Fix lack of proper loading of best_prec1 from the checkpoint (#1000)

- resume() is a nested function and when it loads best_prec1
  it creates a local variable that hides the one from the parent
  function (which refers to the global one). This PR adds `global`
  to modify the global variable as intended
Signed-off-by: Janusz Lisiecki <jlisiecki@nvidia.com>

examples/imagenet/main_amp.py

@@ -182,6 +182,7 @@ def main():
                 print("=> loading checkpoint '{}'".format(args.resume))
                 checkpoint = torch.load(args.resume, map_location = lambda storage, loc: storage.cuda(args.gpu))
                 args.start_epoch = checkpoint['epoch']
+                global best_prec1
                 best_prec1 = checkpoint['best_prec1']
                 model.load_state_dict(checkpoint['state_dict'])
                 optimizer.load_state_dict(checkpoint['optimizer'])