Merge master into dev-retiarii (#3178)

docs/tools/restoremd.py

+import os
+import shutil
+from pathlib import Path
+
+
+for root, dirs, files in os.walk('archive_en_US'):
+    root = Path(root)
+    for file in files:
+        moved_root = Path('en_US') / root.relative_to('archive_en_US')
+        shutil.move(root / file, moved_root / file)
+        os.remove(moved_root / (Path(file).stem + '.rst'))

examples/model_compress/BNN_quantizer_cifar10.py

@@ -5,7 +5,7 @@ import torch
 import torch.nn as nn
 import torch.nn.functional as F
 from torchvision import datasets, transforms
-from nni.compression.torch import BNNQuantizer
+from nni.algorithms.compression.pytorch.quantization import BNNQuantizer
 
 
 class VGG_Cifar10(nn.Module):

examples/model_compress/DoReFaQuantizer_torch_mnist.py

 import torch
 import torch.nn.functional as F
 from torchvision import datasets, transforms
-from nni.compression.torch import DoReFaQuantizer
+from nni.algorithms.compression.pytorch.quantization import DoReFaQuantizer
 
 
 class Mnist(torch.nn.Module):
@@ -86,4 +86,4 @@ def main():
 
 
 if __name__ == '__main__':
-    main()
\ No newline at end of file
+    main()

examples/model_compress/L1_torch_cifar10.py

@@ -3,7 +3,7 @@ import torch
 import torch.nn as nn
 import torch.nn.functional as F
 from torchvision import datasets, transforms
-from nni.compression.torch import L1FilterPruner
+from nni.algorithms.compression.pytorch.pruning import L1FilterPruner
 from models.cifar10.vgg import VGG
 
 

examples/model_compress/QAT_torch_quantizer.py

 import torch
 import torch.nn.functional as F
 from torchvision import datasets, transforms
-from nni.compression.torch import QAT_Quantizer
+from nni.algorithms.compression.pytorch.quantization import QAT_Quantizer
 
 
 class Mnist(torch.nn.Module):

examples/model_compress/amc/amc_search.py

@@ -8,7 +8,7 @@ import time
 import torch
 import torch.nn as nn
 from torchvision.models import resnet
-from nni.compression.torch import AMCPruner
+from nni.algorithms.compression.pytorch.pruning import AMCPruner
 from data import get_split_dataset
 from utils import AverageMeter, accuracy
 

examples/model_compress/amc/amc_train.py

@@ -15,9 +15,9 @@ import torch.optim as optim
 from tensorboardX import SummaryWriter
 from torchvision.models import resnet
 
-from nni.compression.torch.pruning.amc.lib.net_measure import measure_model
-from nni.compression.torch.pruning.amc.lib.utils import get_output_folder
-from nni.compression.torch import ModelSpeedup
+from nni.algorithms.compression.pytorch.pruning.amc.lib.net_measure import measure_model
+from nni.algorithms.compression.pytorch.pruning.amc.lib.utils import get_output_folder
+from nni.compression.pytorch import ModelSpeedup
 
 from data import get_dataset
 from utils import AverageMeter, accuracy, progress_bar

examples/model_compress/auto_pruners_torch.py

@@ -14,10 +14,10 @@ from torchvision import datasets, transforms
 from models.mnist.lenet import LeNet
 from models.cifar10.vgg import VGG
 from models.cifar10.resnet import ResNet18, ResNet50
-from nni.compression.torch import L1FilterPruner, L2FilterPruner, FPGMPruner
-from nni.compression.torch import SimulatedAnnealingPruner, ADMMPruner, NetAdaptPruner, AutoCompressPruner
-from nni.compression.torch import ModelSpeedup
-from nni.compression.torch.utils.counter import count_flops_params
+from nni.algorithms.compression.pytorch.pruning import L1FilterPruner, L2FilterPruner, FPGMPruner
+from nni.algorithms.compression.pytorch.pruning import SimulatedAnnealingPruner, ADMMPruner, NetAdaptPruner, AutoCompressPruner
+from nni.compression.pytorch import ModelSpeedup
+from nni.compression.pytorch.utils.counter import count_flops_params
 
 
 def get_data(dataset, data_dir, batch_size, test_batch_size):

examples/model_compress/lottery_torch_mnist_fc.py

@@ -6,7 +6,7 @@ import torch.nn.functional as F
 import torch.utils.data
 import torchvision.datasets as datasets
 import torchvision.transforms as transforms
-from nni.compression.torch import LotteryTicketPruner
+from nni.algorithms.compression.pytorch.pruning import LotteryTicketPruner
 
 class fc1(nn.Module):
 

examples/model_compress/model_prune_torch.py

@@ -9,8 +9,16 @@ from torchvision import datasets, transforms
 
 from models.cifar10.vgg import VGG
 import nni
-from nni.compression.torch import LevelPruner, SlimPruner, FPGMPruner, L1FilterPruner, \
-    L2FilterPruner, AGPPruner, ActivationMeanRankFilterPruner, ActivationAPoZRankFilterPruner
+from nni.algorithms.compression.pytorch.pruning import (
+    LevelPruner,
+    SlimPruner,
+    FPGMPruner,
+    L1FilterPruner,
+    L2FilterPruner,
+    AGPPruner,
+    ActivationMeanRankFilterPruner,
+    ActivationAPoZRankFilterPruner
+)
 
 prune_config = {
     'level': {

examples/model_compress/model_speedup.py

@@ -6,7 +6,7 @@ import torch.nn as nn
 import torch.nn.functional as F
 from torchvision import datasets, transforms
 from models.cifar10.vgg import VGG
-from nni.compression.torch import apply_compression_results, ModelSpeedup
+from nni.compression.pytorch import apply_compression_results, ModelSpeedup
 
 torch.manual_seed(0)
 use_mask = True

examples/model_compress/pruning_kd.py

@@ -3,7 +3,7 @@ import torch
 import torch.nn as nn
 import torch.nn.functional as F
 from torchvision import datasets, transforms
-from nni.compression.torch import L1FilterPruner
+from nni.algorithms.compression.pytorch.pruning import L1FilterPruner
 from knowledge_distill.knowledge_distill import KnowledgeDistill
 from models.cifar10.vgg import VGG
 

examples/model_compress/slim_torch_cifar10.py

@@ -5,7 +5,7 @@ import torch
 import torch.nn as nn
 import torch.nn.functional as F
 from torchvision import datasets, transforms
-from nni.compression.torch import SlimPruner
+from nni.algorithms.compression.pytorch.pruning import SlimPruner
 from models.cifar10.vgg import VGG
 
 def updateBN(model):

examples/nas/classic_nas/mnist.py

@@ -18,7 +18,7 @@ import torch.optim as optim
 from torchvision import datasets, transforms
 
 from nni.nas.pytorch.mutables import LayerChoice, InputChoice
-from nni.nas.pytorch.classic_nas import get_and_apply_next_architecture
+from nni.algorithms.nas.pytorch.classic_nas import get_and_apply_next_architecture
 
 
 logger = logging.getLogger('mnist_AutoML')

examples/trials/cifar10_pytorch/config_adl.yml

@@ -17,10 +17,13 @@ tuner:
     #choice: maximize, minimize
     optimize_mode: maximize
 trial:
-  command: python3 main_adl.py
-  codeDir: .
+  command: python3 /cifar10/main_adl.py
+  codeDir: /cifar10
   gpuNum: 1
   image: {replace_with_the_image_that_has_adaptdl_installed}
+  # optional
+  imagePullSecrets:
+    - name: {secret}
   adaptive: true
   checkpoint:
     storageClass: dfs

examples/trials/cifar10_pytorch/main_adl.py

@@ -146,7 +146,7 @@ def valid(epoch):
         writer.add_scalar("Accuracy/Valid", stats["accuracy"], epoch)
 
         if adaptdl.env.replica_rank() == 0:
-            nni.report_intermediate_result(stats["accuracy"], accum=stats)
+            nni.report_intermediate_result(stats["accuracy"])
 
         print("Valid:", stats)
         return stats["accuracy"]

examples/trials/mnist-tfv2/launch.py

@@ -22,5 +22,6 @@ experiment.config.max_trial_number = 5
 experiment.config.search_space = search_space
 experiment.config.trial_command = 'python3 mnist.py'
 experiment.config.trial_code_directory = Path(__file__).parent
+experiment.config.training_service.use_active_gpu = True
 
-experiment.run(8081, debug=True)
+experiment.run(8081)

nni/algorithms/compression/pytorch/pruning/__init__.py

@@ -3,7 +3,6 @@
 
 from .finegrained_pruning import *
 from .structured_pruning import *
-from .apply_compression import apply_compression_results
 from .one_shot import *
 from .agp import *
 from .lottery_ticket import LotteryTicketPruner

nni/algorithms/compression/pytorch/pruning/structured_pruning.py

@@ -278,7 +278,8 @@ class StructuredWeightMasker(WeightMasker):
                 sparsity, _w, _w_idx)
             num_total = current_weight.size(0)
             if num_total < 2 or num_prune < 1:
-                return base_mask
+                masks[name] = base_mask
+                continue
             _tmp_mask = self.get_mask(
                 base_mask, current_weight, num_prune, _w, _w_idx, channel_masks)
 

nni/compression/pytorch/__init__.py

@@ -3,3 +3,4 @@
 
 from .speedup import ModelSpeedup
 from .compressor import Compressor, Pruner, Quantizer
+from .pruning import apply_compression_results