wabbit_wappa
wabbit_wappa copied to clipboard
mokelly
Wabbit Wappa is a full-featured Python wrapper for the Vowpal Wabbit machine learning utility.
This PR primarily allows connections to a deamonized vw if an ip and port are only given. The current repo also has an older name for the active learning parameter,...
The timing (e.g. https://github.com/robinsonkwame/wabbit_wappa/blob/master/examples/capitalization_demo.py#L111-L117) in the example files could be improved (maybe by even an order of magnitude) by consolidating multiple examples and transmitting as one packet; here the examples...
I am looking to have a wrapper that essentially places my existing numpy 2D arrays and unicode vector labels into the VW format and allows me to perform fast cost...
Hello, My setup: Linux Mint: 17.2 Python: 3.4.3 VW: 7.3.0 I'm getting the traceback below when I run the tests. I'm using the Linux repo version of VW (7.3.0), so...
Traceback (most recent call last): File "test_wabbit_wappa.py", line 146, in training() File "test_wabbit_wappa.py", line 73, in training ('b', -1 - random.random())] File "/Library/Frameworks/Python.framework/Versions/3.4/lib/python3.4/site-packages/wabbit_wappa/**init**.py", line 290, in send_example result = self.send_line(line,...
Can a release be done and a push to pypi so we get the fix in setup.py for session and can therefore install from latest pip, otherwise I get root@28781e790b3c:/#...
Addresses https://github.com/mokelly/wabbit_wappa/issues/3: --`VW()` can connect to existing VW daemons --`daemon_mode` can be used to launch a daemon VW instance and connect automatically
I started a bugfix branch with a failing unit test for this issue: https://github.com/mokelly/wabbit_wappa/tree/bugfix/long-examples As a workaround for now, `active_mode` supports much longer lines without difficulty.
install_requires should be the minimal set of packages needed for installation; as far as I read the code correctly, it should only contain pexpect.
