Fix pytest collection warnings (openvinotoolkit#1097)

tests/torch/pruning/filter_pruning/test_algo.py

@@ -34,7 +34,7 @@
 from tests.torch.pruning.helpers import get_basic_pruning_config
 from tests.torch.pruning.helpers import PruningTestModel
 from tests.torch.pruning.helpers import BigPruningTestModel
-from tests.torch.pruning.helpers import TestModelMultipleForward
+from tests.torch.pruning.helpers import MultipleForwardModel
 from tests.torch.pruning.helpers import PruningTestModelConcatBN
 from tests.torch.pruning.helpers import DisconectedGraphModel
 
@@ -415,7 +415,7 @@ def test_clusters_for_multiple_forward(repeat_seq_of_shared_convs,
     config['compression']['params']['all_weights'] = False
     config['compression']['params']['prune_first_conv'] = True
     config['compression']['pruning_init'] = 0.5
-    model = TestModelMultipleForward(repeat_seq_of_shared_convs, additional_last_shared_layers)
+    model = MultipleForwardModel(repeat_seq_of_shared_convs, additional_last_shared_layers)
     _, pruning_algo = create_compressed_model_and_algo_for_test(model, config)
 
     clusters = pruning_algo.pruned_module_groups_info.clusters

tests/torch/pruning/filter_pruning/test_layers.py

@@ -22,7 +22,7 @@
 from tests.torch.helpers import fill_conv_weight, fill_bias
 
 
-class TestFilterPruningBlockModel(nn.Module):
+class FilterPruningBlockModel(nn.Module):
     def __init__(self, layer):
         super().__init__()
         self.layer = layer
@@ -55,7 +55,7 @@ def test_can_infer_magnitude_pruned_conv(weights_val, bias_val):
     nncf_module = NNCFConv2d(1, 1, 2)
     pytorch_module = nn.Conv2d(1, 1, 2)
 
-    sparse_model = TestFilterPruningBlockModel(nncf_module)
+    sparse_model = FilterPruningBlockModel(nncf_module)
 
     fill_conv_weight(nncf_module, weights_val)
     fill_bias(nncf_module, bias_val)

tests/torch/pruning/helpers.py

@@ -43,7 +43,7 @@ def forward(self, x):
         return x
 
 
-class TestModelDiffConvs(nn.Module):
+class DiffConvsModel(nn.Module):
     def __init__(self):
         super().__init__()
         # Usual conv
@@ -67,7 +67,7 @@ def forward(self, x):
         return x
 
 
-class TestModelBranching(nn.Module):
+class BranchingModel(nn.Module):
     def __init__(self):
         super().__init__()
         self.conv1 = create_conv(1, 3, 2, 1, -2)
@@ -84,7 +84,7 @@ def forward(self, x):
         return x
 
 
-class TestModelResidualConnection(nn.Module):
+class ResidualConnectionModel(nn.Module):
     def __init__(self):
         super().__init__()
         self.conv1 = create_conv(1, 8, 3, 1, -2, padding=1)
@@ -105,7 +105,7 @@ def forward(self, x):
         return x
 
 
-class TestModelEltwiseCombination(nn.Module):
+class EltwiseCombinationModel(nn.Module):
     def __init__(self):
         super().__init__()
         self.conv1 = create_conv(1, 8, 3, 1, -2, padding=1)
@@ -290,7 +290,7 @@ def forward(self, x):
         return x
 
 
-class TestModelShuffleNetUnit(nn.Module):
+class ShuffleNetUnitModel(nn.Module):
     def __init__(self):
         super().__init__()
         self.conv = create_conv(1, 16, 1, 1, -2)
@@ -302,7 +302,7 @@ def forward(self, x):
         return x
 
 
-class TestModelShuffleNetUnitDW(nn.Module):
+class ShuffleNetUnitModelDW(nn.Module):
     def __init__(self):
         super().__init__()
         self.conv = create_conv(1, 16, 1, 1, -2)
@@ -314,7 +314,7 @@ def forward(self, x):
         return x
 
 
-class TestModelMultipleForward(nn.Module):
+class MultipleForwardModel(nn.Module):
     def __init__(self, repeat_seq_of_shared_convs=False, additional_last_shared_layers=False):
         super().__init__()
         self.num_iter_shared_convs = 2 if repeat_seq_of_shared_convs else 1
@@ -351,7 +351,7 @@ def forward(self, x):
         return x1, x2, x3
 
 
-class TestModelGroupNorm(nn.Module):
+class GroupNormModel(nn.Module):
     def __init__(self):
         super().__init__()
         self.conv1 = create_conv(1, 16, 1, 1, -2)

tests/torch/pruning/test_onnx_export.py

@@ -13,7 +13,7 @@
 import pytest
 
 from tests.torch.pruning.helpers import BigPruningTestModel, get_basic_pruning_config, \
-    PruningTestModelConcat, PruningTestModelEltwise, TestModelDiffConvs, TestModelGroupNorm
+    PruningTestModelConcat, PruningTestModelEltwise, DiffConvsModel, GroupNormModel
 from tests.torch.helpers import load_exported_onnx_version
 
 pytestmark = pytest.mark.skip(reason="Export as actually deleting filters from the model is currently disabled.")
@@ -99,7 +99,7 @@ def test_pruning_export_eltwise_model(tmp_path, prune_first, ref_shapes):
                           (True, [[[16, 1, 2, 2], [16]], [[16, 1, 1, 1], [16]], [[32, 16, 3, 3], [32]],
                                   [[16, 4, 1, 1], [16]]])])
 def test_pruning_export_diffconvs_model(tmp_path, prune_first, ref_shapes):
-    model = TestModelDiffConvs()
+    model = DiffConvsModel()
     nncf_config = get_basic_pruning_config(input_sample_size=[1, 1, 8, 8])
     nncf_config['compression']['algorithm'] = 'filter_pruning'
 
@@ -113,7 +113,7 @@ def test_pruning_export_diffconvs_model(tmp_path, prune_first, ref_shapes):
 
 
 def test_pruning_export_groupnorm_model(tmp_path):
-    model = TestModelGroupNorm()
+    model = GroupNormModel()
     nncf_config = get_basic_pruning_config(input_sample_size=[1, 1, 8, 8])
     nncf_config['compression']['algorithm'] = 'filter_pruning'
 

tests/torch/pruning/test_utils.py

@@ -18,7 +18,7 @@
 from nncf.common.pruning.utils import get_last_nodes_of_type
 from nncf.torch.pruning.utils import get_bn_for_conv_node_by_name
 from tests.torch.pruning.helpers import get_basic_pruning_config, BigPruningTestModel, \
-    TestModelBranching
+    BranchingModel
 from tests.torch.helpers import create_compressed_model_and_algo_for_test
 
 
@@ -65,7 +65,7 @@ def test_get_bn_for_conv_node():
 
 @pytest.mark.parametrize(('model', 'ref_first_module_names'),
                          [(BigPruningTestModel, ['conv1']),
-                          (TestModelBranching, ['conv1', 'conv2', 'conv3']),
+                          (BranchingModel, ['conv1', 'conv2', 'conv3']),
                           ],
                          )
 def test_get_first_pruned_layers(model, ref_first_module_names):
@@ -83,7 +83,7 @@ def test_get_first_pruned_layers(model, ref_first_module_names):
 
 @pytest.mark.parametrize(('model', 'ref_last_module_names'),
                          [(BigPruningTestModel, ['conv3']),
-                          (TestModelBranching, ['conv4', 'conv5']
+                          (BranchingModel, ['conv4', 'conv5']
                            ),
                           ],
                          )

tests/torch/pytest.ini

@@ -0,0 +1,6 @@
+[pytest]
+markers = 
+    oveval
+    eval
+    convert
+    train

tests/torch/quantization/test_manual_precision_init.py

@@ -181,7 +181,7 @@ def test_manual_single_conv(params):
         check_bitwidth_graph(ctrl, model, path_to_dot, graph_dir)
 
 
-class TestPrecisionInitDesc:
+class PrecisionInitTestDesc:
     def __init__(self):
         self.model_creator = AddTwoConv
         config = get_quantization_config_without_range_init()
@@ -251,7 +251,7 @@ def check_precision_init(self, compression_ctrl: QuantizationController):
 
 
 def test_quantization_configs__with_precisions_list():
-    desc = TestPrecisionInitDesc()
+    desc = PrecisionInitTestDesc()
     model = desc.model_creator()
     config = desc.config
     register_bn_adaptation_init_args(config)

tests/torch/quantization/test_quantization_metric.py

@@ -55,12 +55,12 @@ def as_dict(obj):
     return obj
 
 
-TestStruct = namedtuple(
-    'TestStruct', ('initializers', 'activations', 'weights', 'ignored_scopes', 'target_device', 'expected'))
+CaseStruct = namedtuple(
+    'CaseStruct', ('initializers', 'activations', 'weights', 'ignored_scopes', 'target_device', 'expected'))
 
 
 QUANTIZATION_SHARE_AND_BITWIDTH_DISTR_STATS_TEST_CASES = [
-    TestStruct(
+    CaseStruct(
         initializers={},
         activations={},
         weights={},
@@ -95,7 +95,7 @@ def as_dict(obj):
             }
         }
     ),
-    TestStruct(
+    CaseStruct(
         initializers={},
         activations={},
         weights={},
@@ -152,7 +152,7 @@ def test_quantization_share_and_bitwidth_distribution_stats(data):
 
 
 MEMORY_CONSUMPTION_STATS_TEST_CASES = [
-    TestStruct(
+    CaseStruct(
         initializers={},
         activations={},
         weights={},
@@ -166,7 +166,7 @@ def test_quantization_share_and_bitwidth_distribution_stats(data):
             'weight_memory_consumption_decrease': 4.0
         }
     ),
-    TestStruct(
+    CaseStruct(
         initializers={
             'precision': {
                 'bitwidth_per_scope': [
@@ -189,7 +189,7 @@ def test_quantization_share_and_bitwidth_distribution_stats(data):
             'weight_memory_consumption_decrease': 4.05
         }
     ),
-    TestStruct(
+    CaseStruct(
         initializers={},
         activations={},
         weights={},
@@ -227,7 +227,7 @@ def test_memory_consumption_stats(data):
 
 
 QUANTIZATION_CONFIGURATION_STATS_TEST_CASES = [
-    TestStruct(
+    CaseStruct(
         initializers={},
         activations={},
         weights={},
@@ -238,7 +238,7 @@ def test_memory_consumption_stats(data):
             'total_edges_in_cfg': 177
         }
     ),
-    TestStruct(
+    CaseStruct(
         initializers={},
         activations={},
         weights={},

tests/torch/sparsity/magnitude/test_modules.py

@@ -22,7 +22,7 @@
 from tests.torch.helpers import fill_conv_weight, fill_linear_weight, fill_bias
 
 
-class TestModel(nn.Module):
+class SingleLayerModel(nn.Module):
     def __init__(self, layer):
         super().__init__()
         self.layer = layer
@@ -53,7 +53,7 @@ def forward(self, x):
 )
 def test_can_infer_magnitude_sparse_conv(weight_importance, threshold, ref_output):
     nncf_module = NNCFConv2d(1, 1, 2)
-    sparse_model = TestModel(nncf_module)
+    sparse_model = SingleLayerModel(nncf_module)
     sparsifier = sparse_model.sparsifier
     fill_conv_weight(nncf_module, 9)
     fill_bias(nncf_module, 0)
@@ -83,7 +83,7 @@ def test_can_infer_magnitude_sparse_conv(weight_importance, threshold, ref_outpu
 )
 def test_can_infer_magnitude_sparse_linear(weight_importance, threshold, ref_output):
     nncf_module = NNCFLinear(4, 1)
-    sparse_model = TestModel(nncf_module)
+    sparse_model = SingleLayerModel(nncf_module)
     sparsifier = sparse_model.sparsifier
     fill_linear_weight(nncf_module, 9)
     fill_bias(nncf_module, 0)

tests/torch/sparsity/rb/test_components.py

@@ -21,7 +21,7 @@
 from nncf.torch.sparsity.rb.loss import SparseLoss
 
 
-class TestModel(nn.Module):
+class SingleLayerModel(nn.Module):
     def __init__(self, layer, frozen, size=1):
         super().__init__()
         self.size = size
@@ -42,7 +42,7 @@ def forward(self, x):
 
 def sparse_model(module, frozen, size=1):
     layer = module(size, size, size)
-    return TestModel(layer, frozen, size)
+    return SingleLayerModel(layer, frozen, size)
 
 
 @pytest.mark.parametrize('module',

tests/torch/sparsity/test_common.py

@@ -317,14 +317,14 @@ def test_adaptive_scheduler_per_epoch_step(self, rb_algo_mock, ref_sparsity_leve
 }
 
 
-class TestLoss:
+class MockLoss:
     def __init__(self):
         self.current_sparsity = 0.5
 
 
-class TestCompressionController:
+class MockCompressionController:
     def __init__(self):
-        self.loss = TestLoss()
+        self.loss = MockLoss()
 
     def set_sparsity_level(self, level):
         pass
@@ -334,7 +334,7 @@ def set_sparsity_level(self, level):
                                            MultiStepSparsityScheduler, AdaptiveSparsityScheduler],
                          ids=['Polynomial', 'Exponential', 'Multistep', 'Adaptive'])
 def test_scheduler_get_state(scheduler_cls):
-    args = (TestCompressionController(), {'sparsity_init': 0.3, 'update_per_optimizer_step': True, 'patience': 2})
+    args = (MockCompressionController(), {'sparsity_init': 0.3, 'update_per_optimizer_step': True, 'patience': 2})
     scheduler = scheduler_cls(*args)
 
     # Test init state