switch to total column prediction for now - need more R&D

modules/deeplearning/icing_cnn.py

@@ -1103,7 +1103,7 @@ def run_evaluate_static(data_dct, ll, cc, ckpt_dir_s_path, day_night='DAY', flig
 def run_evaluate_static_new(data_dct, num_lines, num_elems, ckpt_dir_s_path, day_night='DAY', flight_levels=[0, 1, 2, 3, 4], prob_thresh=0.5):
 
     ckpt_dir_s = os.listdir(ckpt_dir_s_path)
-    ckpt_dir = ckpt_dir_s[0]
+    ckpt_dir = ckpt_dir_s_path + ckpt_dir_s[0]
 
     probs_2d_dct = {flvl: None for flvl in flight_levels}
     preds_2d_dct = {flvl: None for flvl in flight_levels}
@@ -1115,12 +1115,7 @@ def run_evaluate_static_new(data_dct, num_lines, num_elems, ckpt_dir_s_path, day
         nn.build_model()
         nn.build_training()
         nn.build_evaluation()
-
-        ckpt = tf.train.Checkpoint(step=tf.Variable(1), model=nn.model)
-        ckpt_manager = tf.train.CheckpointManager(ckpt, ckpt_dir, max_to_keep=3)
-        ckpt.restore(ckpt_manager.latest_checkpoint)
-
-        nn.do_evaluate()
+        nn.do_evaluate(ckpt_dir)
 
         probs = nn.test_probs
         if NumClasses == 2: