13:40 < dan_dan> rcurtin: one last question regarding the Python bindings before Christmas break. I am trying to use hmm_train and I can't seem to find the type of input I need to give it. I gave it an np.ndarray and a string with the filename of my data and got 'expected bytes, found str/np.ndarray/etc". Whenever I pass it the string encoded as bytes, it gives a "Fatal Error". Checking the line, it seems to be on a function that accepts a Python strin

18:48 < rcurtin> dan_dan: I haven't perfected the input for the hmm_train python binding yet

18:49 < rcurtin> right now you should pass in a filename, and this filename should either contain a list of other filenames (if you have many training sequences), in which case you should also give batch=True

18:49 < rcurtin> or the file should just contain the data you are training on

18:50 < rcurtin> let me run a quick test on it

18:51 < rcurtin> >>> hmm_train(input_file='build/test_data_3_1000.csv', type='gaussian', states=5, verbose=True)

18:51 < rcurtin> that works for me (although the HMM it trains is kind of meaningless, since that's just some random data file)

18:52 < rcurtin> and if I create a file called 'build/files.csv' which just contains three lines, all of which are 'build/test_data_3_1000.csv', this works:

18:52 < rcurtin> >>> hmm_train(input_file='build/files.csv', batch=True, type='gaussian', states=5, verbose=True)

20:13 < saurav> hii, i would like to contribute mlpack can you tell me where to start from ?