refactor: rasa/shared/core/training_data/visualization.py #12544
Conversation
In the refactored version, In function(_remove_auxiliary_nodes), I've replaced the conversion of graph.predecessors(i) into a list with the direct usage of the generator. Additionally, I've introduced the predecessors_seen set to efficiently keep track of seen predecessors. When a duplicated predecessor is found, we can remove the node and break out of the inner loop. This optimization reduces the time complexity of checking for duplicated nodes to approximately O((TMP_NODE_ID - special_node_idx) + out_degree(node)).
|
|
In this commit, I have implemented example test cases to test method: _remove_auxiliary_nodes in @pytest.mark.parametrize. Also,fixed the import error of networkx.
Hey @Urkem, Done! |
In this version, I have removed break statement from the loop so it can remove all the predecessors nodes which are in predecessors_seen without breaking the loop.
|
Hey @Urkem, |
|
Hey @Urkem, |
|
Hey @arjun-234 I think these changes are making a regression test fail at, |
|
Hey @vcidst, |
|
@arjun-234 I see, it might be a flaky test in that case. I'll re-run the CI and see if the problem persists |
@vcidst Thanks :) |
In the refactored version, In function(_remove_auxiliary_nodes), I've replaced the conversion of graph.predecessors(i) into a list with the direct usage of the generator. Additionally, I've introduced the predecessors_seen set to efficiently keep track of seen predecessors. When a duplicated predecessor is found, we can remove the node and break out of the inner loop.
Proposed changes:
Status (please check what you already did):