Difference between revisions of "RNNs with Darknet"

From Interaction Station Wiki
Jump to navigation Jump to search
Line 11: Line 11:
  
 
*Generating new text
 
*Generating new text
cd darknet
 
 
  ./darknet rnn generate cfg/rnn.cfg shakespeare.weights -srand 0
 
  ./darknet rnn generate cfg/rnn.cfg shakespeare.weights -srand 0
  
Line 19: Line 18:
  
 
*Training with other corpus of text
 
*Training with other corpus of text
*We can modify file cfg/rnn.train.cfg file  (batch = 50 is the value we set by default, the original value was 256)
 
  
*Some online corpus of text:
+
*Download a text file. Some corpus of text online:
 
  George Meredith
 
  George Meredith
 
  http://www.gutenberg.org/ebooks/4500
 
  http://www.gutenberg.org/ebooks/4500
Line 28: Line 26:
 
  More
 
  More
 
  http://www.gutenberg.org/ebooks/search/?query=The+Complete+Works+of
 
  http://www.gutenberg.org/ebooks/search/?query=The+Complete+Works+of
 +
 +
*We train the neural network with the new text (This might take a while):
 +
./darknet rnn train cfg/rnn.train.cfg -file filename.txt
 +
 +
 +
*Generating new text
 +
./darknet rnn generate cfg/rnn.cfg mark.weights -srand 0
 +
 +
 +
*Optional: We can modify file cfg/rnn.train.cfg file  (batch = 50 is the value we set by default, the original value was 256)
  
  
 
*More info:
 
*More info:
 
*https://pjreddie.com/darknet/rnns-in-darknet/
 
*https://pjreddie.com/darknet/rnns-in-darknet/

Revision as of 21:15, 7 January 2019

Installing Darknet

Using RNNs with Darknet

  • Getting some calculated weights from several writers:
wget https://pjreddie.com/media/files/shakespeare.weights
wget https://pjreddie.com/media/files/kant.weights
wget https://pjreddie.com/media/files/tolstoy.weights
  • Generating new text
./darknet rnn generate cfg/rnn.cfg shakespeare.weights -srand 0
  • Parameters:
-srand N -> N determines the random seed


  • Training with other corpus of text
  • Download a text file. Some corpus of text online:
George Meredith
http://www.gutenberg.org/ebooks/4500
Mark Twain
http://www.gutenberg.org/ebooks/3200
More
http://www.gutenberg.org/ebooks/search/?query=The+Complete+Works+of
  • We train the neural network with the new text (This might take a while):
./darknet rnn train cfg/rnn.train.cfg -file filename.txt


  • Generating new text
./darknet rnn generate cfg/rnn.cfg mark.weights -srand 0


  • Optional: We can modify file cfg/rnn.train.cfg file (batch = 50 is the value we set by default, the original value was 256)