-
It might be better introducing a bit more generic fix, like .close() method in an AbstractEventStream, so the resources the stream allocated during creation will be released explicitly upon .close() method call.
2009-10-02 14:17:40 UTC by autayeu
-
After training, temporary files like events37537375696294525.tmp remain in the temporary directory.
AFAIK, the problem
TwoPassDataIndexer -> open reader in the FileEventStream -> reader remains open -> file.delete() does not work.
A possible solution might be - remove deleteOnExit() due unreliabilty and memory consumption, add explicit .close() call:
Left base folder...
2009-10-02 12:21:26 UTC by autayeu
-
Hi,
take a look at this thread:
https://sourceforge.net/projects/maxent/forums/forum/18385/topic/1925312
Also note that a feature with a value of "0" is ignored by the model.
Hope this helps...Tom.
2009-09-20 13:26:41 UTC by tsmorton
-
I tried to learn how to use this package these days via run the examples firstly. Surprisingly, I found that the results of the examples are not right.
I do the testing using the training data, the testing results are very bad. I also tried my own data, which have an abvious difference among classes, and the results are also bad.
I am wondering if there's something wrong with the examples...
2009-09-20 12:21:35 UTC by yangenid
-
I have train a ME model using the data:
sim=1 chain=0 b
sim=1.1 chain=0.1 b
sim=1.0 chain=0.1 b
sim=1.1 chain=0 b
sim=0 chain=1 nb
sim=0.1 chain=1.1 nb
sim=0 chain=1.1 nb
sim=0.1 chain=1.0 nb
and then use the same data without "b" or "nb" to test, the results are:
For context: sim=1
b[0.9990] nb[0.0010]
For context: sim=1.1
b[0.9995]...
2009-09-19 09:18:47 UTC by yangenid
-
Hi,
The features you have will work as is with the RealValueFileEventStream. I would recommend however that you change your non-real-valued feature to a slightly different format. As they are the code will try and turn true and false into a float, generate an exception, catch that exception and treat them as binary features. With the "=" sign if won't try to make a float out of...
2009-09-06 20:04:00 UTC by tsmorton
-
hi, Tom:
My event are like this: "time=2.5 sim=0.6 mus=true sil=false".
It seems I can not employ this event directly to train the model.
Do you mean that I can change this event into the form like "time=2.5 sim=0.6 mus", and if there is anotoher event with both 'mus' and 'sil' true, the event will be changed to be "time=2.5 sim=0.6 mus sil"?
Looking...
2009-09-06 13:39:43 UTC by yangenid
-
Hi, everybody.
According the introduction of maxent-2.5.2, it could process both binary and real data, via the option of -real.
but If I have both binary and real data in the same event, how could I use them? or do I have to transfer the binary data into real or the real data to binary?
Your answers will be greatly appreciated, thanks a lot.
2009-09-06 13:25:01 UTC by yangenid
-
Hi,
Yes they can. I'm not sure how you are creating your events but in RealValueFileEventStream binary events look like "name=1.2". Binary event don't have a value at the end. A binary event is just like a real-valued event with a value of 1.0.
Hope this helps...Tom.
2009-09-06 10:52:47 UTC by tsmorton
-
Hi, everybody.
According the introduction of maxent-2.5.2, it could process both binary and real data, via the option of -real.
but If I have both binary and real data in the same event, how could I use them? or do I have to transfer the binary data into real or the real data to binary?
Your answers will be greatly appreciated, thanks a lot.
2009-09-06 10:04:13 UTC by yangenid