Start accumulate gradients schedule at epoch 0 (continued)

pytorch_lightning/trainer/training_tricks.py

@@ -99,7 +99,7 @@ def configure_accumulated_gradients(self, accumulate_grad_batches):
         if isinstance(accumulate_grad_batches, dict):
             self.accumulation_scheduler = GradientAccumulationScheduler(accumulate_grad_batches)
         elif isinstance(accumulate_grad_batches, int):
-            schedule = {1: accumulate_grad_batches}
+            schedule = {0: accumulate_grad_batches}
             self.accumulation_scheduler = GradientAccumulationScheduler(schedule)
         else:
             raise TypeError("Gradient accumulation supports only int and dict types")

tests/trainer/test_lr_finder.py

@@ -154,7 +154,7 @@ def test_accumulation_and_early_stopping(tmpdir):
         'Learning rate was not altered after running learning rate finder'
     assert len(lrfinder.results['lr']) == 100, \
         'Early stopping for learning rate finder did not work'
-    assert lrfinder._total_batch_idx == 190, \
+    assert lrfinder._total_batch_idx == 100 * 2, \
         'Accumulation parameter did not work'
 
 

tests/trainer/test_trainer.py

@@ -103,7 +103,8 @@ def test_no_val_end_module(monkeypatch, tmpdir, tmpdir_server, url_ckpt):
     model_2.eval()
 
 
-def test_gradient_accumulation_scheduling(tmpdir):
+@pytest.mark.parametrize('schedule_expected', [({1: 2, 3: 4}, [1, 2, 4]), (3, [3, 3, 3]), (4, [4, 4, 4])])
+def test_gradient_accumulation_scheduling(tmpdir, schedule_expected):
     """
     Test grad accumulation by the freq of optimizer updates
     """
@@ -123,59 +124,61 @@ def test_gradient_accumulation_scheduling(tmpdir):
     with pytest.raises(TypeError):
         assert Trainer(accumulate_grad_batches={1: 2.5, 3: 5})
 
+    model = EvalModelTemplate()
+    schedule = schedule_expected[0]
+    expected = schedule_expected[1]
+
+    trainer = Trainer(accumulate_grad_batches=schedule,
+                      limit_train_batches=0.8,
+                      limit_val_batches=0.8,
+                      max_epochs=4,
+                      default_root_dir=tmpdir)
+
     # test optimizer call freq matches scheduler
-    def _optimizer_step(self, epoch, batch_idx, optimizer,
-                        optimizer_idx, second_order_closure=None):
+    def _optimizer_step(epoch, batch_idx, optimizer, optimizer_idx,
+                        second_order_closure=None, on_tpu=False,
+                        using_native_amp=False, using_lbfgs=False):
         # only test the first 12 batches in epoch
         if batch_idx < 12:
             if epoch == 0:
                 # reset counter when starting epoch
-                if batch_idx == 0:
-                    self.prev_called_batch_idx = 0
+                if batch_idx == expected[0] - 1:
+                    model.prev_called_batch_idx = expected[0] - 1
 
                     # use this opportunity to test once
-                    assert self.trainer.accumulate_grad_batches == 1
+                    assert trainer.accumulate_grad_batches == expected[0]
 
-                assert batch_idx == self.prev_called_batch_idx
-                self.prev_called_batch_idx += 1
+                assert batch_idx == model.prev_called_batch_idx
+                model.prev_called_batch_idx += expected[0]
 
             elif 1 <= epoch <= 2:
                 # reset counter when starting epoch
-                if batch_idx == 1:
-                    self.prev_called_batch_idx = 1
+                if batch_idx == expected[1] - 1:
+                    model.prev_called_batch_idx = expected[1] - 1
 
                     # use this opportunity to test once
-                    assert self.trainer.accumulate_grad_batches == 2
+                    assert trainer.accumulate_grad_batches == expected[1]
 
-                assert batch_idx == self.prev_called_batch_idx
-                self.prev_called_batch_idx += 2
+                assert batch_idx == model.prev_called_batch_idx
+                model.prev_called_batch_idx += expected[1]
 
             else:
-                if batch_idx == 3:
-                    self.prev_called_batch_idx = 3
+                if batch_idx == expected[2] - 1:
+                    model.prev_called_batch_idx = expected[2] - 1
 
                     # use this opportunity to test once
-                    assert self.trainer.accumulate_grad_batches == 4
+                    assert trainer.accumulate_grad_batches == expected[2]
 
-                assert batch_idx == self.prev_called_batch_idx
-                self.prev_called_batch_idx += 3
+                assert batch_idx == model.prev_called_batch_idx
+                model.prev_called_batch_idx += expected[2]
 
         optimizer.step()
 
         # clear gradients
         optimizer.zero_grad()
 
-    model = EvalModelTemplate()
-    schedule = {1: 2, 3: 4}
-
-    trainer = Trainer(accumulate_grad_batches=schedule,
-                      limit_train_batches=0.1,
-                      limit_val_batches=0.1,
-                      max_epochs=2,
-                      default_root_dir=tmpdir)
-
     # for the test
-    trainer.optimizer_step = _optimizer_step
+    model.optimizer_step = _optimizer_step
     model.prev_called_batch_idx = 0
 
     trainer.fit(model)