vseledkin

import numpy as np

# vectors is an array of vectors of any length
vectors = np.array(vectors, dtype=np.float32)
print('normalize %s vectors' % len(vectors))
vectors = vectors / np.linalg.norm(vectors, ord=2, axis=1, keepdims=True)

vseledkin

import pickle as pk

with open('filename.pk', 'rb') as filehandler:
  obj = pk.load(filehandler)

vseledkin

def knn_cosine_search(x, D, K):
    """ find K nearest neighbours of data among D """
    ndata = len(D)
    K = K if K < ndata else ndata
    # euclidean distances from the other points
    x = x.reshape(1, -1)
    cos = cdist(D, x, 'cosine').reshape(-1)
    idx = np.argsort(cos)  # sorting
    # return the indexes of K nearest neighbours

vseledkin

def knn_euclidean_search(x, D, K):
    """ find K nearest neighbours of data among D """
    ndata = len(D)
    K = K if K < ndata else ndata
    # euclidean distances from the other points
    sqd = np.sqrt(((D - x)**2).sum(axis=1))
    idx = np.argsort(sqd)  # sorting
    # return the indexes of K nearest neighbours
    idx = idx[:K]
    d = sqd[idx]

vseledkin

x#! /bin/bash
# use XCode 8.2.1
export TF_TYPE=”cpu”
export TF_NEED_GCP=0
export TF_NEED_HDFS=0
export TF_NEED_AWS=0
export TF_NEED_KAFKA=0
export TF_ENABLE_XLA=1
export TF_NEED_GDR=0
export TF_NEED_VERBS=0

vseledkin

Build tensorflow on OSX with NVIDIA CUDA support (GPU acceleration)

These instructions are based on Mistobaan's gist but expanded and updated to work with the latest tensorflow OSX CUDA PR.

Requirements

OS X 10.10 (Yosemite) or newer

vseledkin

How to compile TensorFlow 1.6-rc1 on macOS High Sierra 10.13.3 with GPU acceleration and optionally eGPU support.

source: https://byai.io/howto-tensorflow-1-6-on-mac-with-gpu-acceleration/

Requirements

Caution - you have to disable SIP (System Integrity Protection)

• NVIDIA Web-Drivers
• CUDA-Drivers

vseledkin

Keybase proof

I hereby claim:

• I am vseledkin on github.
• I am apfelson (https://keybase.io/apfelson) on keybase.
• I have a public key ASDIO3bGsJnLc8uayvwOjez7HcHNdbVDjDRGKhmGdbTKrgo

To claim this, I am signing this object:

vseledkin

func substitute(text, before, after string, positions [][]int) string {
	// flatten
	flatten_positions := make([]int, 2*len(positions))
	for i, pos := range positions {
		flatten_positions[2*i] = pos[0]
		flatten_positions[2*i+1] = pos[1]
	}
	sort.Ints(flatten_positions)

	var count int

vseledkin

func SplitToNGrams(s string, n []int) []string {
	words := strings.Fields(s)
	var tokens []string
	for cursor, _ := range words {
		for _, ni := range n {
			// get suffix of length ni in words back from ni position in words array
			if cursor+ni <= len(words) {
				tokens = append(tokens, strings.Join(words[cursor:cursor+ni], "•"))
			}
		}