Unskip some unit tests related to issue #82 (#98)

* Unskip some unit tests related to issue #82

* Ensure test_state_dict to use capturable=True for torch.optim.Adam

* Fix TestFusedAdam tests in test_fused_optimizer.py

tests/L0/run_amp/test_checkpointing.py

@@ -228,7 +228,7 @@ class TestCheckpointing(unittest.TestCase):
                 continue
 
             model = MyModel().to('cuda')
-            optimizer = optim.Adam(model.parameters(), lr=1e-3)
+            optimizer = optim.Adam(model.parameters(), lr=1e-3, capturable=True)
             model, optimizer = amp.initialize(
                 model, optimizer, opt_level=opt_level, verbosity=0)
 

tests/L0/run_optimizers/test_fused_optimizer.py

@@ -91,11 +91,12 @@ class TestFusedAdam(TestFusedOptimizer):
     def setUp(self):
         super().setUp()
         self.options = {'lr':5e-4, 'betas':(0.9, 0.999), 'eps':1e-08,
+            'weight_decay': 0, 'amsgrad': False, "capturable": True}
+        self.tst_options = {'lr':5e-4, 'betas':(0.9, 0.999), 'eps':1e-08,
             'weight_decay': 0, 'amsgrad': False}
         self.ref_optim = torch.optim.Adam
         self.fused_optim = apex.optimizers.FusedAdam
 
-    @unittest.skip("Skipped the test since a regression introduced from PyTorch upstream: due to https://github.com/pytorch/pytorch/issues/80809#issuecomment-1175211598. Please also refer to https://github.com/ROCmSoftwarePlatform/apex/issues/82")
     def test_float(self):
         self.gen_single_type_test(param_type=torch.float)
 
@@ -105,11 +106,9 @@ class TestFusedAdam(TestFusedOptimizer):
     def test_half(self):
         self.gen_single_type_test(param_type=torch.float16, skip_assert=True)
 
-    @unittest.skip("Skipped the test since a regression introduced from PyTorch upstream: due to https://github.com/pytorch/pytorch/issues/80809#issuecomment-1175211598. Please also refer to https://github.com/ROCmSoftwarePlatform/apex/issues/82")
     def test_bfloat16(self):
         self.gen_single_type_test(param_type=torch.bfloat16, skip_assert=True)
 
-    @unittest.skip("Skipped the test since a regression introduced from PyTorch upstream: due to https://github.com/pytorch/pytorch/issues/80809#issuecomment-1175211598. Please also refer to https://github.com/ROCmSoftwarePlatform/apex/issues/82")
     @unittest.skipIf(torch.cuda.device_count()<2, "more than 1 GPU required")
     def test_multi_device(self):
         devices = ("cuda:0", "cuda:1")
@@ -176,15 +175,17 @@ class TestFusedAdam(TestFusedOptimizer):
             self.assertLessEqual(max_abs_diff, self.max_abs_diff)
             self.assertLessEqual(max_rel_diff, self.max_rel_diff)
 
-    @unittest.skip("Skipped the test since a regression introduced from PyTorch upstream: due to https://github.com/pytorch/pytorch/issues/80809#issuecomment-1175211598. Please also refer to https://github.com/ROCmSoftwarePlatform/apex/issues/82")
     def test_adam_option(self):
         nelem = 1
         adam_option = {'lr':0.01, 'betas':(0.6, 0.9), 'eps':3e-06,
+            'weight_decay':0, 'amsgrad':False, 'capturable':True}
+
+        adam_option_tst = {'lr':0.01, 'betas':(0.6, 0.9), 'eps':3e-06,
             'weight_decay':0, 'amsgrad':False}
 
         tensor = torch.rand(nelem, dtype=torch.float, device='cuda')
         ref_param, tst_param, ref_optim, tst_optim = \
-            self.gen_param_optim([tensor], adam_option)
+            self.gen_param_optim([tensor], adam_option, adam_option_tst)
 
         for i in range(self.iters):
             self.gen_grad(ref_param, tst_param)