model files

levent

Hi,

I have an experiment that I have done with 4.01 beta2. I have no problems with the previous version but I have upgraded my old computer and now I cannot read old model files. Is there a solution to this?

My second question is reading weka ml perceptron structure from a model file. I would like to know how many hidden layers were used in my model (I had a parameter iterator so I am not sure what was the value saved into model file).

Thank you in advance !!
Levent

levent

I found my solution, I checked my experiment files and found that I have written the model as xml zipped.
Renamed model as zip, unzipped, renamed file as xml and opened it with explorer.

Then I searched for hidden keyword and found hidden layer node number inside. Just in case you ever need it, it seems a wise idea not to write models as binaries as this saved a lot of trouble for me.

Levent

land

Hi,
the 4.01 beta2 is now over a year old. Many things have changed since then and even the models might differ in some cases. So loading old models might not work. But if you have the processes, which generated the models the first time, you should be able to load them and recompute the model in a new, probably improved way

By the way: Saving Models in the binary format is faster and smaller but even small changes in the models over between two versions result in incompatible files. The XML way is much more robust, although it gets really big if not compressed. So you should always use the XML-Zipped format...

Greetings,
  Sebastian

levent

Thank you. I had the experiments for my thesis work and needed to reanalyze the results. I will use the newest version definitely when I continue.