Unable to forwardDims() of lstm Operator

I am trying to test inference of an lstm_model.onnx.

The model has 6 inputs :

• W, R, B and initial_h as initializers
  
• X and initial_c as dynamic inputs
  

I am using this script to test the model on Aidge:

import aidge_core
import aidge_backend_cpu
import aidge_onnx

import numpy as np

aidge_model = aidge_onnx.load_onnx('lstm_model.onnx')

input_x = np.random.randn(sequence_length, batch_size, input_size).astype(np.float32)

input_tensor = aidge_core.Tensor(input_x)
input_node = aidge_core.Producer(input_tensor, "X")
input_node.get_operator().set_datatype(aidge_core.dtype.float32)
input_node.get_operator().set_backend("cpu")
input_node.add_child(aidge_model, 0, aidge_model.get_ordered_inputs()[0])
aidge_model.add(input_node)

input_initial_c = np.random.randn(num_directions, batch_size, hidden_size).astype(np.float32)
input_tensor2 = aidge_core.Tensor(input_initial_c)
input_node2 = aidge_core.Producer(input_tensor2, "initial_c")
input_node2.get_operator().set_datatype(aidge_core.dtype.float32)
input_node2.get_operator().set_backend("cpu")
input_node2.add_child(aidge_model, 0, aidge_model.get_ordered_inputs()[1])
aidge_model.add(input_node2)

aidge_model.set_datatype(aidge_core.dtype.float32)
aidge_model.set_backend("cpu")

scheduler = aidge_core.SequentialScheduler(aidge_model)

scheduler.forward()

I am getting a segmentation fault in Aidge::GraphView::forwardDims: https://gitlab.eclipse.org/eclipse/aidge/aidge_core/-/blob/dev/src/graph/GraphView.cpp?ref_type=heads#L502

nodePtr is a null pointer.
when I tried to change the condition in the if to:

if (nodePtr && nodePtr->getOperator()->operatorType() == OperatorType::Tensor)

I get another error later:

Assertion failed: (getInput(1)->nbDims() == 2) in /tmp/pip-req-build-cckhbi1f/src/operator/FC.cpp:46
Wrong weight Tensor dimension: 3 for FC operator (should have 2 dimensions).