search

mlpack / benchmarks

Machine Learning Benchmark Scripts
101 stars 49 forks source link

Weka not being called correctly #75

Closed gut closed 7 years ago

gut commented 7 years ago

After installing everything I noticed that the weka version downloaded was 3.8.1 (libraries/weka/README).

When I run the suite as make run BLOCK=weka METHODBLOCK=KMEANS, I get the following error:

[FATAL] Can't parse the data: wrong format

Deeper check was narrowed down to this explicit call:

$ java -classpath /home/gut/benchmarks/libraries/share//weka.jar:methods/weka KMeans -i d -c 6
This program performs K-Means clustering on the given dataset.

Required options:
-c [int]         Number of clusters to find.
-i [string]      Input dataset to perform clustering on.

Options:
-m [int]         Maximum number of iterations before K-Means
                 terminates. Default value 1000.
-s [int]         Random seed.

And it didn't work because the regex \n .*?total_time: (?P<total_time>.*?)s.*?\n didn't match. (In fact weka didn't run, it just exits by printing the "help" output as showed above)

Did I miss something?

zoq commented 7 years ago

Hello thanks for the really helpful error report, it's fixed in 81346dfe5f73e7f9a6a5724 and is basically the same issue as you fixed in #76. It looks like the issue was part of the recent install update. Thanks again for the report.