search

karpathy / char-rnn

Multi-layer Recurrent Neural Networks (LSTM, GRU, RNN) for character-level language models in Torch
11.52k stars 2.58k forks source link

Sampling Text #217

Open invaderxyz opened 6 years ago

invaderxyz commented 6 years ago

So, in order to sample something, you need to put: $ th sample.lua cv/lm_lstm_epoch0.02_3.3272.t7 -gpuid -1

But... where, exactly? Every time I run sample.lua in ZeroBraneStudio; it starts and ends the program before I get the chance to input anything. Am I doing something wrong?

(Keep in mind, I am completely new to this; I just wanted to mess around with one of these...)

Zemaaan commented 4 years ago

First you need to train the network using train.lua file.

Then you use terminal on Linux, and you type in your command and press enter.

invaderxyz commented 4 years ago

Ah, so this is Linux only then? I managed to get it running on Windows (though on my old computer since this was two years ago :p) but I couldn't figure out how to perform the Windows equivalent.

Zemaaan commented 4 years ago

Well it's certainly easier to make it run on Linux than on Windows.

You should also be aware that you cant compile Torch on CUDA10 since Torch doesen't support it, so you will have to use CUDA 9.