update readme, file names, removing TF code, moving tests

tests/optimization_test.py

@@ -0,0 +1,50 @@
+# coding=utf-8
+# Copyright 2018 The Google AI Language Team Authors.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+from __future__ import absolute_import
+from __future__ import division
+from __future__ import print_function
+
+import unittest
+
+import torch
+
+import optimization as optimization
+
+class OptimizationTest(unittest.TestCase):
+
+    def assertListAlmostEqual(self, list1, list2, tol):
+        self.assertEqual(len(list1), len(list2))
+        for a, b in zip(list1, list2):
+            self.assertAlmostEqual(a, b, delta=tol)
+
+    def test_adam(self):
+        w = torch.tensor([0.1, -0.2, -0.1], requires_grad=True)
+        target = torch.tensor([0.4, 0.2, -0.5])
+        criterion = torch.nn.MSELoss(reduction='elementwise_mean')
+        # No warmup, constant schedule, no gradient clipping
+        optimizer = optimization.BERTAdam(params=[w], lr=2e-1,
+                                          weight_decay_rate=0.0,
+                                          max_grad_norm=-1)
+        for _ in range(100):
+            loss = criterion(w, target)
+            loss.backward()
+            optimizer.step()
+            w.grad.detach_() # No zero_grad() function on simple tensors. we do it ourselves.
+            w.grad.zero_()
+        self.assertListAlmostEqual(w.tolist(), [0.4, 0.2, -0.5], tol=1e-2)
+
+
+if __name__ == "__main__":
+    unittest.main()