serial-recall task

handwriting/logs/train_scribe.txt

@@ -4,4 +4,4 @@ Scribe: Realistic Handriting in Tensorflow
 
 	loaded dataset:
 		11895 individual data points
-		5947 batches
+		743 batches

handwriting/models/checkpoint

@@ -0,0 +1,6 @@
+model_checkpoint_path: "model.ckpt-34000"
+all_model_checkpoint_paths: "model.ckpt-30000"
+all_model_checkpoint_paths: "model.ckpt-31000"
+all_model_checkpoint_paths: "model.ckpt-32000"
+all_model_checkpoint_paths: "model.ckpt-33000"
+all_model_checkpoint_paths: "model.ckpt-34000"

handwriting/rnn_controller.py

@@ -4,13 +4,13 @@ from controller import Controller
 from tensorflow.contrib.rnn.python.ops.core_rnn_cell import LSTMStateTuple
 
 """
-A 1-Layer recurrent neural network (LSTM) with 64 hidden nodes
+RNN (cell type LSTM) with 128 hidden layers
 """
 
 class RNNController(Controller):
 
     def init_controller_params(self):
-        self.rnn_dim = 150
+        self.rnn_dim = 300
         self.lstm_cell = tf.contrib.rnn.core_rnn_cell.BasicLSTMCell(self.rnn_dim)
         self.state = tf.Variable(tf.zeros([self.batch_size, self.rnn_dim]), trainable=False)
         self.output = tf.Variable(tf.zeros([self.batch_size, self.rnn_dim]), trainable=False)

handwriting/train.py

@@ -15,11 +15,11 @@ tf.app.flags.DEFINE_integer("ylen", 3, "output dimension")
 tf.app.flags.DEFINE_integer("stroke_steps", stroke_steps, "Number of time steps for stroke data")
 tf.app.flags.DEFINE_integer("ascii_steps", stroke_steps/25, "Sequence length")
 tf.app.flags.DEFINE_integer("data_scale", 50, "How to scale stroke data")
-tf.app.flags.DEFINE_integer("batch_size", 2, "Size of batch in minibatch gradient descent")
+tf.app.flags.DEFINE_integer("batch_size", 16, "Size of batch in minibatch gradient descent")
 
 tf.app.flags.DEFINE_integer("R", 1, "Number of DNC read heads")
-tf.app.flags.DEFINE_integer("W", 16, "Word length for DNC memory")
-tf.app.flags.DEFINE_integer("N", 10, "Number of words the DNC memory can store")
+tf.app.flags.DEFINE_integer("W", 100, "Word length for DNC memory")
+tf.app.flags.DEFINE_integer("N", 8, "Number of words the DNC memory can store")
 
 tf.app.flags.DEFINE_integer("train", True, "Train or sample???")
 tf.app.flags.DEFINE_integer("print_every", 100, "Print training info after this number of train steps")

repeat-copy/rnn_controller.py

@@ -4,13 +4,13 @@ from controller import Controller
 from tensorflow.contrib.rnn.python.ops.core_rnn_cell import LSTMStateTuple
 
 """
-A 1-Layer recurrent neural network (LSTM) with 64 hidden nodes
+RNN (cell type LSTM) with 128 hidden layers
 """
 
 class RNNController(Controller):
 
     def init_controller_params(self):
-        self.rnn_dim = 64
+        self.rnn_dim = 128
         self.lstm_cell = tf.contrib.rnn.core_rnn_cell.BasicLSTMCell(self.rnn_dim)
         self.state = tf.Variable(tf.zeros([self.batch_size, self.rnn_dim]), trainable=False)
         self.output = tf.Variable(tf.zeros([self.batch_size, self.rnn_dim]), trainable=False)

repeat-copy/rnn_models/checkpoint

@@ -1,4 +1,4 @@
-model_checkpoint_path: "model.ckpt-10000"
-all_model_checkpoint_paths: "model.ckpt-8000"
-all_model_checkpoint_paths: "model.ckpt-9000"
-all_model_checkpoint_paths: "model.ckpt-10000"
+model_checkpoint_path: "model.ckpt-6000"
+all_model_checkpoint_paths: "model.ckpt-4000"
+all_model_checkpoint_paths: "model.ckpt-5000"
+all_model_checkpoint_paths: "model.ckpt-6000"

serial-recall/.ipynb_checkpoints/debug-checkpoint.ipynb

@@ -0,0 +1,6 @@
+{
+ "cells": [],
+ "metadata": {},
+ "nbformat": 4,
+ "nbformat_minor": 1
+}

serial-recall/rnn_controller.py

@@ -0,0 +1,26 @@
+import numpy as np
+import tensorflow as tf
+from controller import Controller
+from tensorflow.contrib.rnn.python.ops.core_rnn_cell import LSTMStateTuple
+
+"""
+RNN (cell type LSTM) with 128 hidden units
+"""
+
+class RNNController(Controller):
+
+    def init_controller_params(self):
+        self.rnn_dim = 128
+        self.lstm_cell = tf.contrib.rnn.core_rnn_cell.BasicLSTMCell(self.rnn_dim)
+        self.state = tf.Variable(tf.zeros([self.batch_size, self.rnn_dim]), trainable=False)
+        self.output = tf.Variable(tf.zeros([self.batch_size, self.rnn_dim]), trainable=False)
+
+    def nn_step(self, X, state):
+        X = tf.convert_to_tensor(X)
+        return self.lstm_cell(X, state)
+
+    def update_state(self, update):
+        return tf.group(self.output.assign(update[0]), self.state.assign(update[1]))
+
+    def get_state(self):
+        return LSTMStateTuple(self.output, self.state)

serial-recall/rnn_models/checkpoint

@@ -0,0 +1,4 @@
+model_checkpoint_path: "model.ckpt-10000"
+all_model_checkpoint_paths: "model.ckpt-8000"
+all_model_checkpoint_paths: "model.ckpt-9000"
+all_model_checkpoint_paths: "model.ckpt-10000"