NLP篇【02补】TensorFlow 代码实现负采样（coding）

一、前言

在NLP篇【02】白话Word2vec原理以及层softmax、负采样的实现 文章中介绍了层softmax与负采样，有兴趣可以根据介绍，自己实现一下负采样或层softmax。当然，对于只想拿来直接使用的，可以看看这篇文章，直接利用TensorFlow 的api实现负采样，也是非常的简单。

二、TensorFlow 代码

api介绍：

tf.nn.sampled_softmax_loss(weights, # Shape (num_classes, dim)     - floatXX
                     biases,        # Shape (num_classes)          - floatXX 
                     labels,        # Shape (batch_size, num_true) - int64
                     inputs,        # Shape (batch_size, dim)      - floatXX  
                     num_sampled,   # - int
                     num_classes,   # - int
                     num_true=1,  
                     sampled_values=None,
                     remove_accidental_hits=True,
                     partition_strategy="mod",
                     name="sampled_softmax_loss")

具体代码： 

import tensorflow as tf

# Network Parameters
n_hidden_1 = 256  # 1st layer number of features
n_input = 784     # MNIST data input (img shape: 28*28)
n_classes = 10    # MNIST total classes (0-9 digits)

# Dependent & Independent Variable Placeholders
x = tf.placeholder("float", [None, n_input])
y = tf.placeholder("float", [None, n_classes]) #

# Weights and Biases
weights = {
    'h1': tf.Variable(tf.random_normal([n_input, n_hidden_1])),
    'out': tf.Variable(tf.random_normal([n_hidden_1, n_classes]))
}
biases = {
    'b1': tf.Variable(tf.random_normal([n_hidden_1])),
    'out': tf.Variable(tf.random_normal([n_classes]))
}

# Super simple model builder
def tiny_perceptron(x, weights, biases):
    layer_1 = tf.add(tf.matmul(x, weights['h1']), biases['b1'])
    out_layer = tf.nn.relu(layer_1)
    # out_layer = tf.matmul(layer_1, weights['out']) + biases['out']
    return out_layer  #(bt,dim)

# Create the model
pred = tiny_perceptron(x, weights, biases)

# Set up loss function inputs and inspect their shapes
w = tf.transpose(weights['out'])  #(num_class,dim)
b = biases['out']
labels = tf.reshape(tf.argmax(y, 1), [-1,1])
inputs = pred
num_sampled = 3
num_true = 1
num_classes = n_classes

print('Shapes\n------\nw:\t%s\nb:\t%s\nlabels:\t%s\ninputs:\t%s' % (w.shape, b.shape, labels.shape, inputs.shape))
# Shapes
# ------
# w:      (10, 256)  # Requires (num_classes, dim)     - CORRECT
# b:      (10,)      # Requires (num_classes)          - CORRECT
# labels: (?, 1)     # Requires (batch_size, num_true) - CORRECT
# inputs: (?, 256)    # Requires (batch_size, dim)     - CORRECT

loss_function = tf.reduce_mean(tf.nn.sampled_softmax_loss(
                     weights=w,         #(num_class,dim)
                     biases=b,
                     labels=labels,
                     inputs=inputs,     #(bt,dim)
                     num_sampled=num_sampled,
                     num_true=num_true,
                     num_classes=num_classes)
)

注意：这里的labels如果是one-hot类型编码，需要labels=tf.reshape(tf.argmax(labels_one_hot, 1), [-1,1])

示例：

import tensorflow as tf

a = tf.nn.sampled_softmax_loss(weights=tf.constant([[1, 2], [3, 4], [5, 6], [7, 8], [9, 10], [11, 12], [13, 14], [15, 16], [17, 18], [19, 20]]),
                               # [num_classes, dim] = [10, 2]
                               biases=tf.constant([0., 0.1, 0.2, 0.3, 0.4, 0.5, 0.6, 0.7, 0.8, 0.9]),
                               # [num_classes] = [10]
                               labels=tf.constant([[2], [3], [5]]),
                               # [batch_size, num_true] = [3, 1]
                               inputs=tf.constant([[0.2, 0.1], [0.4, 0.1], [0.22, 0.12]]),
                               # [batch_size, dim] = [3, 2]
                               
                               num_sampled=3,
                               num_classes=10,
                               num_true = 1,
                               seed = 2020,
                               name = "sampled_softmax_loss"
                          )

结果：

with tf.compat.v1.Session() as sess:
    print(sess.run(a))

# [0.37478584 0.2859666  0.0703702 ]

在源码的介绍中，train的方式和eval的方式是分开的：

if mode == "train":
    loss = tf.nn.sampled_softmax_loss(
        weights=weights,
        biases=biases,
        labels=labels,
        inputs=inputs,
        ...)

elif mode == "eval":
    logits = tf.matmul(inputs, tf.transpose(weights))
    logits = tf.nn.bias_add(logits, biases)
    labels_one_hot = tf.one_hot(labels, n_classes)
    loss = tf.nn.softmax_cross_entropy_with_logits(
        labels=labels_one_hot,
        logits=logits)

 train步骤中除了选择一批label作为负样本外，实现方式与eval中描述的一致。