Reproducability of the classifier with all the parameters exported and reassigned #2574
Replies: 3 comments 2 replies
-
|
Hi @Akash08naik I'm converting this issue into a Discussion (under the Discussions tab above). |
Beta Was this translation helpful? Give feedback.
-
|
@Akash08naik Do you have a short code example of your use case? |
Beta Was this translation helpful? Give feedback.
-
|
The actual issue is after building the classifier for example take pytorchclassifier itself so it contains various attributes such as model,loss, optimizer , preprocessing defences etc. so my issue number one is 1). I am not able to save this classifier. For loading it back again whenver useful. 2) So to save the attributes of classifier and rebuild it again i am saving the attributes of the classifier into a yaml file and then when required creating a new instance of the classifier and reassigning the saved attributes . But when i do this i am not able to get the same results as produced by the original classifier. And when comparing both the classifiers also it is returning both are not same But i did not change anything. |
Beta Was this translation helpful? Give feedback.
-
|
Hello, nothing after that i did not find any solution for the question? |
Beta Was this translation helpful? Give feedback.
-
Once a classifier is build and then performed attack with it then the process is ended. But to use this classifier again somewhere we need to do the whole process of building again . So how to conserve or serialize the above classifier which we get during the runtime. And reproduce the same results.
Beta Was this translation helpful? Give feedback.
All reactions