CODEWITHSUNDEEP
tensorflowsoftmax
Kirnap
Kirnap

Reputation: 17

Fast softmax regression implementation in tensorflow

I am trying to implement the softmax regression model in tensorflow in order to make a benchmark with other mainstream deep-learning frameworks. The official documentation code is slow because of the feed_dict issue in tensorflow. I am trying to serve the data as tensorflow constant but I don't know the most efficient way to do that. For now I just use the single batch as constant and trained through that batch. What are the efficient solutions of making minibatched solution of that code? Here is my code:

from tensorflow.examples.tutorials.mnist import input_data

import tensorflow as tf
import numpy as np

mnist = input_data.read_data_sets(FLAGS.data_dir, one_hot=True)
batch_xs, batch_ys = mnist.train.next_batch(100)

x = tf.constant(batch_xs, name="x")
W = tf.Variable(0.1*tf.random_normal([784, 10]))
b = tf.Variable(tf.zeros([10]))
logits = tf.matmul(x, W) + b

batch_y = batch_ys.astype(np.float32)
y_ = tf.constant(batch_y, name="y_")

loss = tf.reduce_mean(tf.nn.softmax_cross_entropy_with_logits(logits, y_))
train_step = tf.train.GradientDescentOptimizer(0.5).minimize(loss)
....
# Minitbatch is never updated during that for loop
for i in range(5500):
    sess.run(train_step)

Upvotes: 0

Views: 500

Answers (1)

Dmytro Danevskyi
Dmytro Danevskyi

Reputation: 3159

Just as follows.

from tensorflow.examples.tutorials.mnist import input_data

import tensorflow as tf
import numpy as np

batch_size = 32 #any size you want

mnist = input_data.read_data_sets(FLAGS.data_dir, one_hot=True)


x = tf.placeholder(tf.float32, shape = [None, 784])
y = tf.placeholder(tf.float32, shape = [None, 10])

W = tf.Variable(0.1*tf.random_normal([784, 10]))
b = tf.Variable(tf.zeros([10]))

logits = tf.matmul(x, W) + b

loss = tf.reduce_mean(tf.nn.softmax_cross_entropy_with_logits(logits, y))
train_step = tf.train.GradientDescentOptimizer(0.5).minimize(loss)
....
# Minitbatch is never updated during that for loop
for i in range(1000):
    batch_xs, batch_ys = mnist.train.next_batch(batch_size)
    l, _ = sess.run([loss, train_step], feed_dict = {x: batch_x, y: batch_ys})
    print l #loss for every minibatch

Shape like [None, 784] allows you to feed any value of shape [?, 784].

I haven't tested this code, but I hope it would work.

Upvotes: 1

Related Questions