Demo.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "from training_function import *"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "name_loss:\n",
      "logistic-regression-sum-loss\n",
      "Model created.\n",
      "Begin laoding data...\n",
      "Extracting data/MNIST_data/train-images-idx3-ubyte.gz\n",
      "Extracting data/MNIST_data/t10k-images-idx3-ubyte.gz\n",
      "Begin training...\n",
      "Learning rate: 0.30000\n",
      "Epoch-1.000\n",
      "Training loss: 169.018\n",
      "Training accuracy: 0.051\n",
      "Testing loss: 167.848\n",
      "Testing accuracy: 0.050\n",
      "\n",
      "\n",
      "Learning rate: 0.30000\n",
      "Epoch-2.000\n",
      "Training loss: 109.642\n",
      "Training accuracy: 0.025\n",
      "Testing loss: 108.669\n",
      "Testing accuracy: 0.025\n",
      "\n",
      "\n",
      "Learning rate: 0.30000\n",
      "Epoch-3.000\n",
      "Training loss: 85.177\n",
      "Training accuracy: 0.015\n",
      "Testing loss: 84.535\n",
      "Testing accuracy: 0.015\n",
      "\n",
      "\n",
      "Learning rate: 0.30000\n",
      "Epoch-4.000\n",
      "Training loss: 73.753\n",
      "Training accuracy: 0.010\n",
      "Testing loss: 73.583\n",
      "Testing accuracy: 0.010\n",
      "\n",
      "\n",
      "Learning rate: 0.30000\n",
      "Epoch-5.000\n",
      "Training loss: 68.034\n",
      "Training accuracy: 0.008\n",
      "Testing loss: 68.228\n",
      "Testing accuracy: 0.008\n",
      "\n",
      "\n",
      "Learning rate: 0.30000\n",
      "Epoch-6.000\n",
      "Training loss: 64.970\n",
      "Training accuracy: 0.007\n",
      "Testing loss: 65.480\n",
      "Testing accuracy: 0.007\n",
      "\n",
      "\n",
      "Learning rate: 0.30000\n",
      "Epoch-7.000\n",
      "Training loss: 63.170\n",
      "Training accuracy: 0.006\n",
      "Testing loss: 63.899\n",
      "Testing accuracy: 0.007\n",
      "\n",
      "\n",
      "Learning rate: 0.30000\n",
      "Epoch-8.000\n",
      "Training loss: 61.737\n",
      "Training accuracy: 0.006\n",
      "Testing loss: 62.659\n",
      "Testing accuracy: 0.006\n",
      "\n",
      "\n",
      "Learning rate: 0.30000\n",
      "Epoch-9.000\n",
      "Training loss: 60.630\n",
      "Training accuracy: 0.005\n",
      "Testing loss: 61.680\n",
      "Testing accuracy: 0.006\n",
      "\n",
      "\n",
      "Learning rate: 0.30000\n",
      "Epoch-10.000\n",
      "Training loss: 59.808\n",
      "Training accuracy: 0.005\n",
      "Testing loss: 60.976\n",
      "Testing accuracy: 0.005\n",
      "\n",
      "\n",
      "Learning rate: 0.30000\n",
      "Epoch-11.000\n",
      "Training loss: 59.162\n",
      "Training accuracy: 0.005\n",
      "Testing loss: 60.457\n",
      "Testing accuracy: 0.005\n",
      "\n",
      "\n",
      "Learning rate: 0.30000\n",
      "Epoch-12.000\n",
      "Training loss: 58.627\n",
      "Training accuracy: 0.004\n",
      "Testing loss: 60.051\n",
      "Testing accuracy: 0.005\n",
      "\n",
      "\n",
      "Learning rate: 0.30000\n",
      "Epoch-13.000\n",
      "Training loss: 58.174\n",
      "Training accuracy: 0.004\n",
      "Testing loss: 59.708\n",
      "Testing accuracy: 0.005\n",
      "\n",
      "\n",
      "Learning rate: 0.30000\n",
      "Epoch-14.000\n",
      "Training loss: 57.824\n",
      "Training accuracy: 0.004\n",
      "Testing loss: 59.474\n",
      "Testing accuracy: 0.005\n",
      "\n",
      "\n",
      "Learning rate: 0.30000\n",
      "Epoch-15.000\n",
      "Training loss: 57.502\n",
      "Training accuracy: 0.004\n",
      "Testing loss: 59.251\n",
      "Testing accuracy: 0.005\n",
      "\n",
      "\n",
      "Learning rate: 0.30000\n",
      "Epoch-16.000\n",
      "Training loss: 57.196\n",
      "Training accuracy: 0.004\n",
      "Testing loss: 59.030\n",
      "Testing accuracy: 0.005\n",
      "\n",
      "\n",
      "Learning rate: 0.30000\n",
      "Epoch-17.000\n",
      "Training loss: 56.934\n",
      "Training accuracy: 0.004\n",
      "Testing loss: 58.874\n",
      "Testing accuracy: 0.005\n",
      "\n",
      "\n",
      "Learning rate: 0.30000\n",
      "Epoch-18.000\n",
      "Training loss: 56.699\n",
      "Training accuracy: 0.004\n",
      "Testing loss: 58.721\n",
      "Testing accuracy: 0.005\n",
      "\n",
      "\n",
      "Learning rate: 0.30000\n",
      "Epoch-19.000\n",
      "Training loss: 56.502\n",
      "Training accuracy: 0.004\n",
      "Testing loss: 58.601\n",
      "Testing accuracy: 0.005\n",
      "\n",
      "\n",
      "Learning rate: 0.30000\n",
      "Epoch-20.000\n",
      "Training loss: 56.297\n",
      "Training accuracy: 0.003\n",
      "Testing loss: 58.474\n",
      "Testing accuracy: 0.004\n",
      "\n",
      "\n",
      "Learning rate: 0.30000\n",
      "Epoch-21.000\n",
      "Training loss: 56.092\n",
      "Training accuracy: 0.003\n",
      "Testing loss: 58.331\n",
      "Testing accuracy: 0.004\n",
      "\n",
      "\n",
      "Learning rate: 0.30000\n",
      "Epoch-22.000\n",
      "Training loss: 55.956\n",
      "Training accuracy: 0.003\n",
      "Testing loss: 58.262\n",
      "Testing accuracy: 0.004\n",
      "\n",
      "\n",
      "Learning rate: 0.30000\n",
      "Epoch-23.000\n",
      "Training loss: 55.800\n",
      "Training accuracy: 0.003\n",
      "Testing loss: 58.171\n",
      "Testing accuracy: 0.004\n",
      "\n",
      "\n",
      "Learning rate: 0.30000\n",
      "Epoch-24.000\n",
      "Training loss: 55.652\n",
      "Training accuracy: 0.003\n",
      "Testing loss: 58.088\n",
      "Testing accuracy: 0.004\n",
      "\n",
      "\n",
      "Learning rate: 0.30000\n",
      "Epoch-25.000\n",
      "Training loss: 55.510\n",
      "Training accuracy: 0.003\n",
      "Testing loss: 58.002\n",
      "Testing accuracy: 0.004\n",
      "\n",
      "\n",
      "Learning rate: 0.30000\n",
      "Epoch-26.000\n",
      "Training loss: 55.385\n",
      "Training accuracy: 0.003\n",
      "Testing loss: 57.931\n",
      "Testing accuracy: 0.004\n",
      "\n",
      "\n",
      "Learning rate: 0.30000\n",
      "Epoch-27.000\n",
      "Training loss: 55.283\n",
      "Training accuracy: 0.003\n",
      "Testing loss: 57.914\n",
      "Testing accuracy: 0.004\n",
      "\n",
      "\n",
      "Learning rate: 0.30000\n",
      "Epoch-28.000\n",
      "Training loss: 55.180\n",
      "Training accuracy: 0.003\n",
      "Testing loss: 57.864\n",
      "Testing accuracy: 0.004\n",
      "\n",
      "\n",
      "Learning rate: 0.30000\n",
      "Epoch-29.000\n",
      "Training loss: 55.099\n",
      "Training accuracy: 0.003\n",
      "Testing loss: 57.837\n",
      "Testing accuracy: 0.004\n",
      "\n",
      "\n",
      "Learning rate: 0.30000\n",
      "Epoch-30.000\n",
      "Training loss: 54.992\n",
      "Training accuracy: 0.003\n",
      "Testing loss: 57.787\n",
      "Testing accuracy: 0.004\n",
      "\n",
      "\n",
      "Learning rate: 0.30000\n",
      "Epoch-31.000\n",
      "Training loss: 54.902\n",
      "Training accuracy: 0.003\n",
      "Testing loss: 57.755\n",
      "Testing accuracy: 0.004\n",
      "\n",
      "\n",
      "Learning rate: 0.30000\n",
      "Epoch-32.000\n",
      "Training loss: 54.822\n",
      "Training accuracy: 0.003\n",
      "Testing loss: 57.727\n",
      "Testing accuracy: 0.004\n",
      "\n",
      "\n",
      "Learning rate: 0.30000\n",
      "Epoch-33.000\n",
      "Training loss: 54.746\n",
      "Training accuracy: 0.003\n",
      "Testing loss: 57.708\n",
      "Testing accuracy: 0.004\n",
      "\n",
      "\n",
      "Learning rate: 0.30000\n",
      "Epoch-34.000\n",
      "Training loss: 54.655\n",
      "Training accuracy: 0.003\n",
      "Testing loss: 57.681\n",
      "Testing accuracy: 0.004\n",
      "\n",
      "\n",
      "Learning rate: 0.30000\n",
      "Epoch-35.000\n",
      "Training loss: 54.596\n",
      "Training accuracy: 0.003\n",
      "Testing loss: 57.684\n",
      "Testing accuracy: 0.004\n",
      "\n",
      "\n",
      "Learning rate: 0.30000\n",
      "Epoch-36.000\n",
      "Training loss: 54.534\n",
      "Training accuracy: 0.003\n",
      "Testing loss: 57.658\n",
      "Testing accuracy: 0.004\n",
      "\n",
      "\n",
      "Learning rate: 0.30000\n",
      "Epoch-37.000\n",
      "Training loss: 54.456\n",
      "Training accuracy: 0.003\n",
      "Testing loss: 57.646\n",
      "Testing accuracy: 0.004\n",
      "\n",
      "\n",
      "Learning rate: 0.30000\n",
      "Epoch-38.000\n",
      "Training loss: 54.403\n",
      "Training accuracy: 0.003\n",
      "Testing loss: 57.632\n",
      "Testing accuracy: 0.004\n",
      "\n",
      "\n",
      "Learning rate: 0.30000\n",
      "Epoch-39.000\n",
      "Training loss: 54.341\n",
      "Training accuracy: 0.003\n",
      "Testing loss: 57.626\n",
      "Testing accuracy: 0.004\n",
      "\n",
      "\n",
      "Learning rate: 0.30000\n",
      "Epoch-40.000\n",
      "Training loss: 54.293\n",
      "Training accuracy: 0.003\n",
      "Testing loss: 57.619\n",
      "Testing accuracy: 0.004\n",
      "\n",
      "\n",
      "Learning rate: 0.30000\n",
      "Epoch-41.000\n",
      "Training loss: 54.220\n",
      "Training accuracy: 0.003\n",
      "Testing loss: 57.608\n",
      "Testing accuracy: 0.004\n",
      "\n",
      "\n",
      "Learning rate: 0.30000\n",
      "Epoch-42.000\n",
      "Training loss: 54.178\n",
      "Training accuracy: 0.003\n",
      "Testing loss: 57.606\n",
      "Testing accuracy: 0.004\n",
      "\n",
      "\n",
      "Learning rate: 0.30000\n",
      "Epoch-43.000\n",
      "Training loss: 54.136\n",
      "Training accuracy: 0.003\n",
      "Testing loss: 57.609\n",
      "Testing accuracy: 0.004\n",
      "\n",
      "\n",
      "Learning rate: 0.30000\n",
      "Epoch-44.000\n",
      "Training loss: 54.088\n",
      "Training accuracy: 0.003\n",
      "Testing loss: 57.610\n",
      "Testing accuracy: 0.004\n",
      "\n",
      "\n",
      "Learning rate: 0.30000\n",
      "Epoch-45.000\n",
      "Training loss: 54.035\n",
      "Training accuracy: 0.003\n",
      "Testing loss: 57.602\n",
      "Testing accuracy: 0.004\n",
      "\n",
      "\n",
      "Learning rate: 0.30000\n",
      "Epoch-46.000\n",
      "Training loss: 53.998\n",
      "Training accuracy: 0.003\n",
      "Testing loss: 57.591\n",
      "Testing accuracy: 0.004\n",
      "\n",
      "\n",
      "Learning rate: 0.30000\n",
      "Epoch-47.000\n",
      "Training loss: 53.955\n",
      "Training accuracy: 0.003\n",
      "Testing loss: 57.614\n",
      "Testing accuracy: 0.004\n",
      "\n",
      "\n",
      "Learning rate: 0.30000\n",
      "Epoch-48.000\n",
      "Training loss: 53.915\n",
      "Training accuracy: 0.003\n",
      "Testing loss: 57.612\n",
      "Testing accuracy: 0.004\n",
      "\n",
      "\n",
      "Learning rate: 0.30000\n",
      "Epoch-49.000\n",
      "Training loss: 53.853\n",
      "Training accuracy: 0.003\n",
      "Testing loss: 57.601\n",
      "Testing accuracy: 0.004\n",
      "\n",
      "\n",
      "Learning rate: 0.30000\n",
      "Epoch-50.000\n",
      "Training loss: 53.822\n",
      "Training accuracy: 0.003\n",
      "Testing loss: 57.608\n",
      "Testing accuracy: 0.004\n",
      "\n",
      "\n",
      "Learning rate: 0.30000\n",
      "Epoch-51.000\n",
      "Training loss: 53.785\n",
      "Training accuracy: 0.003\n",
      "Testing loss: 57.607\n",
      "Testing accuracy: 0.004\n",
      "\n",
      "\n",
      "Learning rate: 0.30000\n",
      "Epoch-52.000\n",
      "Training loss: 53.751\n",
      "Training accuracy: 0.003\n",
      "Testing loss: 57.622\n",
      "Testing accuracy: 0.004\n",
      "\n",
      "\n",
      "Learning rate: 0.30000\n",
      "Epoch-53.000\n",
      "Training loss: 53.718\n",
      "Training accuracy: 0.003\n",
      "Testing loss: 57.615\n",
      "Testing accuracy: 0.004\n",
      "\n",
      "\n",
      "Learning rate: 0.30000\n",
      "Epoch-54.000\n",
      "Training loss: 53.681\n",
      "Training accuracy: 0.002\n",
      "Testing loss: 57.625\n",
      "Testing accuracy: 0.004\n",
      "\n",
      "\n",
      "Learning rate: 0.30000\n",
      "Epoch-55.000\n",
      "Training loss: 53.649\n",
      "Training accuracy: 0.002\n",
      "Testing loss: 57.628\n",
      "Testing accuracy: 0.004\n",
      "\n",
      "\n",
      "Learning rate: 0.30000\n",
      "Epoch-56.000\n",
      "Training loss: 53.621\n",
      "Training accuracy: 0.002\n",
      "Testing loss: 57.655\n",
      "Testing accuracy: 0.004\n",
      "\n",
      "\n",
      "Learning rate: 0.30000\n",
      "Epoch-57.000\n",
      "Training loss: 53.578\n",
      "Training accuracy: 0.002\n",
      "Testing loss: 57.637\n",
      "Testing accuracy: 0.004\n",
      "\n",
      "\n",
      "Learning rate: 0.30000\n",
      "Epoch-58.000\n",
      "Training loss: 53.543\n",
      "Training accuracy: 0.002\n",
      "Testing loss: 57.650\n",
      "Testing accuracy: 0.004\n",
      "\n",
      "\n",
      "Learning rate: 0.30000\n",
      "Epoch-59.000\n",
      "Training loss: 53.523\n",
      "Training accuracy: 0.002\n",
      "Testing loss: 57.654\n",
      "Testing accuracy: 0.004\n",
      "\n",
      "\n",
      "Learning rate: 0.30000\n",
      "Epoch-60.000\n",
      "Training loss: 53.492\n",
      "Training accuracy: 0.002\n",
      "Testing loss: 57.666\n",
      "Testing accuracy: 0.004\n",
      "\n",
      "\n",
      "Learning rate: 0.30000\n",
      "Epoch-61.000\n",
      "Training loss: 53.456\n",
      "Training accuracy: 0.002\n",
      "Testing loss: 57.673\n",
      "Testing accuracy: 0.004\n",
      "\n",
      "\n",
      "Learning rate: 0.30000\n",
      "Epoch-62.000\n",
      "Training loss: 53.427\n",
      "Training accuracy: 0.002\n",
      "Testing loss: 57.675\n",
      "Testing accuracy: 0.004\n",
      "\n",
      "\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Learning rate: 0.30000\n",
      "Epoch-63.000\n",
      "Training loss: 53.408\n",
      "Training accuracy: 0.002\n",
      "Testing loss: 57.687\n",
      "Testing accuracy: 0.004\n",
      "\n",
      "\n",
      "Learning rate: 0.30000\n",
      "Epoch-64.000\n",
      "Training loss: 53.365\n",
      "Training accuracy: 0.002\n",
      "Testing loss: 57.686\n",
      "Testing accuracy: 0.004\n",
      "\n",
      "\n",
      "Learning rate: 0.30000\n",
      "Epoch-65.000\n",
      "Training loss: 53.348\n",
      "Training accuracy: 0.002\n",
      "Testing loss: 57.699\n",
      "Testing accuracy: 0.004\n",
      "\n",
      "\n",
      "Learning rate: 0.30000\n",
      "Epoch-66.000\n",
      "Training loss: 53.323\n",
      "Training accuracy: 0.002\n",
      "Testing loss: 57.704\n",
      "Testing accuracy: 0.004\n",
      "\n",
      "\n",
      "Learning rate: 0.30000\n",
      "Epoch-67.000\n",
      "Training loss: 53.289\n",
      "Training accuracy: 0.002\n",
      "Testing loss: 57.708\n",
      "Testing accuracy: 0.004\n",
      "\n",
      "\n",
      "Learning rate: 0.30000\n",
      "Epoch-68.000\n",
      "Training loss: 53.262\n",
      "Training accuracy: 0.002\n",
      "Testing loss: 57.713\n",
      "Testing accuracy: 0.004\n",
      "\n",
      "\n",
      "Learning rate: 0.30000\n",
      "Epoch-69.000\n",
      "Training loss: 53.241\n",
      "Training accuracy: 0.002\n",
      "Testing loss: 57.725\n",
      "Testing accuracy: 0.004\n",
      "\n",
      "\n",
      "Learning rate: 0.30000\n",
      "Epoch-70.000\n",
      "Training loss: 53.216\n",
      "Training accuracy: 0.002\n",
      "Testing loss: 57.737\n",
      "Testing accuracy: 0.004\n",
      "\n",
      "\n",
      "Learning rate: 0.30000\n",
      "Epoch-71.000\n",
      "Training loss: 53.203\n",
      "Training accuracy: 0.002\n",
      "Testing loss: 57.747\n",
      "Testing accuracy: 0.004\n",
      "\n",
      "\n",
      "Learning rate: 0.30000\n",
      "Epoch-72.000\n",
      "Training loss: 53.169\n",
      "Training accuracy: 0.002\n",
      "Testing loss: 57.765\n",
      "Testing accuracy: 0.004\n",
      "\n",
      "\n",
      "Learning rate: 0.30000\n",
      "Epoch-73.000\n",
      "Training loss: 53.146\n",
      "Training accuracy: 0.002\n",
      "Testing loss: 57.760\n",
      "Testing accuracy: 0.004\n",
      "\n",
      "\n",
      "Learning rate: 0.30000\n",
      "Epoch-74.000\n",
      "Training loss: 53.121\n",
      "Training accuracy: 0.002\n",
      "Testing loss: 57.770\n",
      "Testing accuracy: 0.004\n",
      "\n",
      "\n",
      "Learning rate: 0.30000\n",
      "Epoch-75.000\n",
      "Training loss: 53.111\n",
      "Training accuracy: 0.002\n",
      "Testing loss: 57.790\n",
      "Testing accuracy: 0.004\n",
      "\n",
      "\n",
      "Learning rate: 0.30000\n",
      "Epoch-76.000\n",
      "Training loss: 53.090\n",
      "Training accuracy: 0.002\n",
      "Testing loss: 57.785\n",
      "Testing accuracy: 0.004\n",
      "\n",
      "\n",
      "Learning rate: 0.30000\n",
      "Epoch-77.000\n",
      "Training loss: 53.068\n",
      "Training accuracy: 0.002\n",
      "Testing loss: 57.798\n",
      "Testing accuracy: 0.004\n",
      "\n",
      "\n",
      "Learning rate: 0.30000\n",
      "Epoch-78.000\n",
      "Training loss: 53.043\n",
      "Training accuracy: 0.002\n",
      "Testing loss: 57.814\n",
      "Testing accuracy: 0.004\n",
      "\n",
      "\n",
      "Learning rate: 0.30000\n",
      "Epoch-79.000\n",
      "Training loss: 53.024\n",
      "Training accuracy: 0.002\n",
      "Testing loss: 57.826\n",
      "Testing accuracy: 0.004\n",
      "\n",
      "\n",
      "Learning rate: 0.30000\n",
      "Epoch-80.000\n",
      "Training loss: 52.994\n",
      "Training accuracy: 0.002\n",
      "Testing loss: 57.827\n",
      "Testing accuracy: 0.004\n",
      "\n",
      "\n",
      "Learning rate: 0.30000\n",
      "Epoch-81.000\n",
      "Training loss: 52.980\n",
      "Training accuracy: 0.002\n",
      "Testing loss: 57.837\n",
      "Testing accuracy: 0.004\n",
      "\n",
      "\n",
      "Learning rate: 0.30000\n",
      "Epoch-82.000\n",
      "Training loss: 52.965\n",
      "Training accuracy: 0.002\n",
      "Testing loss: 57.851\n",
      "Testing accuracy: 0.004\n",
      "\n",
      "\n",
      "Learning rate: 0.30000\n",
      "Epoch-83.000\n",
      "Training loss: 52.946\n",
      "Training accuracy: 0.002\n",
      "Testing loss: 57.858\n",
      "Testing accuracy: 0.004\n",
      "\n",
      "\n",
      "Learning rate: 0.30000\n",
      "Epoch-84.000\n",
      "Training loss: 52.925\n",
      "Training accuracy: 0.002\n",
      "Testing loss: 57.869\n",
      "Testing accuracy: 0.004\n",
      "\n",
      "\n",
      "Learning rate: 0.30000\n",
      "Epoch-85.000\n",
      "Training loss: 52.909\n",
      "Training accuracy: 0.002\n",
      "Testing loss: 57.887\n",
      "Testing accuracy: 0.004\n",
      "\n",
      "\n",
      "Learning rate: 0.30000\n",
      "Epoch-86.000\n",
      "Training loss: 52.892\n",
      "Training accuracy: 0.002\n",
      "Testing loss: 57.894\n",
      "Testing accuracy: 0.004\n",
      "\n",
      "\n",
      "Learning rate: 0.30000\n",
      "Epoch-87.000\n",
      "Training loss: 52.885\n",
      "Training accuracy: 0.002\n",
      "Testing loss: 57.913\n",
      "Testing accuracy: 0.004\n",
      "\n",
      "\n",
      "Learning rate: 0.30000\n",
      "Epoch-88.000\n",
      "Training loss: 52.861\n",
      "Training accuracy: 0.002\n",
      "Testing loss: 57.920\n",
      "Testing accuracy: 0.004\n",
      "\n",
      "\n",
      "Learning rate: 0.30000\n",
      "Epoch-89.000\n",
      "Training loss: 52.846\n",
      "Training accuracy: 0.002\n",
      "Testing loss: 57.922\n",
      "Testing accuracy: 0.004\n",
      "\n",
      "\n",
      "Learning rate: 0.30000\n",
      "Epoch-90.000\n",
      "Training loss: 52.821\n",
      "Training accuracy: 0.002\n",
      "Testing loss: 57.935\n",
      "Testing accuracy: 0.004\n",
      "\n",
      "\n",
      "Learning rate: 0.30000\n",
      "Epoch-91.000\n",
      "Training loss: 52.803\n",
      "Training accuracy: 0.002\n",
      "Testing loss: 57.951\n",
      "Testing accuracy: 0.004\n",
      "\n",
      "\n",
      "Learning rate: 0.30000\n",
      "Epoch-92.000\n",
      "Training loss: 52.793\n",
      "Training accuracy: 0.002\n",
      "Testing loss: 57.971\n",
      "Testing accuracy: 0.004\n",
      "\n",
      "\n",
      "Learning rate: 0.30000\n",
      "Epoch-93.000\n",
      "Training loss: 52.780\n",
      "Training accuracy: 0.002\n",
      "Testing loss: 57.967\n",
      "Testing accuracy: 0.004\n",
      "\n",
      "\n",
      "Learning rate: 0.30000\n",
      "Epoch-94.000\n",
      "Training loss: 52.761\n",
      "Training accuracy: 0.002\n",
      "Testing loss: 57.983\n",
      "Testing accuracy: 0.004\n",
      "\n",
      "\n",
      "Learning rate: 0.30000\n",
      "Epoch-95.000\n",
      "Training loss: 52.742\n",
      "Training accuracy: 0.002\n",
      "Testing loss: 57.998\n",
      "Testing accuracy: 0.004\n",
      "\n",
      "\n",
      "Learning rate: 0.30000\n",
      "Epoch-96.000\n",
      "Training loss: 52.724\n",
      "Training accuracy: 0.002\n",
      "Testing loss: 58.006\n",
      "Testing accuracy: 0.004\n",
      "\n",
      "\n",
      "Learning rate: 0.30000\n",
      "Epoch-97.000\n",
      "Training loss: 52.710\n",
      "Training accuracy: 0.002\n",
      "Testing loss: 58.017\n",
      "Testing accuracy: 0.004\n",
      "\n",
      "\n",
      "Learning rate: 0.30000\n",
      "Epoch-98.000\n",
      "Training loss: 52.698\n",
      "Training accuracy: 0.002\n",
      "Testing loss: 58.027\n",
      "Testing accuracy: 0.004\n",
      "\n",
      "\n",
      "Learning rate: 0.30000\n",
      "Epoch-99.000\n",
      "Training loss: 52.680\n",
      "Training accuracy: 0.002\n",
      "Testing loss: 58.037\n",
      "Testing accuracy: 0.004\n",
      "\n",
      "\n",
      "Learning rate: 0.30000\n",
      "Epoch-100.000\n",
      "Training loss: 52.663\n",
      "Training accuracy: 0.002\n",
      "Testing loss: 58.047\n",
      "Testing accuracy: 0.004\n",
      "\n",
      "\n",
      "Learning rate: 0.30000\n",
      "Epoch-101.000\n",
      "Training loss: 52.654\n",
      "Training accuracy: 0.002\n",
      "Testing loss: 58.072\n",
      "Testing accuracy: 0.004\n",
      "\n",
      "\n",
      "Learning rate: 0.30000\n",
      "Epoch-102.000\n",
      "Training loss: 52.635\n",
      "Training accuracy: 0.002\n",
      "Testing loss: 58.072\n",
      "Testing accuracy: 0.004\n",
      "\n",
      "\n",
      "Learning rate: 0.30000\n",
      "Epoch-103.000\n",
      "Training loss: 52.629\n",
      "Training accuracy: 0.002\n",
      "Testing loss: 58.084\n",
      "Testing accuracy: 0.004\n",
      "\n",
      "\n",
      "Learning rate: 0.30000\n",
      "Epoch-104.000\n",
      "Training loss: 52.617\n",
      "Training accuracy: 0.002\n",
      "Testing loss: 58.095\n",
      "Testing accuracy: 0.004\n",
      "\n",
      "\n",
      "Learning rate: 0.30000\n",
      "Epoch-105.000\n",
      "Training loss: 52.595\n",
      "Training accuracy: 0.002\n",
      "Testing loss: 58.105\n",
      "Testing accuracy: 0.004\n",
      "\n",
      "\n",
      "Learning rate: 0.30000\n",
      "Epoch-106.000\n",
      "Training loss: 52.591\n",
      "Training accuracy: 0.002\n",
      "Testing loss: 58.132\n",
      "Testing accuracy: 0.004\n",
      "\n",
      "\n",
      "Learning rate: 0.30000\n",
      "Epoch-107.000\n",
      "Training loss: 52.577\n",
      "Training accuracy: 0.002\n",
      "Testing loss: 58.123\n",
      "Testing accuracy: 0.004\n",
      "\n",
      "\n",
      "Learning rate: 0.30000\n",
      "Epoch-108.000\n",
      "Training loss: 52.556\n",
      "Training accuracy: 0.002\n",
      "Testing loss: 58.126\n",
      "Testing accuracy: 0.004\n",
      "\n",
      "\n",
      "Learning rate: 0.30000\n",
      "Epoch-109.000\n",
      "Training loss: 52.548\n",
      "Training accuracy: 0.002\n",
      "Testing loss: 58.145\n",
      "Testing accuracy: 0.004\n",
      "\n",
      "\n",
      "Learning rate: 0.30000\n",
      "Epoch-110.000\n",
      "Training loss: 52.544\n",
      "Training accuracy: 0.002\n",
      "Testing loss: 58.156\n",
      "Testing accuracy: 0.004\n",
      "\n",
      "\n",
      "Learning rate: 0.30000\n",
      "Epoch-111.000\n",
      "Training loss: 52.528\n",
      "Training accuracy: 0.002\n",
      "Testing loss: 58.154\n",
      "Testing accuracy: 0.004\n",
      "\n",
      "\n",
      "Learning rate: 0.30000\n",
      "Epoch-112.000\n",
      "Training loss: 52.514\n",
      "Training accuracy: 0.002\n",
      "Testing loss: 58.167\n",
      "Testing accuracy: 0.004\n",
      "\n",
      "\n",
      "Learning rate: 0.30000\n",
      "Epoch-113.000\n",
      "Training loss: 52.511\n",
      "Training accuracy: 0.002\n",
      "Testing loss: 58.184\n",
      "Testing accuracy: 0.004\n",
      "\n",
      "\n",
      "Learning rate: 0.30000\n",
      "Epoch-114.000\n",
      "Training loss: 52.486\n",
      "Training accuracy: 0.002\n",
      "Testing loss: 58.183\n",
      "Testing accuracy: 0.004\n",
      "\n",
      "\n",
      "Learning rate: 0.30000\n",
      "Epoch-115.000\n",
      "Training loss: 52.481\n",
      "Training accuracy: 0.002\n",
      "Testing loss: 58.214\n",
      "Testing accuracy: 0.004\n",
      "\n",
      "\n",
      "Learning rate: 0.30000\n",
      "Epoch-116.000\n",
      "Training loss: 52.465\n",
      "Training accuracy: 0.002\n",
      "Testing loss: 58.221\n",
      "Testing accuracy: 0.004\n",
      "\n",
      "\n",
      "Learning rate: 0.30000\n",
      "Epoch-117.000\n",
      "Training loss: 52.457\n",
      "Training accuracy: 0.002\n",
      "Testing loss: 58.230\n",
      "Testing accuracy: 0.004\n",
      "\n",
      "\n",
      "Learning rate: 0.30000\n",
      "Epoch-118.000\n",
      "Training loss: 52.439\n",
      "Training accuracy: 0.002\n",
      "Testing loss: 58.237\n",
      "Testing accuracy: 0.004\n",
      "\n",
      "\n",
      "Learning rate: 0.30000\n",
      "Epoch-119.000\n",
      "Training loss: 52.421\n",
      "Training accuracy: 0.002\n",
      "Testing loss: 58.245\n",
      "Testing accuracy: 0.004\n",
      "\n",
      "\n",
      "Learning rate: 0.30000\n",
      "Epoch-120.000\n",
      "Training loss: 52.420\n",
      "Training accuracy: 0.002\n",
      "Testing loss: 58.272\n",
      "Testing accuracy: 0.004\n",
      "\n",
      "\n",
      "name_loss:\n",
      "logistic-regression-sum-loss\n",
      "Model created.\n",
      "Begin laoding data...\n",
      "Extracting data/MNIST_data/train-images-idx3-ubyte.gz\n",
      "Extracting data/MNIST_data/t10k-images-idx3-ubyte.gz\n",
      "Begin training...\n",
      "Learning rate: 1.00000\n",
      "Epoch-1.000\n",
      "Training loss: 183.166\n",
      "Training accuracy: 0.058\n",
      "Testing loss: 182.719\n",
      "Testing accuracy: 0.058\n",
      "\n",
      "\n",
      "Learning rate: 1.00000\n",
      "Epoch-2.000\n",
      "Training loss: 135.135\n",
      "Training accuracy: 0.037\n",
      "Testing loss: 134.332\n",
      "Testing accuracy: 0.036\n",
      "\n",
      "\n",
      "Learning rate: 1.00000\n",
      "Epoch-3.000\n",
      "Training loss: 99.707\n",
      "Training accuracy: 0.021\n",
      "Testing loss: 98.860\n",
      "Testing accuracy: 0.021\n",
      "\n",
      "\n",
      "Learning rate: 1.00000\n",
      "Epoch-4.000\n",
      "Training loss: 82.502\n",
      "Training accuracy: 0.014\n",
      "Testing loss: 81.846\n",
      "Testing accuracy: 0.014\n",
      "\n",
      "\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Learning rate: 1.00000\n",
      "Epoch-5.000\n",
      "Training loss: 73.966\n",
      "Training accuracy: 0.010\n",
      "Testing loss: 73.678\n",
      "Testing accuracy: 0.010\n",
      "\n",
      "\n",
      "Learning rate: 1.00000\n",
      "Epoch-6.000\n",
      "Training loss: 69.086\n",
      "Training accuracy: 0.008\n",
      "Testing loss: 69.104\n",
      "Testing accuracy: 0.009\n",
      "\n",
      "\n",
      "Learning rate: 1.00000\n",
      "Epoch-7.000\n",
      "Training loss: 66.013\n",
      "Training accuracy: 0.007\n",
      "Testing loss: 66.287\n",
      "Testing accuracy: 0.007\n",
      "\n",
      "\n",
      "Learning rate: 1.00000\n",
      "Epoch-8.000\n",
      "Training loss: 64.084\n",
      "Training accuracy: 0.006\n",
      "Testing loss: 64.563\n",
      "Testing accuracy: 0.007\n",
      "\n",
      "\n",
      "Learning rate: 1.00000\n",
      "Epoch-9.000\n",
      "Training loss: 62.747\n",
      "Training accuracy: 0.006\n",
      "Testing loss: 63.417\n",
      "Testing accuracy: 0.006\n",
      "\n",
      "\n",
      "Learning rate: 1.00000\n",
      "Epoch-10.000\n",
      "Training loss: 61.645\n",
      "Training accuracy: 0.006\n",
      "Testing loss: 62.476\n",
      "Testing accuracy: 0.006\n",
      "\n",
      "\n",
      "Learning rate: 1.00000\n",
      "Epoch-11.000\n",
      "Training loss: 60.686\n",
      "Training accuracy: 0.005\n",
      "Testing loss: 61.633\n",
      "Testing accuracy: 0.006\n",
      "\n",
      "\n",
      "Learning rate: 1.00000\n",
      "Epoch-12.000\n",
      "Training loss: 59.933\n",
      "Training accuracy: 0.005\n",
      "Testing loss: 61.008\n",
      "Testing accuracy: 0.005\n",
      "\n",
      "\n",
      "Learning rate: 1.00000\n",
      "Epoch-13.000\n",
      "Training loss: 59.316\n",
      "Training accuracy: 0.005\n",
      "Testing loss: 60.502\n",
      "Testing accuracy: 0.005\n",
      "\n",
      "\n",
      "Learning rate: 1.00000\n",
      "Epoch-14.000\n",
      "Training loss: 58.773\n",
      "Training accuracy: 0.004\n",
      "Testing loss: 60.054\n",
      "Testing accuracy: 0.005\n",
      "\n",
      "\n",
      "Learning rate: 1.00000\n",
      "Epoch-15.000\n",
      "Training loss: 58.315\n",
      "Training accuracy: 0.004\n",
      "Testing loss: 59.694\n",
      "Testing accuracy: 0.005\n",
      "\n",
      "\n",
      "Learning rate: 1.00000\n",
      "Epoch-16.000\n",
      "Training loss: 57.909\n",
      "Training accuracy: 0.004\n",
      "Testing loss: 59.372\n",
      "Testing accuracy: 0.005\n",
      "\n",
      "\n",
      "Learning rate: 1.00000\n",
      "Epoch-17.000\n",
      "Training loss: 57.563\n",
      "Training accuracy: 0.004\n",
      "Testing loss: 59.124\n",
      "Testing accuracy: 0.005\n",
      "\n",
      "\n",
      "Learning rate: 1.00000\n",
      "Epoch-18.000\n",
      "Training loss: 57.254\n",
      "Training accuracy: 0.004\n",
      "Testing loss: 58.903\n",
      "Testing accuracy: 0.005\n",
      "\n",
      "\n",
      "Learning rate: 1.00000\n",
      "Epoch-19.000\n",
      "Training loss: 56.975\n",
      "Training accuracy: 0.004\n",
      "Testing loss: 58.697\n",
      "Testing accuracy: 0.005\n",
      "\n",
      "\n",
      "Learning rate: 1.00000\n",
      "Epoch-20.000\n",
      "Training loss: 56.723\n",
      "Training accuracy: 0.004\n",
      "Testing loss: 58.522\n",
      "Testing accuracy: 0.004\n",
      "\n",
      "\n",
      "Learning rate: 1.00000\n",
      "Epoch-21.000\n",
      "Training loss: 56.486\n",
      "Training accuracy: 0.004\n",
      "Testing loss: 58.370\n",
      "Testing accuracy: 0.004\n",
      "\n",
      "\n",
      "Learning rate: 1.00000\n",
      "Epoch-22.000\n",
      "Training loss: 56.283\n",
      "Training accuracy: 0.003\n",
      "Testing loss: 58.240\n",
      "Testing accuracy: 0.004\n",
      "\n",
      "\n",
      "Learning rate: 1.00000\n",
      "Epoch-23.000\n",
      "Training loss: 56.097\n",
      "Training accuracy: 0.003\n",
      "Testing loss: 58.120\n",
      "Testing accuracy: 0.004\n",
      "\n",
      "\n",
      "Learning rate: 1.00000\n",
      "Epoch-24.000\n",
      "Training loss: 55.926\n",
      "Training accuracy: 0.003\n",
      "Testing loss: 58.028\n",
      "Testing accuracy: 0.004\n",
      "\n",
      "\n",
      "Learning rate: 1.00000\n",
      "Epoch-25.000\n",
      "Training loss: 55.768\n",
      "Training accuracy: 0.003\n",
      "Testing loss: 57.944\n",
      "Testing accuracy: 0.004\n",
      "\n",
      "\n",
      "Learning rate: 1.00000\n",
      "Epoch-26.000\n",
      "Training loss: 55.626\n",
      "Training accuracy: 0.003\n",
      "Testing loss: 57.857\n",
      "Testing accuracy: 0.004\n",
      "\n",
      "\n",
      "Learning rate: 1.00000\n",
      "Epoch-27.000\n",
      "Training loss: 55.498\n",
      "Training accuracy: 0.003\n",
      "Testing loss: 57.814\n",
      "Testing accuracy: 0.004\n",
      "\n",
      "\n",
      "Learning rate: 1.00000\n",
      "Epoch-28.000\n",
      "Training loss: 55.367\n",
      "Training accuracy: 0.003\n",
      "Testing loss: 57.755\n",
      "Testing accuracy: 0.004\n",
      "\n",
      "\n",
      "Learning rate: 1.00000\n",
      "Epoch-29.000\n",
      "Training loss: 55.250\n",
      "Training accuracy: 0.003\n",
      "Testing loss: 57.707\n",
      "Testing accuracy: 0.004\n",
      "\n",
      "\n",
      "Learning rate: 1.00000\n",
      "Epoch-30.000\n",
      "Training loss: 55.140\n",
      "Training accuracy: 0.003\n",
      "Testing loss: 57.652\n",
      "Testing accuracy: 0.004\n",
      "\n",
      "\n",
      "Learning rate: 1.00000\n",
      "Epoch-31.000\n",
      "Training loss: 55.043\n",
      "Training accuracy: 0.003\n",
      "Testing loss: 57.624\n",
      "Testing accuracy: 0.004\n",
      "\n",
      "\n",
      "Learning rate: 1.00000\n",
      "Epoch-32.000\n",
      "Training loss: 54.950\n",
      "Training accuracy: 0.003\n",
      "Testing loss: 57.586\n",
      "Testing accuracy: 0.004\n",
      "\n",
      "\n",
      "Learning rate: 1.00000\n",
      "Epoch-33.000\n",
      "Training loss: 54.838\n",
      "Training accuracy: 0.003\n",
      "Testing loss: 57.541\n",
      "Testing accuracy: 0.004\n",
      "\n",
      "\n",
      "Learning rate: 1.00000\n",
      "Epoch-34.000\n",
      "Training loss: 54.754\n",
      "Training accuracy: 0.003\n",
      "Testing loss: 57.526\n",
      "Testing accuracy: 0.004\n",
      "\n",
      "\n",
      "Learning rate: 1.00000\n",
      "Epoch-35.000\n",
      "Training loss: 54.674\n",
      "Training accuracy: 0.003\n",
      "Testing loss: 57.502\n",
      "Testing accuracy: 0.004\n",
      "\n",
      "\n",
      "Learning rate: 1.00000\n",
      "Epoch-36.000\n",
      "Training loss: 54.582\n",
      "Training accuracy: 0.003\n",
      "Testing loss: 57.477\n",
      "Testing accuracy: 0.004\n",
      "\n",
      "\n",
      "Learning rate: 1.00000\n",
      "Epoch-37.000\n",
      "Training loss: 54.508\n",
      "Training accuracy: 0.003\n",
      "Testing loss: 57.462\n",
      "Testing accuracy: 0.004\n",
      "\n",
      "\n",
      "Learning rate: 1.00000\n",
      "Epoch-38.000\n",
      "Training loss: 54.438\n",
      "Training accuracy: 0.003\n",
      "Testing loss: 57.448\n",
      "Testing accuracy: 0.004\n",
      "\n",
      "\n",
      "Learning rate: 1.00000\n",
      "Epoch-39.000\n",
      "Training loss: 54.376\n",
      "Training accuracy: 0.003\n",
      "Testing loss: 57.438\n",
      "Testing accuracy: 0.004\n",
      "\n",
      "\n",
      "Learning rate: 1.00000\n",
      "Epoch-40.000\n",
      "Training loss: 54.297\n",
      "Training accuracy: 0.003\n",
      "Testing loss: 57.415\n",
      "Testing accuracy: 0.004\n",
      "\n",
      "\n",
      "Learning rate: 1.00000\n",
      "Epoch-41.000\n",
      "Training loss: 54.236\n",
      "Training accuracy: 0.003\n",
      "Testing loss: 57.406\n",
      "Testing accuracy: 0.004\n",
      "\n",
      "\n",
      "Learning rate: 1.00000\n",
      "Epoch-42.000\n",
      "Training loss: 54.177\n",
      "Training accuracy: 0.003\n",
      "Testing loss: 57.407\n",
      "Testing accuracy: 0.004\n",
      "\n",
      "\n",
      "Learning rate: 1.00000\n",
      "Epoch-43.000\n",
      "Training loss: 54.115\n",
      "Training accuracy: 0.003\n",
      "Testing loss: 57.391\n",
      "Testing accuracy: 0.004\n",
      "\n",
      "\n",
      "Learning rate: 1.00000\n",
      "Epoch-44.000\n",
      "Training loss: 54.063\n",
      "Training accuracy: 0.003\n",
      "Testing loss: 57.399\n",
      "Testing accuracy: 0.004\n",
      "\n",
      "\n",
      "Learning rate: 1.00000\n",
      "Epoch-45.000\n",
      "Training loss: 54.008\n",
      "Training accuracy: 0.003\n",
      "Testing loss: 57.390\n",
      "Testing accuracy: 0.004\n",
      "\n",
      "\n",
      "Learning rate: 1.00000\n",
      "Epoch-46.000\n",
      "Training loss: 53.953\n",
      "Training accuracy: 0.003\n",
      "Testing loss: 57.382\n",
      "Testing accuracy: 0.004\n",
      "\n",
      "\n",
      "Learning rate: 1.00000\n",
      "Epoch-47.000\n",
      "Training loss: 53.895\n",
      "Training accuracy: 0.003\n",
      "Testing loss: 57.391\n",
      "Testing accuracy: 0.004\n",
      "\n",
      "\n",
      "Learning rate: 1.00000\n",
      "Epoch-48.000\n",
      "Training loss: 53.849\n",
      "Training accuracy: 0.003\n",
      "Testing loss: 57.392\n",
      "Testing accuracy: 0.004\n",
      "\n",
      "\n",
      "Learning rate: 1.00000\n",
      "Epoch-49.000\n",
      "Training loss: 53.800\n",
      "Training accuracy: 0.003\n",
      "Testing loss: 57.370\n",
      "Testing accuracy: 0.004\n",
      "\n",
      "\n",
      "Learning rate: 1.00000\n",
      "Epoch-50.000\n",
      "Training loss: 53.748\n",
      "Training accuracy: 0.003\n",
      "Testing loss: 57.383\n",
      "Testing accuracy: 0.004\n",
      "\n",
      "\n",
      "Learning rate: 1.00000\n",
      "Epoch-51.000\n",
      "Training loss: 53.716\n",
      "Training accuracy: 0.003\n",
      "Testing loss: 57.391\n",
      "Testing accuracy: 0.004\n",
      "\n",
      "\n",
      "Learning rate: 1.00000\n",
      "Epoch-52.000\n",
      "Training loss: 53.666\n",
      "Training accuracy: 0.002\n",
      "Testing loss: 57.393\n",
      "Testing accuracy: 0.004\n",
      "\n",
      "\n",
      "Learning rate: 1.00000\n",
      "Epoch-53.000\n",
      "Training loss: 53.618\n",
      "Training accuracy: 0.002\n",
      "Testing loss: 57.391\n",
      "Testing accuracy: 0.004\n",
      "\n",
      "\n",
      "Learning rate: 1.00000\n",
      "Epoch-54.000\n",
      "Training loss: 53.573\n",
      "Training accuracy: 0.002\n",
      "Testing loss: 57.382\n",
      "Testing accuracy: 0.004\n",
      "\n",
      "\n",
      "Learning rate: 1.00000\n",
      "Epoch-55.000\n",
      "Training loss: 53.530\n",
      "Training accuracy: 0.002\n",
      "Testing loss: 57.390\n",
      "Testing accuracy: 0.004\n",
      "\n",
      "\n",
      "Learning rate: 1.00000\n",
      "Epoch-56.000\n",
      "Training loss: 53.498\n",
      "Training accuracy: 0.002\n",
      "Testing loss: 57.422\n",
      "Testing accuracy: 0.004\n",
      "\n",
      "\n",
      "Learning rate: 1.00000\n",
      "Epoch-57.000\n",
      "Training loss: 53.457\n",
      "Training accuracy: 0.002\n",
      "Testing loss: 57.418\n",
      "Testing accuracy: 0.004\n",
      "\n",
      "\n",
      "Learning rate: 1.00000\n",
      "Epoch-58.000\n",
      "Training loss: 53.420\n",
      "Training accuracy: 0.002\n",
      "Testing loss: 57.423\n",
      "Testing accuracy: 0.004\n",
      "\n",
      "\n",
      "Learning rate: 1.00000\n",
      "Epoch-59.000\n",
      "Training loss: 53.384\n",
      "Training accuracy: 0.002\n",
      "Testing loss: 57.424\n",
      "Testing accuracy: 0.004\n",
      "\n",
      "\n",
      "Learning rate: 1.00000\n",
      "Epoch-60.000\n",
      "Training loss: 53.353\n",
      "Training accuracy: 0.002\n",
      "Testing loss: 57.446\n",
      "Testing accuracy: 0.004\n",
      "\n",
      "\n",
      "Learning rate: 1.00000\n",
      "Epoch-61.000\n",
      "Training loss: 53.319\n",
      "Training accuracy: 0.002\n",
      "Testing loss: 57.454\n",
      "Testing accuracy: 0.004\n",
      "\n",
      "\n",
      "Learning rate: 1.00000\n",
      "Epoch-62.000\n",
      "Training loss: 53.275\n",
      "Training accuracy: 0.002\n",
      "Testing loss: 57.451\n",
      "Testing accuracy: 0.004\n",
      "\n",
      "\n",
      "Learning rate: 1.00000\n",
      "Epoch-63.000\n",
      "Training loss: 53.242\n",
      "Training accuracy: 0.002\n",
      "Testing loss: 57.451\n",
      "Testing accuracy: 0.004\n",
      "\n",
      "\n",
      "Learning rate: 1.00000\n",
      "Epoch-64.000\n",
      "Training loss: 53.215\n",
      "Training accuracy: 0.002\n",
      "Testing loss: 57.473\n",
      "Testing accuracy: 0.004\n",
      "\n",
      "\n",
      "Learning rate: 1.00000\n",
      "Epoch-65.000\n",
      "Training loss: 53.180\n",
      "Training accuracy: 0.002\n",
      "Testing loss: 57.476\n",
      "Testing accuracy: 0.004\n",
      "\n",
      "\n",
      "Learning rate: 1.00000\n",
      "Epoch-66.000\n",
      "Training loss: 53.151\n",
      "Training accuracy: 0.002\n",
      "Testing loss: 57.493\n",
      "Testing accuracy: 0.004\n",
      "\n",
      "\n",
      "Learning rate: 1.00000\n",
      "Epoch-67.000\n",
      "Training loss: 53.118\n",
      "Training accuracy: 0.002\n",
      "Testing loss: 57.501\n",
      "Testing accuracy: 0.004\n",
      "\n",
      "\n",
      "Learning rate: 1.00000\n",
      "Epoch-68.000\n",
      "Training loss: 53.091\n",
      "Training accuracy: 0.002\n",
      "Testing loss: 57.500\n",
      "Testing accuracy: 0.004\n",
      "\n",
      "\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Learning rate: 1.00000\n",
      "Epoch-69.000\n",
      "Training loss: 53.068\n",
      "Training accuracy: 0.002\n",
      "Testing loss: 57.526\n",
      "Testing accuracy: 0.004\n",
      "\n",
      "\n",
      "Learning rate: 1.00000\n",
      "Epoch-70.000\n",
      "Training loss: 53.031\n",
      "Training accuracy: 0.002\n",
      "Testing loss: 57.533\n",
      "Testing accuracy: 0.004\n",
      "\n",
      "\n",
      "Learning rate: 1.00000\n",
      "Epoch-71.000\n",
      "Training loss: 53.009\n",
      "Training accuracy: 0.002\n",
      "Testing loss: 57.555\n",
      "Testing accuracy: 0.004\n",
      "\n",
      "\n",
      "Learning rate: 1.00000\n",
      "Epoch-72.000\n",
      "Training loss: 52.978\n",
      "Training accuracy: 0.002\n",
      "Testing loss: 57.551\n",
      "Testing accuracy: 0.004\n",
      "\n",
      "\n",
      "Learning rate: 1.00000\n",
      "Epoch-73.000\n",
      "Training loss: 52.959\n",
      "Training accuracy: 0.002\n",
      "Testing loss: 57.571\n",
      "Testing accuracy: 0.004\n",
      "\n",
      "\n",
      "Learning rate: 1.00000\n",
      "Epoch-74.000\n",
      "Training loss: 52.930\n",
      "Training accuracy: 0.002\n",
      "Testing loss: 57.590\n",
      "Testing accuracy: 0.004\n",
      "\n",
      "\n",
      "Learning rate: 1.00000\n",
      "Epoch-75.000\n",
      "Training loss: 52.902\n",
      "Training accuracy: 0.002\n",
      "Testing loss: 57.595\n",
      "Testing accuracy: 0.004\n",
      "\n",
      "\n"
     ]
    }
   ],
   "source": [
    "list_results = []\n",
    "\n",
    "alg1_result = train_model(\n",
    "    dataset_name = 'MNIST', # ['MNIST', 'FACES', 'CURVES']\n",
    "    home_path = '/home/jupyter/',\n",
    "    algorithm = 'K-BFGS', # ['KFAC', 'K-BFGS', 'K-BFGS(L)', 'RMSprop', 'Adam', 'SGD-momentum']\n",
    "    lr = 0.3,\n",
    "    lambda_damping = 0.3,\n",
    "    batch_size = 1000,\n",
    "    RMSprop_epsilon = 1e-4,\n",
    "    if_gpu = True,\n",
    "    if_max_epoch = 0,\n",
    "    max_epoch = 100,\n",
    "    max_cpu_time = 500,\n",
    "    verbose = True)\n",
    "\n",
    "list_results += [alg1_result]\n",
    "\n",
    "alg2_result = train_model(\n",
    "    dataset_name = 'MNIST', # ['MNIST', 'FACES', 'CURVES']\n",
    "    home_path = '/home/jupyter/',\n",
    "    algorithm = 'KFAC', # ['KFAC', 'K-BFGS', 'K-BFGS(L)', 'RMSprop', 'Adam', 'SGD-momentum']\n",
    "    lr = 1,\n",
    "    lambda_damping = 3,\n",
    "    batch_size = 1000,\n",
    "    RMSprop_epsilon = 1e-4,\n",
    "    if_gpu = True,\n",
    "    if_max_epoch = 0,\n",
    "    max_epoch = 100,\n",
    "    max_cpu_time = 500,\n",
    "    verbose = True)\n",
    "\n",
    "list_results += [alg2_result]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "%matplotlib inline\n",
    "from plot_utils import *"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x720 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x720 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plot_results(list_results)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.7.6"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}