update

tests/test_runtime/test_optimizer.py

@@ -0,0 +1,51 @@
+# Copyright (c) OpenMMLab. All rights reserved.
+import pytest
+import torch
+import torch.nn as nn
+
+from mmselfsup.core import build_optimizer, LARS
+
+
+class ExampleModel(nn.Module):
+
+    def __init__(self):
+        super(ExampleModel, self).__init__()
+        self.test_cfg = None
+        self.predictor = nn.Linear(2, 1)
+
+    def forward(self, img, img_metas, test_mode=False, **kwargs):
+        res = self.predictor(img)
+        return res
+
+    def train_step(self, data_batch, optimizer):
+        loss = self.forward(**data_batch)
+        return dict(loss=loss.sum())
+
+
+def test_lars():
+    optimizer = dict(
+        type='LARS',
+        lr=0.3,
+        momentum=0.9,
+        weight_decay=1e-6,
+        paramwise_options={'bias': dict(weight_decay=0., lars_exclude=True)})
+
+    model = ExampleModel()
+    optimizer = build_optimizer(model, optimizer)
+
+    for i in range(2):
+        loss = model.train_step(
+            dict(img=torch.ones(2, 2), img_metas=None), optimizer)
+
+        optimizer.zero_grad()
+        loss['loss'].backward()
+        optimizer.step()
+
+    with pytest.raises(ValueError):
+        optimizer = LARS(model.parameters(), lr=-1)
+    with pytest.raises(ValueError):
+        optimizer = LARS(model.parameters(), lr=0.1, momentum=-1)
+    with pytest.raises(ValueError):
+        optimizer = LARS(model.parameters(), lr=0.1, weight_decay=-1)
+    with pytest.raises(ValueError):
+        optimizer = LARS(model.parameters(), lr=0.1, eta=-1)