utils.py

# -*- coding: utf-8 -*-
"""Utils.ipynb

Automatically generated by Colaboratory.

Original file is located at
    https://colab.research.google.com/drive/1pCbkGljQ9r5BJbrp8KlmAOrkvmzlDQ4c
"""

from __future__ import division
from collections import Counter, defaultdict
import os
from random import shuffle
import tensorflow as tf

def _context_windows(region, left_size, right_size):
    for i, word in enumerate(region):
        start_index = i - left_size
        end_index = i + right_size
        left_context = _window(region, start_index, i - 1)
        right_context = _window(region, i + 1, end_index)
        yield (left_context, word, right_context)


def _window(region, start_index, end_index):
    """
    Returns the list of words starting from `start_index`, going to `end_index`
    taken from region. If `start_index` is a negative number, or if `end_index`
    is greater than the index of the last word in region, this function will pad
    its return value with `NULL_WORD`.
    """
    last_index = len(region) + 1
    selected_tokens = region[max(start_index, 0):min(end_index, last_index) + 1]
    return selected_tokens


def _device_for_node(n):
    if n.type == "MatMul":
        return "/gpu:0"
    else:
        return "/cpu:0"


def _batchify(batch_size, *sequences):
    for i in range(0, len(sequences[0]), batch_size):
        yield tuple(sequence[i:i+batch_size] for sequence in sequences)


def _plot_with_labels(low_dim_embs, labels, path, size):
    import matplotlib.pyplot as plt
    assert low_dim_embs.shape[0] >= len(labels), "More labels than embeddings"
    figure = plt.figure(figsize=size)  # in inches
    for i, label in enumerate(labels):
        x, y = low_dim_embs[i, :]
        plt.scatter(x, y)
        plt.annotate(label, xy=(x, y), xytext=(5, 2), textcoords='offset points', ha='right',
                     va='bottom')
    if path is not None:
        figure.savefig(path)
        plt.close(figure)