add some small assertions

tensorpack/tfutils/gradproc.py

@@ -67,7 +67,7 @@ class GlobalNormClip(GradientProcessor):
         Args:
             global_norm(float): the threshold to clip with.
         """
-        self._norm = global_norm
+        self._norm = float(global_norm)
 
     def _process(self, grads):
         g = [k[0] for k in grads]
@@ -176,6 +176,7 @@ class ScaleGradient(MapGradient):
         if not isinstance(multipliers, list):
             multipliers = [multipliers]
         self.multipliers = multipliers
+        assert log in [True, False], log
         self._log = log
         super(ScaleGradient, self).__init__(self._mapper)
 

tensorpack/tfutils/optimizer.py

@@ -16,6 +16,7 @@ class ProxyOptimizer(tf.train.Optimizer):
     A transparent proxy which delegates all methods of :class:`tf.train.Optimizer`
     """
     def __init__(self, opt, name='ProxyOptimizer'):
+        assert isinstance(opt, tf.train.Optimizer), opt
         super(ProxyOptimizer, self).__init__(False, name)
         self._opt = opt
 
@@ -44,6 +45,7 @@ def apply_grad_processors(opt, gradprocs):
         a :class:`tf.train.Optimizer` instance which runs the gradient
         processors before updating the variables.
     """
+    assert isinstance(gradprocs, (list, tuple)), gradprocs
 
     class _ApplyGradientProcessor(ProxyOptimizer):
         def __init__(self, opt, gradprocs):

tensorpack/train/multigpu.py

@@ -82,7 +82,7 @@ class MultiGPUTrainerBase(Trainer):
                 if idx == t:
                     logger.info("Building graph for training tower {}...".format(idx))
                 else:
-                    logger.info("Building graph for training tower {} on device {}...".format(idx, t))
+                    logger.info("Building graph for training tower {} on device {}...".format(idx, device))
 
                 ret.append(func())
 
@@ -264,7 +264,7 @@ class SyncMultiGPUTrainerReplicated(MultiGPUTrainerBase, SingleCostFeedfreeTrain
             run_before=True, run_as_trigger=True))
 
 
-# Copied from https://github.com/tensorflow/benchmarks/blob/master/scripts/tf_cnn_benchmarks/variable_mgr.py
+# Adopt from https://github.com/tensorflow/benchmarks/blob/master/scripts/tf_cnn_benchmarks/variable_mgr.py
     @staticmethod
     def get_post_init_ops():
         # Copy initialized values for variables on GPU 0 to other GPUs.