Fix CGO import error (#4010)

dependencies/required.txt

@@ -13,6 +13,7 @@ websockets
 filelock
 prettytable
 dataclasses ; python_version < "3.7"
+typing_extensions ; python_version < "3.8"
 numpy < 1.19.4 ; sys_platform == "win32"
 numpy < 1.20 ; sys_platform != "win32" and python_version < "3.7"
 numpy ; sys.platform != "win32" and python_version >= "3.7"

examples/nas/oneshot/spos/multi_trial.py

@@ -13,7 +13,7 @@ from torchvision.datasets import CIFAR10
 
 from blocks import ShuffleNetBlock, ShuffleXceptionBlock
 
-from nn_meter import load_latency_predictors
+from nn_meter import load_latency_predictor
 
 
 class ShuffleNetV2(nn.Module):
@@ -142,7 +142,7 @@ class LatencyFilter:
             if reverse is `False`, then the model returns `True` when `latency < threshold`,
             else otherwisse
         """
-        self.predictors = load_latency_predictors(predictor, predictor_version)
+        self.predictors = load_latency_predictor(predictor, predictor_version)
         self.threshold = threshold
 
     def __call__(self, ir_model):

nni/common/device.py

+# Copyright (c) Microsoft Corporation.
+# Licensed under the MIT license.
 
 from dataclasses import dataclass
-from typing import Literal
+try:
+    from typing import Literal
+except ImportError:
+    from typing_extensions import Literal
+
 
 @dataclass
 class GPUDevice:

nni/retiarii/evaluator/pytorch/lightning.py

@@ -13,7 +13,7 @@ from torch.utils.data import DataLoader
 
 import nni
 try:
-    import nni.retiarii.evaluator.pytorch.cgo.trainer as cgo_trainer
+    from .cgo import trainer as cgo_trainer
     cgo_import_failed = False
 except ImportError:
     cgo_import_failed = True