read courpus must be doing something wrong

[masterchop]

If i tried to train my own data it fails, I follow the instructions build the folders with the TXT files in it. I ran the script to rename all the files as explained in the DOC. your data set works but mine just wont trained with the error of bad shape.

The TXT files are used in another code with Keras and works just fine. I wonder why its not working with Low Level API. Another thing to notice it that i complained about the unicode you had like latin-1 i remove that parameter or force it to utf8 but the problem persisted.

numFeatures = trainX.shape[1] IndexError: tuple index out of range

[MamtaBhagia]

@masterchop I had a similar issue - the delimiter used to convert to csv in read_corpus is "," while in logistic_regression_train, the delimiter used to covert the csv to numpy is "\t"

So, your training data created using read corpus when imported in logistic_regression_train.py is imported like it has many rows and just 1 column, i.e shape = (478,)