Revert "[update] Rename tests"

This reverts commit 0183a73b75.

tests/test_evaluation/functional/test_hmean.py

@@ -0,0 +1,23 @@
+# Copyright (c) OpenMMLab. All rights reserved.
+from unittest import TestCase
+
+from mmocr.evaluation.functional import compute_hmean
+
+
+class TestHmean(TestCase):
+
+    def test_compute_hmean(self):
+        with self.assertRaises(AssertionError):
+            compute_hmean(0, 0, 0.0, 0)
+        with self.assertRaises(AssertionError):
+            compute_hmean(0, 0, 0, 0.0)
+        with self.assertRaises(AssertionError):
+            compute_hmean([1], 0, 0, 0)
+        with self.assertRaises(AssertionError):
+            compute_hmean(0, [1], 0, 0)
+
+        _, _, hmean = compute_hmean(2, 2, 2, 2)
+        self.assertEqual(hmean, 1)
+
+        _, _, hmean = compute_hmean(0, 0, 2, 2)
+        self.assertEqual(hmean, 0)

tests/test_models/test_common/layers/test_transformer_layers.py

@@ -0,0 +1,37 @@
+# Copyright (c) OpenMMLab. All rights reserved.
+from unittest import TestCase
+
+import torch
+
+from mmocr.models.common.layers.transformer_layers import (TFDecoderLayer,
+                                                           TFEncoderLayer)
+
+
+class TestTFEncoderLayer(TestCase):
+
+    def test_forward(self):
+        encoder_layer = TFEncoderLayer()
+        in_enc = torch.rand(1, 20, 512)
+        out_enc = encoder_layer(in_enc)
+        self.assertEqual(out_enc.shape, torch.Size([1, 20, 512]))
+
+        encoder_layer = TFEncoderLayer(
+            operation_order=('self_attn', 'norm', 'ffn', 'norm'))
+        out_enc = encoder_layer(in_enc)
+        self.assertEqual(out_enc.shape, torch.Size([1, 20, 512]))
+
+
+class TestTFDecoderLayer(TestCase):
+
+    def test_forward(self):
+        decoder_layer = TFDecoderLayer()
+        in_dec = torch.rand(1, 30, 512)
+        out_enc = torch.rand(1, 128, 512)
+        out_dec = decoder_layer(in_dec, out_enc)
+        self.assertEqual(out_dec.shape, torch.Size([1, 30, 512]))
+
+        decoder_layer = TFDecoderLayer(
+            operation_order=('self_attn', 'norm', 'enc_dec_attn', 'norm',
+                             'ffn', 'norm'))
+        out_dec = decoder_layer(in_dec, out_enc)
+        self.assertEqual(out_dec.shape, torch.Size([1, 30, 512]))

tests/test_models/test_common/modules/test_transformer_module.py

@@ -0,0 +1,15 @@
+# Copyright (c) OpenMMLab. All rights reserved.
+from unittest import TestCase
+
+import torch
+
+from mmocr.models.common.modules import PositionalEncoding
+
+
+class TestPositionalEncoding(TestCase):
+
+    def test_forward(self):
+        pos_encoder = PositionalEncoding()
+        x = torch.rand(1, 30, 512)
+        out = pos_encoder(x)
+        assert out.size() == x.size()

tests/test_models/test_textrecog/layers/test_conv_layer.py

@@ -0,0 +1,42 @@
+# Copyright (c) OpenMMLab. All rights reserved.
+from unittest import TestCase
+
+import torch
+
+from mmocr.models.textrecog.layers.conv_layer import (BasicBlock, Bottleneck,
+                                                      conv1x1, conv3x3)
+
+
+class TestUtils(TestCase):
+
+    def test_conv3x3(self):
+        conv = conv3x3(3, 6)
+        self.assertEqual(conv.in_channels, 3)
+        self.assertEqual(conv.out_channels, 6)
+        self.assertEqual(conv.kernel_size, (3, 3))
+
+    def test_conv1x1(self):
+        conv = conv1x1(3, 6)
+        self.assertEqual(conv.in_channels, 3)
+        self.assertEqual(conv.out_channels, 6)
+        self.assertEqual(conv.kernel_size, (1, 1))
+
+
+class TestBasicBlock(TestCase):
+
+    def test_forward(self):
+        x = torch.rand(1, 64, 224, 224)
+        basic_block = BasicBlock(64, 64)
+        self.assertEqual(basic_block.expansion, 1)
+        out = basic_block(x)
+        self.assertEqual(out.shape, torch.Size([1, 64, 224, 224]))
+
+
+class TestBottleneck(TestCase):
+
+    def test_forward(self):
+        x = torch.rand(1, 64, 224, 224)
+        bottle_neck = Bottleneck(64, 64, downsample=True)
+        self.assertEqual(bottle_neck.expansion, 4)
+        out = bottle_neck(x)
+        self.assertEqual(out.shape, torch.Size([1, 256, 224, 224]))