LSTM-Autoencoder icon indicating copy to clipboard operation
LSTM-Autoencoder copied to clipboard

Error when running

Open yijinhua opened this issue 8 years ago • 1 comments

Hi Cheng,

I am testing out implementations of auto-encoders for text but am getting error messages when I tried to run your code from the command line. It says: /Users/yhua/torch/install/bin/luajit: ./util/BatchLoaderC.lua:98: attempt to index local 'f' (a nil value) stack traceback: ./util/BatchLoaderC.lua:98: in function 'text_to_tensor' ./util/BatchLoaderC.lua:22: in function 'create' LSTMAutoencoder.lua:61: in main chunk [C]: in function 'dofile' ...yhua/torch/install/lib/luarocks/rocks/trepl/scm-1/bin/th:150: in main chunk [C]: at 0x010b5cfbd0

Just wondering if I need to be loading in my own data set and if so how can I get the corpus? Thanks

yijinhua avatar Mar 06 '17 23:03 yijinhua

I have the same issue... I suppose that following parameters in command line are required, see LSTMAutoencoder.lua : cmd:text('Options') cmd:option('-data_dir', 'corpus', 'path of the dataset') cmd:option('-sample_dir', 'samples', 'path of the sampled sentences') Corpus probably you can get from: http://yknzhu.wixsite.com/mbweb

mglowacki100 avatar Mar 10 '18 10:03 mglowacki100