set recursion limit on the Machine Learning Server
I am now deployed my model in the Microsoft Machine Learning Server. However, it achieved the recursion limit (about only 150 in the Server) while processing the model. If it happens in my local, I can fix this simply by running python: sys.setrecursionlimit(150). Thus, could you please advice how to change the recursion limit in the Server? Maybe use 'az ml command?
Mary
shared this idea
2 comments
-
Mary
commented
Problem Solved: the reason for the error is that override of CountVectorizer().build_analyzer() the callable call itself result in more recursion times, which does not reach the limit of my local but dose reach the limit of the server:
Before:
class StemmedCountVectorizer(CountVectorizer):
def build_analyzer_new(self):
analyzer = super(CountVectorizer, self).build_analyzer()
...After: analyzer = CountVectorizer().build_analyzer()
Create a new callable rather than itself's solved the problem. -
Mary
commented
Make it simple: Could you please advise how to increase the 'maximum recursion depth' in the server ??? (e.g. if it happened in my local, it can be solved as simple as run python code: sys.setrecursionlimit(50000), but I don't know how to do it in the server of 'azureml.deploy.DeployClient.service')
Thank youlog contain error here:
ValueError: Error serializing clf_NB
Pipeline(steps=[('SnowbStemed_TF', StemmedCountVectorizer(analyzer='word', binary=False, decode_error='strict',
dtype=<class 'numpy.int64'>, encoding='utf-8', input='content',
lowercase=True, max_df=1.0, max_features=None, min_df=1,
ngram_range=(1, 1), preprocessor=None, s...e, use_idf=True)), ('NB_Classifier', MultinomialNB(alpha=0.0001, class_prior=None, fit_prior=True))])
RecursionError
maximum recursion depth exceeded