Awesome work on the library. I have discovered it recently and it seems to have a lot of good features.
I wanted to try out LIME for tree structured data (ASTs). I have trained a model called TBCNN. Now I want to check which nodes are the ones causing the prediction to be 1 or 0. I cannot use the method described here
def new_predict(x): data = strings_to_embeddings(x) return model.predict_proba(data)
In my case x would be a list of nested dictionaries.
eg.
[{'node': 'ci_root', 'children': [{'node': 'ci_class_decl', 'children': [{'node': 'ci_modifiers', 'children': [{'node': 'ci_modifier', 'children': []}]}, {'node': 'ci_type', 'children': [{'node': 'SimpleType', 'children': [{'node': 'SimpleName', 'children': []}]}]}, ... ]
NOTE: This is just a part of one data point.
So to sum up, I have two questions
1) Is it possible to use LIME on such data?
2) If yes, please provide some hints on how it can be done.
Awesome work on the library. I have discovered it recently and it seems to have a lot of good features. I wanted to try out LIME for tree structured data (ASTs). I have trained a model called TBCNN. Now I want to check which nodes are the ones causing the prediction to be 1 or 0. I cannot use the method described here
def new_predict(x): data = strings_to_embeddings(x) return model.predict_proba(data)In my case x would be a list of nested dictionaries.eg.
[{'node': 'ci_root', 'children': [{'node': 'ci_class_decl', 'children': [{'node': 'ci_modifiers', 'children': [{'node': 'ci_modifier', 'children': []}]}, {'node': 'ci_type', 'children': [{'node': 'SimpleType', 'children': [{'node': 'SimpleName', 'children': []}]}]}, ... ]NOTE: This is just a part of one data point.So to sum up, I have two questions 1) Is it possible to use LIME on such data? 2) If yes, please provide some hints on how it can be done.
Thank you