Release/0.2.6.1 (#512)

* Version 0.2.6

* Fix multiplying predictions when evaluate called mulitple times

* Add test

* Update changelog

* Update changelog

* Update CHANGELOG.md

* Bump version number

* Update CHANGELOG.md

CHANGELOG.md

@@ -20,7 +20,6 @@ The format is based on [Keep a Changelog](http://keepachangelog.com/en/1.0.0/).
 - Added documentation about serialization.
 - Added support for indefinite data loading. Iterators can now be run until complete independent of epochs or iterators can be refreshed during an epoch if complete. 
 - Added support for batch intervals in interval checkpointer
-
 ### Changed
 - Changed the default behaviour of the std metric to compute the sample std, in line with torch.std
 - Tqdm precision argument now rounds to decimal places rather than significant figures
@@ -34,6 +33,14 @@ The format is based on [Keep a Changelog](http://keepachangelog.com/en/1.0.0/).
 - Fixed a bug in the cite decorator causing the citation to not show up correctly
 - Fixed a memory leak in the mse primitive metric
 
+## [0.2.6.1] - 2019-02-25
+### Added
+### Changed
+### Deprecated
+### Removed
+### Fixed
+- Fixed a bug where predictions would multiply when predict was called more than once
+
 ## [0.2.6] - 2018-12-19
 ### Added
 ### Changed

tests/callbacks/test_aggregate_predictions.py

@@ -26,4 +26,22 @@ def test_aggreate_predictions(self):
         aggregator.on_end_validation(final_state)
         self.assertTrue(list(final_state[tb.FINAL_PREDICTIONS].numpy()) == list(aggregate.numpy()))
 
+    def test_aggreate_predictions_multiple_calls(self):
+        aggregator = AggregatePredictions()
+
+        y_pred_1 = torch.Tensor([1,2,3])
+        y_pred_2 = torch.Tensor([3,4,5])
+
+        state_1 = {tb.Y_PRED: y_pred_1}
+        state_2 = {tb.Y_PRED: y_pred_2}
+
+        aggregator.on_step_validation(state_1)
+        self.assertTrue(list(aggregator.predictions_list[0].numpy()) == list(y_pred_1.numpy()))
+
+        aggregator.on_step_validation(state_2)
+        self.assertTrue(list(aggregator.predictions_list[1].numpy()) == list(y_pred_2.numpy()))
+
+        aggregator.on_end_epoch(state_2)
+        self.assertTrue(list(aggregator.predictions_list) == [])
+
 

torchbearer/callbacks/aggregate_predictions.py

@@ -14,3 +14,7 @@ def on_step_validation(self, state):
 
     def on_end_validation(self, state):
         state[torchbearer.FINAL_PREDICTIONS] = torch.cat(self.predictions_list, 0)
+
+    def on_end_epoch(self, state):
+        super(AggregatePredictions, self).on_end_epoch(state)
+        self.predictions_list = []