diff --git a/hls4ml/backends/fpga/passes/clone.py b/hls4ml/backends/fpga/passes/clone.py
index 306e839900..0c1f7f2e07 100644
--- a/hls4ml/backends/fpga/passes/clone.py
+++ b/hls4ml/backends/fpga/passes/clone.py
@@ -87,7 +87,7 @@ def transform(self, model, node):
                 )
                 for i in range(len(output_map[output])):
                     key = output + '_cpy' + str(i + 1)
-                    clone_layer.attributes[key].type = node.attributes['result_t']
+                    clone_layer.attributes[key].type = node.get_output_variable().type
                 model.insert_node(clone_layer)
                 transformed = True
 
diff --git a/hls4ml/converters/onnx_to_hls.py b/hls4ml/converters/onnx_to_hls.py
index 75850fa93e..f60659f404 100644
--- a/hls4ml/converters/onnx_to_hls.py
+++ b/hls4ml/converters/onnx_to_hls.py
@@ -63,6 +63,8 @@ def get_input_shape(graph, node):
     """
     rv = []
     for inp in node.input:
+        if inp == '':
+            continue
         try:
             value_info_idx = next((i for i, x in enumerate(graph.value_info) if x.name == inp))
             dim = list(d.dim_value for d in graph.value_info[value_info_idx].type.tensor_type.shape.dim)
diff --git a/hls4ml/model/optimizer/passes/batchnorm_opt.py b/hls4ml/model/optimizer/passes/batchnorm_opt.py
index 0dde6b77a9..60e87dc670 100644
--- a/hls4ml/model/optimizer/passes/batchnorm_opt.py
+++ b/hls4ml/model/optimizer/passes/batchnorm_opt.py
@@ -102,7 +102,7 @@ def transform(self, model, node):
         const_prec = const_node.get_output_variable().type.precision
 
         new_val = (
-            const_node.attributes['value'] * node.weights['scale'].data_unquantized + node.weights['bias'].data_unquantized
+            const_node.get_attr('value') * node.weights['scale'].data_unquantized + node.weights['bias'].data_unquantized
         )
 
         const_node.set_attr('value', new_val)
diff --git a/hls4ml/model/optimizer/passes/linear.py b/hls4ml/model/optimizer/passes/linear.py
index b1aee7adc7..2a3e2b3b9d 100644
--- a/hls4ml/model/optimizer/passes/linear.py
+++ b/hls4ml/model/optimizer/passes/linear.py
@@ -1,4 +1,17 @@
-from hls4ml.model.layers import Activation, BatchNormalization, Conv1D, Conv2D, Dense
+from hls4ml.model.layers import (
+    Activation,
+    BatchNormalization,
+    Concatenate,
+    Conv1D,
+    Conv2D,
+    Dense,
+    DepthwiseConv1D,
+    DepthwiseConv2D,
+    Input,
+    Pooling1D,
+    Pooling2D,
+    Resize,
+)
 from hls4ml.model.optimizer import OptimizerPass
 from hls4ml.model.types import UnspecifiedPrecisionType
 
@@ -15,7 +28,20 @@ def transform(self, model, node):
         return True
 
 
-_safe_parents = (Dense, Conv1D, Conv2D, BatchNormalization, Activation)
+_safe_parents = (
+    Input,
+    Dense,
+    Conv1D,
+    Conv2D,
+    DepthwiseConv1D,
+    DepthwiseConv2D,
+    BatchNormalization,
+    Activation,
+    Pooling1D,
+    Pooling2D,
+    Resize,
+    Concatenate,
+)
 
 
 class MergeLinearActivation(OptimizerPass):
diff --git a/hls4ml/model/optimizer/passes/quant_opt.py b/hls4ml/model/optimizer/passes/quant_opt.py
index 69e9ca7685..dcb8ca2a63 100644
--- a/hls4ml/model/optimizer/passes/quant_opt.py
+++ b/hls4ml/model/optimizer/passes/quant_opt.py
@@ -187,7 +187,7 @@ def transform(self, model, node):
         integer = bitwidth
         scale = node.get_attr('scale')
         if _ALSO_MATCH_PO2 and not (scale == np.ones_like(scale)).all():
-            _, exp = np.frexp(np.squeeze(scale))
+            _, exp = np.frexp(np.unique(scale.ravel()).item())
             integer = bitwidth + exp - 1
 
         precision, quantizer = _calculate_precision_quantizer(bitwidth, integer, signed, narrow, rounding_mode)
@@ -336,7 +336,7 @@ def transform(self, model, node):
 
         inshape = node.get_input_variable().shape
 
-        attributes_rescale = {'n_filt': -1}
+        attributes_rescale = {'n_filt': -1, 'quantizer': quantizer}
 
         rescale_config = copy.deepcopy(model.config.get_layer_config(node))
         rescale_name = f'{node.name}_rescale'