addressing Jovan's comments

Author: JanFSchulte

hls4ml/backends/quartus/passes/recurrent_templates.py

@@ -93,7 +93,7 @@ def format(self, node):
         params['config_mult_h'] = f'config{node.index}_h_mult'
         params['act_t'] = '{}_config{}'.format(node.get_attr('activation'), str(node.index) + '_act')
         params['act_recurrent_t'] = '{}_config{}'.format(node.get_attr('recurrent_activation'), str(node.index) + '_rec_act')
-        params['pytorch'] = 'true' if "pytorch" in node.attributes.keys() else 'false'
+        params['pytorch'] = 'true' if node.get_attr('pytorch', False) else 'false'
         gru_config = self.gru_template.format(**params)
 
         # Activation is on candidate hidden state, dimensionality (1, n_units)
@@ -306,7 +306,7 @@ def __init__(self):
 
     def format(self, node):
         params = self._default_function_params(node)
-        if "pytorch" in node.attributes.keys():
+        if node.get_attr('pytorch', False):
             self.template = simple_rnn_pytorch_function_template
             params['weights'] = 'w{0}, wr{0}, b{0}, br{0}'.format(str(node.index))
         else:

hls4ml/backends/vivado/passes/recurrent_templates.py

@@ -98,7 +98,7 @@ def format(self, node):
         params['act_t'] = '{}_config{}'.format(node.get_attr('activation'), node.index)
         params['strategy'] = node.get_attr('strategy')
         params['static'] = 'true' if node.attributes['static'] else 'false'
-        params['pytorch'] = 'true' if "pytorch" in node.attributes.keys() else 'false'
+        params['pytorch'] = 'true' if node.get_attr('pytorch', False) else 'false'
         params['recr_type'] = node.class_name.lower()
         params['RECR_TYPE'] = node.class_name
 

test/pytest/test_pytorch_api.py

@@ -846,7 +846,7 @@ def forward(self, x):
 
     # X_input is channels last
     X_input = np.ascontiguousarray(X_input.transpose(0, 2, 1))
-    config = config_from_pytorch_model(model, inputs_channel_last=True, transpose_outputs=False)
+    config = config_from_pytorch_model(model, channels_last_conversion="internal", transpose_outputs=False)
 
     output_dir = str(test_root_path / f'hls4mlprj_pytorch_view_{backend}_{io_type}')
     hls_model = convert_from_pytorch_model(