diff --git a/onnx2keras/converter.py b/onnx2keras/converter.py
index 4f6780ca..5e9d8d5d 100644
--- a/onnx2keras/converter.py
+++ b/onnx2keras/converter.py
@@ -76,7 +76,14 @@ def onnx_to_keras(onnx_model, input_names,
 
     logger.debug('Gathering weights to dictionary.')
     weights = {}
+
     for onnx_w in onnx_weights:
+        # Simpler solution.  Seems to work with all keras2onnx-made models (July 28 2020)
+        # Unsure if it works with others. See below alternate solution
+        onnx_extracted_weights_name = onnx_w.ListFields()[-1][1]
+        weights[onnx_extracted_weights_name] = numpy_helper.to_array(onnx_w)
+        # Previous code, does not work for models from keras2onnx as of July 28 2020, and likely earlier
+        '''
         try:
             if len(onnx_w.ListFields()) < 4:
                 onnx_extracted_weights_name = onnx_w.ListFields()[1][1]
@@ -86,7 +93,24 @@ def onnx_to_keras(onnx_model, input_names,
         except:
             onnx_extracted_weights_name = onnx_w.ListFields()[3][1]
             weights[onnx_extracted_weights_name] = numpy_helper.to_array(onnx_w)
-
+        '''
+        # Alternate (ugly) solution, based on previous code
+        '''
+        try:
+            if len(onnx_w.ListFields()) < 4:
+                onnx_extracted_weights_name = onnx_w.ListFields()[1][1]
+            else:
+                onnx_extracted_weights_name = onnx_w.ListFields()[2][1]
+            weights[onnx_extracted_weights_name] = numpy_helper.to_array(onnx_w)
+        except:
+            # Ugly fix
+            try:
+                onnx_extracted_weights_name = onnx_w.ListFields()[3][1]
+                weights[onnx_extracted_weights_name] = numpy_helper.to_array(onnx_w)
+            except:
+                onnx_extracted_weights_name = onnx_w.ListFields()[-1][1]
+                weights[onnx_extracted_weights_name] = numpy_helper.to_array(onnx_w)
+        '''
         logger.debug('Found weight {0} with shape {1}.'.format(
                      onnx_extracted_weights_name,
                      weights[onnx_extracted_weights_name].shape))
diff --git a/onnx2keras/operation_layers.py b/onnx2keras/operation_layers.py
index 97b8a4a5..653518b7 100644
--- a/onnx2keras/operation_layers.py
+++ b/onnx2keras/operation_layers.py
@@ -27,12 +27,20 @@ def convert_clip(node, params, layers, lambda_func, node_name, keras_name):
         assert AttributeError('More than 1 input for clip layer.')
 
     input_0 = ensure_tf_type(layers[node.input[0]], name="%s_const" % keras_name)
-
-    if params['min'] == 0:
-        logger.debug("Using ReLU({0}) instead of clip".format(params['max']))
-        layer = keras.layers.ReLU(max_value=params['max'], name=keras_name)
+    clip_min = None
+    clip_max = None
+    for key in layers.keys():
+        if   'clip_min' in key:
+            clip_min = layers[key]
+        elif 'clip_max' in key:
+            clip_max = layers[key]
+    if clip_min == 0:
+        logger.debug("Using ReLU({0}) instead of clip".format(clip_max))
+        layer = keras.layers.ReLU(max_value=clip_max)
+        # previous code, caused scope name errors for keras2onnx-made models
+        #layer = keras.layers.ReLU(max_value=clip_max, name=keras_name)
     else:
-        def target_layer(x, vmin=params['min'], vmax=params['max']):
+        def target_layer(x, vmin=clip_min, vmax=clip_max):
             import tensorflow as tf
             return tf.clip_by_value(x, vmin, vmax)
         layer = keras.layers.Lambda(target_layer, name=keras_name)