Python 2.7.6 with Numpy 1.8.1, running on Windows 7:AMD64 in 32bit mode, crashes when I try to reduce a Pattern model.
Little or big dataset pruned in document building with thresholds and stopwords, or not at all.
The documents and model apparently build just fine on an otherwise stable machine as I can examine their content..
The Trace file is huge of course and ends with:
That certainly looks like it's a data driven memory leak, iterating through a vector (735, 736, 737) then back to the 'for loop' at 735 where it abruptly fails.
Any quick ideas about my configuration?
I used easy_install for both Numpy and Pattern.