Merge branch 'update-readme' of https://github.com/fastmachinelearning/hls4ml into update-readme

Author: JanFSchulte

docs/backend/quartus.rst

@@ -9,4 +9,4 @@ The **Quartus** backend of hls4ml is designed for deploying NNs on Intel/Altera
 The **oneAPI** backend contains the migrated the HLS code from this backend, with significantly better io_stream support, though the **oneAPI** backend does not yet support profiling, tracing,
 or the BramFactor option supported by the **Quartus** backend.  Nevertheless, little or no further development is expected for this backend.
 
-The **Quartus** backend only implements the ``Resource`` strategy for the layers. There is no ``Latency`` implementation of any of the layers.
+The **Quartus** backend only implements the ``Resource`` strategy for the layers. There is no ``Latency`` implementation of any of the layers.

docs/backend/vitis.rst

@@ -9,4 +9,3 @@ backend can be easily incorporated into Vitis kernel.
 
 Users should generally use the **Vitis** backend for new designs that target AMD/Xilinx FPGAs; new ``hls4ml`` developments will not necessarily be backported to
 the **Vivado** backend.
-

hls4ml/converters/pytorch/pooling.py

@@ -90,15 +90,19 @@ def parse_pooling_layer(operation, layer_name, input_names, input_shapes, node,
                 layer['stride_height'] = node.kwargs['stride'][0]
                 layer['stride_width'] = node.kwargs['stride'][1]
             else:
-                layer['stride_height'] = node.kwargs['stride']
-                layer['stride_width'] = node.kwargs['stride']
-            if type(node.kwargs['kernel_size']) is tuple:
-                layer['pool_height'] = node.kwargs['kernel_size'][0]
-                layer['pool_width'] = node.kwargs['kernel_size'][1]
+                if node.kwargs['stride'] is None:
+                    # if stride is not set it is supposed to default to the kernel size
+                    layer['stride_height'] = node.args[1]
+                    layer['stride_width'] = node.args[1]
+                else:
+                    layer['stride_height'] = node.kwargs['stride']
+                    layer['stride_width'] = node.kwargs['stride']
+            if type(node.args[1]) is tuple:
+                layer['pool_height'] = node.args[1][0]
+                layer['pool_width'] = node.args[1][1]
             else:
-                layer['pool_height'] = node.kwargs['kernel_size']
-                layer['pool_width'] = node.kwargs['kernel_size']
-
+                layer['pool_height'] = node.args[1]
+                layer['pool_width'] = node.args[1]
             if type(node.kwargs['padding']) is tuple:
                 padding = node.kwargs['padding']
             else:

hls4ml/converters/pytorch/reshape.py

@@ -93,13 +93,23 @@ def parse_flatten_layer(operation, layer_name, input_names, input_shapes, node,
     layer['class_name'] = 'Reshape'
     layer['name'] = layer_name
     layer['inputs'] = input_names
-
-    start_dim = class_object.start_dim
-    end_dim = class_object.end_dim
-    if end_dim + 1 == 0 or end_dim + 1 > len(input_shapes[0]):
-        end_dim = len(input_shapes[0])
+    if node.op == 'call_module':
+        start_dim = class_object.start_dim
+        end_dim = class_object.end_dim
+        if end_dim + 1 == 0 or end_dim + 1 > len(input_shapes[0]):
+            end_dim = len(input_shapes[0])
+        else:
+            end_dim = end_dim + 1
     else:
-        end_dim = end_dim + 1
+        start_dim = node.args[1]
+        if len(node.args) == 3:
+            end_dim = node.args[2]
+        else:
+            end_dim = -1
+        if end_dim + 1 == 0 or end_dim + 1 > len(input_shapes[0]):
+            end_dim = len(input_shapes[0])
+        else:
+            end_dim = end_dim + 1
 
     layer['target_shape'] = (
         input_shapes[0][0:start_dim] + [np.prod(input_shapes[0][start_dim:end_dim])] + input_shapes[0][end_dim:]