:param k: number of nearest neighbours :param pool: optional pooling function (e.g. np.mean or np.max) :return: KSG similarity with given

Computes Hilbert-Schmidt independence criterion (HSIC) between word embedding matrices X and Y :param X: word embedding matri

APSynP similarity measure (Santus et al., 2018) applied to sentences represented as averaged word vectors :param x: list of word embeddin

Builds a Centered Kernel Alignment (CKA) similarity function with the specified kernel :param kernel: kernel function for CKA :return

Returns a centering matrix of dimension d :param d: dimension of the matrix :return: centering matrix of dimension d

Computes a linear kernel for X :param X: word embedding matrix X with shape (k x D) :return: linear kernel for X

Cosine similarity between two avg. word vectors. :param x: list of word embeddings for the first sentence :param y: list of word embeddin

Computes Centered Kernel Alignment (CKA) between word embedding matrices X and Y :param X: word embedding matrix X with shape

Computes Distance Correlation (dCorr) between word embedding matrices X and Y :param x: X: word embedding matrix X with shape (k x D)

Computes a Gaussian kernel for X :param X: word embedding matrix X with shape (k x D) :param sigma: standard deviation of the kernel

Loads words and word vectors from a text file :param path_to_vec: path to word vector file in word2vec format :param word2id: words to lo

Kendall correlation coefficient between two sentences represented as averaged word vectors :param x: list of word embeddings for the firs

Kraskov–Stogbauer–Grassberger (KSG) estimator of mutual information between two sentences represented as word embedding matrices x an

Spearman correlation coefficient between two sentences represented as max-pooled word vectors :param x: list of word embeddings for the f

Pearson correlation coefficient between two sentences represented as averaged word vectors :param x: list of word embeddings for the firs

Spearman correlation coefficient between two sentences represented as averaged word vectors :param x: list of word embeddings for the fir