Enable `fixed_arch` on Retiarii (#3972)

docs/en_US/NAS/ApiReference.rst

@@ -105,4 +105,6 @@ Retiarii Experiments
 Utilities
 ---------
 
-..  autofunction:: nni.retiarii.serialize
\ No newline at end of file
+..  autofunction:: nni.retiarii.serialize
+
+..  autofunction:: nni.retiarii.fixed_arch

docs/en_US/NAS/OneshotTrainer.rst

@@ -34,4 +34,10 @@ See `API reference <./ApiReference.rst>`__ for detailed usages. Here, we show an
   trainer.fit()
   final_architecture = trainer.export()
 
-**Format of the exported architecture.** TBD.
+After the searching is done, we can use the exported architecture to instantiate the full network for retraining. Here is an example:
+
+.. code-block:: python
+
+    from nni.retiarii import fixed_arch
+    with fixed_arch('/path/to/checkpoint.json'):
+        model = Model()

docs/en_US/NAS/WriteOneshot.rst

@@ -16,7 +16,7 @@ A typical example is DartsTrainer, where learnable-parameters are used to combin
     class DartsLayerChoice(nn.Module):
         def __init__(self, layer_choice):
             super(DartsLayerChoice, self).__init__()
-            self.name = layer_choice.key
+            self.name = layer_choice.label
             self.op_choices = nn.ModuleDict(layer_choice.named_children())
             self.alpha = nn.Parameter(torch.randn(len(self.op_choices)) * 1e-3)
 

examples/nas/oneshot/darts/model.py

@@ -7,7 +7,7 @@ import torch
 import torch.nn as nn
 
 import ops
-from nni.nas.pytorch import mutables
+from nni.retiarii.nn.pytorch import LayerChoice, InputChoice
 
 
 class AuxiliaryHead(nn.Module):
@@ -45,7 +45,7 @@ class Node(nn.Module):
             stride = 2 if i < num_downsample_connect else 1
             choice_keys.append("{}_p{}".format(node_id, i))
             self.ops.append(
-                mutables.LayerChoice(OrderedDict([
+                LayerChoice(OrderedDict([
                     ("maxpool", ops.PoolBN('max', channels, 3, stride, 1, affine=False)),
                     ("avgpool", ops.PoolBN('avg', channels, 3, stride, 1, affine=False)),
                     ("skipconnect", nn.Identity() if stride == 1 else ops.FactorizedReduce(channels, channels, affine=False)),
@@ -53,9 +53,9 @@ class Node(nn.Module):
                     ("sepconv5x5", ops.SepConv(channels, channels, 5, stride, 2, affine=False)),
                     ("dilconv3x3", ops.DilConv(channels, channels, 3, stride, 2, 2, affine=False)),
                     ("dilconv5x5", ops.DilConv(channels, channels, 5, stride, 4, 2, affine=False))
-                ]), key=choice_keys[-1]))
+                ]), label=choice_keys[-1]))
         self.drop_path = ops.DropPath()
-        self.input_switch = mutables.InputChoice(choose_from=choice_keys, n_chosen=2, key="{}_switch".format(node_id))
+        self.input_switch = InputChoice(n_candidates=len(choice_keys), n_chosen=2, label="{}_switch".format(node_id))
 
     def forward(self, prev_nodes):
         assert len(self.ops) == len(prev_nodes)

examples/nas/oneshot/darts/retrain.py

@@ -12,8 +12,8 @@ from torch.utils.tensorboard import SummaryWriter
 import datasets
 import utils
 from model import CNN
-from nni.nas.pytorch.fixed import apply_fixed_architecture
 from nni.nas.pytorch.utils import AverageMeter
+from nni.retiarii import fixed_arch
 
 logger = logging.getLogger('nni')
 
@@ -119,8 +119,8 @@ if __name__ == "__main__":
     args = parser.parse_args()
     dataset_train, dataset_valid = datasets.get_dataset("cifar10", cutout_length=16)
 
-    model = CNN(32, 3, 36, 10, args.layers, auxiliary=True)
-    apply_fixed_architecture(model, args.arc_checkpoint)
+    with fixed_arch(args.arc_checkpoint):
+        model = CNN(32, 3, 36, 10, args.layers, auxiliary=True)
     criterion = nn.CrossEntropyLoss()
 
     model.to(device)

nni/retiarii/__init__.py

@@ -4,5 +4,6 @@
 from .operation import Operation
 from .graph import *
 from .execution import *
+from .fixed import fixed_arch
 from .mutator import *
 from .serializer import basic_unit, json_dump, json_dumps, json_load, json_loads, serialize, serialize_cls, model_wrapper

nni/retiarii/fixed.py

+import json
+import logging
+from pathlib import Path
+from typing import Union, Dict, Any
+
+from .utils import ContextStack
+
+_logger = logging.getLogger(__name__)
+
+
+def fixed_arch(fixed_arch: Union[str, Path, Dict[str, Any]], verbose=True):
+    """
+    Load architecture from ``fixed_arch`` and apply to model. This should be used as a context manager. For example,
+
+    .. code-block:: python
+
+        with fixed_arch('/path/to/export.json'):
+            model = Model(3, 224, 224)
+
+    Parameters
+    ----------
+    fixed_arc : str, Path or dict
+        Path to the JSON that stores the architecture, or dict that stores the exported architecture.
+    verbose : bool
+        Print log messages if set to True
+
+    Returns
+    -------
+    ContextStack
+        Context manager that provides a fixed architecture when creates the model.
+    """
+
+    if isinstance(fixed_arch, (str, Path)):
+        with open(fixed_arch) as f:
+            fixed_arch = json.load(f)
+
+    if verbose:
+        _logger.info(f'Fixed architecture: %s', fixed_arch)
+
+    return ContextStack('fixed', fixed_arch)

nni/retiarii/oneshot/pytorch/darts.py

@@ -3,6 +3,7 @@
 
 import copy
 import logging
+from collections import OrderedDict
 
 import torch
 import torch.nn as nn
@@ -18,8 +19,8 @@ _logger = logging.getLogger(__name__)
 class DartsLayerChoice(nn.Module):
     def __init__(self, layer_choice):
         super(DartsLayerChoice, self).__init__()
-        self.name = layer_choice.key
-        self.op_choices = nn.ModuleDict(layer_choice.named_children())
+        self.name = layer_choice.label
+        self.op_choices = nn.ModuleDict(OrderedDict([(name, layer_choice[name]) for name in layer_choice.names]))
         self.alpha = nn.Parameter(torch.randn(len(self.op_choices)) * 1e-3)
 
     def forward(self, *args, **kwargs):
@@ -38,13 +39,13 @@ class DartsLayerChoice(nn.Module):
             yield name, p
 
     def export(self):
-        return torch.argmax(self.alpha).item()
+        return list(self.op_choices.keys())[torch.argmax(self.alpha).item()]
 
 
 class DartsInputChoice(nn.Module):
     def __init__(self, input_choice):
         super(DartsInputChoice, self).__init__()
-        self.name = input_choice.key
+        self.name = input_choice.label
         self.alpha = nn.Parameter(torch.randn(input_choice.n_candidates) * 1e-3)
         self.n_chosen = input_choice.n_chosen or 1