How to use the implicit.nearest_neighbours.ItemItemRecommender function in implicit

class CosineRecommender(ItemItemRecommender):
    """ An Item-Item Recommender on Cosine distances between items """
    def fit(self, counts, show_progress=True):
        # cosine distance is just the dot-product of a normalized matrix
        ItemItemRecommender.fit(self, normalize(counts), show_progress)


class TFIDFRecommender(ItemItemRecommender):
    """ An Item-Item Recommender on TF-IDF distances between items """
    def fit(self, counts, show_progress=True):
        weighted = normalize(tfidf_weight(counts))
        ItemItemRecommender.fit(self, weighted, show_progress)


class BM25Recommender(ItemItemRecommender):
    """ An Item-Item Recommender on BM25 distance between items """
    def __init__(self, K=20, K1=1.2, B=.75, num_threads=0):
        super(BM25Recommender, self).__init__(K, num_threads)
        self.K1 = K1
        self.B = B

    def fit(self, counts, show_progress=True):
        weighted = bm25_weight(counts, self.K1, self.B)
        ItemItemRecommender.fit(self, weighted, show_progress)


def tfidf_weight(X):
    """ Weights a Sparse Matrix by TF-IDF Weighted """
    X = coo_matrix(X)

    # calculate IDF

ret = cls()
        ret.similarity = similarity
        ret.scorer = NearestNeighboursScorer(similarity)
        ret.K = m['K']
        return ret


class CosineRecommender(ItemItemRecommender):
    """ An Item-Item Recommender on Cosine distances between items """
    def fit(self, counts, show_progress=True):
        # cosine distance is just the dot-product of a normalized matrix
        ItemItemRecommender.fit(self, normalize(counts), show_progress)


class TFIDFRecommender(ItemItemRecommender):
    """ An Item-Item Recommender on TF-IDF distances between items """
    def fit(self, counts, show_progress=True):
        weighted = normalize(tfidf_weight(counts))
        ItemItemRecommender.fit(self, weighted, show_progress)


class BM25Recommender(ItemItemRecommender):
    """ An Item-Item Recommender on BM25 distance between items """
    def __init__(self, K=20, K1=1.2, B=.75, num_threads=0):
        super(BM25Recommender, self).__init__(K, num_threads)
        self.K1 = K1
        self.B = B

    def fit(self, counts, show_progress=True):
        weighted = bm25_weight(counts, self.K1, self.B)
        ItemItemRecommender.fit(self, weighted, show_progress)

def load(cls, filename):
        # numpy.save automatically appends a npz suffic, numpy.load doesn't apparently
        if not filename.endswith(".npz"):
            filename = filename + ".npz"

        m = numpy.load(filename)
        similarity = csr_matrix((m['data'], m['indices'], m['indptr']), shape=m['shape'])

        ret = cls()
        ret.similarity = similarity
        ret.scorer = NearestNeighboursScorer(similarity)
        ret.K = m['K']
        return ret


class CosineRecommender(ItemItemRecommender):
    """ An Item-Item Recommender on Cosine distances between items """
    def fit(self, counts, show_progress=True):
        # cosine distance is just the dot-product of a normalized matrix
        ItemItemRecommender.fit(self, normalize(counts), show_progress)


class TFIDFRecommender(ItemItemRecommender):
    """ An Item-Item Recommender on TF-IDF distances between items """
    def fit(self, counts, show_progress=True):
        weighted = normalize(tfidf_weight(counts))
        ItemItemRecommender.fit(self, weighted, show_progress)


class BM25Recommender(ItemItemRecommender):
    """ An Item-Item Recommender on BM25 distance between items """
    def __init__(self, K=20, K1=1.2, B=.75, num_threads=0):