fix bugs

tensorneat/examples/interpret_visualize/genome_sympy.py

@@ -1,14 +1,14 @@
 import jax, jax.numpy as jnp
 
 from algorithm.neat import *
-from algorithm.neat.genome.hidden import AdvanceInitialize
+from algorithm.neat.genome.dense import DenseInitialize
 from utils.graph import topological_sort_python
+from utils import *
 
-if __name__ == '__main__':
-    genome = AdvanceInitialize(
-        num_inputs=17,
-        num_outputs=6,
-        hidden_cnt=8,
+if __name__ == "__main__":
+    genome = DenseInitialize(
+        num_inputs=3,
+        num_outputs=1,
         max_nodes=50,
         max_conns=500,
     )
@@ -19,16 +19,19 @@ if __name__ == '__main__':
     nodes, conns = genome.initialize(state, randkey)
 
     network = genome.network_dict(state, nodes, conns)
-    print(set(network["nodes"]), set(network["conns"]))
-    order, _ = topological_sort_python(set(network["nodes"]), set(network["conns"]))
-    print(order)
 
     input_idx, output_idx = genome.get_input_idx(), genome.get_output_idx()
-    print(input_idx, output_idx)
 
-    print(genome.repr(state, nodes, conns))
-    print(network)
+    res = genome.sympy_func(state, network, sympy_input_transform=lambda x: 999999999*x, sympy_output_transform=SympyStandardSigmoid)
+    (symbols,
+    args_symbols,
+    input_symbols,
+    nodes_exprs,
+    output_exprs,
+    forward_func,) = res
 
-    res = genome.sympy_func(state, network, precision=3)
-    print(res)
+    print(symbols)
+    print(output_exprs[0].subs(args_symbols))
 
+    inputs = jnp.zeros(3)
+    print(forward_func(inputs))