argument("--resume", type=str, help="Resume path.") 14 | parser.add_argument("--init_model", type=str, help="The initialization model path.") 15 | parser.add_argument( 16 | "--model_config", type=str, help="The path to the model configuration" 17 | ) 18 | parser.add_argument( 19 | "--optim_config", type=str, help="The path to the optimizer configuration" 20 | ) 21 | parser.add_argument("--procedure", type=str, help="The procedure basic prefix.") 22 | parser.add_argument( 23 | "--model_source", 24 | type=str, 25 | default="normal", 26 | help="The source of model defination.", 27 | ) 28 | parser.add_argument( 29 | "--extra_model_path", 30 | type=str, 31 | default=None, 32 | help="The extra model ckp file (help to indicate the searched architecture).", 33 | ) 34 | add_shared_args(parser) 35 | # Optimization options 36 | parser.add_argument( 37 | "--batch_size", type=int, default=2, help="Batch size for training." 38 | ) 39 | args = parser.parse_args() 40 | 41 | if args.rand_seed is None or args.rand_seed < 0: 42 | args.rand_seed = random.randint(1, 100000) 43 | assert args.save_dir is not None, "save-path argument can not be None" 44 | return args 45 | -------------------------------------------------------------------------------- /xautodl/config_utils/cls_init_args.py: -------------------------------------------------------------------------------- 1 | import random, argparse 2 | from .share_args import add_shared_args 3 | 4 | 5 | def obtain_cls_init_args(): 6 | parser = argparse.ArgumentParser( 7 | description="Train a classification model on typical image classification datasets.", 8 | formatter_class=argparse.ArgumentDefaultsHelpFormatter, 9 | ) 10 | parser.add_argument("--resume", type=str, help="Resume path.") 11 | parser.add_argument("--init_model", type=str, help="The initialization model path.") 12 | parser.add_argument( 13 | "--model_config", type=str, help="The path to the model configuration" 14 | ) 15 | parser.add_argument( 16 | "--optim_config", type=str, help="The path to the optimizer configuration" 17 | ) 18 | parser.add_argument("--procedure", type=str, help="The procedure basic prefix.") 19 | parser.add_argument( 20 | "--init_checkpoint", type=str, help="The checkpoint path to the initial model." 21 | ) 22 | add_shared_args(parser) 23 | # Optimization options 24 | parser.add_argument( 25 | "--batch_size", type=int, default=2, help="Batch size for training." 26 | ) 27 | args = parser.parse_args() 28 | 29 | if args.rand_seed is None or args.rand_seed < 0: 30 | args.rand_seed = random.randint(1, 100000) 31 | assert args.save_dir is not None, "save-path argument can not be None" 32 | return args 33 | -------------------------------------------------------------------------------- /xautodl/config_utils/cls_kd_args.py: -------------------------------------------------------------------------------- 1 | import random, argparse 2 | from .share_args import add_shared_args 3 | 4 | 5 | def obtain_cls_kd_args(): 6 | parser = argparse.ArgumentParser( 7 | description="Train a classification model on typical image classification datasets.", 8 | formatter_class=argparse.ArgumentDefaultsHelpFormatter, 9 | ) 10 | parser.add_argument("--resume", type=str, help="Resume path.") 11 | parser.add_argument("--init_model", type=str, help="The initialization model path.") 12 | parser.add_argument( 13 | "--model_config", type=str, help="The path to the model configuration" 14 | ) 15 | parser.add_argument( 16 | "--optim_config", type=str, help="The path to the optimizer configuration" 17 | ) 18 | parser.add_argument("--procedure", type=str, help="The procedure basic prefix.") 19 | parser.add_argument( 20 | "--KD_checkpoint", 21 | type=str, 22 | help="The teacher checkpoint in knowledge distillation.", 23 | ) 24 | parser.add_argument( 25 | "--KD_alpha", type=float, help="The alpha parameter in knowledge distillation." 26 | ) 27 | parser.add_argument( 28 | "--KD_temperature", 29 | type=float, 30 | help="The temperature parameter in knowledge distillation.", 31 | ) 32 | # parser.add_argument('--KD_feature', type=float, help='Knowledge distillation at the feature level.') 33 | add_shared_args(parser) 34 | # Optimization options 35 | parser.add_argument( 36 | "--batch_size", type=int, default=2, help="Batch size for training." 37 | ) 38 | args = parser.parse_args() 39 | 40 | if args.rand_seed is None or args.rand_seed < 0: 41 | args.rand_seed = random.randint(1, 100000) 42 | assert args.save_dir is not None, "save-path argument can not be None" 43 | return args 44 | -------------------------------------------------------------------------------- /xautodl/config_utils/pruning_args.py: -------------------------------------------------------------------------------- 1 | import os, sys, time, random, argparse 2 | from .share_args import add_shared_args 3 | 4 | 5 | def obtain_pruning_args(): 6 | parser = argparse.ArgumentParser( 7 | description="Train a classification model on typical image classification datasets.", 8 | formatter_class=argparse.ArgumentDefaultsHelpFormatter, 9 | ) 10 | parser.add_argument("--resume", type=str, help="Resume path.") 11 | parser.add_argument("--init_model", type=str, help="The initialization model path.") 12 | parser.add_argument( 13 | "--model_config", type=str, help="The path to the model configuration" 14 | ) 15 | parser.add_argument( 16 | "--optim_config", type=str, help="The path to the optimizer configuration" 17 | ) 18 | parser.add_argument("--procedure", type=str, help="The procedure basic prefix.") 19 | parser.add_argument( 20 | "--keep_ratio", 21 | type=float, 22 | help="The left channel ratio compared to the original network.", 23 | ) 24 | parser.add_argument("--model_version", type=str, help="The network version.") 25 | parser.add_argument( 26 | "--KD_alpha", type=float, help="The alpha parameter in knowledge distillation." 27 | ) 28 | parser.add_argument( 29 | "--KD_temperature", 30 | type=float, 31 | help="The temperature parameter in knowledge distillation.", 32 | ) 33 | parser.add_argument("--Regular_W_feat", type=float, help="The .") 34 | parser.add_argument("--Regular_W_conv", type=float, help="The .") 35 | add_shared_args(parser) 36 | # Optimization options 37 | parser.add_argument( 38 | "--batch_size", type=int, default=2, help="Batch size for training." 39 | ) 40 | args = parser.parse_args() 41 | 42 | if args.rand_seed is None or args.rand_seed < 0: 43 | args.rand_seed = random.randint(1, 100000) 44 | assert args.save_dir is not None, "save-path argument can not be None" 45 | assert ( 46 | args.keep_ratio > 0 and args.keep_ratio <= 1 47 | ), "invalid keep ratio : {:}".format(args.keep_ratio) 48 | return args 49 | -------------------------------------------------------------------------------- /xautodl/config_utils/random_baseline.py: -------------------------------------------------------------------------------- 1 | import os, sys, time, random, argparse 2 | from .share_args import add_shared_args 3 | 4 | 5 | def obtain_RandomSearch_args(): 6 | parser = argparse.ArgumentParser( 7 | description="Train a classification model on typical image classification datasets.", 8 | formatter_class=argparse.ArgumentDefaultsHelpFormatter, 9 | ) 10 | parser.add_argument("--resume", type=str, help="Resume path.") 11 | parser.add_argument("--init_model", type=str, help="The initialization model path.") 12 | parser.add_argument( 13 | "--expect_flop", type=float, help="The expected flop keep ratio." 14 | ) 15 | parser.add_argument( 16 | "--arch_nums", 17 | type=int, 18 | help="The maximum number of running random arch generating..", 19 | ) 20 | parser.add_argument( 21 | "--model_config", type=str, help="The path to the model configuration" 22 | ) 23 | parser.add_argument( 24 | "--optim_config", type=str, help="The path to the optimizer configuration" 25 | ) 26 | parser.add_argument( 27 | "--random_mode", 28 | type=str, 29 | choices=["random", "fix"], 30 | help="The path to the optimizer configuration", 31 | ) 32 | parser.add_argument("--procedure", type=str, help="The procedure basic prefix.") 33 | add_shared_args(parser) 34 | # Optimization options 35 | parser.add_argument( 36 | "--batch_size", type=int, default=2, help="Batch size for training." 37 | ) 38 | args = parser.parse_args() 39 | 40 | if args.rand_seed is None or args.rand_seed < 0: 41 | args.rand_seed = random.randint(1, 100000) 42 | assert args.save_dir is not None, "save-path argument can not be None" 43 | # assert args.flop_ratio_min < args.flop_ratio_max, 'flop-ratio {:} vs {:}'.format(args.flop_ratio_min, args.flop_ratio_max) 44 | return args 45 | -------------------------------------------------------------------------------- /xautodl/config_utils/search_single_args.py: -------------------------------------------------------------------------------- 1 | import os, sys, time, random, argparse 2 | from .share_args import add_shared_args 3 | 4 | 5 | def obtain_search_single_args(): 6 | parser = argparse.ArgumentParser( 7 | description="Train a classification model on typical image classification datasets.", 8 | formatter_class=argparse.ArgumentDefaultsHelpFormatter, 9 | ) 10 | parser.add_argument("--resume", type=str, help="Resume path.") 11 | parser.add_argument( 12 | "--model_config", type=str, help="The path to the model configuration" 13 | ) 14 | parser.add_argument( 15 | "--optim_config", type=str, help="The path to the optimizer configuration" 16 | ) 17 | parser.add_argument("--split_path", type=str, help="The split file path.") 18 | parser.add_argument("--search_shape", type=str, help="The shape to be searched.") 19 | # parser.add_argument('--arch_para_pure', type=int, help='The architecture-parameter pure or not.') 20 | parser.add_argument( 21 | "--gumbel_tau_max", type=float, help="The maximum tau for Gumbel." 22 | ) 23 | parser.add_argument( 24 | "--gumbel_tau_min", type=float, help="The minimum tau for Gumbel." 25 | ) 26 | parser.add_argument("--procedure", type=str, help="The procedure basic prefix.") 27 | parser.add_argument("--FLOP_ratio", type=float, help="The expected FLOP ratio.") 28 | parser.add_argument("--FLOP_weight", type=float, help="The loss weight for FLOP.") 29 | parser.add_argument( 30 | "--FLOP_tolerant", type=float, help="The tolerant range for FLOP." 31 | ) 32 | add_shared_args(parser) 33 | # Optimization options 34 | parser.add_argument( 35 | "--batch_size", type=int, default=2, help="Batch size for training." 36 | ) 37 | args = parser.parse_args() 38 | 39 | if args.rand_seed is None or args.rand_seed < 0: 40 | args.rand_seed = random.randint(1, 100000) 41 | assert args.save_dir is not None, "save-path argument can not be None" 42 | assert args.gumbel_tau_max is not None and args.gumbel_tau_min is not None 43 | assert ( 44 | args.FLOP_tolerant is not None and args.FLOP_tolerant > 0 45 | ), "invalid FLOP_tolerant : {:}".format(FLOP_tolerant) 46 | # assert args.arch_para_pure is not None, 'arch_para_pure is not None: {:}'.format(args.arch_para_pure) 47 | # args.arch_para_pure = bool(args.arch_para_pure) 48 | return args 49 | -------------------------------------------------------------------------------- /xautodl/config_utils/share_args.py: -------------------------------------------------------------------------------- 1 | import os, sys, time, random, argparse 2 | 3 | 4 | def add_shared_args(parser): 5 | # Data Generation 6 | parser.add_argument("--dataset", type=str, help="The dataset name.") 7 | parser.add_argument("--data_path", type=str, help="The dataset name.") 8 | parser.add_argument( 9 | "--cutout_length", type=int, help="The cutout length, negative means not use." 10 | ) 11 | # Printing 12 | parser.add_argument( 13 | "--print_freq", type=int, default=100, help="print frequency (default: 200)" 14 | ) 15 | parser.add_argument( 16 | "--print_freq_eval", 17 | type=int, 18 | default=100, 19 | help="print frequency (default: 200)", 20 | ) 21 | # Checkpoints 22 | parser.add_argument( 23 | "--eval_frequency", 24 | type=int, 25 | default=1, 26 | help="evaluation frequency (default: 200)", 27 | ) 28 | parser.add_argument( 29 | "--save_dir", type=str, help="Folder to save checkpoints and log." 30 | ) 31 | # Acceleration 32 | parser.add_argument( 33 | "--workers", 34 | type=int, 35 | default=8, 36 | help="number of data loading workers (default: 8)", 37 | ) 38 | # Random Seed 39 | parser.add_argument("--rand_seed", type=int, default=-1, help="manual seed") 40 | -------------------------------------------------------------------------------- /xautodl/datasets/__init__.py: -------------------------------------------------------------------------------- 1 | ################################################## 2 | # Copyright (c) Xuanyi Dong [GitHub D-X-Y], 2019 # 3 | ################################################## 4 | from .get_dataset_with_transform import get_datasets, get_nas_search_loaders 5 | from .SearchDatasetWrap import SearchDataset 6 | -------------------------------------------------------------------------------- /xautodl/datasets/landmark_utils/__init__.py: -------------------------------------------------------------------------------- 1 | from .point_meta import PointMeta2V, apply_affine2point, apply_boundary 2 | -------------------------------------------------------------------------------- /xautodl/datasets/math_base_funcs.py: -------------------------------------------------------------------------------- 1 | ##################################################### 2 | # Copyright (c) Xuanyi Dong [GitHub D-X-Y], 2021.03 # 3 | ##################################################### 4 | import math 5 | import abc 6 | import copy 7 | import numpy as np 8 | 9 | 10 | class MathFunc(abc.ABC): 11 | """The math function -- a virtual class defining some APIs.""" 12 | 13 | def __init__(self, freedom: int, params=None, xstr="x"): 14 | # initialize as empty 15 | self._params = dict() 16 | for i in range(freedom): 17 | self._params[i] = None 18 | self._freedom = freedom 19 | if params is not None: 20 | self.set(params) 21 | self._xstr = str(xstr) 22 | self._skip_check = True 23 | 24 | def set(self, params): 25 | for key in range(self._freedom): 26 | param = copy.deepcopy(params[key]) 27 | self._params[key] = param 28 | 29 | def check_valid(self): 30 | if not self._skip_check: 31 | for key in range(self._freedom): 32 | value = self._params[key] 33 | if value is None: 34 | raise ValueError("The {:} is None".format(key)) 35 | 36 | @property 37 | def xstr(self): 38 | return self._xstr 39 | 40 | def reset_xstr(self, xstr): 41 | self._xstr = str(xstr) 42 | 43 | def output_shape(self, input_shape): 44 | return input_shape 45 | 46 | @abc.abstractmethod 47 | def __call__(self, x): 48 | raise NotImplementedError 49 | 50 | @abc.abstractmethod 51 | def noise_call(self, x, std): 52 | clean_y = self.__call__(x) 53 | if isinstance(clean_y, np.ndarray): 54 | noise_y = clean_y + np.random.normal(scale=std, size=clean_y.shape) 55 | else: 56 | raise ValueError("Unkonwn type: {:}".format(type(clean_y))) 57 | return noise_y 58 | 59 | def __repr__(self): 60 | return "{name}(freedom={freedom})".format( 61 | name=self.__class__.__name__, freedom=freedom 62 | ) 63 | -------------------------------------------------------------------------------- /xautodl/datasets/math_core.py: -------------------------------------------------------------------------------- 1 | ##################################################### 2 | # Copyright (c) Xuanyi Dong [GitHub D-X-Y], 2021.05 # 3 | ##################################################### 4 | from .math_static_funcs import ( 5 | LinearSFunc, 6 | QuadraticSFunc, 7 | CubicSFunc, 8 | QuarticSFunc, 9 | ConstantFunc, 10 | ComposedSinSFunc, 11 | ComposedCosSFunc, 12 | ) 13 | from .math_dynamic_funcs import ( 14 | LinearDFunc, 15 | QuadraticDFunc, 16 | SinQuadraticDFunc, 17 | BinaryQuadraticDFunc, 18 | ) 19 | from .math_dynamic_generator import UniformDGenerator, GaussianDGenerator 20 | -------------------------------------------------------------------------------- /xautodl/datasets/test_utils.py: -------------------------------------------------------------------------------- 1 | ################################################## 2 | # Copyright (c) Xuanyi Dong [GitHub D-X-Y], 2019 # 3 | ################################################## 4 | import os 5 | 6 | 7 | def test_imagenet_data(imagenet): 8 | total_length = len(imagenet) 9 | assert ( 10 | total_length == 1281166 or total_length == 50000 11 | ), "The length of ImageNet is wrong : {}".format(total_length) 12 | map_id = {} 13 | for index in range(total_length): 14 | path, target = imagenet.imgs[index] 15 | folder, image_name = os.path.split(path) 16 | _, folder = os.path.split(folder) 17 | if folder not in map_id: 18 | map_id[folder] = target 19 | else: 20 | assert map_id[folder] == target, "Class : {} is not {}".format( 21 | folder, target 22 | ) 23 | assert image_name.find(folder) == 0, "{} is wrong.".format(path) 24 | print("Check ImageNet Dataset OK") 25 | -------------------------------------------------------------------------------- /xautodl/log_utils/__init__.py: -------------------------------------------------------------------------------- 1 | ################################################## 2 | # Copyright (c) Xuanyi Dong [GitHub D-X-Y], 2019 # 3 | ################################################## 4 | # every package does not rely on pytorch or tensorflow 5 | # I tried to list all dependency here: os, sys, time, numpy, (possibly) matplotlib 6 | ################################################## 7 | from .logger import Logger, PrintLogger 8 | from .meter import AverageMeter 9 | from .time_utils import ( 10 | time_for_file, 11 | time_string, 12 | time_string_short, 13 | time_print, 14 | convert_secs2time, 15 | ) 16 | from .pickle_wrap import pickle_save, pickle_load 17 | -------------------------------------------------------------------------------- /xautodl/log_utils/pickle_wrap.py: -------------------------------------------------------------------------------- 1 | ##################################################### 2 | # Copyright (c) Xuanyi Dong [GitHub D-X-Y], 2019.01 # 3 | ##################################################### 4 | import pickle 5 | from pathlib import Path 6 | 7 | 8 | def pickle_save(obj, path): 9 | file_path = Path(path) 10 | file_dir = file_path.parent 11 | file_dir.mkdir(parents=True, exist_ok=True) 12 | with file_path.open("wb") as f: 13 | pickle.dump(obj, f) 14 | 15 | 16 | def pickle_load(path): 17 | if not Path(path).exists(): 18 | raise ValueError("{:} does not exists".format(path)) 19 | with Path(path).open("rb") as f: 20 | data = pickle.load(f) 21 | return data 22 | -------------------------------------------------------------------------------- /xautodl/log_utils/time_utils.py: -------------------------------------------------------------------------------- 1 | ##################################################### 2 | # Copyright (c) Xuanyi Dong [GitHub D-X-Y], 2019.01 # 3 | ##################################################### 4 | import time, sys 5 | import numpy as np 6 | 7 | 8 | def time_for_file(): 9 | ISOTIMEFORMAT = "%d-%h-at-%H-%M-%S" 10 | return "{:}".format(time.strftime(ISOTIMEFORMAT, time.gmtime(time.time()))) 11 | 12 | 13 | def time_string(): 14 | ISOTIMEFORMAT = "%Y-%m-%d %X" 15 | string = "[{:}]".format(time.strftime(ISOTIMEFORMAT, time.gmtime(time.time()))) 16 | return string 17 | 18 | 19 | def time_string_short(): 20 | ISOTIMEFORMAT = "%Y%m%d" 21 | string = "{:}".format(time.strftime(ISOTIMEFORMAT, time.gmtime(time.time()))) 22 | return string 23 | 24 | 25 | def time_print(string, is_print=True): 26 | if is_print: 27 | print("{} : {}".format(time_string(), string)) 28 | 29 | 30 | def convert_secs2time(epoch_time, return_str=False): 31 | need_hour = int(epoch_time / 3600) 32 | need_mins = int((epoch_time - 3600 * need_hour) / 60) 33 | need_secs = int(epoch_time - 3600 * need_hour - 60 * need_mins) 34 | if return_str: 35 | str = "[{:02d}:{:02d}:{:02d}]".format(need_hour, need_mins, need_secs) 36 | return str 37 | else: 38 | return need_hour, need_mins, need_secs 39 | 40 | 41 | def print_log(print_string, log): 42 | # if isinstance(log, Logger): log.log('{:}'.format(print_string)) 43 | if hasattr(log, "log"): 44 | log.log("{:}".format(print_string)) 45 | else: 46 | print("{:}".format(print_string)) 47 | if log is not None: 48 | log.write("{:}\n".format(print_string)) 49 | log.flush() 50 | -------------------------------------------------------------------------------- /xautodl/models/SharedUtils.py: -------------------------------------------------------------------------------- 1 | ##################################################### 2 | # Copyright (c) Xuanyi Dong [GitHub D-X-Y], 2019.01 # 3 | ##################################################### 4 | import torch 5 | import torch.nn as nn 6 | 7 | 8 | def additive_func(A, B): 9 | assert A.dim() == B.dim() and A.size(0) == B.size(0), "{:} vs {:}".format( 10 | A.size(), B.size() 11 | ) 12 | C = min(A.size(1), B.size(1)) 13 | if A.size(1) == B.size(1): 14 | return A + B 15 | elif A.size(1) < B.size(1): 16 | out = B.clone() 17 | out[:, :C] += A 18 | return out 19 | else: 20 | out = A.clone() 21 | out[:, :C] += B 22 | return out 23 | 24 | 25 | def change_key(key, value): 26 | def func(m): 27 | if hasattr(m, key): 28 | setattr(m, key, value) 29 | 30 | return func 31 | 32 | 33 | def parse_channel_info(xstring): 34 | blocks = xstring.split(" ") 35 | blocks = [x.split("-") for x in blocks] 36 | blocks = [[int(_) for _ in x] for x in blocks] 37 | return blocks 38 | -------------------------------------------------------------------------------- /xautodl/models/cell_infers/__init__.py: -------------------------------------------------------------------------------- 1 | ##################################################### 2 | # Copyright (c) Xuanyi Dong [GitHub D-X-Y], 2019.01 # 3 | ##################################################### 4 | from .tiny_network import TinyNetwork 5 | from .nasnet_cifar import NASNetonCIFAR 6 | -------------------------------------------------------------------------------- /xautodl/models/cell_searchs/__init__.py: -------------------------------------------------------------------------------- 1 | ################################################## 2 | # Copyright (c) Xuanyi Dong [GitHub D-X-Y], 2019 # 3 | ################################################## 4 | # The macro structure is defined in NAS-Bench-201 5 | from .search_model_darts import TinyNetworkDarts 6 | from .search_model_gdas import TinyNetworkGDAS 7 | from .search_model_setn import TinyNetworkSETN 8 | from .search_model_enas import TinyNetworkENAS 9 | from .search_model_random import TinyNetworkRANDOM 10 | from .generic_model import GenericNAS201Model 11 | from .genotypes import Structure as CellStructure, architectures as CellArchitectures 12 | 13 | # NASNet-based macro structure 14 | from .search_model_gdas_nasnet import NASNetworkGDAS 15 | from .search_model_gdas_frc_nasnet import NASNetworkGDAS_FRC 16 | from .search_model_darts_nasnet import NASNetworkDARTS 17 | 18 | 19 | nas201_super_nets = { 20 | "DARTS-V1": TinyNetworkDarts, 21 | "DARTS-V2": TinyNetworkDarts, 22 | "GDAS": TinyNetworkGDAS, 23 | "SETN": TinyNetworkSETN, 24 | "ENAS": TinyNetworkENAS, 25 | "RANDOM": TinyNetworkRANDOM, 26 | "generic": GenericNAS201Model, 27 | } 28 | 29 | nasnet_super_nets = { 30 | "GDAS": NASNetworkGDAS, 31 | "GDAS_FRC": NASNetworkGDAS_FRC, 32 | "DARTS": NASNetworkDARTS, 33 | } 34 | -------------------------------------------------------------------------------- /xautodl/models/cell_searchs/_test_module.py: -------------------------------------------------------------------------------- 1 | ################################################## 2 | # Copyright (c) Xuanyi Dong [GitHub D-X-Y], 2019 # 3 | ################################################## 4 | import torch 5 | from search_model_enas_utils import Controller 6 | 7 | 8 | def main(): 9 | controller = Controller(6, 4) 10 | predictions = controller() 11 | 12 | 13 | if __name__ == "__main__": 14 | main() 15 | -------------------------------------------------------------------------------- /xautodl/models/initialization.py: -------------------------------------------------------------------------------- 1 | import torch 2 | import torch.nn as nn 3 | 4 | 5 | def initialize_resnet(m): 6 | if isinstance(m, nn.Conv2d): 7 | nn.init.kaiming_normal_(m.weight, mode="fan_out", nonlinearity="relu") 8 | if m.bias is not None: 9 | nn.init.constant_(m.bias, 0) 10 | elif isinstance(m, nn.BatchNorm2d): 11 | nn.init.constant_(m.weight, 1) 12 | if m.bias is not None: 13 | nn.init.constant_(m.bias, 0) 14 | elif isinstance(m, nn.Linear): 15 | nn.init.normal_(m.weight, 0, 0.01) 16 | nn.init.constant_(m.bias, 0) 17 | -------------------------------------------------------------------------------- /xautodl/models/shape_infers/__init__.py: -------------------------------------------------------------------------------- 1 | ##################################################### 2 | # Copyright (c) Xuanyi Dong [GitHub D-X-Y], 2019.01 # 3 | ##################################################### 4 | from .InferCifarResNet_width import InferWidthCifarResNet 5 | from .InferImagenetResNet import InferImagenetResNet 6 | from .InferCifarResNet_depth import InferDepthCifarResNet 7 | from .InferCifarResNet import InferCifarResNet 8 | from .InferMobileNetV2 import InferMobileNetV2 9 | from .InferTinyCellNet import DynamicShapeTinyNet 10 | -------------------------------------------------------------------------------- /xautodl/models/shape_infers/shared_utils.py: -------------------------------------------------------------------------------- 1 | def parse_channel_info(xstring): 2 | blocks = xstring.split(" ") 3 | blocks = [x.split("-") for x in blocks] 4 | blocks = [[int(_) for _ in x] for x in blocks] 5 | return blocks 6 | -------------------------------------------------------------------------------- /xautodl/models/shape_searchs/__init__.py: -------------------------------------------------------------------------------- 1 | ################################################## 2 | # Copyright (c) Xuanyi Dong [GitHub D-X-Y], 2019 # 3 | ################################################## 4 | from .SearchCifarResNet_width import SearchWidthCifarResNet 5 | from .SearchCifarResNet_depth import SearchDepthCifarResNet 6 | from .SearchCifarResNet import SearchShapeCifarResNet 7 | from .SearchSimResNet_width import SearchWidthSimResNet 8 | from .SearchImagenetResNet import SearchShapeImagenetResNet 9 | from .generic_size_tiny_cell_model import GenericNAS301Model 10 | -------------------------------------------------------------------------------- /xautodl/nas_infer_model/DXYs/__init__.py: -------------------------------------------------------------------------------- 1 | # Performance-Aware Template Network for One-Shot Neural Architecture Search 2 | from .CifarNet import NetworkCIFAR as CifarNet 3 | from .ImageNet import NetworkImageNet as ImageNet 4 | from .genotypes import Networks 5 | from .genotypes import build_genotype_from_dict 6 | -------------------------------------------------------------------------------- /xautodl/nas_infer_model/__init__.py: -------------------------------------------------------------------------------- 1 | ##################################################### 2 | # Copyright (c) Xuanyi Dong [GitHub D-X-Y], 2019.01 # 3 | ##################################################### 4 | # I write this package to make AutoDL-Projects to be compatible with the old GDAS projects. 5 | # Ideally, this package will be merged into lib/models/cell_infers in future. 6 | # Currently, this package is used to reproduce the results in GDAS (Searching for A Robust Neural Architecture in Four GPU Hours, CVPR 2019). 7 | ################################################## 8 | 9 | import os, torch 10 | 11 | 12 | def obtain_nas_infer_model(config, extra_model_path=None): 13 | 14 | if config.arch == "dxys": 15 | from .DXYs import CifarNet, ImageNet, Networks 16 | from .DXYs import build_genotype_from_dict 17 | 18 | if config.genotype is None: 19 | if extra_model_path is not None and not os.path.isfile(extra_model_path): 20 | raise ValueError( 21 | "When genotype in confiig is None, extra_model_path must be set as a path instead of {:}".format( 22 | extra_model_path 23 | ) 24 | ) 25 | xdata = torch.load(extra_model_path) 26 | current_epoch = xdata["epoch"] 27 | genotype_dict = xdata["genotypes"][current_epoch - 1] 28 | genotype = build_genotype_from_dict(genotype_dict) 29 | else: 30 | genotype = Networks[config.genotype] 31 | if config.dataset == "cifar": 32 | return CifarNet( 33 | config.ichannel, 34 | config.layers, 35 | config.stem_multi, 36 | config.auxiliary, 37 | genotype, 38 | config.class_num, 39 | ) 40 | elif config.dataset == "imagenet": 41 | return ImageNet( 42 | config.ichannel, 43 | config.layers, 44 | config.auxiliary, 45 | genotype, 46 | config.class_num, 47 | ) 48 | else: 49 | raise ValueError("invalid dataset : {:}".format(config.dataset)) 50 | else: 51 | raise ValueError("invalid nas arch type : {:}".format(config.arch)) 52 | -------------------------------------------------------------------------------- /xautodl/procedures/__init__.py: -------------------------------------------------------------------------------- 1 | ################################################## 2 | # Copyright (c) Xuanyi Dong [GitHub D-X-Y], 2019 # 3 | ###################################################################### 4 | # This folder is deprecated, which is re-organized in "xalgorithms". # 5 | ###################################################################### 6 | from .starts import prepare_seed 7 | from .starts import prepare_logger 8 | from .starts import get_machine_info 9 | from .starts import save_checkpoint 10 | from .starts import copy_checkpoint 11 | from .optimizers import get_optim_scheduler 12 | from .funcs_nasbench import evaluate_for_seed as bench_evaluate_for_seed 13 | from .funcs_nasbench import pure_evaluate as bench_pure_evaluate 14 | from .funcs_nasbench import get_nas_bench_loaders 15 | 16 | 17 | def get_procedures(procedure): 18 | from .basic_main import basic_train, basic_valid 19 | from .search_main import search_train, search_valid 20 | from .search_main_v2 import search_train_v2 21 | from .simple_KD_main import simple_KD_train, simple_KD_valid 22 | 23 | train_funcs = { 24 | "basic": basic_train, 25 | "search": search_train, 26 | "Simple-KD": simple_KD_train, 27 | "search-v2": search_train_v2, 28 | } 29 | valid_funcs = { 30 | "basic": basic_valid, 31 | "search": search_valid, 32 | "Simple-KD": simple_KD_valid, 33 | "search-v2": search_valid, 34 | } 35 | 36 | train_func = train_funcs[procedure] 37 | valid_func = valid_funcs[procedure] 38 | return train_func, valid_func 39 | -------------------------------------------------------------------------------- /xautodl/procedures/eval_funcs.py: -------------------------------------------------------------------------------- 1 | ##################################################### 2 | # Copyright (c) Xuanyi Dong [GitHub D-X-Y], 2020.04 # 3 | ##################################################### 4 | import abc 5 | 6 | 7 | def obtain_accuracy(output, target, topk=(1,)): 8 | """Computes the precision@k for the specified values of k""" 9 | maxk = max(topk) 10 | batch_size = target.size(0) 11 | 12 | _, pred = output.topk(maxk, 1, True, True) 13 | pred = pred.t() 14 | correct = pred.eq(target.view(1, -1).expand_as(pred)) 15 | 16 | res = [] 17 | for k in topk: 18 | correct_k = correct[:k].contiguous().view(-1).float().sum(0, keepdim=True) 19 | res.append(correct_k.mul_(100.0 / batch_size)) 20 | return res 21 | -------------------------------------------------------------------------------- /xautodl/spaces/__init__.py: -------------------------------------------------------------------------------- 1 | ##################################################### 2 | # Copyright (c) Xuanyi Dong [GitHub D-X-Y], 2021.01 # 3 | ##################################################### 4 | # Define complex searc space for AutoDL # 5 | ##################################################### 6 | 7 | from .basic_space import Categorical 8 | from .basic_space import Continuous 9 | from .basic_space import Integer 10 | from .basic_space import Space 11 | from .basic_space import VirtualNode 12 | from .basic_op import has_categorical 13 | from .basic_op import has_continuous 14 | from .basic_op import is_determined 15 | from .basic_op import get_determined_value 16 | from .basic_op import get_min 17 | from .basic_op import get_max 18 | -------------------------------------------------------------------------------- /xautodl/trade_models/__init__.py: -------------------------------------------------------------------------------- 1 | ##################################################### 2 | # Copyright (c) Xuanyi Dong [GitHub D-X-Y], 2021.03 # 3 | ##################################################### 4 | from .transformers import get_transformer 5 | -------------------------------------------------------------------------------- /xautodl/utils/__init__.py: -------------------------------------------------------------------------------- 1 | ##################################################### 2 | # Copyright (c) Xuanyi Dong [GitHub D-X-Y], 2019.01 # 3 | ##################################################### 4 | # This directory contains some ad-hoc functions, classes, etc. 5 | # It will be re-formulated in the future. 6 | ##################################################### 7 | from .evaluation_utils import obtain_accuracy 8 | from .gpu_manager import GPUManager 9 | from .flop_benchmark import get_model_infos, count_parameters, count_parameters_in_MB 10 | from .affine_utils import normalize_points, denormalize_points 11 | from .affine_utils import identity2affine, solve2theta, affine2image 12 | from .hash_utils import get_md5_file 13 | from .str_utils import split_str2indexes 14 | from .str_utils import show_mean_var 15 | -------------------------------------------------------------------------------- /xautodl/utils/evaluation_utils.py: -------------------------------------------------------------------------------- 1 | import torch 2 | 3 | 4 | def obtain_accuracy(output, target, topk=(1,)): 5 | """Computes the precision@k for the specified values of k""" 6 | maxk = max(topk) 7 | batch_size = target.size(0) 8 | 9 | _, pred = output.topk(maxk, 1, True, True) 10 | pred = pred.t() 11 | correct = pred.eq(target.view(1, -1).expand_as(pred)) 12 | 13 | res = [] 14 | for k in topk: 15 | correct_k = correct[:k].view(-1).float().sum(0, keepdim=True) 16 | res.append(correct_k.mul_(100.0 / batch_size)) 17 | return res 18 | -------------------------------------------------------------------------------- /xautodl/utils/hash_utils.py: -------------------------------------------------------------------------------- 1 | import os 2 | import hashlib 3 | 4 | 5 | def get_md5_file(file_path, post_truncated=5): 6 | md5_hash = hashlib.md5() 7 | if os.path.exists(file_path): 8 | xfile = open(file_path, "rb") 9 | content = xfile.read() 10 | md5_hash.update(content) 11 | digest = md5_hash.hexdigest() 12 | else: 13 | raise ValueError("[get_md5_file] {:} does not exist".format(file_path)) 14 | if post_truncated is None: 15 | return digest 16 | else: 17 | return digest[-post_truncated:] 18 | -------------------------------------------------------------------------------- /xautodl/utils/str_utils.py: -------------------------------------------------------------------------------- 1 | import numpy as np 2 | 3 | 4 | def split_str2indexes(string: str, max_check: int, length_limit=5): 5 | if not isinstance(string, str): 6 | raise ValueError("Invalid scheme for {:}".format(string)) 7 | srangestr = "".join(string.split()) 8 | indexes = set() 9 | for srange in srangestr.split(","): 10 | srange = srange.split("-") 11 | if len(srange) != 2: 12 | raise ValueError("invalid srange : {:}".format(srange)) 13 | if length_limit is not None: 14 | assert ( 15 | len(srange[0]) == len(srange[1]) == length_limit 16 | ), "invalid srange : {:}".format(srange) 17 | srange = (int(srange[0]), int(srange[1])) 18 | if not (0 <= srange[0] <= srange[1] < max_check): 19 | raise ValueError( 20 | "{:} vs {:} vs {:}".format(srange[0], srange[1], max_check) 21 | ) 22 | for i in range(srange[0], srange[1] + 1): 23 | indexes.add(i) 24 | return indexes 25 | 26 | 27 | def show_mean_var(xlist): 28 | values = np.array(xlist) 29 | print( 30 | "{:.2f}".format(values.mean()) 31 | + "$_{{\pm}{" 32 | + "{:.2f}".format(values.std()) 33 | + "}}$" 34 | ) 35 | -------------------------------------------------------------------------------- /xautodl/utils/temp_sync.py: -------------------------------------------------------------------------------- 1 | # To be deleted. 2 | import copy 3 | import torch 4 | 5 | from xlayers.super_core import SuperSequential, SuperMLPv1 6 | from xlayers.super_core import SuperSimpleNorm 7 | from xlayers.super_core import SuperLinear 8 | 9 | 10 | def optimize_fn(xs, ys, device="cpu", max_iter=2000, max_lr=0.1): 11 | xs = torch.FloatTensor(xs).view(-1, 1).to(device) 12 | ys = torch.FloatTensor(ys).view(-1, 1).to(device) 13 | 14 | model = SuperSequential( 15 | SuperSimpleNorm(xs.mean().item(), xs.std().item()), 16 | SuperLinear(1, 200), 17 | torch.nn.LeakyReLU(), 18 | SuperLinear(200, 100), 19 | torch.nn.LeakyReLU(), 20 | SuperLinear(100, 1), 21 | ).to(device) 22 | model.train() 23 | optimizer = torch.optim.Adam(model.parameters(), lr=max_lr, amsgrad=True) 24 | loss_func = torch.nn.MSELoss() 25 | lr_scheduler = torch.optim.lr_scheduler.MultiStepLR( 26 | optimizer, 27 | milestones=[ 28 | int(max_iter * 0.25), 29 | int(max_iter * 0.5), 30 | int(max_iter * 0.75), 31 | ], 32 | gamma=0.3, 33 | ) 34 | 35 | best_loss, best_param = None, None 36 | for _iter in range(max_iter): 37 | preds = model(xs) 38 | 39 | optimizer.zero_grad() 40 | loss = loss_func(preds, ys) 41 | loss.backward() 42 | optimizer.step() 43 | lr_scheduler.step() 44 | 45 | if best_loss is None or best_loss > loss.item(): 46 | best_loss = loss.item() 47 | best_param = copy.deepcopy(model.state_dict()) 48 | 49 | # print('loss={:}, best-loss={:}'.format(loss.item(), best_loss)) 50 | model.load_state_dict(best_param) 51 | return model, loss_func, best_loss 52 | 53 | 54 | def evaluate_fn(model, xs, ys, loss_fn, device="cpu"): 55 | with torch.no_grad(): 56 | inputs = torch.FloatTensor(xs).view(-1, 1).to(device) 57 | ys = torch.FloatTensor(ys).view(-1, 1).to(device) 58 | preds = model(inputs) 59 | loss = loss_fn(preds, ys) 60 | preds = preds.view(-1).cpu().numpy() 61 | return preds, loss.item() 62 | -------------------------------------------------------------------------------- /xautodl/xlayers/__init__.py: -------------------------------------------------------------------------------- 1 | ##################################################### 2 | # Copyright (c) Xuanyi Dong [GitHub D-X-Y], 2019.01 # 3 | ##################################################### 4 | # This file is expected to be self-contained, expect 5 | # for importing from spaces to include search space. 6 | ##################################################### 7 | from .super_core import * 8 | -------------------------------------------------------------------------------- /xautodl/xlayers/super_core.py: -------------------------------------------------------------------------------- 1 | ##################################################### 2 | # Copyright (c) Xuanyi Dong [GitHub D-X-Y], 2021.03 # 3 | ##################################################### 4 | from .super_module import SuperRunMode 5 | from .super_module import IntSpaceType 6 | from .super_module import LayerOrder 7 | 8 | from .super_module import SuperModule 9 | from .super_container import SuperSequential 10 | from .super_linear import SuperLinear 11 | from .super_linear import SuperMLPv1, SuperMLPv2 12 | 13 | from .super_norm import SuperSimpleNorm 14 | from .super_norm import SuperLayerNorm1D 15 | from .super_norm import SuperSimpleLearnableNorm 16 | from .super_norm import SuperIdentity 17 | from .super_dropout import SuperDropout 18 | from .super_dropout import SuperDrop 19 | 20 | super_name2norm = { 21 | "simple_norm": SuperSimpleNorm, 22 | "simple_learn_norm": SuperSimpleLearnableNorm, 23 | "layer_norm_1d": SuperLayerNorm1D, 24 | "identity": SuperIdentity, 25 | } 26 | 27 | from .super_attention import SuperSelfAttention 28 | from .super_attention import SuperQKVAttention 29 | from .super_attention_v2 import SuperQKVAttentionV2 30 | from .super_transformer import SuperTransformerEncoderLayer 31 | 32 | from .super_activations import SuperReLU 33 | from .super_activations import SuperLeakyReLU 34 | from .super_activations import SuperTanh 35 | from .super_activations import SuperGELU 36 | from .super_activations import SuperSigmoid 37 | 38 | super_name2activation = { 39 | "relu": SuperReLU, 40 | "sigmoid": SuperSigmoid, 41 | "gelu": SuperGELU, 42 | "leaky_relu": SuperLeakyReLU, 43 | "tanh": SuperTanh, 44 | } 45 | 46 | 47 | from .super_trade_stem import SuperAlphaEBDv1 48 | from .super_positional_embedding import SuperDynamicPositionE 49 | from .super_positional_embedding import SuperPositionalEncoder 50 | 51 | from .super_rearrange import SuperReArrange 52 | -------------------------------------------------------------------------------- /xautodl/xmisc/__init__.py: -------------------------------------------------------------------------------- 1 | ##################################################### 2 | # Copyright (c) Xuanyi Dong [GitHub D-X-Y], 2021.06 # 3 | ##################################################### 4 | """The module and yaml related functions.""" 5 | from .module_utils import call_by_dict 6 | from .module_utils import call_by_yaml 7 | from .module_utils import nested_call_by_dict 8 | from .module_utils import nested_call_by_yaml 9 | from .yaml_utils import load_yaml 10 | 11 | from .torch_utils import count_parameters 12 | 13 | from .logger_utils import Logger 14 | 15 | """The data sampler related classes.""" 16 | from .sampler_utils import BatchSampler 17 | 18 | """The meter related classes.""" 19 | from .meter_utils import AverageMeter 20 | 21 | """The scheduler related classes.""" 22 | from .scheduler_utils import CosineParamScheduler, WarmupParamScheduler, LRMultiplier 23 | 24 | 25 | def get_scheduler(indicator, lr): 26 | if indicator == "warm-cos": 27 | multiplier = WarmupParamScheduler( 28 | CosineParamScheduler(lr, lr * 1e-3), 29 | warmup_factor=0.001, 30 | warmup_length=0.05, 31 | warmup_method="linear", 32 | ) 33 | 34 | else: 35 | raise ValueError("Unknown indicator: {:}".format(indicator)) 36 | return multiplier 37 | -------------------------------------------------------------------------------- /xautodl/xmisc/logger_utils.py: -------------------------------------------------------------------------------- 1 | ##################################################### 2 | # Copyright (c) Xuanyi Dong [GitHub D-X-Y], 2021.06 # 3 | ##################################################### 4 | import sys 5 | from pathlib import Path 6 | 7 | from .time_utils import time_for_file, time_string 8 | 9 | 10 | class Logger: 11 | """A logger used in xautodl.""" 12 | 13 | def __init__(self, root_dir, prefix="", log_time=True): 14 | """Create a summary writer logging to log_dir.""" 15 | self.root_dir = Path(root_dir) 16 | self.log_dir = self.root_dir / "logs" 17 | self.log_dir.mkdir(parents=True, exist_ok=True) 18 | 19 | self._prefix = prefix 20 | self._log_time = log_time 21 | self.logger_path = self.log_dir / "{:}{:}.log".format( 22 | self._prefix, time_for_file() 23 | ) 24 | self._logger_file = open(self.logger_path, "w") 25 | 26 | @property 27 | def logger(self): 28 | return self._logger_file 29 | 30 | def log(self, string, save=True, stdout=False): 31 | string = "{:} {:}".format(time_string(), string) if self._log_time else string 32 | if stdout: 33 | sys.stdout.write(string) 34 | sys.stdout.flush() 35 | else: 36 | print(string) 37 | if save: 38 | self._logger_file.write("{:}\n".format(string)) 39 | self._logger_file.flush() 40 | 41 | def close(self): 42 | self._logger_file.close() 43 | if self.writer is not None: 44 | self.writer.close() 45 | 46 | def __repr__(self): 47 | return "{name}(dir={log_dir}, prefix={_prefix}, log_time={_log_time})".format( 48 | name=self.__class__.__name__, **self.__dict__ 49 | ) 50 | -------------------------------------------------------------------------------- /xautodl/xmisc/sampler_utils.py: -------------------------------------------------------------------------------- 1 | ##################################################### 2 | # Copyright (c) Xuanyi Dong [GitHub D-X-Y], 2021.06 # 3 | ##################################################### 4 | import random 5 | 6 | 7 | class BatchSampler: 8 | """A batch sampler used for single machine training.""" 9 | 10 | def __init__(self, dataset, batch, steps): 11 | self._num_per_epoch = len(dataset) 12 | self._iter_per_epoch = self._num_per_epoch // batch 13 | self._steps = steps 14 | self._batch = batch 15 | if self._num_per_epoch < self._batch: 16 | raise ValueError( 17 | "The dataset size must be larger than batch={:}".format(batch) 18 | ) 19 | self._indexes = list(range(self._num_per_epoch)) 20 | 21 | def __iter__(self): 22 | """ 23 | yield a batch of indexes using random sampling 24 | """ 25 | for i in range(self._steps): 26 | if i % self._iter_per_epoch == 0: 27 | random.shuffle(self._indexes) 28 | j = i % self._iter_per_epoch 29 | yield self._indexes[j * self._batch : (j + 1) * self._batch] 30 | 31 | def __len__(self): 32 | return self._steps 33 | -------------------------------------------------------------------------------- /xautodl/xmisc/time_utils.py: -------------------------------------------------------------------------------- 1 | ##################################################### 2 | # Copyright (c) Xuanyi Dong [GitHub D-X-Y], 2021.06 # 3 | ##################################################### 4 | import time 5 | 6 | 7 | def time_for_file(): 8 | ISOTIMEFORMAT = "%d-%h-at-%H-%M-%S" 9 | return "{:}".format(time.strftime(ISOTIMEFORMAT, time.gmtime(time.time()))) 10 | 11 | 12 | def time_string(): 13 | ISOTIMEFORMAT = "%Y-%m-%d %X" 14 | string = "[{:}]".format(time.strftime(ISOTIMEFORMAT, time.gmtime(time.time()))) 15 | return string 16 | 17 | 18 | def convert_secs2time(epoch_time, return_str=False): 19 | need_hour = int(epoch_time / 3600) 20 | need_mins = int((epoch_time - 3600 * need_hour) / 60) 21 | need_secs = int(epoch_time - 3600 * need_hour - 60 * need_mins) 22 | if return_str: 23 | str = "[{:02d}:{:02d}:{:02d}]".format(need_hour, need_mins, need_secs) 24 | return str 25 | else: 26 | return need_hour, need_mins, need_secs 27 | -------------------------------------------------------------------------------- /xautodl/xmisc/torch_utils.py: -------------------------------------------------------------------------------- 1 | ##################################################### 2 | # Copyright (c) Xuanyi Dong [GitHub D-X-Y], 2021.06 # 3 | ##################################################### 4 | import torch 5 | import torch.nn as nn 6 | import numpy as np 7 | 8 | 9 | def count_parameters(model_or_parameters, unit="mb"): 10 | if isinstance(model_or_parameters, nn.Module): 11 | counts = sum(np.prod(v.size()) for v in model_or_parameters.parameters()) 12 | elif isinstance(model_or_parameters, nn.Parameter): 13 | counts = models_or_parameters.numel() 14 | elif isinstance(model_or_parameters, (list, tuple)): 15 | counts = sum(count_parameters(x, None) for x in models_or_parameters) 16 | else: 17 | counts = sum(np.prod(v.size()) for v in model_or_parameters) 18 | if unit.lower() == "kb" or unit.lower() == "k": 19 | counts /= 1e3 20 | elif unit.lower() == "mb" or unit.lower() == "m": 21 | counts /= 1e6 22 | elif unit.lower() == "gb" or unit.lower() == "g": 23 | counts /= 1e9 24 | elif unit is not None: 25 | raise ValueError("Unknow unit: {:}".format(unit)) 26 | return counts 27 | -------------------------------------------------------------------------------- /xautodl/xmisc/yaml_utils.py: -------------------------------------------------------------------------------- 1 | ##################################################### 2 | # Copyright (c) Xuanyi Dong [GitHub D-X-Y], 2021.06 # 3 | ##################################################### 4 | import os 5 | import yaml 6 | 7 | 8 | def load_yaml(path): 9 | if not os.path.isfile(path): 10 | raise ValueError("{:} is not a file.".format(path)) 11 | with open(path, "r") as stream: 12 | data = yaml.safe_load(stream) 13 | return data 14 | -------------------------------------------------------------------------------- /xautodl/xmodels/__init__.py: -------------------------------------------------------------------------------- 1 | ##################################################### 2 | # Copyright (c) Xuanyi Dong [GitHub D-X-Y], 2021.06 # 3 | ##################################################### 4 | # The models in this folder is written with xlayers # 5 | ##################################################### 6 | 7 | from .core import * 8 | --------------------------------------------------------------------------------